/ ..^^!" °''% c ^o ■" isn©ii^/ygM ^f^^TES O^ ^ r JUL 1 7 1981 Woods Hole, Mass. ^ Vol. 79, No. 1 January 1981 HOBSON, EDMUND S., WILLIAM N. McFARLAND, and JAMES R. CHESS. Crepuscular and nocturnal activities of Californian nearshore fishes, with con- sideration of their scotopic visual pigments and the photic environment 1 GOODING, REGINALD M., WILLIAM H. NEILL, and ANDREW E. DIZON. Res- piration rates and low-oxygen tolerance limits in skipjack tuna, Katsuwonus pelamis 31 BEARDSLEY, GRANT L., and RAMON J. CONSER. An analysis of catch and effort data from the U.S. recreational fishery for billfishes (Istiophoridae) in the western North Atlantic Ocean and Gulf of Mexico, 1971-78 49 FERNHOLM, BO, and CARL L. HUBBS. Western Atlantic hagfishes of the genus Eptatretus (Myxinidae) with description of two new species 69 BLACKBURN, MAURICE, and D. L. SERVENTY. Observations on distribution and life history of skipjack tuna, Katsuwonus pelamis, in Australian waters 85 KAPPENMAN, RUSSELL F. A method for growth curve comparisons 95 RICHARDSON, SALLY L. Current knowledge of larvae of sculpins (Pisces: Cottidae and allies) in northeast Pacific genera with notes on intergeneric relationships 103 TOWNSEND, DAVID W, and JOSEPH J. GRAHAM. Growth and age structure of larval Atlantic herring, Clupea harengus harengus, in the Sheepscot River estuary, Maine, as determined by daily growth increments in otoliths 123 KOSLOW, J. ANTHONY. Feeding selectivity of schools of northern anchovy, Engraulis mordax, in the Southern California Bight 131 WEBB, P. W, and R. T. COROLLA. Burst swimming performance of northern anchovy, Engraulis mordax, larvae 143 UCHIYAMA, JAMES H., and PAUL STRUHSAKER. Age and growth of skipjack tuna, Katsuwonus pelamis, and yellowfin tuna, Thunnus albacares, as indicated by daily growth increments of sagittae 151 RICHARDSON, SALLY L. Pelagic eggs and larve of the deepsea sole, Embas- sichthys bathybius (Pisces: Pleuronectidae), with comments on generic affinities . 163 Notes WEIHS, DANIEL. Effects of swimming path curvature on the energetics of fish motion 171 (Continued on back cover) Seattle, Washington U.S. DEPARTMENT OF COMMERCE Malcolm Baldrlge, Secretary NATIONAL OCEANIC AND ATMOSPHERIC ADMINISTRATION Terry L. Leitzell, Assistant Administrator for Fisheries NATIONAL MARINE FISHERIES SERVICE Fishery Bulletin The Fishery Bulletin carries original research reports and technical notes on investigations in fishery science, engineering, and economics. The Bulletin of the United States Fish Commission was begun in 1881; it became the Bulletin of the Bureau of Fisheries in 1904 and the Fishery Bulletin of the Fish and Wildlife Service in 1941. Separates were issued as documents through volume 46; the last document was No. 1103. Beginning with volume 47 in 1931 and continuing through volume 62 in 1963, each separate appeared as a numbered bulletin. A new system began in 1963 with volume 63 in which papers are bound together in a single issue of the bulletin instead of being issued individually. Beginning with volume 70, number 1, January 1972, the Fishery Bulletin became a periodical, issued quarterly. In this form, it is available by subscription from the Superintendent of Documents, U.S. Government Printing Ofi&ce, Washington, DC 20402. It is also available free in limited numbers to libraries, research institutions, State and Federal agencies, and in exchange for other scientific publications. EDITOR Dr. Jay C. Quast Scientific Editor, Fishery Bulletin Northwest and Alaska Fisheries Center Auke Bay Laboratory National Marine Fisheries Service, NOAA P.O. Box 155, Auke Bay, AK 99821 Editorial Committee Dr. Bruce B. Collette Dr. Reuben Lasker National Marine Fisheries Service National Marine Fisheries Service Dr. Edward D. Houde Dr. Jerome J. Pella Chesapeake Biological Laboratory National Marine Fisheries Service Dr. Merton C. Ingham Dr. Sally L. Richardson National Marine Fisheries Service Gulf Coast Research Laboratory Kiyoshi G. Fukano, Managing Editor The Fishery Bulletin (USPS 090-870) is published quarterly by Scientific Publications Office, National Marine Fisheries Service, NOAA, Room 336, 1 700 Westlake Avenue North, Seattle, WA 98109. Controlled circulation paid to Finance Department, USPS, Washington, DC 20260. Although the contents have not been copyrighted and may be reprinted entirely, reference to source is appreciated. The Secretary of Commerce has determined that the publication of this periodical is necessary in the transaction of the public business required by law of this Department. Use of funds for printing of this periodical has been approved by the Director of the Office of Management and Budget through 31 March 1982. Fishery Bulletin CONTENTS Vol.79, No. 1 January 1981 HOBSON, EDMUND S., WILLIAM N. McFARLAND, and JAMES R. CHESS. Crepuscular and nocturnal activities of Californian nearshore fishes, with con- sideration of their scotopic visual pigments and the photic environment 1 GOODING, REGINALD M., WILLIAM H. NEILL, and ANDREW E. DIZON. Res- piration rates and low-oxygen tolerance limits in skipjack tuna, Katsuwonus pelamis 31 BEARDSLEY, GRANT L., and RAMON J. CONSER. An analysis of catch and effort data from the U.S. recreational fishery for billfishes (Istiophoridae) in the western North Atlantic Ocean and Gulf of Mexico, 1971-78 49 FERNHOLM, BO, and CARL L. HUBBS. Western Atlantic hagfishes of the genus Eptatretus (Myxinidae) with description of two new species 69 BLACKBURN, MAURICE, and D. L. SERVENTY. Observations on distribution and life history of skipjack tuna, Katsuwonus pelamis, in Australian waters 85 KAPPENMAN, RUSSELL F. A method for growth curve comparisons 95 RICHARDSON, SALLY L. Current knowledge of larvae of sculpins (Pisces: Cottidae and allies) in northeast Pacific genera with notes on intergeneric relationships 103 TOWNSEND, DAVID W, and JOSEPH J. GRAHAM. Growth and age structure of larval Atlantic herring, Clupea harengus harengus, i-n the Sheepscot River estuary, Maine, as determined by daily growth increments in otoliths 123 KOSLOW, J. ANTHONY. Feeding selectivity of schools of northern anchovy, Engraulis mordax, in the Southern California Bight 131 WEBB, P. W, and R. T. COROLLA. Burst swimming performance of northern anchovy, Engraulis mordax, larvae 143 UCHIYAMA, JAMES H., and PAUL STRUHSAKER. Age and growth of skipjack tuna, Katsuwonus pelamis, and yellowfin tuna, Thunnus albacares, as indicated by daily growth increments of sagittae 151 RICHARDSON, SALLY L. Pelagic eggs and larve of the deepsea sole, Emhas- sichthys bathybius (Pisces: Pleuronectidae), with comments on generic affinities . 163 Notes WEIHS, DANIEL. Effects of swimming path curvature on the energetics of fish motion 171 (Continued on next page) Seattle, Washington 1981 For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402 — Subscription price per year: $14.00 domestic and $17.50 foreign. Cost per single issue; $4.00 domestic and $5.00 foreign. Contents-continued HAYNES, EVAN. Description of Stage II zoeae of snow crab, Chionoecetes bairdi, (Oxyrhyncha, Majidae) from plankton of lower Cook Inlet, Alaska 177 SPOTTE, STEPHEN, and GARY ADAMS. Feeding rate of captive adult female northern fur seals, Callorhinus ursinus 182 KAYA, CALVIN M., ANDREW E. DIZON, and SHARON D. HENDRIX. Induced spawning of a tuna, Euthynnus affinus 185 FROST, KATHRYN J., and LLOYD F LOWRY Trophic importance of some ma- rine gadids in northern Alaska and their body-otolith size relationships 187 WILLIAMS, AUSTIN B., and DAVID McN. WILLIAMS. Carolinian records for American lobster, Homarus americanus, and tropical swimming crab, Callinectes bocourti. Postulated means of dispersal 192 BARKER, SETH L., DAVID W TOWNSEND, and JOHN S. HACUNDA. Mortal- ities of Atlantic herring, Clupea h. harengus, smooth flounder, Liopsetta putnami, and rainbow smelt, Osmerus mordax, larvae exposed to acute thermal shock .... 198 BOWMAN, RAY E. Food of 10 species of northwest Atlantic juvenile groundfish .. 200 LIBBY, DAVID A. Difference in sex ratios of the anadromous alewife, Alosa pseudoharengus , between the top and bottom of a fishway at Damariscotta Lake, Maine 207 AL-JUDAIMI, MANAL M., A. K. JAFRI, and K. A. GEORGE. Proximate compo- sition and nutritive value of some important food fishes from the Arabian Gulf . . 211 Notices NOAA Technical Reports NMFS published during the last 6 months of 1980 213 Vol. 78, No. 4 was published on 28 April 1981. The National Marine Fisheries Service (NMFS) does not approve, rec- ommend or endorse any proprietary product or proprietary material mentioned in this publication. No reference shall be made to NMFS, or to this publication furnished by NMFS, in any advertising or sales pro- motion which would indicate or imply that NMFS approves, recommends or endorses any proprietary product or proprietary material mentioned herein, or which has as its purpose an intent to cause directly or indirectly the advertised product to be used or purchased because of this NMFS publication. CREPUSCULAR AND NOCTURNAL ACTIVITIES OF CALIFORNIAN NEARSHORE FISHES, WITH CONSIDERATION OF THEIR SCOTOPIC VISUAL PIGMENTS AND THE PHOTIC ENVIRONMENT^ Edmund S. Hobson,^ William N. McFarland,^ and James R. Chess^ ABSTRACT Activities in 27 of the major southern Califomian nearshore fish species, with emphasis on trophic relationships, were studied between 1972 and 1975 at Santa Catalina Island. Because these fishes orient primarily by vision, they are strongly influenced by the underwater photic environment, which we define with representative spectra. We center on crepuscular and nocturnal events, but also describe daytime events for comparison. The species that feed mostly by day include Atherinops affinis, Paralabrax clathratus, Girella nigricans, Medialuna califomiensis , Brachyistius frenatus, Cymatogaster aggregata, Damalichthys vacca, Embiotoca Jacksoni, Chromis punctipinnis, Hypsypops rubicunda, Halichoeres semicinctus, Oxyjulis califomica, Semicossyphus pulcher, Alloclinus holderi, Gibbonsia elegans, Heterostichus rostratus, and Coryphopterus nicholsi. Those that feed mostly at night include Scorpaena guttata, Sebastes atrovirens, S. serranoides (subadult), S. serriceps, Xenistius califomiensis, Seriphus politus , Umbrina roncador, and Hyperprosopon argenteum . Those that show no clear diurnal or nocturnal mode include Leiocottus hirundo and Pleuronichthys coenosus. Activity patterns tend to be defined less clearly in the warm-temperate fish communities of Califor- nia than in fish communities of tropical reefs. Included are the twilight patterns of transition between diurnal and nocturnal modes, which are considered to be defined by predation pressures. The lesser definition of twilight patterns in California could mean reduced crepuscular predation there, but we believe that Califomian fishes, too, have evolved under severe threats from crepuscular and nocturnal predators. We suggest this is evidenced in the spectral sensitivities of their scotopic visual pigments, which cluster around 500 nm — the best position for vision during twilight and at night in Califomian coastal waters. Although the scotopic system dominates vision in dim light, the spectral sensitivities of the scotopic pigments are poorly matched to the major forms of incident light at night — moonlight and starlight. Rather, they match twilight and bioluminescence, which favor similar spectral sensitivities. We believe this benefits these fishes most on defense. The match with twilight, when the low levels of incident light shift briefly to shorter wavelengths, enhances vision during the crepuscular periods of intensified threats from predators. And the match with bioluminescence permits fishes to react to threatening moves in nocturnal predators by responding to luminescing plankton that fire in the turbulence generated by these moves. Most fishes that live in southern Califomian coastal waters orient by vision, and so are strongly influenced by the cheiracteristics of underwater light at different times of the diel cycle. Knowing that these variations in light are accompanied by differing behavior patterns in the fishes (Hobson and Chess 1976; Ebeling and Bray 1976), we con- sider here circumstances during twilight and at night, when light is reduced and the fishes' 'Contribution No. 45 from the Catalina Marine Science Center, University of Southern California. ^Southwest Fisheries Center Tiburon Laboratory, National Marine Fisheries Service, NOAA, 3150 Paradise Drive, Tiburon, CA 94920. ^Section of Ecology and Systematics, Division of Biological Sciences, Cornell University, Ithaca, NY 14853. scotopic (dim-light sensitive) visual systems are operating (McFarland and Munz 1975c). A later report will consider circumstances during day- light. We relate the crepuscular and nocturnal ac- tivities of the fishes and their scotopic visual pig- ments to the spectral composition of light in their warm-temperate habitat, and compare these rela- tionships with the similar ties among activities, visual pigments, and light among fishes in tropical waters. We stress trophic relationships, because we con- sider these the major forces shaping activity pat- terns and related sensory systems in these fishes. The species studied are among the more numerous and readily observed in the nearshore warm- temperate eastern Pacific Ocean. Our accounts of Manuscript accepted June 1980. FISHERY BULLETIN; VOL. 79, No. 1, 1981. FISHERY BULLETIN: VOL. 79, NO. 1 their activities cover observations over 15 yr in southern California — from San Diego north to Point Conception — but our more detailed obser- vations, along with the light measurements and analysis of visual pigments, refer to Santa Catalina Island (lat. 33°28 ' N, long. 118°29 ' W), 35 km from the mainland (Figure 1). Here the water is consistently warmer and more transparent than on the adjacent mainland; during our study sur- face temperatures ranged between about 11 ° and 20° C, and underwater visibility generally ex- ceeded 10 m. Thus, when related to comparable data collected earlier in the tropics (Hobson 1968a, 1972, 1974; Munz and McFarland 1973; McFarland and Munz 1975a), these results offer a conserva- tive measure of differences between warm- temperate and tropical habitats. METHODS Determining the Spectral Composition of Submarine Sunlight The spectral distribution of submarine light was measured with a Gtunma 3000R spectroradiome- ter^ mounted in an underwater housing (Munz and McFarland 1973). The instrument, fitted with ■•Reference to trade names does not imply endorsement by the National Marine Fisheries Service, NOAA. a cosine receptor head and calibrated in photons per square centimeter per nanometer per second, draws a quantal irradiance spectrum. Radiance of the backlighting along a particular line of sight was measured by restricting the angle of view of the receptor head to a narrow cone (ca. 0.008 steradians). Usually radiance was determined along the zenith, horizontal, and nadir lines of sight. Because we were interested in comparing the spectral distribution of submarine light for differ- ent water conditions and along different lines of sight, results have been normalized and are pre- sented in terms of relative number of photons. The light levels that occur at twilight were beyond the spectroradiometer's sensitivity for measurement of spectral radiance. At twilight, therefore, spec- tral irradiance and not spectral radiance was measured. Irradiance data are reported in terms of absolute numbers of photons. To facilitate comparisons, several of the spectral curves were indexed by calculating their A.P50 val- ues (Munz and McFarland 1973; McFarland and Munz 1975a). The AP50 value represents the wavelength within the visible spectrum (400-700 nm) that halves the total number of photons under a spectral curve. Because underwater light is usu- ally homochromatic and fairly symmetrical in dis- tribution, XP50 provides a useful single index to a spectrum. 34' 33° Figure 1.— The study area in southern California. Santa Catalina Island was the site of detailed observations, includ- ing light measurements and analysis of visual pigments in fishes. 32° I I i^i^^j I J )■ T-IJll"! J It ri 1 tl { 11 t T f I I I ( I I 1 T't* T! r 15 T f J ^.SANTA BARBARA PT CONCEPTION LOS ANGELES 50 100 Ul I— I I— I U-l 1— I J KILOMETERS I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 120° 119° 118° HOBSON ET AL.: CREPUSCULAR AND NOCTURNAL ACTIVITIES OF CALIFORNIA FISHES Determining Activity Patterns in Fishes Our accounts of activity patterns in the fishes stem from direct underwater observations and from study of gut contents. The underwater obser- vations were made using scuba and by snorkehng during all hours of day and night. The gut contents were from fishes speared at all hours of day and night, but primarily during late afternoon and within 2 h before first morning light — times which best distinguish diurnal and nocturnal habits. To study the gut contents, the digestive tract of each fish specimen was removed im- mediately after collection and preserved in a 10% formaldehyde solution. Analysis under a binocu- lar dissecting scope was performed later in the laboratory. We note in this report only major food items that we believe might add insight to our accounts of diel activity patterns. More detailed accounts of the food habits are given elsewhere (Hobson and Chess 1976; in prep.). All mea- surements offish size are of standard length (SL). Although our accounts center on crepuscular and nocturnal events, we describe enough of what happens in daylight to consider these events in the context of diel patterns. Determining Spectral Photosensitivity of Fishes each retinal extract was homogeneous or con- tained more than one visual pigment. Pigment analysis was assisted by a computer program (Munz and Allen 1968) designed to test for homogeneity and also to characterize each visual pigment by estimation of the wavelength of peak absorbance (Amax)- Generally, the major photo- labile component in a vertebrate retinal extract is the rod visual pigment, and the minor compo- nent(s) is the cone visual pigment(s) (Munz and McFarland 1975; McFarland and Munz 1975b). Thus, in each retinal extract from the Catalina samples the dominant pigment is considered the scotopic (or rod) visual pigment. UNDERWATER PHOTIC ENVIRONMENT Coastal waters characteristically absorb light of shorter wavelengths than do oceanic waters be- cause they contain more dissolved organic matter. They also scatter more light due to higher con- centrations of suspended particulate matter. As a result, they transmit light of longer wavelengths, and, therefore, under a midday sun appear blue- green, rather than blue like the open sea (see Jer- lov 1968 for classification of water tj^es). Starting with these well-established facts, we attempted to characterize the underwater photic environment at Santa Catalina Island. Two techniques were used to obtain both fresh, and dark-adapted retinae from the fishes. Some of the specimens were captured alive and dark- adapted under laboratory conditions, whereas others were speared at night and immediately placed in dark containers before being returned to the laboratory for additional treatment. Spectral absorbance characteristics of the visual pigments from these retinae were determined by standard procedures. After each fish was dark-adapted, its eyes were enucleated, the retinae were removed under deep red light (Wratten #2 filter), and then frozen in 4% alum. Later the retinae were thawed, washed in triplicate, centrifuged, and the pellet extracted in 2% digitonin. Sonication of the pellet at 0° C for 1 min assisted solubilization of the visual pigment. After centrifugation, 10% by vol- ume of saturated sodium borate and 10% by vol- ume of 0.2 M hydroxylamine were added to the supernatant and the spectral absorbance of the extracted visual pigment recorded with a Cary 14 spectrophotometer. The method of partial bleach- ing (Dartnall 1952) was applied to test whether Submarine Daylight Midday Spectra Essentially all submarine daylight meaningful to fishes is produced by the sun. Although photic conditions during midday are not our concern in this paper (they will be considered in a later re- port), midday spectra effectively illustrate some fundamental aspects of the photic environment that are needed to understand scotopic vision in fishes. In particular, midday spectra can be used to define the spectral transmission characteristics of a given water mass, and so provide means to com- pare the photic environment in Californian coast- al waters with the photic conditions elsewhere, including comparisons of crepuscular and noctur- nal circumstances between different habitats. Californian coastal waters vary greatly in the way they transmit light, and while some of this variation is seasonal, much is shorter term and irregular. At times during our study at Santa Catalina very little suspended material was pres- FISHERY BULLETIN: VOL. 79, NO. 1 ent, and underwater visibility exceeded 20 m, but at other times heavy phytoplankton blooms re- duced visibility to <3 m. (We define underwater visibility as the horizontal distance over which we could see major environmental features in day- light.) Most of the time, however, conditions were intermediate between these extremes. Although our irregular observations of visibility do not permit a precise figure, we estimate that at least 70% of the time horizontal visibility in daylight was between 8 and 12 m. The images that we saw in these more typical conditions were relatively sharp — more like images seen in clear water than the relatively fuzzy images seen in turbid water. So in characterizing the photic environment in the surface waters, we recognize three sets of con- ditions: 1) clear — when visibility exceeded about 15 m amid relatively small amounts of visible sus- pended or dissolved materials, 2) bloom — when visibility was less than about 5 m owing to dense phytoplankton, and 3) typical — the more usual condition, when circumstances were intermediate to the above. Thus, when visibility was about 20 m in July 1976, the spectral radiance showed the essential blue-greenness of the water (Figure 2: clear curve), whereas when visibility was about 3 m during a phytoplankton bloom in May 1974, the spectra showed a shift toward the yellow-green ^TROPICAL SEA H— 3m 100 ./' » / \ / \^ BLOOM ""/ * \ / \ ^.-^- TYPICAL (f) o / ^ \ / /T ^'^ CLEAR 1- / ^ N. / / \y^^ o / ^ N. / A I ' X ^ Tt /\ a / ^ /v / \ 50 r * / \ / 1 ~ UJ V v/ 1 > A \ \ 1- / \ \ \ < _i / \ \ \ UJ / "S, \ \ a: * / "^ >v \ \ y >v ^^ \. " u/^ ^ ^V^ ^"^'''*''^^ ^-<^:.: t: : :.:•: .......^^ ,-: ....... A...:: \ ^ — H ' -^ ^^. 400 500 600 7C WAVELENGTH (nm) Figure 2. — Horizontal spectral radiance in warm-temperate coastal waters (Santa Catalina Island) under typical, bloom, and clear conditions, and in a tropical sea (Enewetak Atoll) under clear conditions. Cosine detector was 3 m below surface oriented horizontally (90° from the zenith). All values normalized, with typical condition at Santa Catalina stippled for emphasis. wavelengths (with >50% of the photons in the visible spectrum located between 500 and 600 nm; Figure 2: bloom curve). Under bloom conditions, therefore, the radiance was similar to that of lakes rich in plants (McFarland and Munz 1975c). The more usual intermediate condition, however, was closer to the clear than to the bloom condition (Figure 2: typical curve). Nevertheless, even under the clearest conditions encountered at Santa Catalina, the water was greener than it was under similar circumstances in a tropical lagoon (Figure 2: tropical-sea curve). The spectra depicted in Figure 2 represent a horizontal view, which effectively measures the background spacelight (or horizontal backlight- ing) against which the fishes studied here see most objects. At the same time we also measured downwelling and upwelling spectra, but they add nothing to the topics considered in this paper that is not illustrated by the horizontal readings. So they will be included in our later paper on photic conditions during midday. Twilight Spectra The broad spectrum of downwelling light in near-surface waters at Santa Catalina shifted to- ward the blue during twilight (Figures 3,4), even though skylight acquires relatively more red photons at this time. Fading daylight characteris- tically loses photons between 550 and 700 nm more rapidly than it loses photons below 550 nm, so that as twilight progresses the proportion of photons at the shorter wavelengths steadily in- creases (McFarland and Munz 1975c; McFarland et al. 1979). The pattern varies in response to changing local conditions, however. For example, on 21 November 1974, events at day's end pro- ceeded typically under a clear sky until shortly after sunset (Figure 3, top and middle panels). At this time, the sky suddenly was covered by a layer of cirrostratus clouds and immediately acquired a red-orange hue (through refraction of the sun's rays). Although we did not record atmospheric spectra at this time, an underwater spectrum re- corded 10 min after sunset (Figure 3, bottom panel) was essentially flattened across the visible wavelengths. Other variations in the twilight shift to shorter wavelengths occur under differing water conditions, as exemplified by a weakening of the phenomenon during phytoplankton blooms (Figure 4). The extent of the blue shift during underwater twilight can be measured by the HOBSON ET AL.: CREPUSCULAR AND NOCTURNAL ACTIVITIES OF CALIFORNIA FISHES 8 o .3 CM E .^ .2 N CO z o I- o X a. .08 .04 PRESUNSET-30min,Z at 3m POSTSUNSET- lOmin M 400 500 600 WAVELENGTH (nm) 700 Figure 3. — Changes in down welling submarine spectral irradiance through sunset near the surface at Santa Catalina Island under typical conditions. The shift toward the shorter wavelengths typical of surface waters at sunset (McFarland and Munz 1975c) is apparent in the upper and center graphs, where those wavelengths that divide the number of visible photons into equal quarters are identified by P25, P50, and P75. The bottom graph represents the underwater spectrum when a thin layer of clouds moved over the sky at about 300 m altitude and demonstrates some of the variability that can stem from meteorological events. changes in APso- Thus, during our observations under typical conditions, A.P5Q shifted from a pre- sunset value of 527 to 509 nm (Figure 3), and during bloom conditions from 540 to 528 nm (Fig- ure 4). Presumably the APgQ would have been closer to 500 nm had we made comparable mea- surements during twilight when the water was exceptionally clear. Submarine Nightlight Moonlight, starlight, and bioluminescence are the major forms of submarine nightlight meaning- ful to fishes. We did not measure these during the present study, but data available in the literature permit a comparison of the spectra each would be expected to produce in water like that typical of Santa Catalina (Figure 5). The influence of moon- light and starlight on activities in fishes and other aquatic predators has been discussed (e.g., Hobson 1965, 1966). This report considers the role of bioluminescent emissions of epipelagic plankton in predator- prey relationships. Epipelagic bioluminescent plankton, especially dinoflagellates, are widespread in most neritic and oceanic seas (Tett and Kelley 1973). Sailing narra- tives and logbooks of open water mariners are replete with descriptions of the "phosphorescent FISHERY BULLETIN: VOL. 79, NO. 1 400 500 600 WAVELENGTH (nm) 700 Figure 4. — Changes in downwelling submarine light through sunset near the surface at Santa Catalina Island during a phyto- plankton bloom. Although there is relatively less blue light under these conditions, a blue shift at and following sunset is nonetheless evident when compared with daylight under bloom conditions (Figure 2). P25, Pgf,, and P^j identify those wavelengths that divide the number of visible photons into equal quarters. fire" of the ship's wake, the luminescent shroud about porpoises running before the bow, and the showers of sparks that trail fishes dashing below the hull. At Santa Catalina Island, biolumines- cence from plankton was visible in the water at all times of the year, more so at some times than at others. Most marine bioluminescent plankton emit light in the blue region of the spectrum (Tett and Kelley 1973). For example, light from Gonyaulax polyedra and Noctiluca miliaris (which is repre- sentative of most dinoflagellates) peaks near 475 nm, and more than half of the photons are emitted below 500 nm (Hastings and Sweeney 1957; Nicol 1958). Because of the skewed emission spectra from these organisms (Figure 6), however, fishes close to the luminescent source would absorb more photons with visual pigments that have k^^^ val- ues nearer 490 nm than 475 nm. In any event, the spectral quality of biolumines- cence received by the fishes is modified by two variables — distance the fishes are from the light's source, and clarity of the intervening water. Be- cause the water more effectively absorbs the longer than the shorter wavelengths, a fish farther from a target in clear water will receive relatively more photons at the shorter wavelengths (Figure 6, upper panel). Water clarity, however, has an HOBSON ET AL.: CREPUSCULAR AND NOCTURNAL ACTIVITIES OF CALIFORNIA FISHES BIOLUMIN. - MOONLIGHT 00 ~ \ r\ /''^^ \\ ~ \ / \ / / \\ "" ^\j \ / / / \ / / / / / / \ * STARLIGHT 1 / / \ •y 1 \ / / \ I 1 y^ / 1 1 «. 1 -/— J^ \ \ 50 — r "S^ \ \ \ — / 1 / \ / 1/ \ \ N / /f \ \^ \ — / /\ \ \\ / / j \ ^ X / / \ v\. ~ " / 1 \ -/ 1 \ Wv ^ / 1 1 1 \ 1 1 " - 100 400 500 600 WAVELENGTH (nm) 700 Figure 5. — Underwater spectral distributions of moonlight, starlight, and bioluminescence. Values for bioluminescence are for Noctiluca miliaris, as given by Nicol (1958), at zero range. Values for moonlight and starlight, based on measurements in Munz and McFarland (1977), are for down welling light at zero range from a flat spectral reflector at a depth of 3 m in water equivalent to typical conditions at Santa Catalina Island ( Jerlov 1968, Coastal Type 1). even greater effect than distance on the attenua- tion and spectral modification of submarine light. In the typical water we encountered at Santa Catalina, for example, there was a fairly high transmission of light between 425 and 575 nm. As light travels from a source like N. miliaris under these conditions its radiance attenuates slowly and its spectrum shifts only slightly (Figure 6, middle panel). On the other hand, as the same light travels through water heavily loaded with phytoplankton it attenuates rapidly and there is a marked and continuous spectral shift toward the green (Figure 6, lower panel). ACTIVITY PATTERNS AND VISUAL PIGMENTS IN FISHES Because the photic environment contrasts sharply between day and night, those visually orienting fishes that are adapted to diurnal condi- tions should be less suited to feed after dark, while those adapted to nocturnal conditions should be less suited to feed by day. This expectation has been supported in studies of temperate species, both marine (Hobson and Chess 1976; Ebeling and Bray 1976) and freshwater (Emery 1973; Helfman 1979), just as it has in studies of marine fishes in 400 500 600 WAVELENGTH (nm) Figure 6. — Attenuation and spectral distribution of light emit- ted by Noctiluca miliaris over distance in water of differing clarities. The three panels of the figure each represent a different water type (clarity), as defined by Jerlov (1968): Type lA (upper) is equivalent to clear tropical seas; Coastal Type 1 (middle) is equivalent to typical conditions at Santa Catalina Island; and Coastal Type 7 (lower) is equivalent to conditions of heavy phytoplankton bloom at Santa Catalina. The heavy outer curve in each panel represents the light emitted by A'^. miliaris (left axes) at zero range, and so is the same in each water type. The inner curves in each panel represent relative attenuation of light at distances ( in meters ) indicated by the accompanying numbers. The broken line in each panel represents the transmission values/meter (right axes) for that water type, as given by Jerlov (1968). the tropics (Hobson 1965, 1968a, 1972, 1974, 1975; Starck and Davis 1966; Collette and Talbot 1972; Smith and Tyler 1972; Vivien 1973). Thus, in con- sidering the impact of diel variations in the photic environment it is meaningful to distinguish diur- nal and nocturnal species, even though some near- shore fishes feed at all hours — many by changing their food or tactics between day and night, e.g., the serranids Epinephelus labriformis in the Gulf of California (Hobson 1968a) and E. merra in the Indian Ocean (Harmelin- Vivien and Bouchon 1976) and the mullid Parupeneus bifasciatus in Hawaii (Hobson 1974). FISHERY BULLETIN: VOL. 79, NO. 1 We consider 27 of the most abundant fish species in the nearshore waters at Santa Catalina Island, describing what each does during twilight and at night, and noting the wavelengths of light to which the scotopic system of each is most sensitive (A^^^). In examining the retinae from the fishes, we noted whether the visual pigments were homogeneous. Significantly, there were no sec- ondary pigments in 12 species and only a trace in 5. Secondary pigments, which presumably are cone pigments (Munz and McFarland 1975), were present in 10 species and abundant in only 3. The data are given below, grouped according to that segment of the diel cycle when the species obtains most of its food. It is important that we observed only slight variation in the A^iax of any one species. Fishes That Feed Primarily by Day Some of the fishes that feed primarily by day are known to be inactive at night, but evidence of nocturnal inactivity remains lacking for others, and still others are known to feed routinely after dark. The predominantly diurnal Californian fishes considered in this paper, along with certain of their visual characteristics, are listed in Table 1. The following accounts of diel activities emphasize crepuscular and nocturnal habits. Table L — Some southern Californian marine fishes that feed primarily by day, with the spectral absorbance maximum (^niax^ of pigments extracted from their retinae. Other Family and species ^max^95%C.I. N pigments' Atherinidae: Atherinops affinis^ 505.8 2 Serranidae: Paralabrax clathratus' 498.8±2.0 4 + Kyphosidae: Girella nigricans 498 .3± 1.0 3 Medialuna californiensis 496.9 2 + Embiotocidae: Brachyistius frenatus 500.9±0.5 3 Cymatogaster aggregata 500.4±4.0 3 + Damalichthys vacca 500.9 2 + Embiotoca jacksoni^ 500.8 2 + Pomacentridae: Chromis punctipinnis^ 496.1 1 + Hypsypops rubicunda 496.3±0.1 3 Labridae: Halichoeres semicinctus^ 513.2±7.4 3 + + Oxyjuiis californica' 511.8 1 + + Semicossyphus pulcher^ 496.7±3.5 3 + + Clinidae: Alloclinus holderi 496.5 2 Gibbonsia elegans 499.7+8.9 3 + Heterostichus rostratus 499.7±2.0 4 T Gobiidae: Coryphopterus nichoisi' 497.9 1 Mean" 499.1 Range" 496.1-505.8 'Other than the primary pigment: = none; T = trace; + = <10%; ++ = >10%. ^Visual pigments for these species were also studied by Munz (1957, 1958b, c, 1964). He reported similar A^iax values for all but £ jacksoni, which he listed (Munz 1958b) as 506 nm. The difference can be attributed to varying amounts of secondary pigments in his extracts, which can bias the \max estimates if not taken into account ^Visual pigments in these species are porphyropsins, which are based on the aldehyde of Vitamin A2. Pigments in all other species are rhodopsins. which are based on the aldehyde of Vitamin At. ^Halichoeres semicinctus and O. californica excluded owing to basic differences in their pigments (see footnote 3, above). Atherinidae: Atherinops affinis The topsmelt aggregates by day in the surface waters close to kelp forests, but at night most larger individuals move away from the kelp and disperse close beneath the water's surface over adjacent deeper water. At first we suspected these larger individuals might feed after dark. Their movements are similar to those of tropical Pacific atherinids of the genus Pranesus, which are known to be nocturnal feeders (Hobson and Chess 1973; Hobson 1974; Major 1977), and nocturnal habits are widespread in other planktivorous atherinids, including A //anei^a harringtonensis in the tropical Atlantic Ocean (Starck and Davis 1966). Furthermore, we have often seen Atherinops affinis feed at night next to illumi- nated piers, although we consider this an artificial situation. Despite the evidence of nocturnal feeding in other atherinids, however, our suspicions concern- ing A. affinis were contradicted by examination of gut contents. Of 22 individuals (129-219 mm SL, x = 168.8) collected at davni as they reassembled in schools along the outer edge of kelp forests, 19 were empty, 2 contained just a few fish scales, and 1 contained calanoids and cyphonautes larvae that appeared recently ingested — probably since sun- rise that morning. At least some of the smaller A. affinis remain close to the kelp at night, but there is little evidence that they feed during that period. Of 10 (82-160 mm SL, jc = 102.4) collected close to kelp during the hour before dawn, 9 were empty. The one with food, however, contained three gam- marids and one isopod that obviously had been taken at night. In contrast, there is ample evi- dence that A. affinis feeds intensively during the day. We routinely observed this species feeding in the surface waters during all daylight hours, and only 1 of 10 (126-190 mm SL, 3c = 158.6) collected from a large aggregation during midafternoon lacked food in its gut; the other 9 contained x = 1,325 prey items, mostly cladocerans and copepods. 8 HOBSON ET AL.: CREPUSCULAR AND NOCTURNAL ACTIVITIES OF CALIFORNIA FISHES Serranidae: Paralabrax dathratus The kelp bass progresses through three major ontogenetic phases based on trophic relationships. The first phase includes juveniles up to about 65 mm SL that feed primarily on zooplankton during the day and pass the night sheltered amid vegeta- tion. The third phase includes the largest individuals — those exceeding about 165 mm SL — which are increasingly piscivorous with growth and may be primarily crepuscular (al- though limitations in our data leave the feeding chronology at this larger size in question). In this paper we consider individuals represent- ing the second phase — subadult fish between 65 and 165 mm SL. Individuals of this size feed mostly on crustaceans that live on or close to a substrate by day and swim in the water column after dark, including certain gammarid and caprellid am- phipods, isopods, cumaceans, mysids, and cari- deans. The subadult P. dathratus capture these crustaceans mainly by day close to benthic cover. Most subadult P. dathratus do not feed at night. Only 42% (13 of 31, 72-163 mm SL, x = 110.3) of those collected during the hour before sunrise con- tained prey, whereas there was prey in 96% (51 of 53, 68-153 mm SL, x = 107.9) of those collected during the afternoon. Whether or not the subadult P. dathratus feeds at night, however, seems related to their location. Most of them are amid rocks and vegetation at night, just as during the day, and here they seem to feed little, if at all. Of 16 (95-146 mm SL,x = 108.6) collected in these surroundings during the hour before dawn, only 1 (6%) had food in its stomach (a moderately digested caridean). Nocturnal feeding in subadult P. dathratus seems to occur mostly in those individuals that move after dark out over open sand (Figure 7) — a habitat only infrequently occupied by them during the day. Of the 16 (72-148 mm SL, X = 97.4) collected in such places during the hour before dawn 12 (75%) contained food, much of it fresh. Major prey were the cumacean Cyclaspis nuhila (4 mm), the gammarids Am- pelisca cristata (3-4 mm) and Amphideutopus oculatus (2-3 mm), and the caprellid Caprella californica (6-8 mm) — all species that are active on or close above the sand at night. Clearly, the nocturnal move over the sand is a well established feeding pattern in subadult P. dathratus. Nevertheless, even under these special cir- cumstances predatory success after dark seems limited. Among specimens from open sand at night the stomachs containing food averaged only 30% full, compared with 66% full for specimens from a wide range of diurnal circumstances. FIGURE 7. — A subadult Paralabrax clathmtus , about 150 mm SL, alert on open sand at night. Some individuals of this size feed under these circumstances even while larger and smaller conspecifics are inactive close to rocks and algae. Apparently the sand reflects enough moonlight and starlight to permit some predominantly diurnal fishes to feed in these surroundings at night. FISHERY BULLETIN: VOL. 79, NO. 1 Kyphosidae: Girella nigricans The opaleye does not seem to feed extensively at night, although our limited observations on this point are somewhat ambiguous. This rather tenu- ous opinion rests heavily on two points. First, the species feeds largely on algae, and various or- ganisms that live on algae — a diet generally indi- cative of diurnal feeding (Hobson 1965, 1974; Viv- ien 1973). Second, kyphosids generally have been found to feed by day (e.g., Randall 1967). On the other hand, observations of kyphosids at night have indicated a variable condition. Kyphosus in- cisor reportedly rests ". . .in sheltered, though not confined, locations on the reef-top" in the Florida Keys (Starck and Davis 1966), andK. elegans has been noted to behave similarly in the Gulf of California (E. S. Hobson unpubl. obs.). ButK. ele- gans, at least, is alert in these shelters, and K. cinerascens in Hawaii not only swims above the reef at night, but may feed at this time as well (Hobson 1974). The presence of G. nigricans ". . .in holes or on the bottom . . ." at night led Ebeling and Bray (1976) to consider it diurnal. We agree with them even though we frequently saw this species swimming in the water column after dark, espe- cially in the kelp forest. Limited study of gut contents suggest reduced feeding in some G. nigricans at night. One (201 mm) sampled among the rocks during the hour before sunrise had an empty stomach, but the stomach of another (264 mm) that was swimming above the bottom at this time was 20% full. Sixty- seven percent of the gut contents in this second individual consisted of motile animals, including gammarids, caprellids, carideans, and the gas- tropod Tricolia sp., whereas only 10% consisted of algae. This material, much of it fresh, differed sharply from that in gut contents of individuals that had been feeding by day. Benthic algae consti- tuted 81% of the material in all 11 individuals (173-255 mm SL, x = 206) that were collected during the afternoon, and whose stomachs aver- aged 75% full; essentially all other items in these individuals were sessile organisms that encrust, or live attached to, algae — principally bryozoans and hydroids. Kyphosidae: Medialuna calif orniensis We consider the halfmoon primarily diurnal for essentially the same tenuous reasons that led us to this conclusion for G. nigricans: many of its close relatives reportedly are diurnal, as documented above, and it feeds heavily on plants (Limbaugh 1955; Quast 1968), a diet widely associated with diurnal foraging. Furthermore, we saw M. californiensis , like G. nigricans, in the water col- umn in far fewer numbers at night than during the day. But although we saw G. nigricans in larger numbers close to rocky substrata at night, we saw M. californiensis, which was exceptionally numerous in the kelp forests by day, only in sharp- ly reduced numbers there after dark. Reporting a similar situation in a kelp forest at Santa Barbara, Ebeling and Bray (1976) observed about half as many M. californiensis on their transect line at night as during the day. They summarized their nocturnal observations by stating that this species ". . . often appeared to be more sensitive to our pres- ence than were individuals of other species near the bottom, and we cannot deny the possibility that Medialuna feeds at night." Leading to much the same position, our study of gut contents from 14 specimens (146-243 mm SL,x = 196) collected during the afternoon (Hobson and Chess in prep.) shows that this largely herbivorous species un- questionably feeds by day, but leaves unanswered whether or not it also feeds at night. Embiotocidae: Brachyistius frenatus Although the kelp perch is basically diurnal throughout life, this characteristic can be some- what variable. When less than about 100 mm SL it feeds primarily on zooplankters in the water col- umn, and as it grows larger it increasingly turns to tiny prey — mostly crustaceans — that it picks from vegetation (Hobson and Chess 1976). At night it occurs in most of the same places that it occupies by day, but is more numerous in midwater aggre- gations fully exposed along the outer edges of the kelp forests (Hobson and Chess 1976). Describing the nocturnal condition of this species, Bray and Ebeling (1975) noticed that it "...tended to hang motionlessly along the kelp stipes or even in open water" and added that it was "...quiescent at night and easily caught with a small hand net . . . ." They concluded that B. frenatus feeds ". . . mostly, if not exclusively, during the day" — an opinion based on study of gut contents from specimens collected every 2 h throughout the night. But Hob- son and Chess (1976) reported that while B. fre- natus is primarily diurnal, larger individuals also feed to a limited extent at night. In addition to the data presented in that paper, all of which involved 10 HOBSON ET AL.: CREPUSCULAR AND NOCTURNAL ACTIVITIES OF CALIFORNIA FISHES specimens <100 mm long, 6 of 14 (102-114 mm SL, X = 106.8) collected during the hour before dawn contained prey, many of them fresh. The major nocturnal prey, which included the gammarid Batea transversa (2-4 mm), the caprellid Caprella californica (8-14 mm), and the isopod Paracercies cordata (1-6 mm), were organisms that rise into the water column at night. We believe these prey were captured in the water column because most were in fish that had been aggregated in midwater outside the seaward edge of a kelp forest. Perhaps Bray and Ebeling (1975) found no evidence of noc- turnal feeding because their sample comprised mostly smaller fish. Significantly, the major or- ganisms apparently taken in the water column at night include the same species picked from the surface of algae by day. Embiotocidae: Cymatogaster aggregata The shiner perch, which is even more variable in its diel behavior than Brachyistius frenatus, has two basic feeding modes: it captures zooplankton, mostly crustaceans, in the water column, and it captures organisms, again mostly crustaceans, that are in, on, or close above a sandy bottom. The planktivorous habit predominates among indi- viduals smaller than about 65 mm SL and con- tinues to be important throughout life, whereas feeding on sand-dwelling forms becomes increas- ingly important to individuals >65 mm SL until it predominates among the largest individuals. The planktivorous habit is diurnal, whereas feeding on or close to a sandy bottom occurs during both day and night, but mostly at night. Thus, the small juveniles are primarily day feeders: of 23 (53-64 mm SL, x - 58.2) collected during the afternoon, 12 (53%) contained food, the major items being zooplankters (mostly copepods). Only two (12%) also included prey that may have been taken from the seafloor (gammarid frag- ments). In comparison, only 2 of 17 (37-64 mm SL, X = 46.5) collected during the hour before dawn contained food: one (58 mm) contained just a few cumacean fragments, but the other (54 mm) con- tained a variety of sand-dwelling crustaceans, some fresh, including the cumacean Cyclaspis nubila (2-4 mm), the gammarid Acuminodeutopus heteruropus (1-2 mm), and the tanaid Leptochelia duhia (2-4 mm), along with sand. The changes in food habits that appear among individuals >65 mm SL were similarly defined. Of 34 (67-110 mm SL, x = 91.8) collected during the afternoon, 27 (79%) contained food presumably taken by day: 19 (70%) of these had fed exclusively on zooplankton (primarily calanoid and cyclopoid copepods and cladocerans), 4 (15% ) had taken only sand-dwellers (primarily tanaids and gam- marids), and 4 (15%) had fed on both zooplankton and sand-dwellers in large numbers (combina- tions of the above forms, with the two types sharp- ly separated in the guts). In comparison, of 46 (66-120 mm SL,x = 85.9) collected during the hour before dawn, 37 (80%) contained prey, many fresh, that appeared for the most part to have been taken at night. Significantly, all these prey were sand- dwellers, the major forms being the cumacean Cyclaspis nubila (2-4 mm), the gammarids Acuminodeutopus heteruropus (1-2 mm) and Am- pelisca christata (2-4 mm), the tanaid Leptochelia duhia (3-4 mm), and the ostracod Euphilomedes carcharondonta (1-2 mm), along with sand. Embiotocidae: Damalichthys vacca and Embiotoca jacksoni Damalichthys vacca, the pile perch, and E. jacksoni, the black perch, both appear to be strictly diurnal feeders. During daytime, adults of D. vacca feed primarily on moUusks and other heav- ily shelled prey, whereas adults of E. jacksoni take an exceptionally wide variety of benthic or- ganisms, including polychaetes, mollusks, gam- marids, caprellids, isopods, and mysids (Lim- baugh 1955; Quast 1968). At night we observed both species hovering close to the seafloor, gener- ally in exposed positions. On the other hand, Ebe- ling and Bray (1976) reported D. vacca "... scat- tered in the water column at night." That neither species feeds after dark is evidenced by the ab- sence of fresh food in their guts at that time (Ebe- ling and Bray 1976). Pomacentridae: Hypsypops rubicunda and Chromis punctipinnis Although the Californian pomacentrids — H. rubicunda, the garibaldi, and C . punctipinnis , the blacksmith — are strictly diurnal fishes that re- main relatively inactive in their nocturnal shel- ters, they nevertheless remain alert throughout the night. Hypsypops rubicunda, in fact, often ap- pears restless as it moves in its shelter place. This species is solitary during both day and night, and its nocturnal shelter is a specific hole or crevice in the well-defined territory that also includes its 11 FISHERY BULLETIN: VOL. 79, NO. 1 diurnal foraging area (Clarke 1970). In contrast, C. punctipinnis is highly gregarious during both day and night, and often individuals crowd noc- turnal shelters among the rocks. When its noctur- nal resting places are far from its diurnal feeding grounds, C punctipinnis migrates between the two locations during twilight in prominent pro- cessions (Hobson and Chess 1976). These migra- tions tend to be better defined in the evening than in the morning. At times evening migrations began up to 30 min or more before sunset, while at other times they were not evident until about sun- set. Generally, however, the migrations peaked from shortly before sunset until about 15 min af- ter, and then continued at greatly reduced levels for another 10 min or so before ending. Chromis punctipinnis is among the most numerous species on many nearshore reefs in southern California (Limbaugh 1955; Quastl968), and in some places apparently there is insufficient rocky shelter to accommodate at night the vast numbers that forage in the water column by day. In such places excess individuals sometimes clus- ter after dark in dense numbers on the sand next to the reef. Occasionally C. punctipinnis is in the water column at night, but does not seem to feed at this time. This is attested by the empty stomachs of all 11 specimens examined from predawn collec- tions by Hobson and Chess (1976). Labridae: Oxyjulis caltfornica, Halichoeres semicinctus, and Semicossyphus pulch&r The Californian labrids are so obviously quies- cent at night that we have no doubt that, like tropical labrids (Hobson 1965, 1974), they do not feed at this time. Also, like their tropical relatives, they follow precise patterns when shifting be- tween diurnal and nocturnal modes (Hobson 1972; Domm and Domm 1973). Oxyjulis californica, the senorita, buries in the sand at nightfall (Herald 1961) , a habit also attributed to H. semicinctus , the rock wrasse, by Limbaugh (1955), who noted that this species "sleeps buried with head protruding." Feder et al. (1974) repeated this statement, as did Fitch and Lavenberg (1975), who reported that it also "burrows between or under rocks to escape predators or to sleep." We found that H. semicinctus consistently took nocturnal shelter amid low benthic algae (from which it often was unintentionally flushed by our nocturnal ac- tivities). Both O. californica and H. semicinctus were consistent in timing their descent to shelter in the evening, and their rise into the water the next morning. Data on the timing of these events were collected during 1973 and 1974 at Fisher- man's Cove, Santa Catalina Island, in a sand- bottomed habitat dominated by the low brown alga Dictyopteris zonaroides. The last O. califor- nica seen entering the sand on four evenings at this site slipped from view 11-22, x = 16.8, min after sunset, which agrees with Bray and Ebeling (1975), who on three occasions observed this species entering sand and rubble "About 15 min after sunset " The first O. californica seen here on 11 mornings appeared 11-22, x = 15.8, min be- fore sunrise. During the same period at this site the last H. semicinctus seen taking cover on six evenings disappeared 20-24, x = 22.0, min after sunset, and the first to appear on 10 mornings emerged 18-25, x = 22.8, min before sunrise. It may be significant that H. semicinctus , which grows to a larger size, tended to be active later in the evening and earlier in the morning. Among diurnal fishes on tropical reefs, the larger individu- als tend to retire later and rise earlier (Hobson 1972). Comparable data are lacking forS. pulcher, the sheephead (the largest of the three Californian labrids), even though this species was numerous at the observation site during the day. Semicos- syphus pulcher shelters among rocks at night (Hobson 1968b), and probably because the obser- vation site lacked rocks, this species left the area sometime before going under cover. On 4 evenings the last S. pulcher departed the observation site 10-21, X = 18.0, min after sunset, and on 11 morn- ings the first to return arrived 11-24, x = 17.0, min before sunrise. On just one evening, in a nearby rocky area where the species found shelter, the last active S. pulcher was seen 30 min after sunset. Because S. pulcher rested in distinctive shelters and was visible throughout the night (Figure 8), it offered the best opportunity to investigate consis- tency in resting places among individuals. Reportedly, at least some tropical wrasses return each evening to specific resting sites (Winn and Bardach 1960; Starck and Davis 1966; Hobson 1972). So the resting places of nine S. pulcher were located at midnight during November 1973, and then revisited at the same hour once each week for 3 wk. On the first return only two of the nine positions were occupied. No. 3 and No. 5 — both by what appeared to be the same individuals that had been there before. On the second return, again only two positions were occupied: No. 5 seemed to 12 HOBSON ET AL.: CREPUSCULAR AND NOCTURNAL ACTIVITIES OF CALIFORNIA FISHES Figure 8. — a female Semicossyphus pulcher at rest on the seafloor at night, showing the typical pattern of its nocturnal hues. Its exposed position is common in this and many other noctumally resting diurnal fishes in California. harbor the same fish as before, but this time posi- tion No. 1 contained a small individual not seen before. On the final return, position No. 5 once more was occupied by what seemed to be the same fish. Again, just one other position was filled — No. 6, which harbored the fish seen in position No. 9 on the first night (recognized by a notch in its dorsal fin), but which had gone unseen since then. Thus, only one position, No. 5, sheltered a fish each time. And contrary to what one might expect, of the nine positions. No. 5 offered the least cover. It was sim- ply a shallow depression on the reeftop where the resting fish was largely exposed, and certainly would not seem an effective shelter. Clinidae: Alloclinus holderi, Gibbonsia elegans, and Heterostichus rostratus The three Californian clinids studied here — A. holderi, the island kelpfish; G. elegans , the spotted kelpfish; andH. rostratus, the giant kelpfish — are known to feed regularly by day, based on fresh food in specimens collected during the afternoon (Hob- son and Chess in prep.). But it is difficult to deter- mine relative activity in these highly cryptic fishes because they move so infrequently and, therefore, often go unnoticed even when ftilly ex- posed. Both A. holderi, which sits on rocks, and G. elegans, which sits amid benthic algae, retire to shelter at nightfall, as do various tropical clinids (Starck and Davis 1966; Smith and Tyler 1972). Heterostichus rostratus, on the other hand, often hovers among columns of giant kelp during both day and night (Figure 9). Our data comparing relative feeding activity in H. rostratus between day and night are limited, but indicate that day- time feeding predominates. The one individual (184 mm) collected during the hour before sunrise 13 FISHERY BULLETIN: VOL. 79, NO. 1 Figure 9. — While most of its smaller blennioid relatives go under cover at nightfall, the large clinid Heterostichus rostratus, here amid rising kelp stipes at night, shows a similar attitude at all hours, even though it seems to feed primarily by day. was empty, and the two (163 and 207 mm) collected at midnight contained only well digested frag- ments. In comparison, among eight individuals (234-385 mm SL, jc = 280) collected during midaf- ternoon, four contained fresh food, one contained only well digested fragments, and three were empty. Gobiidae: Coryphoptertts nicholsi Large numbers of the blackeye goby rested in exposed positions on sand bottoms in and around rocks throughout the day, when intermittently, they darted forward, or a short distance into the water column, and snapped at tiny prey. Few, how- ever, were visible at night. Presumably most shel- tered in the reef after dark — a pattern reportedly followed by four species of Coryphopterus in the tropical Atlantic Ocean (Smith and Tyler 1972). Gut contents indicate this species feeds little, if at all, after dark. Among the few individuals seen at night, seven (43-83 mm SL, x = 59.3) were col- lected during the hour before the first morning light; the gut of one was empty, and the other six contained only well digested items. In comparison, all 69 specimens (36-90 mm SL,x = 64.1) collected 14 lOBSON ET AL.; CREPUSCULAR AND NOCTURNAL ACTIVITIES OF CALIFORNIA FISHES luring the afternoon had food in their guts, with 31 of these (88%) containing recently ingested material. Fishes That Feed Primarily at Night The fishes that feed primarily at night clearly are specialized to detect and capture prey in the dark; nevertheless, under appropriate cir- cumstances some also take prey during the day. The predominantly nocturnal species considered in this paper, w^ith certain of their visual charac- teristics, are listed in Table 2. The following ac- counts of diel activities highlights major features of their crepuscular and nocturnal habits. Scorpaenidae: Scorpaena guttata The sculpin rests immobile among rocks during the day (Figure 10) and generally is difficult to discern owing to its cr5^tic features. Although Table 2. — Some southern Califomian fishes that feed primarily at night, with the spectral absorbance maximum 400 m from the nearest point where the species had been seen during the day (Hobson and Chess 1976). On four evenings we noted when the first salema arrived on this feeding ground, and this 17 FISHERY BULLETIN: VOL. 79, NO. 1 proved to be betv^een 34 and 40, x = 37.3, min after sunset. Once there, its behavior was much like that described above for nocturnally feeding olive rockfish. Its diet, too, proved similar to that of the rockfish. All 13 individuals (163-170 mm SL, x = 173.7) collected more than 3 h after sunset were full of food, much of it fresh. Prey were organisms that are in the nearshore water column only at night, with major forms being the gammarids Batea transversa (2-4 mm) and Ampelisca cristata (3-8 mm), the caprellid Caprella pilidigita (6-12 mm), the cumacean Cyclaspis nubila (2-4 mm), the mysid Siriella pacifica (4-10 mm), and epitok- ous nereid polychaetes (15 mm). Sciaenidae: Seriphus politus The queenfish schools in relatively inactive as- semblages near shore during the day and dis- perses to feed in the water column at night after moving away from its daytime schooling sites (Hobson and Chess 1976). Thus, its diel activity pattern is similar to that of subadult Sebastes ser- ranoides and X. californiensis, described above. The first Seriphus politus appeared at a nocturnal feeding site on four occasions at 38-60, x = 44, min after sunset. Food and feeding behavior of S. politus also are similar to the other two species. All 31 individuals (114-193 mm SL, x = 151) sampled later than 3 h after sunset contained prey, much of it fresh. All were larger zooplankton that are in the water column only at night, wdth major forms being mysids (Siriella pacifica, 3-11 mm, and Acanthomysis sculpta, 6-11 mm), a gammarid (Batea transversa, 2-4 mm), and an isopod (Paracercies sp., 2-7 mm). Sciaenidae: Umbrina roncador The yellowfin croaker schools close to sandy beaches during the day, and at nightfall disperses here and also to the regions immediately offshore. It feeds on organisms in the sediment, often prob- ing with its snout to make the capture. Most of its foraging seems to occur at night. Of 20 individuals (191-255 mm SL, x - 210.8) collected more than 3 h after sunset, all but 1 contained prey, much of it fresh. Major items were sand-dwelling polychaetes, many of them tubicolous, with Onuphis sp. (15-40 mm) and Nothria stigmaeus (10-20 mm) predominating; other important prey were sand-dwelling gammarids, especially Am- pelisca cristata (2-12 mm), Acuminodeutopus heteruropus (2-3 mm), and Paraphoxus heterocus- pidatus (2-3 mm). Only limited feeding occurs by day, as attested by eight individuals (210-239 mm SL, X = 222.5) collected during the afternoon. Of these, only two contained fresh material (sand- dwelling amphipods, most of them Ampelisca cristata). These same two, and two others, also con- tained extensively digested polychaetes that obvi- ously had been in the guts for some time; the other four (50% of the sample) were empty. Embiotocidae: Hyperprosopon argenteum The walleye surfperch is the only predomi- nantly nocturnal species among the five em- biotocids considered in this paper. It schools inac- tively by day, often close to shore, then disperses at nightfall and moves to feeding grounds some dis- tance away (Hobson and Chess 1976; Ebeling and Bray 1976). It forages in the water column, where it takes the larger zooplankton that are numerous there only after dark. Thus, H. argenteum has habits similar to those of the other nocturnal planktivores described above. Although it usually forages lower in the water column than these others, its diet and feeding behavior also are simi- lar: of 29 individuals (100-157 mm SL, x = 126) collected over nocturnal feeding grounds at night, or from recently formed schools before sunrise, 28 (97%) contained food, much of it fresh. Major prey items were the gammarids Batea transversa (2-4 mm), Ampelisca cristata (3-4 mm), and Ampithoe sp. (4-6 mm); the cumacean Cyclaspis nubila (2-4 mm); the isopod Paracercies sp. (2-5 mm); and the caprellid Caprella pilidigita (4-10 mm). Fishes That Feed Day and Night Only two of the species studied resist classifica- tion as being either primarily diurnal or primarily nocturnal in their feeding activities. Both seem equipped to exploit circumstances that permit ef- fective feeding during all hours of day and night. They are listed in Table 3, along with certain of their visual characteristics. The following ac- counts of their diel foraging activities puts their diurnal and nocturnal habits in perspective. Cottidae: Leiocottus hirundo The lavender sculpin rests immobile in exposed locations on sandy substrata, usually near rocks and algae, at all hours of day and night. On the 18 HOBSON ET AL.: CREPUSCULAR AND NOCTURNAL ACTIVITIES OF CALIFORNIA FISHES Table 3. — Some southern Califomian fishes that feed day and night, with the spectral absorbance maximum (X^jax' o^ P'8" ments extracted from their retinae. Family and species ^max^95%C.I. N Other pigments' Cottidae: Leiocottus hirundo PleuronectJdae: Pleuronichthys coenosus Mean 500.0 500.9^0.5 500.5 2 3 T 'See Table 1, footnote 1. rare occasions that feeding was observed, this fish moved only a few centimeters to snatch an object in the sand. Eight specimens (130-196 mm SL, jc = 177) were collected from a variety of habitats dur- ing the afternoon. Six (75%) contained food in their stomachs, with the predominant prey being the polychaetes Glycera capitata (30-85 mm), Lumberineris sp. (20-90 mm), and terrebellid ten- tacles. One had taken the gammarid Ampelisca cristata (12 mm), and one a holothurian (17 mm). Comparable data on nocturnal feeding was ob- tained from eight specimens (66-194 mm SL, x = 136) collected during the hour before dawn. Again, six (75%) contained food in their stomachs, with polychaetes — Glycera sp. (35 mm), Nothria stig- matis (8 mm), and a terrebellid (30 mm) — the major prey, though less so than during the day. Gammarids, especially Ampelisca cristata (2-6 mm) were important to these nocturnal individu- als, as was the clam Solemya valvus (8-10 mm). But these differences in prey selection between day and night may relate to fish size rather than to time of feeding: two Leiocottus hirundo collected during the night were smaller (66 and 84 mm) than any taken by day, and it was these that had preyed mostly on gammarids. Aside from these minor differences in food composition, feeding habits appear similar day and night: stomachs of the day feeders averaged 75% full, and contained x = 4.2 items, compared with an average of 69% full andic = 4.8 items for the night feeders. Pleuronectidae: Pleuronichthys coenosus The C-0 turbot rests immobile on sandy sub- strata at all hours of day and night — usually ex- posed but sometimes under a thin layer of sedi- ment. Often it occurs in the same habitat as L. hirundo, but more so than the cottid it ranges into regions of open sand, where its highly variable coloration often matches the surroundings. We have observed feeding only in daylight, when typi- cally this species rests motionless, with body somewhat elevated on dorsal and anal fins and head poised above the substratum (Figure 12). Its mobile, closely set eyes are oriented vertically on a bony ridge, and function almost as if set in a tur- ret. This arrangement permits the fish to scan the seafloor close at hand, probably for moving sedi- ments or other signs of prey that aire just below the surface. Occasionally we have seen individuals that had been immobile in one spot for some time move a meter or so across the seafloor, pause for a moment, and then drive their heads into the sedi- ment. Usually we were unable to see what they had taken, but daytime quarry were identified in the 11 specimens that contained food out of 14 (161-212 mm SL, x = 186) collected from sandy substrata during the afternoon (stomachs aver- aged 47% full). The major prey were polychaetes, especially terrebellid tentacles. Although we did not observe feeding at night, prey were identified in all 11 specimens (159-220 mm SL, x = 183.5) collected during the hour before dawn (stomachs averaged 72% full). Again, polychaetes, especially terrebellid tentacles, predominated. Clearly its trophic relationships are similar to those of L. hirundo, except that it may be more able to feed at night. DISCUSSION Events during twilight and at night in Califor- nian marine habitats can be compared with equivalents on tropical reefs. Tropical activity patterns have been described (Hobson 1965, 1968a, 1972, 1974; Starck and Davis 1966; Collette and Talbot 1972; Smith and Tyler 1972; Vivien 1973), as has scotopic vision in tropical fishes (Munz and McFarland 1973). Below we relate our findings with Californian coastal fishes to these and other studies made elsewhere. First we consider crepus- cular and nocturnal activity patterns and then scotopic spectral sensitivity, first in relation to ambient light, and then to bioluminescence. Activity Patterns In relating diel activity patterns of fishes in Californian waters near Santa Barbara to fishes on tropical reefs, Ebeling and Bray (1976) referred to the Californian species as "kelp-bed" fishes. We assume they implied a broad concept of this term that includes fishes sometimes in kelp forests, but more characteristic of other habitats. This is be- cause the tropical side of their comparison (which 19 FISHERY BULLETIN: VOL. 79, NO. 1 Figure 12. — Pleuronichthys coenosus, which feeds day and night, largely on sand-dwelling polychaetes, has eyes on either side of a bony ridge set almost as if in a turret. This arrangement increases its ability to scan the surrounding seafloor for prey and threatening predators. is based on observations by Hobson 1965, 1968a, 1974; Starck and Davis 1966; and others) involves species (often referred to as coral-reef fishes) from a variety of contiguous habitats. A comparison of diurnal and nocturnal behavior requires a mul- tihabitat view because so many fishes move from one habitat to another between day and night. The general scene in Californian nearshore habitats differs dramatically between day and night. A major aspect of this difference is the sharp drop in observed activity among fishes on reefs after dark. Describing an "aura of desolation . . ." in the "notably lackluster night life, . . ." Ebeling and Bray (1976) considered this feature of kelp forests to be in contrast to tropical reefs. But on tropical reefs, too, one notes less activity at night than during the day (e.g., Starck and Schroeder 1965). Nevertheless, there may be an especially pro- nounced difference where Ebeling and Bray studied, because the relative dearth of nocturnal activities there led them to conclude: "...in kelp beds there is no broad replacement for the 'day shift' of fishes at night." In particular, they re- ported an absence of fishes that move from day- time assemblages on reefs to nocturnal feeding grounds on adjacent sand, and also to there being relatively few nocturnal planktivores. But the situation they described is unlike that which pre- vails in the more southerly waters around Santa Catalina, where many species are most active at night. Following a pattern widespread in the tropics, for example, Xenistius californiensis , Umbrina roncador, Seriphus politus, and Hyper- prosopon argenteum (to mention species consid- ered in this report) are relatively inactive in schools near shore, reefs, or kelp forests by day, and disperse over feeding grounds elsewhere at night. It may be significant, however, that with the exception of//, argenteum, these are species with close tropical affinities (Table 4). In contrasting the relative absence of nocturnal planktivores at their Santa Barbara study site, vdth the many such forms at Santa Catalina (as reported by Hob- son and Chess 1976), Ebeling and Bray suggested 20 HOBSON ET AL.: CREPUSCULAR AND NOCTURNAL ACTIVITIES OF CALIFORNIA FISHES Table 4. — Geographic affinities' of the fishes studied. I. Warm-temperate representatives of basically tropical families or genera; species tfiat do not range into ttie colder waters northward from central California. Twelve species: Scorpaena guttata Girella nigricans Paralabrax clathralus Chromis punctipinnis Xenistius californiensis l-lypsypops rubicunda Seriphus politus Halichoeres semicinctus Umbrina roncador Oxyjulis californica Medialuna californiensis Semicossyphus pulcher II. Temperate representatives of basically tropical families; species tfiat range widely into the colder waters northward from central California. Two species: Atherinops affinis Coryphopterus nichoisi III. Warm-temperate representatives of temperate families or genera; species that do not range northward from central California. Five species: Sebastes atrovirens Alloclinus holderi S. serriceps Gibbonsia eiegans Leiocottus liirundo IV Representatives of temperate families or genera; species that range widely northward from central California. Eight species: Sebastes serranoides Embiotoca jacksoni Bractiytstius frenatus Hyperprosopon argenteum Cymatogaster aggregata IHeterostichus rostratus Damalichthys vacca Pleuronichthys coenosus 'Based on ranges given in Miller and Lea (1972). that the difference may reflect the proximity of Santa Barbara to Point Conception, the northern boundary of the warm-temperate zoogeographic region (Hubbs 1960; Quast 1968; Briggs 1974). Certainly there is a strong tropical influence in many of the more clearly defined crepuscular and nocturnal activity patterns among southern Californian fishes. Species with the most distinc- tive patterns tend to be warm-temperate represen- tatives of what basically are tropical families. The three Californian labrids, for example, seek and leave nocturnal shelter at precise times relative to sunset and sunrise, just as their tropical relatives do. And the two Californian pomacentrids shelter under reef cover at night in the same manner as tropical pomacentrids. Similarly, of the species listed above that school inactively by day and dis- perse to feed at night, most represent the predom- inantly tropical families Haemulidae and Sci- aenidae. Clearly these behavior patterns are rooted deeply in their tropical ancestry, and are as characteristic of their kind as the more generally recognized morphological features that define their families. Ebeling and Bray recognized the strong influence that ancestral relationships exert on activity patterns, and distinguished "temper- ate derivatives" from "tropical derivatives." (Un- accountably, however, they considered Paralabrax clathratus and Coryphopterus nichoisi to be of temperate stock, even though the affinities of both are predominantly tropical.) These relationships, then, are insightful in understanding how activity patterns are integrated in southern Californian nearshore fish communities. The geographical affinities of the various species (Table 4) are help- ful in gaining an overview of these relationships. Because nearshore communities in warm- temperate southern California mix fishes of tem- perate and tropical affinities, it is tempting to interpret behaviors in terms of interactions be- tween these two lineages. Such comparisons are risky. For example, Ebeling and Bray (1976) stated: "It is paradoxical that the 'tropical deriva- tives' . . .persist in their complex . . . shelter-seeking while many primarily temperate fishes remain exposed." We see no paradox here. On tropical reefs, too, many diurnal fishes remain exposed at night, while others seek cover. Size often influ- ences which strategy is used. For example, while smaller acanthurids (surgeonfishes) and chaetodontids (butterflyfishes) generally are shel- tered, larger members of their families often rest exposed (Hobson 1972, 1974). Ebeling and Bray went on to suggest: ". . . the 'tropical derivatives' may . . .compete more successfully against primar- ily temperate species such as surfperches for shel- ter on the reef." This speculation, too, is unsup- ported by our observations. Most of the temperate species involved here are widespread northward (see Table 4, Group IV), well beyond the ranges of the tropical derivatives, and there too they are exposed at night (E. S. Hobson pers. obs.). We doubt that nocturnal shelter sites are in short supply on California reefs except under ex- ceptional circumstances. The places we identified as resting sites of Semicossyphus pulcher were just sporadically occupied, which seems an un- likely circumstance if there is strong competition for these sites. But clearly there is a shortage of shelter sites where the diurnal planktivore Chromis punctipinnis is so numerous that at night resting individuals overflow from the rocks and actually pile up on the adjacent sand. Apparently this exceptional situation exists where the zoo- plankters on which this fish feeds are abundant by day, but appropriate nocturnal shelter is limited. Significantly, however, the competition for this shelter appears to be intraspecific. A casual appraisal of southern Californian fishes agrees with Ebeling and Bray (1976) that activities among fishes of the kelp-bed community are "... more loosely 'programmed'" than among fishes in tropical reef communities. A similar con- dition has been described for temperate lake fishes 21 FISHERY BULLETIN: VOL. 79, NO. 1 (Helfman 1979). This position is strengthened by the more clearly defined behavior in the Califor- nian representatives of tropical families. But at least two considerations complicate this compari- son. First, activity patterns, no matter how highly structured, will be less evident in temperate fish communities because a greater proportion of the species there are sedentary. As stated for tropical fishes (Hobson 1972), relative activity in sedentary species is difficult to quantify. Second, and perhaps more important, because there are far fewer species in the temperate habitats, community ac- tivity patterns will be less distinct if only because they are defined by fewer forms. It need not follow that activities of each species are less structured. Despite these cautions, however, it is generally accepted that organisms tend to have less specialized habits where species are fewer, and this circumstance should produce more loosely structured activities. Whether or not activities of individual species are less structured in California than in the tropics, certainly the overall community patterns in California are less clearly defined. In examin- ing the changeovers between diurnal and noctur- nal modes, for example, we found little evidence of the detailed community transition-patterns that tj^ically characterize these phenomena on tropi- cal reefs (Hobson 1972; Collette and Talbot 1972; Domm and Domm 1973; McFarland et al. 1979). In particular, we were unable to clearly define a "quiet period," that 15-20 min segment of twilight on many tropical reefs when smaller fishes — both diurnal and nocturnal — have vacated the water column. Based on studies in the tropics (Hobson 1968a, 1972; Munz and McFarland 1973), the quiet periods are considered times of increased danger from predators when smaller fishes find it adap- tive to avoid exposed positions. So if the quiet periods are less evident in California, it could indi- cate reduced crepuscular predation there. When the sequence of twilight events identified at Santa Catalina Island is related to the timing of the quiet period at Kona, Hawaii (Figure 13), there might appear to be more overlap between the diurnal and nocturnal modes in California. For instance, as the nocturnal juveniles of Sebastes serranoides move into exposed locations (Figure 13, event 5), they sometimes pass above active lab- rids (Figure 13, events 2, 3, and 4) and close to the 'I KONA QUIET PERIGD^I Si H^ m m I CIVIL TWILIGHT ' KONA CAT. Jl± ^ 2JI NAUTICAL TWILIGHT KONA CAT. Sunset 10 20 30 40 TIME IN MINUTES 50 60 Figure 13. — Events during the evening changeover between day and night at one southern Califomian site from August to November 1973, relative to timing of the quiet period at Kona, Hawaii ( Hobson 1972) . ( The Califomian site is illustrated in Hobson and Chess 1976: fig. 2, 3.) Circles represent diurnal species, diamonds represent nocturnal species. Event 1 is based on estimates, events 2-7 on counts, with bars encompassing range, and numbers located at mean. 1. Migration of Chromis punctipinnis to nocturnal resting area: progressive disintegration of bar represents decreasing numbers of individuals in migrating groups (n = 6). 2. Time last Oxyjulis californica was seen active (n = 4). 3. Time last Halichoeres semicinctus was seen active (n = 6). 4. Time last Semicossyphus pulcher departed observation site for nocturnal resting places elsewhere (n =4). 5. Time first Sebastes serranoides appeared at open water feeding ground (« = 5). 6. Time first Xenistius californiensis arrived on feeding ground (n = 4). 7. Time first Seriphus politus arrived on feeding ground (n = 4). Times of civil and nautical twalight are from Nautical Almanac (U.S. Naval Observatory, Washington, D.C.) and are means of times on dates observations were made. 22 HOBSON ET AL.: CREPUSCULAR AND NOCTURNAL ACTIVITIES OF CALIFORNIA FISHES last migrating Chromis punctipinnis (Figure 13, event 1). But this apparent overlap can be ex- plained without evoking a more relaxed regime. Although the three species of Californian labrids retire relatively late (even considering the longer twilight at temperate latitudes), they grow larger than most of their tropical relatives, and it is gen- erally true that among diurnal fishes larger indi- viduals retire later (Hobson 1972). Similarly, there are tropical equivalents to the relatively early shift made by juvenile S. serranoides to its noctur- nal mode. In the western tropical Pacific Ocean, for example, the largely transparent nocturnal juveniles of some apogonids move away from shel- ter long before their larger adults. Some of them — many < 30 mm long — move out as early as 10 min after sunset (E. S. Hobson unpubl. obs.). This entry into exposed locations at a time when many piscivorous predators hunt most effectively might seem in conflict with the quiet-period con- cept. But in the dim twilight we are not surprised that these inconspicuous little fishes seem to go unseen by the visual hunters that so seriously threaten the more visible adults. Certainly these juveniles go unseen by human eyes at this time, except upon close inspection with a diving light, and so fail to detract from the aura of inactivity that characterizes the quiet period. So what might appear to be a more loosely struc- tured sequence of events during twilight in California may instead reflect the lesser number of species that define the transition pattern there. At least some semblance of a quiet period is evi- dent. On the occasions depicted in Figure 13, the numbers of migrating C. punctipinnis declined sharply about 10 min before the juvenile S. ser- ranoides first appeared, and after that only scat- tered small groups passed that way. Considering the tremendous numbers of C punctipinnis in that area (we have never seen a single species so dom- inant on coral reefs), numerous exceptions from the norm should be expected. Furthermore, major species that occupy the water column at night — Xenistius californiensis and Seriphus politus — did not arrive until about 20 min after the vast majority of C punctipinnis had passed through. So although crisp definition is lacking, there is evi- dence of a quiet period in Californian waters from about 15 to 35 min after sunset. It remains uncertain whether the dangers small- er fishes face during twilight in southern Califor- nian coastal waters are as intense as those faced on tropical reefs. Limited data from gut contents indicate that such major predators as Paralabrax clathratus and Sebastes serriceps are primarily crepuscular when capturing smaller fishes. Unfor- tunately, we can no longer directly observe much of the predation that has influenced the evolution of coastal fishes in southern California. This is be- cause during recent decades populations of the larger predators involved — including the giant seabass, Stereolepis gigas, and the white seabass, Cynoscion nobilis — have been decimated by fishermen. Nevertheless, as we discuss next, the possibility that Californian fishes have faced in- tensified selection pressures during twilight is also indicated by the nature of their scotopic visual pigments. Significantly, despite the varied forms and habits of these fishes, the maximum absorp- tions (Amax) of their scotopic pigments cluster about 500 nm, which indicates strong selection for enhanced photosensitivity over this segment of the spectrum. Scotopic Spectral Sensitivity and Ambient Light Tropical reef fishes have scotopic pigments that cluster about wavelengths that spectrally match twilight, which underwater is bluer than the light of day or night (Munz and McFarland 1973). In developing their Twilight Hypothesis, Munz and McFarland pointed out that because dawn and dusk are the most dangerous times for fishes on tropical reefs (Hobson 1972), even slightly in- creased photosensitivity during twilight may be crucial. The Twilight Hypothesis is a variation of the Sensitivity Hypothesis (Lythgoe 1966), which de- clares that visual sensitivity is improved when absorption of photo pigments matches ambient light. The scotopic pigments of both aquatic and terrestrial vertebrates are known to cluster about narrow wave bands, but at different regions of the spectrum (for review, see McFarland and Munz 1975b). Among fishes, the scotopic pigments broadly match the spectral transmission of the water in which the fishes live (Lythgoe 1972). Deep sea fishes, for example, have scotopic pigments that tend to be even more blue sensitive ( Amax from 478 to 490 nm: Denton and Warren 1956; Munz 1957), than tropical marine fishes (X^ax from 489 to 500 nm: Munz and McFarland 1973), and the pig- ments of freshwater fishes are green sensitive (\„„. from 503 to 540 nm: McFarland and Munz 23 FISHERY BULLETIN: VOL. 79, NO. 1 1975b). The present study shows that fishes as- sociated with Californian kelp forests have pig- ments that are most sensitive to blue-green light (\max from 496 to 506 nm). If the adaptive advantage of matching ambient light lies in heightened photosensitivity, then the match should be to light that prevails when selec- tion for improved vision is most intense. We are not surprised, therefore, that the k^^ values for fishes in a given habitat match not the light of day or night, but rather the bluer twilight, and thus the fishes are equipped to meet an intensified threat from crepuscular predators. In clear tropi- cal waters, for example, the X^ax values of the scotopic pigments in reef fishes cluster about 492 nm, which in that habitat matches twilight, rather than the greener light of day or night (Munz and McFarland 1977). Of course, the spec- tral position of this match is influenced by the light transmission characteristics of water in that par- ticular habitat. For example, in most fresh waters the match is made above 520 nm, but this position nevertheless approximates the XPsq of twilight in these very green waters and is in fact toward the blue from light that prevails there during day and night (McFarland and Munz 1975c). As could have been predicted from the Sensitiv- ity Hypothesis, the scotopic pigments of fishes in the blue-green coastal water of California cluster at wavelengths intermediate between those of fishes on coral reefs and those of fishes in freshwa- ter, at about 500 nm (Tables 1-3). The match with twilight, however, is less clear in Californian waters than in the other two environments be- cause, we believe, photic conditions in Californian waters are more variable. Nevertheless, the tight clustering of scotopic pigments around 500 nm in the Californian fishes better matches ambient light during twilight than at night. Both moon- light and starlight are richer in red light than daylight or twilight (Munz and McFarland 1977), and for all water conditions we encountered at Santa Catalina Island downwelling light at night would have XP^^ values well above 520 nm (Figure 5). Only during twilight does ambient light un- derwater shift far enough toward the blue-green region of the spectrum (Figures 3, 4) to produce a close match with the visual pigment X^ax- In evaluating the impact of crepuscular pred- ators on the spectral position of scotopic pigments, however, we must not forget that other selection pressures are operating. We would expect scotopic pigments in fishes to be particularly responsive to such alternate pressures at night, which is espe- cially "short of light" (Dartnell 1975). Certainly fishes must be sensitive to the emissions of biolum- inescent organisms, because few visual cues could be more apparent than a flash of light in the dark. Scotopic Spectral Sensitivity and Bioluminescence Bioluminescence often signifies underwater movement. According to Hobson (1966), "In many areas of the sea at night, a moving object is readily observed due to the luminescence of many minute planktonic organisms, mostly protozoans, which light up when disturbed. These organisms are often so numerous that while making observa- tions underwater I have been able to identify, to species, fishes that swam actively among them. This was not because the fish itself was illumi- nated, but rather because there were so many minute luminescent organisms about the fish that its form was essentially traced out in tiny flecks of light." If such cues are evident to human eyes, adapted to a diurnal, terrestrial existence, the ca- pacity to sense and to orient by them must be highly refined in animals like fishes that have evolved in this environment. Bioluminescence offers an especially effective way to detect predators or prey because predator- prey interactions generally involve movement, and luminescence by plankton is greatly increased in the turbulent water around moving objects. We believe, as did Burkenroad (1943), that this fact has had enormous impact on the nocturnal tactics of both predators and prey. The motionless at- titude that characterizes nocturnal planktivorous fishes when they hunt probably is enforced by the need to minimize turbulence in the water about them. By minimizing turbulence they minimize the firing of luminescent organisms that would betray their presence, and so hover unseen in the dark, ready to strike when nearby prey advertise their positions by disturbing the plankton. This tactic is in essence an ambush and probably is effective only at short range. Reasons for this limi- tation are two. First, an attack, once launched, is immediately identified by flashing plankton, thus giving prey more than a short distance away time for evasive maneuvers. Second, a long-range at- tack directed at plankton luminescing around a particular prey may be led to the prey's wake, because the targets that elicited the attack are left behind when the prey darts away. Probably pred- 24 HOBSON ET AL.t CREPUSCULAR AND NOCTURNAL ACTIVITIES OF CALIFORNIA FISHES ators can be led by luminescent plankton to prey over some distance (Nichol 1962), but we suggest the approach must be made with great stealth to avoid turbulence and resulting luminescence. Surely such limitations preclude many kinds of predatory activities after dark, especially when the prey are as agile as most small fishes. In fact, the diminished threat from predators that smaller reef fishes enjoy at night (Hobson 1973, 1975, 1979) may stem largely from the difficulties predators have at this time moving undetected to within striking range. If fishes use luminescent plankton to detect predators and prey, undoubtedly they have experi- enced strong selection pressures to enhance this detection. An obvious adaptive response to such pressures would be a match of the scotopic visual pigments to the emission spectra of the lumines- cent plankton. Certain vertebrates living at great depths, or in the open ocean, reportedly have scotopic pigments that match the luminescent emissions of organisms with which they interact, socially or as predator or prey (Clarke 1936; Munz 1958a; Lythgoe and Dartnall 1970; McFarland 1971; Locket 1977). These animals, however, inter- act directly with the luminescent organisms, whereas we stress indirect interactions. Of course, the principle is the same either way — detection is enhanced by matching visual pigment absorption to a luminescent emission. The emission spectra of most luminescent plankton, as exemplified by Noctiluca miliaris and Gonyaulax polyedra (Figure 5), indicate that fishes would sense the emitted photons best with blue-sensitive visual pigments that have Xmax val- ues near 490 nm. But this holds only before the light has passed through water. As noted above, the spectrum of light changes as it travels through water, with the degree of change sharply affected by the water's clarity. Clearly any such change will favor a different k^^ value in visual pig- ments. The relative effectiveness of visual pigments with differing X^ax values in waters of differing clarities can be estimated with some simple calcu- lations. Given the relative attenuation of light at each wavelength, which is a fiinction of water type, we can compare the relative photoabsorption of different visual pigments at increasing dis- tances from the luminescent source. Let us con- sider, for example, how a series of pigments with ^max values at 10 nm intervals between 450 and 550 nm, each at 0.4 absorbance units, would ab- sorb the light emitted by Noctiluca as this light passes through differing clarities of seawater, as defined by Jerlov (1968). The relative photoabsorp- tion of each pigment at a given distance from the source can be estimated by multiplying that pig- ment's percentage absorption at each wavelength by the relative amount of light available at that wavelength and then integrating the products over all wavelengths (Figure 14). Thus, to see c CO z o I- o I Q. 00 - jf -H r^ ^ / ^.— RANGE • ,-o-l f lA— -"^ ^ -0 / f ^^^ -r- / / ^0"""^ 1 "**-N».^ -S' ^ \ ^"^v.^ - 50 'y \ L "" : — ^^ ""^^ \ 5 ^ '^^ __«__ _\ * •^ r^. *~~^ o 1 1 J L 1 _L. 1 1 1 1 450 500 WAVELENGTH >v 550 max (nmJ FIGURE 14. — Expected relative effectiveness of visual pigments with differing A^iax values (positioned at wavelengths indicated by circles) in sensing luminescent emissions ofNoctiluca miliaris under varying circumstances. Top curve, for reference, repre- sents effectiveness at zero range (no alteration of emission spec- trum by intervening water); dashed curve (with open circles), for comparison, represents effectiveness at distance of 3 m in clear tropical water (water type lA of Jerlov 1968); three lower curves represent relative effectiveness in waters of decreasing clarity (water types 1, 5, and 7 of Jerlov 1968). Vertical line crosses each curve at the optimal >^max position. Method of calculation in text. luminations from an organism like A/", miliaris at a distance of 3 m in coastal waters (as at Santa Catalina), fishes would best have visual pigments with Amax values between 500 and 510 nm. And as range increases, and water clarity decreases, photodetection of this luminant source would be improved by shifting the \^^ position slightly, but continuously, toward the greener wavelengths (Figure 15). In reality, of course, the reduced visibility in turbid water sharply limits the practical extent of such a shift. During heavy ph5rtoplankton blooms, for example, even large objects in full daylight are 25 FISHERY BULLETIN: VOL. 79, NO. 1 520 - 510 \ max 500 490 - 2 3 4 5 RANGE IN METERS Figure 15. — Calculated optimum K^^^ position for absorption of luminescent emissions of Noctiluca miliaris at increasing range in waters of differing clarity. Water types I, 1, 5, and 7 of Jerlov (1968). Solid lines = coastal waters of increasing turbid- ity; dashed line = clear tropical water. invisible beyond 2-3 m. Obviously under such con- ditions bioluminescence, too, would be invisible at these distances. Probably most meaningful in- teractions involving fishes and the bioluminescent emissions of plankton occur over distances of just a fev^^ meters or less, and to see bioluminescence at these ranges in the waters of varying visibility that surround Santa Catalina fishes probably would best have visual pigments with k^^^ values at a little above 500 nm (Figure 15). One might expect diurnal fishes to be less sensi- tive than nocturnal or crepuscular fishes to selec- tion pressures on scotopic pigments because they are less active under low light. But among species studied at Santa Catalina the X^^ values scarcely differ between the two groups (diurnal feeders: x = 499.1 nm, range = 496.1-505.8— Table 1; noctur- nal feeders: x = 501 nm, range = 496.1-505.1 — Table 2). Perhaps there is little difference here because many of these diurnal feeders are exposed at night and need means to detect threatening predators. In fact, if the fishes are grouped accord- ing to their relative exposure after dark, rather than by how active they are at this time, their X^ax values show a clear pattern (Table 5). Those fishes that are exposed at night, compared with those that are sheltered, tend to have scotopic pigments with spectral sensitivities closer to what we con- sider optimum for detecting bioluminescence in Californian coastal waters. This would mean that, despite their apparent quiescence, at least many diurnal fishes remain visually alert for potential threats during the night. So the similar X^ax values in the diurnally feed- ing A^/iermops afflnis (505.8 nm) and the noctur- nally feeding Hyperprosopon argenteum (505.1 nm) may answer the similar threats both face while exposed in the water column at night. We assume the increased sensitivity to biolumines- cence would also benefit H. argenteum in feeding, but suspect this advantage would be less forceful. The need to evade predators should be sharper than the need to capture prey. Both needs are criti- cal, but there is less tolerance for error on defense than on offense. A fish as prey is likely to be elimi- nated the first time it errs in responding defen- sively to an attack, but the same fish as an at- tacker may err many times without serious con- sequences. Bioluminescence, triggered by movement, rep- resents a well-defined indicator of immediate danger that can effectively focus selection pres- sures on a narrowly defined adaptive response. Moonlight and starlight would seem less suited for this because neither so effectively identifies specific threats and because the impact of both Table 5. — Relative exposure at night, andAj^j,^ position of scotopic visual pigments in certain southern Californian marine fishes.' Relative exposure at night Species '^max "^ean ''max''a"9e I. Fully exposed in water column Fully exposed on or near seafloor, often on open sand Partially or fully sfieltered by rocks or algae Atherinops affinis Sebastes serranoides Xenistius californiensis Seriphus politus Umbrina roncador Damalichthys vacca Embiotoca jacksoni Scorpaena guttata Sebastes atrovirens^ S. serriceps Paralabrax clathratus Girella nigricans Ctiromis punctipinnis Brachyistius frenatus 502.1 Cymatogaster aggregata Hyperprosopon argenteum Leiocottus tiirundo 501.1 Pleuronichtttys coenosus Hypsypops rubicunda 497.3 Semicossyphus pulctier Alloclinus holder! Heterostichus rostratus Coryphopterus nichoisi 500.4-505.8 500.0-503.1 496.1-499.7 'Of species studied, Medialuna californiensis is excluded because its nocturnal fiabits remain uncertain (see text), and Halichoeres semiclnctus and Oxyjulis californica are excluded because tfiey have a different type of visual pigment (see Table 1, footnote 3). ^Even though S. atrovirens feeds in the water column at night, these activities occur close to rising kelp stipes or just beneath the surface canopy 26 HOBSON ET AL.; CREPUSCULAR AND NOCTURNAL ACTIVITIES OF CALIFORNIA FISHES tends to be diffused over a greater range of visual circumstances. Bioluminescence is more constant: the spectral compositions of moonlight and star- light change with water depth and atmospheric conditions, but the spectral composition of bioluminescence is independent of these vari- ables. And, of course, there is less light from moon or stars with increased water depth, which, again, is untrue of a bioluminescent emission. So it is not surprising that the narrow range of Xrnax positions in visual pigments of Californian fishes more closely matches bioluminescence than it does moonlight or starlight (Figure 16). Undoubtedly moonlight and starlight have strong influences on nocturnal relationships MOONLIGHT BEST x„„ 400 500 600 WAVELENGTH (nm) 70 Figure 16. — Relationships between the spectral distributions of moonlight, starlight, and bioluminescence (as produced by Noc- tiluca miliaris) in seawater typical of southern California and the spectral sensitivities of fishes that live there. The five curves for each type of light represent the spectral distribution expected to reach a fish at the distance (in meters) indicated by the number that accompemies each curve. The curves depict moon- light and starlight off a flat reflector at a depth of 3 m (using values from Munz and McFarland 1977), and the luminescence of N. miliaris (using values from Nicol 1958), in water equivalent to typical conditions at Santa Catalina Island (Coastal Type 1 of Jerlov 1968). The solid circle on each spectral curve identifies the wavelength to best match with a visual pigment for maximum photosensitivity, and the stippled column represents the spectral range of maximum photosensitivity in scotopic pigments of Californian fishes tXmax^- Note that these coincide only with bioluminescence. among predators and prey. But probably both help fishes more on offense than on defense. Predators can position themselves, and time their attacks, to play both types of light to their advantage, and to the prey's disadvantage. By charging at prey from below, for example, predators view their targets against the water's relatively light surface, while their own movements are masked by the sur- rounding gloom {Hobson 1966). Certainly attacks that so often spring from the shadows would greatly dilute a defensive advantage prey might gain with spectral sensitivities that match moon- light or starlight. Under these circumstances prey face a broad range of threats that calls for a more generalized response. So we can understand why many smaller reef fishes that habitually range into the water column at night stay closer to shel- ter under moonlight (Hobson 1968a). Despite the offensive advantage that certain predators likely gain from moonlight (or star- light), their scotopic visual pigments tend to be better matched to bioluminescence (or twilight), probably because this answers a more pressing need on defense. So it would seem that even those species that have special tactics to use moonlight or starlight to better see their prey must com- promise with visual pigments less than optimal for this task. Included are those nocturnal plank- tivores, like subadult Sebastes serranoides, that characteristically hover tail-down in the water column, where apparently prey are visible to them against moonlight or starlight from above. In- cluded, too, are those predominantly diurnal fishes, like Paralabrax clathratus and Cymatogas- ter aggregata , that apparently are able to hunt at night close to sand where light levels are elevated by reflected moonlight and starlight. We suggest not that their visual pigments are unsuited to see prey by moonlight or starlight, but rather that these pigments simply could have better spectral sensitivities for this particular job (Figure 16). These arguments, favoring bioluminescence over moonlight and starlight as a selective force in determining Xj^ax position, would also favor bioluminescence over twilight. But there is an im- portant difference in this last comparison. Moon- light and starlight would select for spectral posi- tions different from that selected for by bioluminescence, and so a conflict would exist. Twilight, on the other hand, would select for essen- tially the same spectral position as biolumines- cence, so that the two would act in concert (see below). 27 FISHERY BULLETIN: VOL. 79, NO. 1 If scotopic visual pigments with spectral posi- tions slightly above 500 nm are optimally located to detect bioluminescence in Californian coastal waters, one can see why such pigments occur there in fishes exposed to nocturnal predators. But ques- tions remain concerning why fishes apparently less threatened are consistent in having scotopic pigments positioned slightly under 500 nm. Why are these pigments not loosely positioned above, as well as below, the optimum location? The answer to this question might lie in ancestral relation- ships. In grouping the Californian fishes according to whether their geographical affinities are tropi- cal or temperate (Table 4), our concern was with current relationships. In fact, all these fishes be- long to groups that stem from tropical origins (Berg 1940). The radiation of acanthopterygian^ fishes from a relatively few ancestral forms early in the Cenozoic (Patterson 1964; Romer 1966) has been related (by Hobson 1974) to the concurrent development of modern coral reef communities (Newell 1971). And we have seen that conditions under which coral reefs flourish, compared with conditions in temperate Californian waters, favor in fishes' scotopic visual pigments that are more sensitive to slightly shorter wavelengths. The mean k^^^ of 492 nm that Munz and McFarland (1973) found in the scotopic pigments of coral-reef fishes, compared with the values around 500 nm that characterize Californian fishes, is consistent with the fact that water around coral reefs is gen- erally clearer and more transparent to blue light than water around Santa Catalina (Figure 2). The extent that k^^ values of scotopic pigments in Californian coastal fishes have shifted toward the green from what may have been ancestral posi- tions near 490 nm, and, especially, have become located slightly above 500 nm (the optimal posi- tion in California), may roughly measure the rela- tive strength of nocturnal or crepuscular preda- tion pressures on each species in these greener waters. The argument that scotopic pigments may be positioned to detect bioluminescence can be ex- tended to coral reef fishes. Our calculations show that visual pigments with different Amax values would trap percentages of the light from Noctiluca miliaris as follows: P450 nm = 62%, P475 nm = ^All fishes considered in this paper have been included among the acanthopterygians, or spiny-finned teleosts, although one, Atherinops a f finis, would be relegated by some systematists (e.g., Greenwood et al. 1966) to another group. 94%, P490 nm = 100%, P500 nm = 99%, P525 nm = 93.5%, P550 nm = 79%. Although the peak is broad, the central wavelength for maximum ab- sorption at zero range would be near 495 nm (Fig- ure 14), which is close to the k^^^ of 492 nm in the scotopic pigments of coral-reef fishes. And as dis- tance from the source increases the match be- comes even better (Figure 15). Twilight or Bioluminescence? Whether the clustering and spectral position of visual pigments in warm-temperate and tropical reef fishes is the result of natural selection, or is simply fortuitous, is a complex question. If, as we believe, these features have been refined by in- I tense selection pressures from predators, it re- | mains problematical whether the advantage lies in detecting bioluminescence, or in enhancing photoabsorption during twilight — as suggested by the Twilight Hypothesis. Indeed, both may be important. The scotopic pigments have spectral positions that would be effective in both functions, and the benefits of one would complement the other. Unquestionably, there has been ample time to influence evolution. Bioluminescent plankton have existed since before the first fishes (Seliger 1975), and so has twilight's unique spectrum. Be- cause fishes have experienced these features throughout their history, the slightest favorable adjustment could have been adaptive. Although a 5-10 nm shift in k^^^ position would improve photoabsorption by no more than a few percentage units, even this could have been meaningful. And if selection pressures to detect bioluminescence have, in fact, acted in concert with those to en- hance crepuscular vision, which to us seems likely, then their combined impact certainly would have been a powerful evolutionary force. ACKNOWLEDGMENTS We thank Russell Zimmer and his staff at the Catalina Marine Science Center, University of California, especially Robert Given and Larry i Loper, for making facilities available and assist- ' ing us in many ways. For constructive criticism of the manuscript we thank Alfred Ebeling, Univer- sity of California, Santa Barbara; Gene Helfman, University of Georgia; and Richard Rosenblatt, Scripps Institution of Oceanography. Dolores Fussy, Southwest Fisheries Center Tiburon Laboratory, National Marine Fisheries Service 28 HOBSON ET AL.: CREPUSCULAR AND NOCTURNAL ACTIVITIES OF CALIFORNIA FISHES (NMFS), NOAA, typed the manuscript, and Susan Smith, Southwest Fisheries Center Tiburon Laboratory, NMFS, NOAA, drew Figure 1. Por- tions of this study were supported by N.I.H. Re- search Grant EY-00323 from the National Eye In- stitute. LITERATURE CITED BERG, L. S. 1940. Classification of fishes both recent and fossil. Trav. Inst. Zool. Acad. Sci. U.S.S.R. 5:87-517. BRAY, R. N., AND A. W EBELING. 1975. Food, activity, and habitat of three "picker-type" mi- crocamivorous fishes in the kelp forests off Santa Bar- bara, California. Fish. Bull., U.S. 73:815-829. BRIGGS.J. C. 1974. Marine zoogeography. McGraw-Hill, N.Y, 475 p. BURKENROAD, M. D. 1943. A possible function of bioluminescence. J. Mar. Res. 5:161-164. CLARKE, G. L. 1936. On the depth at which fish can see. Ecology 17:452-456. Clarke, T. A. 1970. Territorial behavior and population dynamics of a pomacentrid fish, the garibaldi, Hypsypops rubicun- da. Ecol. Monogr. 40:189-212. COLLETTE, B. B., AND F H. TALBOT. 1972. Activity patterns of coral reef fishes with emphasis on nocturnal-diurnal changeover. In B. B. Collette andS. A. Earle (editors), Results of theTektite Program: ecology of coral reef fishes, p. 98-124. Nat. Hist. Mus. Los Ang. Cty, Sci. Bull. 14. DARTNALL, H. J. A. 1952. Visual pigment 467, a photosensitive pigment pres- ent in tench retinae. J. Physiol. (Lond.) 116:257-289. 1975. Assessing the fitness of visual pigments for their photic environments. In M. A. AH (editor). Vision in fishes, p. 543-563. Plenum Press, N.Y Denton, e. J., and E J. warren. 1956. Visual pigments of deep-sea fish. Nature (Lond.) 178:1059. DOMM, S. B., AND A. J. DOMM. 1973. The sequence of appearance at davm and disap- pearance at dusk of some coral reef fishes. Pac. Sci. 27: 128-135. EBELING, A. W, AND R. N. BRAY. 1976. Day versus night activity of reef fishes in a kelp forest off Santa Barbara, California. Fish. Bull., U.S. 74:703-717. Emery, a. r. 1973. Preliminary comparisons of day and night habits of freshwater fish in Ontario lakes. J. Fish. Res. Board Can. 30:761-774. Feder, h. m., C. H. Turner, and C. Limbaugh. 1974. Observations on the fishes associated wdth kelp beds in southern California. Calif. Dep. Fish Game, Fish Bull. 160, 144 p. Fitch, j. e., and r. j. Lavenberg. 1975. Tidepool and nearshore fishes of California. Univ. Calif Press, Berkeley, 156 p. Greenwood, p h., d. e. Rosen, S. H. weitzman, and G. S. Myers. 1966. Phyletic studies of teleostean fishes, with a provi- sional classification of living forms. Bull. Am. Mus. Nat. Hist. 131:341-455. Harmelin- Vivien, M. L., and C. Bouchon. 1976. Feeding behavior of some carnivorous fishes (Ser- ranidae and Scorpaenidae) from Tulear (Madagas- car). Mar. Biol. (Berl.) 37:329-340. Hastings, j. W, and B. M. Sweeney. 1957. The luminescent reaction in extracts of the marine dinoflagellate, Gonyaulax polyedra. J. Cell. Comp. Physiol. 49:209-225. HELFMAN, G. S. 1979. Temporal relationships in a freshwater fish com- munity Ph.D. Thesis, Cornell Univ, Ithaca, N.Y, 258 p. Herald, E.S. 1961. Living fishes of the world. Doubleday, Garden City, N.Y, 304 p. HOBSON, E. S. 1965. Diumal-noctumal activity of some inshore fishes in the Gulf of California. Copeia 1965:291-302. 1966. Visual orientation and feeding in seals and sea lions. Nature (Lond.) 210:326-327. 1968a. Predatory behavior of some inshore fishes in the Gulf of California. U.S. Fish Wildl. Serv., Res. Rep. 73, 92 p. 1968b. Coloration and activity of fishes, day and night. Underwater Nat. 5:6-11. 1972. Activity of Hawaiian reef fishes during the evening and morning transitions between daylight and darkness. Fish. Bull., U.S. 70:715-740. 1973. Diel feeding migrations in tropical reef fishes. Hel- gol. wiss. Meeresunters. 24:361-370. 1974. Feeding relationships of teleostean fishes on coral reefs in Kona, Hawaii. Fish. Bull., U.S. 72:915-1031. 1975. Feeding patterns among tropical reef fishes. Am. Sci. 63:382-392. 1979. Interactions between piscivorous fishes and their prey. In H. Clepper (editor), Predator-prey systems in fisheries management, p. 231-242. Sport Fish. Inst., Wash.,D.C. HOBSON, E. S., AND J. R. CHESS. 1973. Feeding oriented movements of the atherinid fish Pranesus pinguis at Majuro Atoll, Marshall Islands. Fish. Bull., U.S. 71:777-786. 1976. Trophic interactions among fishes and zooplankters near shore at Santa Catalina Island, California. Fish. Bull., U.S. 74:567-598. HUBBS, C. L. 1960. The marine vertebrates of the outer coast. In Sym- posium: The biogeography of Baja California cuid adjacent seas, p. 134-147. Syst. Zool. JERLOV, N. G. 1968. Optical oceanography. Elsevier, N.Y, 194 p. LIMBAUGH, C. 1955. Fish life in the kelp beds and the effects of kelp harvesting. Univ. Calif. Inst. Mar. Res. Ref 55-9, 158 p. LOCKET, N. A. 1977. Adaptations to the deep-sea environment. In F. Crescitelli (editor), The visual system in vertebrates, p. 67-192. Handb. Sens. Physiol. 7(5). lythgoe, j. N. 1966. Visual pigments and underwater vision. In R. 29 FISHERY BULLETIN: VOL. 79. NO. 1 Bainbridge, G. C. Evans, and O. Rackham (editors), Light as an ecological factor, p. 375-391. Blackwell Sci. Publ., Oxf. 1972. The adaptation of visual pigments to the photic envi- ronment. In H.J. A. Dartnall (editor). Photochemistry of vision, p. 566-603. Handb. Sens. Physiol. 7(1). Lythgoe, J. N., AND H. J. A. Dartnall. 1970. A "Deep Sea Rhodopsin" in a mammal. Nature (Lond.) 227:955-956. Major, P f. 1977. Predator-prey interactions in schooling fishes during periods of twilight: A study of the silverside Pranesus insularum in Hawaii. Fish. Bull., U.S. 75:415-426. MCFARLAND, W. N. 1971. Cetacean visual pigments. Vision Res. 11:1065- 1076. mcFarland, w. n., and F W. MUNZ. 1975a. The photic environment of clear tropical seas dur- ing the day Vision Res. 15:1063-1070. 1975b. The evolution of photopic visual pigments in fishes. Vision Res. 15:1071-1080. 1975c. The visible spectrum during twilight and its impli- cations to vision. In G. C. Evans, R. Bainbridge, and O. Rackham (editors). Light as an ecological factor H, p. 249-270. Blackwell Sci. Publ., Oxf McFarland, W. N., J. C. Ogden, and J. N. Lythgoe. 1979. The influence of light on the twilight migrations of grunts. Environ. Biol. Fish. 4:9-22. MILLER, D. J., AND R. N. LEA. 1972. Guide to the coastal marine fishes of California. Calif Dep. Fish Game, Fish Bull. 157, 235 p. MUNZ, F W. 1957. The photosensitive retinal pigments of marine and euryhaline teleost fishes. Ph.D. Thesis, Univ. California, Los Ang., 172 p. 1958a. Photosensitive pigments from the retinae of certain deep-sea fishes. J. Physiol. (Lond.) 140:220-235. 1958b. The photosensitive retinal pigments of fishes from relatively turbid coastal waters. J. Gen. Physiol. 42:445-459. 1958c. Retinal pigments of a labrid fish. Nature (Lond.) 181:1012-1013. 1964. The visual pigments of epipelagic and rocky-shore fishes. Vision Res. 4:441-454. MUNZ, F. W, AND D. M. ALLEN. 1968. Computer-assisted analysis of visual pigments. Vi- sion Res. 8:983-996. MUNZ, F. W, AND W. N. MCFARLAND. 1973. The significance of spectral position in the rhodop- sins of tropical marine fishes. Vision Res. 13:1829-1874. 1975. Presumptive cone pigments extracted from tropical marine fishes. Vision Res. 15:1045-1062. 1977. Evolutionary adaptation of fishes to the photic envi- ronment. In F. Crescitelli (editor). The visual system in vertebrates, p. 193-274. Handb. Sens. Physiol. 7(5). NEWELL, N. D. 1971. An outline history of tropical organic reefs. Am. Mus. Novit. 2465, 37 p. NICOL, J. A. C. 1958. Observations on luminescence in Afocri/uca. J. Mar. Biol. Assoc. U.K. 37:535-549. 1962. Animal luminescence. Adv. Comp. Physiol. Biochem. 1:217-273. PATTERSON, C. 1964. A review of Mesozoic acanthopterygian fishes, with special reference to those of the English Chalk. Philos. Trans. R. Soc. Lond., Ser. B, Biol. Sci. 247:213-482. QUAST, J. C. 1968. Observations on the food of the kelp-bed fishes. In W. J. North and C. L. Hubbs (editors), Utilization of kelp- bed resources in southern California, p. 109-142. Calif. Dep, Fish Game, Fish Bull. 139. Randall, J. E. 1967. Food habits of reef fishes of the West Indies. Stud. Trop. Oceanogr. (Miami) 5:665-847. romer, a. S. 1966. Vertebrate paleontology., 3d ed. Univ. Chicago Press, Chicago, 468 p. SELIGER, H. H. 1975. The origin of bioluminescence. Photochem. Photo- biol. 21:355-361. SMITH, C. L., and J. C. Tyler. 1972. Space resource sharing in a coral reef fish communi- ty. In B. B. Collette and S. A. Earle (editors). Results of the Tektite Program: ecology of coral reef fishes, p. 125- 170. Nat. Hist. Mus. Los Ang. Cty., Sci. Bull. STARCK, W a., II, AND W. P DAVIS. 1966. Night habits of fishes of Alligator Reef, Florida. Icthyol. Aquarium J. 38:313-356. STARCK, W. A., AND R,. E. SCHROEDER. 1965. A coral reef at night. Sea Front. 11:66-79. tett, p b., and m. g. kelley. 1973. Marine bioluminescence. Oceanogr. Mar. Biol. Annu. Rev 11:89-173. Vivien, M. L. 1973. Contribution a I'etude de I'ethologie alimentaire de I'ichthyofaune du platier interne dos recifs corralliens de Tulear (Madagascar). Tethys, Suppl. 5:221-308. WINN, H. E., AND J. E. BARDACH. 1960. Some aspects of the comparative biology of parrot fishes at Bermuda. Zoologica (N.Y) 45:29-34. 30 RESPIRATION RATES AND LOW-OXYGEN TOLERANCE LIMITS IN SKIPJACK TUNA, KATSUWONUS PELAMIS Reginald M. Gooding/ William H. Neill,'^ and Andrew E. Dizon^ ABSTRACT Oxygen-uptake rates and swimming speeds of voluntarily active skipjack tuna, Katsuwonus pelamis, at 23°-24° C were measured in the laboratory from captivity-habituated fish (0.6-3.8 kg) and at sea from just-caught fish (L8-2.2 kg). In the shipboard tests, skipjack tuna swam 2-5 lengths/s (length = fork length) and consumed 0.9-2.5 (median = 1.3) mg Oa/g per h during their first 2.2 h of captivity. In laboratory tests, skipjack tuna swam at a mean speed of L4 lengths/s and consumed oxygen at a mean rate of 0.52 mg Oa/g per h. For the laboratory fish, routine swimming speed (S, in lengths/ second) was inversely related to fish weight ( W, in grams) — S = 3.12 - 0.53 logio W; oxygen-uptake rate ( Vq^ , in milligrams Oz/gram per hour) was directly related to both weight and speed (i.e., speed independent of weight effects) — logioVOj = -1-20 + 0.19/logioW + 0.21 S. However, laboratory fish swimming at their characteristic (weight dependent) speeds respired at rates independent of weight. Calculations based on the above interrelations among metabolic rate, swimming speed, and body weight indicated that skipjack tuna of all sizes may have an optimum swimming speed (for maximum distance per unit energy expenditure) near 2.1 lengths/s. Captivity-habituated skipjack tuna (0.8-3.4 kg) also were subjected to a step decrease in concen- tration of dissolved oxygen (O2) at 23°-24° C to determine their responses to acute hypoxia. At levels of O2 below 4 mg/1, voluntary swimming speed increased as O2 declined, reaching 3.9 lengths/s at the lowest test value of O2 , 1.4 mg/1. The 4-h median tolerance limit for low O2 proved similar to the O2 level critical for change in swimming speed, about 4 mg/1. Experimental results are analyzed and compared with those from other fishes to arrive at the following conclusions: 1) The skipjack tuna's "standard" metabolic rate is two to five times that of typical fishes of similar size; 2) the weight exponent for "standard" metabolic rate of skipjack tuna is a positive value near 0.2, as opposed to the -0.2 value tjrpical of fishes; 3) but, because the characteristic swdmming speed of routinely active skipjack tuna is inversely related to weight, routine metabolic rate is virtually independent of fish weight; 4) highly active skipjack tuna can consume oxygen from air-saturated sea water at rates exceeding those known from any other fish of similar size; and 5) the skipjack tuna is relatively inefficient in its use of oxygen and food-energy for swdmming (at least at low speeds) and it dies at O2 levels still well above those lethal for other fishes. Until the mid-1960's the environmental require- ments of commercially important tunas (Scom- bridae) were known mainly from correlations between fishery catch rates and oceanographic conditions (see discussions by Robins 1952; Laev- astu and Rosa 1963; Broadhead and Barrett 1964; Blackburn 1965; Williams 1970; Blackburn and Williams 1975; Matsumoto 1975). With the ad- vent of techniques for studying tunas in captivity (Magnuson 1965; Nakamura 1972), many unre- solved issues of tuna biology could be explored such as feeding and gut-evacuation rates (Mag- nuson 1969), auditory perception (Iversen 1967), visual perception (Nakamura 1968; Tamura et al. 'Southwest Fisheries Center Honolulu Laboratory, National Marine Fisheries Service, NOAA, 2570 Dole St., Honolulu, HI 96812. ^Department of Wildlife and Fisheries Sciences, Texas A&M University, College Station, TX 77843. Manuscript accepted September 1980. FISHERY BULLETIN: VOL. 79, NO. 1, 1981. 1972), thermoperception (Dizon et al. 1974, 1976; Steffel et al. 1976), nerve-muscle physiology (Rayner and Keenan 1967), tissue metabolism (Gordon 1968), respiratory physiology (Stevens 1972), body temperature and thermal inertia (Stevens and Fry 1971; Neill et al. 1976), lethal temperatures (Dizon et al. 1977), swimming me- chanics (Magnuson 1970), and swimming speed as a function of water temperature (Stevens and Fry 1971; Dizon et al. 1977), dissolved oxygen (O2 ) concentration, and salinity (Dizon 1977). In addi- tion, several works of a more integrative nature (Magnuson 1973; Barkley et al. 1978; Kitchell et al. 1978; Stevens and Neill 1978) have drawn heavily on these and unpublished laboratory studies. Among the latter are the experiments documented in this paper on oxygen-uptake rates and limits of tolerance to low oxygen in skipjack tuna, Katsuwonus pelamis. 31 FISHERY BULLETIN: VOL. 79, NO. 1 Respiration research on tunas is still quite limited. Besides the present work, there are only four published studies — three involving tuna metabolism (Gordon 1968; Stevens 1972; Brill 1979) and one involving tunas' low^er tolerance limits for O2 (Anonymous 1965). Gordon (1968), using volumetric microrespirometers, determined rates of oxygen uptake in minced preparations of muscle from skipjack tuna and bigeye tuna, Thunnus obesus. Stevens (1972) measured the oxygen concentration of water entering and leaving the gills of restrained, perfused skipjack tuna; from these data he computed oxygen-uptake rate and utilization. Brill (1979), using a similar technique, estimated the relation between stan- dard metabolism and body weight of skipjack tuna. Stevens (1972) also measured oxygen utili- zation in free-swimming skipjack tuna by sam- pling exhaled water collected via opercular can- nulation. Experiments conducted with skipjack tuna at the Kewalo Research Facility provided the earliest estimate of the lower lethal-oxygen limit for a tuna (Anonymous 1965). Our purposes in this work were 1) to determine the magnitude of oxygen-uptake rate in routinely active skipjack tuna, 2) to establish the relation among oxygen-uptake rate, swimming speed, and body weight in skipjack tuna, and 3) to estimate the lowest concentration of O2 that skipjack tuna can withstand for 4 h. The results already have contributed importantly to the development of models of skipjack tuna distribution (Barkley et al. 1978) and bioenergetics (Kitchell et al. 1978). MATERIALS AND METHODS Source and Preexperimental Treatment of Fish for Laboratory Experiments Skipjack tuna were caught by angling in Hawaiian waters at sea-surface temperatures be- tween 23° and 24° C. The fish were transported in 2,400 1 shipboard tanks that were supplied with flowing seawater and supplemental oxygen. Upon arrival at the National Marine Fisheries Service's Kewalo Research Facility in Honolulu, the skipjack tuna were transferred into either 40,000 or 700,000 1 outdoor holding tanks. Naka- mura (1972) described in greater detail the tech- niques that have been developed at the Kewalo Research Facility for transporting and maintain- ing live tunas. The seawater in the holding tanks had 23°-24° M C temperatures, pH 7.4-7.6, 32-33%o salinity, and ~ 6.4-6.7 mg/1 O2. The seawater well that supplied water to the holding tanks was also the source of water for the experimental tanks. At night the holding tanks were illuminated at a low level. Experiments were conducted with fish that had been in captivity 7-26 d. Once the skipjack tuna started feeding (usually within 3-5 d after capture), they were fed to satiation on thawed northern anchovies, Engraulis mordax, or smelt, Allosmerus sp., once a day. Prior to experiments, the fish were fasted for periods ranging from 24 to 27 h, which is more than sufficient time for gut evacuation in skipjack tuna (Magnuson 1969). However, our method of moving fish from the holding to the experimental tanks involved some food ingestion. Two to four hours before data collection, the fish were removed from a holding tank by angling with a baited, barbless hook. Although a small piece of food (1-2 g) was usu- ally swallowed, this transfer technique did select healthy and actively feeding fish. Oxygen- Uptake Experiments in Laboratory Apparatus Two unstirred, sealed respirometers of differ- ent sizes were used. Circulation of water during experiments was provided only by movements of the fish. The larger respirometer was used only during the first of the 10 series of experiments (Table 1). The circular chamber was a vinyl-lined plywood tank, 4.57 m in diameter and 1 m deep, with a cover made of transparent vinyl film bonded to 1 an inflatable tube, 18 cm in cross section, that ' encircled the tank's inner perimeter just below the rim. After the tube was in place and any trapped air had been removed with an electric pump, the clear plastic cover lay over the entire water surface, forming an effective seal. We ini- tially had intended to run all of the experiments with this respirometer. However, it proved to be difficult to operate and visibility of fish within the chamber was poor. Most importantly, a tank of its volume (16,000 1) required a large biomass of fish in each experiment to effect oxygen reduc- tion in a reasonably short period of time. Skipjack tuna are difficult and expensive to capture and maintain; so, to use fish as economically as pos- 32 GOODING ET AL.: RESPIRATION RATES AND LOW-OXYGEN TOLERANCE IN SKIPJACK TUNA sible, we built a smaller respirometer for the nine subsequent series of experiments. The smaller respirometer consisted of a fiber glass tank, elliptical in the horizontal plane; it held 2,400 1 of water and was 2.44 m long by 1.83 m wide by 0.61 m deep (Figure 1). The top edge of the tank had a 5.0 cm wide lip to which was cemented a 0.9 cm thick sponge-neoprene gasket; on this gasketed lip, there was seated, and firmly clamped, a rigid cover made of 0.4 cm thick transparent acrylic plastic strengthened by three 10 cm wide strips of 6.5 cm thick marine plywood cemented to its outer surface. A short length of 5 cm PVC pipe, which served as a vent and access port, was tapped vertically through the plastic and plywood in the center of the cover. Glued flush with the inside surface of the acrylic, the pipe extended about 8 cm above the water level. An inlet and drain allowed fresh sea water to flow through the chamber at rates up to 190 1/min. Both inlet and drain were valved. By slightly overfilling the chamber before closing the inlet valve, we caused the acrylic cover to bulge up- ward 4 cm at the center. The domed conformation permitted the easy removal of bubbles from the chamber through the vent pipe. A sponge rubber ball, fitted snugly into the pipe during experi- ments, completed the seal. Both respirometers were illuminated with overhead fluorescent lights. Indirect natural light coming through windows and doors was not excluded. Visibility of fish in the smaller respirometer was excellent, permitting detailed observation of fish speed and behavior during an experiment. Dissolved oxygen measurements in both res- pirometers were made with a YSI^ model 51A oxygen-temperature meter coupled with a YSI model 5418 oxygen-temperature probe. For mea- surements in the larger respirometer, water was electrically pumped between the tank and a small acrylic chamber in which the probe was mounted. In experiments involving the smaller respirom- eter, the oxygen-temperature probe was placed inside the tank through the vent pipe in the cover. The probe was vigorously jiggled for about 15 s before each reading of the meter; during the intervals between readings, a sponge rubber ball ^Reference to trade names does not imply endorsement by the National Marine Fisheries Service, NOAA, or by Texas A&M University. 2 m STICK CLEAR ACRYLIC TOP 7.5 cm DRAIN Figure l. — The small sized respirometer used in oxygen-uptake experiments 2-10 on skipjack tuna. 33 FISHERY BULLETIN: VOL. 79, NO. 1 was seated tightly against the cable connecting probe and meter. Calibration of the O2 meter was checked against air-saturated seawater at the start, midpoint, and end of each experimental run; and the temperature readings were frequent- ly verified with a mercury thermometer. As an additional check on the O2 meter's accuracy, readings were twice compared with Winkler de- terminations. The instrument gave stable and reliable readings with accuracy ±0.2 mg O2/I. Oxygen was supplied from standard 6,800 1 cylinders through fine (800 grit) 15 cm aluminum oxide grinding stones (Baldwin 1970) to produce minute bubbles. Preliminary and Control Experiments Preliminary experiments conducted in each respirometer and subsequent work by Dizon (1977) indicated that skipjack tuna demonstrate no overt behavioral manifestations of stress as long as O2 levels are maintained above 4.5 mg O2/I. We thus assumed that the fish under our experimental conditions would exhibit respira- tory independence between 7.0 and 5.0 mg O2/I. All subsequent oxygen-uptake experiments were conducted within that range. The preliminary experiments indicated that 1.5-2.0 g of fish per liter of water consumed oxygen at a rate that would limit a run to <3 h, which we rather arbitrarily set as about the maximum time an experiment should last. Experimental Procedure The 10 series of 4 experiments each were made with groups of 2-8 fish (Table 1). The basic procedures for a series of experimental runs were essentially the same for both respirometers. The detailed procedures, herein described, are for experiments with the small respirometer. Each series of experimental runs started be- tween 0900 and 1100 h and continued through the day and night, into the early hours of the follow- TABLE 1. — Respiration rate experiments with laboratory held skipjack tuna. Length measure is fork length. Fish lot Numbef of fish Mean weight and range (g) Mean length and range (cm) Experi- ment Respiration rate (mg 02/g per h) Mean respiration rate of fish lot Swimming speed (Lis) Mean swimming speed 10 8 3,834(3,114-5,222) 58.6(53.2-68.0) 671 (530-844) 632 (475-805) 36.1 (33.3-39.0) 35.5 (31.5-38.1) 1,719 (1,412-2,026) 44.8 (42.4-47.2) 2,539 (2,411-2,667) 52.8 (52.7-52.8) 1,703 (1,496-1,910) 45.1 (43.5-46.7) 2,178 (1,890-2,467) 49.3 (47.1-51.5) 2 2,790(2,523-3,057) 51.6(49.8-53.3) 1,349 (1,161-1,537) 44.6 (42.4-46.7) 2 2,200 (2,132-2,268) 50.2 (48.3-52.1) 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 0.499 0.551 0.506 0.522 0.382 0.555 0.577 0.658 0.395 0.575 Overall means 1,962 46.9 0.522 1.1 1.7 1.8 1.4 1.2 1.6 1.4 1.4 1.3 1.3 1.4 34 GOODING ET AL.: RESPIRATION RATES AND LOW-OXYGEN TOLERANCE IN SKIPJACK TUNA ing morning. The skipjack tuna were transferred from the holding tank to the respiration tank through which water was flowing at about 130 1/ min. The fish were observed for 30 min and any that showed unusual behavior was replaced with another fish from the holding tank. Then the cover of the respirometer was installed and the water flow, now supplemented with oxygen, was continued. The animals were allowed to habit- uate until they were schooling and swimming slowly around the chamber at about 1.5 fork lengths/s (L/s) with no overt signs of stress. During this period (1-3 h) O2 concentration in the chamber was maintained at air saturation, 6.9-7.0 mg O2/I. The outlet water valve was then closed, the oxygen shut off, and the inflowing water was reduced to about 10 1/min. The cham- ber was slowly filled with water until the acrylic cover domed; then the inflow of water was stopped. When all bubbles were excluded, the sponge rubber ball was positioned to seal the vent pipe and the first experimental run was begun. At the start of a run, O2 concentration in the respirometer was between 6.9 and 7.0 mg O2/I. Oxygen concentration, temperature, swimming speed, and general behavior of the fish were monitored and recorded every 15 min. Swimming speed was estimated by measuring the mean time for three passes by the fish over a straight line distance of 1-2 m. (A comparison of values ob- tained from this technique with mean speeds during complete circuits of fish in the respi- rometer showed that speed was essentially con- stant and that the technique yielded an accurate measure of mean speed during a given circuit.) When the O2 level in the respirometer had been reduced to 5.0±0.1 mg O2/I, final observations were recorded and the run terminated. A water flow of about 130 1/min and oxygena- tion were then resumed, which quickly brought the O2 in the respirometer up to about 7.0 mg O2/I where it was maintained for from 1 to 1.5 h until the start of the next run. Following the same procedure, three more runs were made with each group of fish. In all of the experiments, the fish appeared to be in as good condition at the end of a four-run series as they had been at the beginning. Oxygen Consumption by Just-Caught Skipjack Tuna Rates of oxygen uptake were measured in 11 tank-lots of skipjack tuna while they were in shipboard transit between the fishing grounds and the shoreside research facility in Honolulu. These observations were made during three sepa- rate fishing trips during December 1972 (Charles H. Gilbert cruise 129). Sea-surface temperature was uniformly 24° C. Apparatus The five transport tanks served as respirom- eters. Except for the type of cover and differences in plumbing, these tanks were almost identical to the smaller laboratory respirometer described above. The cover of each transport tank consisted of an elliptical fiber glass plate with an open hatchway in its center (see fig. 15 in Nakamura 1972). The hatchway was an 80 by 48 cm oval in cross section and extended, chimneylike, 20 cm above the plane of the tank cover. Each cover was tightly bolted to the gasketed rim of its tank. Seawater was pumped through each tank at a nearly constant rate between 150 and 250 1/min. Water entered a tank near the bottom at one end and exited at the top through an outlet in the hatchway wall. Oxygenation equipment like that described for the laboratory experiments was used in each tank to supplement O2 , which was mea- sured with the same type of meter used in labora- tory experiments. Experimental Procedures The transport tanks were made ready before fishing began by establishing a flow of sea- water and 200-300% supersaturation of oxygen. Each fish caught was lowered on the fishing line through a tank's hatch and allowed to escape the barbless hook. On each day, the entire comple- ment of fish was taken from a frenzy-feeding school (Strasburg and Yuen 1960) within a period of about 10 min. A tank-lot ranged from 7 to 12 fish, averaging 1.83-2.22 kg; estimated mean weight of captive fish was the actual mean weight of 20 other fish caught from the same school. Such an estimate is quite accurate because the size of individual fish within a skipjack tuna school is remarkably uniform (Brock 1954). Within 2 h of capture, oxygen-uptake rate ( Vbz) was measured for each of the 11 tank-lots of fish. In each tank the flow of oxygen was stopped and the time was measured for O2 to decline from the saturation level (6.9 mg O2/I) to a 35 FISHERY BULLETIN: VOL. 79, NO. 1 second level between 6.2 and 5.6 mg O2/I. For three tank-lots of fish, a second measurement of Vbz was made after resaturation of the tank with oxygen. Because the flow of water through the tanks was maintained throughout the 3- to 15-min period of O2 measurement, calculation of Vbz accounted for O2 both supplied to and removed from the tank in the flow of water: d[02]c dt C =([02]/- [02]c)Q-N W Vo, which gives, upon integration and solution for Vo^ , Vo, = ([O2]/- [02]c)-Q N W t)) where Vbz [O2]/ (l-exp(-Q • C"' = oxygen-uptake rate (milligrams 02/gram per hour), = concentration of O2 in incurrent water (milligrams 02/liter), = concentration of O2 in tank and in excurrent water (milligrams 02/liter), Q = water exchange rate ( liter s/hoiir), N = number of fish in tank, W = estimated mean weight of fish (grams), C = operating volume of tank (liters), t = time (hours) for oxygen concentra- tion in tank to decline from [ O2 ]/ toLOzlc- Low-Oxygen Tolerance Experiments Apparatus The tank used for the low-oxygen tolerance experiments was identical to the smaller respi- rometer tank but was uncovered and was supplied with unaerated water (0.5 mg O2/I) directly from the seawater well. During the time it took to fill the tank, the water took up atmospheric oxygen and O2 increased to 1.4 mg O2/I. Experimental Procedure Twelve experiments were made, using 21 skip- jack tuna (Table 2). The fish for six of the experi- ments were the same pair that had been used for a preceding series of oxygen-uptake experiments. For the other six experiments, skipjack tuna were taken directly from a holding tank. The fish were Table 2. — Swimming speed and resistance time to low oxygen of skipjaci? tuna at various oxygen concentrations. Those fish which continued swimming for 240 min were considered to have survived. Length measure is fork length Oxygen Weight Resistance Swimming speed level of fish time range Experiment (mg/l) (g) (mIn) {Us) 1 3.0 1.496 1,910 74 115 1.8-2.0 2 2.0 2,467 1,890 10 20 1.8-3.7 3 2.5 3.057 3,523 57 60 1.8-2.7 4 1.4 1,537 1,161 9 10 2.8-5.0 5 3.0 1,545 1,887 55 70 1.5-2.8 6 3.5 3,430 1,616 65 155 1.5-2.0 7 3.5 890 902 91 Survived 1.1-2.0 8 3.5 812 791 53 65 1.2-2.5 9 4.0 830 1.020 Survived Survived 1.0-1.5 10 3.0 2.132 106 1.5-2.6 11 2.0 1,765 7 2.0-4.0 12 2.5 1.353 28 1.7-2.2 rested in the uncovered respiration tank at air- saturated O2 levels (6.8-7.0 mg O2/I) for about 2 h prior to being transferred to an immediately adjacent low-oxygen test tank. Because of a fish shortage, we were forced to use single rather than paired animals for the last three experiments. The experiments were done with O2 ranging from 1.4 to 4.0 mg O2/I and temperatures between 23° and 24° C. When O2 in the low-oxygen tank reached the required levels, the fish were netted from the resting tank into the test tank. Each transfer took < 5 s, and two fish were transferred within 30 s. The O2 and the behavior and swimming speeds of the fish were observed continuously. During an experiment, the water took up atmospheric oxy- gen at a rate dependent on the air- water pressure gradient. However, at all of the experimental O2 levels there would have been a decrease of O2 concentration due to the fish's respiration had we not gradually introduced oxygen as it was depleted by the fish; by so doing we continuously maintained O2 concentration within ± 0.2 mg/l of the nominal experimental level. A fish's resistance time was the period from introduction into the low-oxygen tank until the animal lost equilibrium and settled to the bottom. A fish was considered to have survived if it was still swimming after 240 min, at which time the experiment was terminated. At the conclusion of each experiment, weight and fork length of each fish were measured. 36 GOODING ET AL.: RESPIRATION RATES AND LOW-OXYGEN TOLERANCE IN SKIPJACK TUNA RESULTS Oxygen-Uptake Experiments in Laboratory Condition Factor It has been our experience that skipjack tuna do not do well in captivity for extended periods. Obviously, valid behavioral and physiological data require that the experimental animals be in good health. As mentioned above, our fish were actively feeding and had been captive for less than a month. Some additional evidence relative to their general condition is provided by com- paring the length-weight relationship of the ex- perimental fish with the relationship obtained by Nakamura and Uchiyama (1966) for freshly caught skipjack tuna. For our captive fish, logW (grams) -= -2.657 + 3.532 /logL (centimeters); for wild skipjack tuna, logW (grams) = -2.317 + 3.368 /logL (centimeters) (log = logio). This com- parison indicates that our experimental fish were, on average, about 14% lighter at a given length than wild fish of the same mean length (ca. 48 cm), the difference in weight-at-length decreased with increasing fork length. Part of the weight dis- crepancy resulted from the near emptiness of the guts of the experimental fish. Raju (1964) reported that pole-and-line caught skipjack tuna weighing about 1.4 kg had stomach contents comprising about 1.5% of body weight on the average and 6.3% of body weight at maximum. General Behavior The behavior of a group or pair of fish over the four experimental runs changed very little, and behavioral VcU"iation among the 10 series of ex- periments was slight. The fish, usually in close company, continuous- ly circled the respirometer 20-30 cm from the sides. Their course in the smaller respirometer was usually elongate but quite frequently shifted to a circle with a radius of 60-70 cm. Direction reversal and figure of eight patterns were not unusual. Rarely one fish would break away and swim separately, sometimes in a direction oppo- site the other fish, but such divergent patterns never persisted for long. Swimming Speed Swimming speed of skipjack tuna, averaged over all experiments, was 1.4 L/s. Speed was independent of O2 over the experimental range (5.0-7.0 mg O2/I), but stepwise multiple regres- sion indicated significant (P^0.05) effects of fish weight, length of time the fish had been in the respirometer, and the time-order of experi- mental series: S = 3.55 - 0.53 nogW - 0.02t - 0.04 • k where S = swimming speed (lengths/second), W = mean fish weight (grams), t = time (hours) the fish had been in the respirometer, k = time-order (1, ... , 10) of the experi- mental series. Thus, length-specific swimming speed decreased with increasing fish size and with increasing values of both time-related variables. Solution of the regression equation at mean values of t (9^8 h) and k (5.5) yielded S = 3.14 - 0.53 /logW (Figure 2). Respiration Rate Mean respiration rate over all experiments was 0.52 mg 02/g per h (Table 1). The data suggested 22 20 1.8 •= 1.6 a. to 1.4 1 12 ? 1.0 MINIMUM SPEED (lengths- sec"') MAGNUS0N(I973) SPEED(length9.sec"') = 3. 14-0.53 log W (C)"' SPEED(cm.sec-') = l2 08ll + l6 0594logW i jq 100 90 UJ a. 80 2 z s 70 60 O (O CD < I I I I I I J L 200 500 1000 WEIGHT (g) 5000 FIGURE 2. — Relation between voluntary swimming speed and body weight in skipjack tuna. Line A connects point-solutions from Magnuson's minimum speed function (Magnuson 1973). Line B is the relation between relative swimming speed and weight observed in the present study. Line C is the relation between absolute speed and weight observed in the present study. Length measure is fork length. 37 FISHERY BULLETIN: VOL. 79, NO. 1 a slight but statistically nonsignificant decrease in respiration rate with decreasing oxygen con- centration. Thus, our assumption of respiratory independence at oxygen concentrations down to 5.0 mg O2/I seems valid; however, our experi- mental design did not permit proper analysis of the relation between respiration rate and O2 . Of the other variables included in a step- wise regression analysis (fish weight, swimming speed, total time in respirometer, time of day, experimental order), only fish weight and swim- ming speed significantly (P^0.05) affected res- piration rate: log Vo, = -1.20 + 0.19 logW + 0.21s body weight (Figure 2): S = 3.14 - 0.53 /\ogW. When this and the previous equation are com- bined to express the relation between oxygen- uptake rate and body weight for skipjack tuna swimming at their characteristic speeds, we are left with log V02 = -0.54 +^0.08 /logW. Thus, the speed-inclusive effect of W on oxygen-uptake rate actually observed in our experiments was quite small (Figure 3). A multiple regression analysis with speed deleted from the independent- variable list yielded no significant relation be- tween V02 and W (P>0.05). In contrast, the weight-inclusive effect of speed on V02 was readily apparent in a simple plot of log Vbj versus S (means) for all experiments (Figure 4). where V02 = oxygen-uptake rate (milligrams O2/ gram per hour), W = mean fish weight (grams), S= swimming speed (lengths/second). We hasten to point out an irregularity in the relation just presented. While we offer the equa- tion as a best available predictor of independent weight and speed effects on oxygen uptake in skipjack tuna, we recognize that W and S in our fish were not independent. As indicated in the last section, the fish tended to swim at a characteristic speed inversely proportional to the logarithm of 600 800 1000 2000 4000 WEIGHT (g) Figure 3. — Lack of significant relation between oxygen-uptake rate and weight of skipjack tuna swimming at voluntary speeds. Weights are means for the fish in each experimental series. 38 O E < < Q. 3 1.0 09 08 07 06 OSl- 04 03 0.2 log O2 UPTAKE RATE -0.4378 +0.1090 -SPEED r •O.ZB 08 1.0 12 14 16 18 SWIMMING SPEED (lengths^sec-') 20 22 Figure 4. — Relation between oxygen-uptake rate and relative swimming speed of skipjack tuna (ranging from 60 to 4,000 g). Data are means for each experiment. Length measure is fork length. Oxygen Consumption and Activity of Just-Caught Skipjack Tuna Skipjack tuna, during their first 2.2 h of captiv- ity, consumed oxygen at rates between 0.9 and 2.5 mg Oa/g per h, the median for the 14 determina- tions being 1.3 mg 02/g per h at 75 min (Figure 5). Swimming speeds were correspondingly great — between 2 and 5 L/s. Rate of oxygen consumption was not significant- ly correlated with time since capture (Kendall's t = -0.20, P-0.18; Siegel 1956) when all 14 rate determinations were considered as random obser- vations from the same bivariate distribution. However, some features of the experiment sug- gested a decline in the rates of both oxygen con- GOODING ET AL.: RESPIRATION RATES AND LOW-OXYGEN TOLERANCE IN SKIPJACK TUNA 40 30 20 O E < < a. 3 10 9 8 7 n 1 r •^^* J I 1 L 20 40 60 80 100 120 140 TIME SINCE CAPTURE (minutes) FIGURE 5.— Relation between the rate of oxygen uptake and the time elapsed after capture of skipjack tuna. Dashed lines connect paired determinations of oxygen-uptake rate for the same tank-lot of fish. sumption and activity during the 2-h interval following capture. An initial period of frantic swimming, lasting 10-15 min, had already ended before we were able to collect our earliest respira- tion data. Still, the first five determinations of oxygen-consumption rate (within about 1 h of fish capture) were 1.5 mg 02/g per h or more. In two cases, a tank-lot of fish respired at a much reduced rate during the second of two sampling intervals separated by about 1 h. (In a third set of such paired observations, oxygen uptake during the second was . 3 mg O2 /g per h more than during the first, but these determinations were separated by only 20 min.) In conclusion, we believe the oxygen- uptake rate of our fish immediately after their capture was underestimated by the overall me- dian value of 1.3 mg 02/g per h and, in fact, may have exceeded 2.0 mg 02/g per h. Low-Oxygen Tolerance Experiments General Behavior Comparative behavioral responses of the fish in low-oxygen («3.5 mg/1) water were quite consis- tent. However, the sequence of behavior was more accelerated and the fish's reactions were often more violent at the lower oxygen concentrations. At the two lowest 02's (1.4 and 2.0 mg/1) the fish showed symptoms of considerable stress within about 30 s of introduction. Stress was manifested as very fast swimming (2.6 L/s), wide mouth-gape. and little or no attempt to school. During the last few minutes before the skipjack tuna died, they assumed a steep angle of attack with their snouts out of the water and swam jerkily, with intermit- tent bursts of speed up to 6 L/s. Complete collapse came abruptly; the fish simply ceased swimming and settled to the bottom. At 2.5 and 3.0 mg O2/I, the initial stress reactions were milder and the fish started schooling within a few minutes after introduction. Swimming speeds were still rela- tively high (1.7-2.8 L/s), and the sequence and types of behavior were similar to those at the lowest O2. At 4.0 mg O2/I, both skipjack tuna swam and otherwise behaved as if they were in oxygen-saturated water. Resistance Time and Swimming Speed There was a marked, direct relation between the logarithms of resistance time and oxygen con- centration at oxygen levels up to 3.5 mg O2/I (Table 2, Figure 6). At 3.5 mg O2/I, four of the experimental fish had resistance times in the same range as the fish exposed to 3.0 mg O2/I, but one fish survived for the 240-min duration of the experiment. The survivor showed few overt signs of stress but did swim faster than the fish in oxygen-saturated water. At 4.0 mg O2/I, both experimental fish survived 240 min. The 21 skipjack tuna used in the 12 experiments ranged in weight from 791 to 3,523 g (Table 2). There was no significant correlation between weight and resistance time to low-oxygen levels. 111 O Z o o o a UJ 21 o a -\ — I — I — I — I I I log RESISTANCE TIME -0.417 + 2.876 log Oa CONG. »• •• 8 10 20 40 60 80 100 RESISTANCE TIME (minutes) 200 Figure 6. — Relation between resistance time to low-oxygen concentration and dissolved oxygen concentration. Resistance time was the period from the fish's introduction into the low- oxygen tank until it stopped swimming and settled to the bottom. Circled points indicate three fish that were still swim- ming after 240 min. The regression line was fitted to the points for fish that died. 39 FISHERY BULLETIN: VOL. 79, NO. 1 Mean swimming speed increased as O2 de- creased (Figure 7), reaching 3.9 L/s at the lowest O2, 1.4 mg O2/I. Mean speed at 4.0 mg O2/I was slightly less than at higher O2 values (oxygen- uptake experiments); therefore, the critical O2 for an increase in swimming speed appeared to lie between 4.0 and 3.5 mg O2/I. a UJ UJ a. z < 1.5 20 2.5 3.0 3.5 40 45 50 5.5 6.0 6.5 DISSOLVED Oz CONCENTRATION (mg-r') FIGURE 7. — Relation between mean swimming speed and dis- solved oxygen concentration. The data for O2 of 6.0 and 5.0 mg 1' are from the oxygen-uptake experiments. Those at lower concentrations are derived from the low-oxygen tolerance experiments. Length measure is fork length. DISCUSSION Terminology Relevant to Tuna Metabolism In this paper we have strived to quantify the activity and respiration levels of our fish. The question of terminology remains. Doudoroff and Shumway (1970) have emphasized that "different meanings have been attached by different authors to the same term or different terms have been used in the same sense " The question of terminology is further complicated in tunas because they, un- like typical fishes, must maintain some minimum forward motion for hydrodynamic lift (Magnuson 1973) and for gill perfusion (Brown and Muir 1970; Stevens 1972); a stationary tuna both sinks and suffocates. Thus, the notion of "resting" metabolic rate ( Doudoroff and Shumway 1970) is not appli- cable to tunas. What we have collected in our laboratory exper- iments were data on "routine" (Fry 1957, 1971) activity and metabolism. Fry (1971) defined rou- tine metabolic rate as "the mean rate observed in fish whose metabolic rate is influenced by random activity under experimental conditions in which movements are presumably somewhat restricted and the fish protected from outside stimuli." Our fish were in a postabsorptive state (except for bits of food they may have eaten during the transfer process) and were as quiescent as tunas are ever likely to be when confined in a small tank. Per- haps, our laboratory data reflect minimum me- tabolism for skipjack tuna in that the fish were swimming at speeds actually below the hydrody- namic minima calculated by Magnuson (1973) for skipjack tuna (Figure 2). On first consideration, it would seem unlikely that tunas — which lack ventilatory pumps and are, therefore, obligate ram-ventilators (Brown and Muir 1970) — could achieve minimum swimming speed without also achieving minimum rate of oxygen uptake. How- ever, Stevens (1972) has shown that skipjack tuna have the capability for doubling the amount of oxygen they extract per unit flow of water irri- gating the gills (utilization efficiency, 0.4-0.8) with only a 17% reduction in ventilation rate (from 3.0 to 2.5 1 H20/kg per min). Thus, it is conceivable that a skipjack tuna could decrease swimming speed and simultaneously increase oxy- gen uptake. We must, therefore, recognize the pos- sibility that "excitement" (Fry 1971) associated with the alien and confining environment of our respirometers resulted in heightened rates of oxy- gen uptake compared with those that might obtain in wild, unexcited skipjack tuna swimming in the sea at the same speeds. However, the data we collected do not permit an objective evaluation of this possibility. For purposes of further dis- cussion, we assume that our laboratory measure- ments of oxygen-uptake rate contained no com- ponent of "excitement" metabolism independent of swimming speed. Lack of change in respiration rate among sequential experiments indicates that any activity-independent excitement component of metabolism that may have been present was habituation-time invariate. This has encouraged us to go so far in the following section as to esti- mate the hypothetical "standard" ( = "basal" — see Fry 1971; Brett 1972) metabolism of skipjack tuna from our respiration data; this we did, as Fry (1971) recommends, by simply extrapolating to zero speed the regression equation relating res- piration rate and swimming speed. Rates of oxygen uptake measured in the "just- caught" fish can scarcely be considered "routine" 40 GOODING ET AL.: RESPIRATION RATES AND LOW-OXYGEN TOLERANCE IN SKIPJACK TUNA but still may have underestimated the skipjack tuna's maximum or "active" (Fry 1971; Brett 1972) rate of oxygen consumption. Wild skipjack tuna similar in size to our experimental fish can swim at sustained speeds probably exceeding 10 L/s (Yuen 1970); our just-caught fish swam at speeds ^5 L/s during the intervals when oxygen-uptake rates were measured. However, the experimental fish may have been repaying an oxygen debt incurred during the feeding frenzy preceding cap- ture or during the early minutes of captivity; recovery from oxygen debt could have heightened oxygen-uptake rates to levels above those com- mensurate with sustained swimming at the ob- served speeds (Brett 1972). "Standard" Metabolism Even though tunas never lie stationary in the water, it is of interest from the bioenergetic and comparative standpoints to separate the routine metabolic rate into standard and activity-related components. From the equation on p. 38, with swimming speed set equal to 0.0, log Vo, - -1.20 + 0.19 logW, where Vba = oxygen-uptake rate (milligrams O2/ gram per hour), W= mean fish weight (grams). Solutions of this equation at our experimental extremes for W are Vbj = 0-21 mg 02/g per h at W = 632 g and Vo^ = 0.30 mg Oj/g per h at W = 3,834 g. These values are extraordinary for two reasons: 1) They are at the extreme upper limit for nontuna (cf. fig. 4 of Brett 1972), a fact that becomes even more remarkable when one considers that other teleost values are almost all for small (10-100 g) individuals, and 2) the weight exponent is a positive 0.19, not a negative value in the neighborhood of -0.2 characteristic of typical fishes (Fry 1957, 1971; Winberg 1960). While weight exponents for active metabolic rate in salmonids may frequently approach 0.0 (Job 1955; Brett 1965; Rao 1968), we know of no data to suggest weight exponents as large as -1-0.2 for metabolic rate in nonscombrid fishes. The valid- ity of a large, positive value for the weight ex- ponent of "standard" metabolic rate in skipjack tuna is supported by independent data, via direct calorimetry, on heat production rates; the red muscle of sedated skipjack tuna (maintained by gill perfusion) metabolized at a rate proportional to W^3(Neilletal.l976). In marked contrast with our estimate of skip- jack tuna's weight exponent for standard metab- olism is that reported by Brill (1979) — negative 0.44, a value at the other extreme for fishes. Considering that Brill's and our groups of fish were similar in size range and preexperimental history, we must deduce that the large discrep- ancy between estimates relates principally to the difference in experimental methodologies: Brill took, as the standard metabolic rate, the stabilized minimum Vq.^ of perfused skipjack tuna that had been first injected with the neuro- muscular blocking agent gallamine triethiodide, then spinalectomized. Activity-Related Metabolism Our respiration experiments estimated only rates of oxygen uptake, not rates of instantaneous metabolic demand for oxygen. Neill et al. (1976) estimated that the oxygen demand of red muscle in highly active (chased) skipjack tuna can reach 7 mg 02/g per h for periods on the order of 1-2 min. For even shorter periods, involving only true burst swimming, the rate of oxygen demand must be even higher. Brett (1972) has estimated that burst-swimming fishes' instantaneous rate of oxygen demand (on a whole-body basis) exceeds the maximum rate of supply by a factor of 10. Any excess of demand over supply accumulates as an oxygen debt that ultimately must be repaid. Our observations on just-caught fish provided (probably conservative) estimates of the maxi- mum rate at which skipjack tuna can supply oxy- gen to meet their metabolic demands. Like the skipjack tuna's "standard" metabolic rate, its maximum (active) rate of oxygen uptake must be substantially beyond that typical of fishes. Just- caught skipjack tuna respired at a median rate of 1.3 mg 02/g per h; the highest five values (those obtained during the fish's first hour of captivity) were between 1.5 and 2.5 mg 02/g per h. Brett (1972), in reviewing his own and others' work, reported that fishes' maximum rates of oxygen consumption reach a "probable ceiling" near 1.0 ±0.2 mg 02/g per h. The activity-respiration relationship obtained at sea for just-caught skipjack tuna was reason- ably consistent with that extrapolated from the laboratory experiments (Figure 8). However, data of the two kinds may have agreed less well had 41 FISHERY BULLETIN: VOL. 79, NO. 1 600 400 2.00 ;= 1.00 'p. 80 ° 60 E a. 3 .02 1.8 kg SKIPJACK TUNA AT aA-C (REGRESSION OF LABORATORY DATA ) 1.8 kg SKIPJACK TUNA AT - 24 "C (JUST-CAUGHT FISH) SOCKEYE SALMON AT IS'C AND GLASS 1973) 2 3 4 SWIMMING SPEED (length»sec-i) Figure 8. — Comparison between respiration-speed relations for 1.8 kg skipjack tuna calculated from the present study and for 1.8 kg sockeye salmon computed from equations given by Brett and Glass (1973). Ranges of observed values are indicated by the lines extending from the median value for just-caught skipjack tuna. Length measures are fork length. we been able to accurately measure swimming speeds in just-caught fish. The basis for our cau- tious appraisal of such apparently good "fit" is suspicion that the linear model, log Voj = a + b speed, which seems adequate for many fishes (Brett 1972; Brett and Glass 1973; Webb 1975), cannot hold for skipjack tuna over the en- tire range of swdmming speeds that they can sus- tain. Personal observations on these fish and Yuen's (1970) report of a school of skipjack tuna (ca. 44 cm fish) that traveled 28 km in 107 min (average minimum speed = 4.4 m/s) convince us that 40-50 cm skipjack tuna can swim for at least an hour at speeds near 10 Lis. If that is so, our linear model predicts oxygen uptake in 1.8 kg skipjack tuna (median size of just-caught fish) at a maximum sustained rate of at least 33.0 mg 02/g per h. Active metabolic rate of skipjack tuna may substantially exceed Brett's (1972) predicted maximum for fishes, but we are confident it does not do so by a factor of nearly 30. The most logical interpretation of this conundrum is nonlinearity in the relation between log Vq^ and speed; as skipjack tuna swim faster, they must become more efficient in their use of oxygen and energy. The same is probably true for other fast-swim- ming fishes, such as Peterson's (1976) striped mullet, Mugil cephalus. Even in the relatively sluggish goldfish, Carassius auratus, oxygen- uptake rate actually declines as the fish pass from spontaneous activity at low apparent speeds to induced swimming (against currents) at higher speeds (Smit 1965). There is, of course, an alternative explanation: Our laboratory experiments overestimated the true coefficient for speed. In fact, taking the lower 95% confidence limit on the speed coefficient — 0.11 — yields a comparatively modest 3.31 mg 02/g per h for predicted Vba at 10 Lis. But a true speed coefficient as low as 0.11 is not only inconsistent with the comparable coefficient in other fishes (Fry 1971; Brett 1972) but also with other, inde- pendent data (Chang et al.^) on metabolism-speed relations in skipjack tuna. The speed coefficient estimated from that study was 0.22, a value re- markably similar to our mean estimate. To close our consideration of activity-related metabolism in skipjack tuna, we offer a compari- son between respiration-speed relations of a 1.8 kg skipjack tuna at 24° C and a 1.8 kg sockeye salmon, Oncorhynchus nerka, at 15° C (Figure 8). We chose the sockeye salmon because its active metabolic rate "is one of the highest [for fishes] on record, exceeding that determined for other salmonids by 30% to 40%" (Brett and Glass 1973). The sockeye salmon respiration-speed relation was computed from equations given by Brett and Glass (1973); 15° C-values were used because this is near the sockeye salmon's thermal optimum for fast swimming and several other vital func- tions (Brett 1971). Skipjack tuna seem to swim and metabolize at rates nearly independent of temperature (Dizon et al. 1977; Chang et al. footnote 4). At all speeds common to the two fishes, skip- jack tuna have the higher metabolic rate — 3.7 times higher at 1.1 Lis (the skipjack tuna's mini- mum speed) decreasing to 1.7 times higher at 3.2 Lis (the sockeye salmon's maximum sustained speed). If the basis of comparison is the energy cost of swimming (oxygen-uptake rate associated with any particular speed minus standard up- take), the difference between these fishes is less- ened but the qualitative relation is unchanged: at "Chang, R. K. C, B. M. Ito, and W. H. Neill. Manuscr in prep. Temperature independence of metabolism and activity in skipjack tuna, Katsuwonus pelamis. Southwest Fish. Cent., Natl. Mar. Fish. Serv., Honolulu, HI 96812. 42 GOODING ET AL.: RESPIRATION RATES AND LOW-OXYGEN TOLERANCE IN SKIPJACK TUNA 1.1 and 3.2 Lis,, the cost for skipjack tuna are, respectively, 2.5 and 1.4 times those for sockeye salmon. We can only conclude that 1.8 kg skip- jack tuna swim at intermediate speeds less effi- ciently than 1.8 kg sockeye salmon — this despite the fact that, among fishes, the skipjack tuna represents the apex of evolutionary engineering for speed (Magnuson 1973; Stevens and Neill 1978). Presumably, the evolution of skipjack tuna (like that of fast cars) has involved sacrifice of energetic efficiency at low speeds in favor of in- creased efficiency at high speeds, permitting a dramatic increase in maximum attainable speed. Interrelation of Metabolic Rate, Swimming Speed, and Body Weight Voluntary speeds {S, lengths/second) of skip- jack tuna swimming in our laboratory respirom- eters were inversely related to fish weight by the relation S = 3.14 - 0.53 /logW. Magnuson (1973), working from basic hydrodynamic rela- tions, predicted minimum speed for steady-state swimming in various tunas; his model for skip- jack tuna yielded a speed versus fish-weight rela- tion very similar in slope to that we observed (Figure 2). The difference in means may be attrib- utable to differences in condition factor and/or body-water content (Kitchell et al. 1977) between our captive fish and the wild skipjack tuna on which Magnuson's calculations were based. Oxygen-uptake rates (Vbj. milligrams O2/ grams per hour) of our laboratory fish were influenced not only by swimming speed but also by fish weight independent of speed: log Vo^ - -1.20 + 0.19 logW + 0.21s. We have concluded above that 1) the intercept value ("standard" rate at any weight) is unusually large for fishes; 2) the weight coefficient is opposite in sign from that typical of fishes (and of organisms, generally); and 3) the interdependency of log Vbz and S on weight is compensatory, resulting in no statistically demonstrable difference among oxygen-uptake rates for skipjack tuna of various weights (600- 4,000 g) swimming at their characteristic speeds. Conclusion (3) led us to explore the relation between oxygen-uptake rate per unit distance (V62, milligrams 02/gram per kilometer) and swimming speed for skipjack tuna of different sizes. Exponentiating the linear regression equa- tion relating V02 in milligrams 02/gram per hour to W and S, Vq^ = 0.063 • W"^^ • 10" "S = 0.063 W°i» • e°^«^. Multiplying the last equation by 27.78 km ^ • S -1 gener- ated an equivalent expression for ^62 in milli- grams 02/gram per kilometer: Vo, = 1.75 • L W 0.19 . _0.48S Finally, we used the exponentiated length-weight relationship for experimental fish: logW = -2.657 + 3.532 logL; thus, W = 0.0022L3^='2 to eliminate W: Vo, = 0.55 • L 0.33 1 . _0.48S Solutions of this equation for V62 at various values of L and S are shown graphically in Figure 9. Small fish are less efficient (higher V62) at any particular speed than are larger fish, but fish of all sizes reach their particular minimum V62 at the same relative speed — about 2.1 L/s. The relation between this, the optimum speed (Sopt) for cover- ing distance, and the value of the coefficient, 0.48, for the exponential term in S is simple — each is the reciprocal of the other: dVo. ^^^ _ ,33 0.48Se«^«^-e»''«« 0.55 • L "^^ per ds at ds s- = 0, 0.48S • e" ^«^ = e" '^^; therefore, Sopt = 0.48 2.08. For skipjack tuna between 30 and 60 cm length, the characteristic speeds and Sopt = 2.08 corre- spond with V62 rates that are maximally (for 60 cm fish) different by only 13% of min V02 (Figure 9). The question arises as to whether the observed characteristic speeds, rather than "Sopt," might be the (evolutionary) "design" speeds that minimize VOz- The characteristic speeds agree remarkably well (better than does "Sopt") with the optimum speed predicted by Weihs' (1973b) model; Weihs, reasoning from thrust and drag relations for fishes, argued that speed is optimized (energy expended per unit distance is minimized) when "the rate of energy expenditure required for pro- pulsion [and associated physiological work?] is equal to the standard (resting) metabolic rate." For our skipjack tuna, S at Vb2 equivalent to twice the hypothetical standard rate was 1.43 L/s, a value that falls midway in the range of speeds 43 FISHERY BULLETIN: VOL. 79, NO. 1 2 3 4 SWIMMING SPEED ( lengths-sec"' ) FIGURE 9.— Relation of rate of oxygen uptake per unit distance (VQj) swam to swimming speed for skipjack tuna of various lengths. The x's indicate characteristic swimming speeds of fish in the present study. Sopt is the optimum speed for covering distance in terms of minimum Vq^. The arrow at 1.43 lengths • sec ' is the optimum speed predicted by Weihs' (1973b) model. characteristic of fish between 30 and 60 cm length (Figure 10). In a subsequent paper, Weihs (1977) showed that fishes' optimum swimming speeds on an absolute basis ought to be proportional to L"^^. The characteristic speeds of our tuna, when com- puted in centimeters per second and treated as a power function of fish length, are proportional to 100,000 100 ;]-3800g 58 cm ;]-2800g 54 cm ;>l800g 48 cm i>-800g 38 cm 3 4 5 6 7 SWIMMING SPEED ( lengths- sec -' ) 10 Figure lO. — Comparison between the measured oxygen uptake relationship extrapolated between and 8.5 lengths sec~* (broken lines) and the theoretically expected power consumption (solid lines) for four skipjack tuna. Triangles (▲) are the theoretical power consumption based on a detailed analysis of drag forces for a 40 cm, 1,003 g skipjack tuna (Magnuson 1978). Points (•) are based upon a detailed analysis of thrust forces (Magnuson 1978). Length measures are fork length. 44 GOODING ET AL.: RESPIRATION RATES AND LOW-OXYGEN TOLERANCE IN SKIPJACK TUNA ^0 45 ^j^^ thus, fit Weihs' model almost perfectly. We conclude by noting that our discussion of optimum swimming speeds for covering distance relates only to skipjack tuna swimming at con- stant depth (as those in our respirometers were required to do). Weihs (1973a) has calculated that negatively buoyant fishes like the skipjack tuna could achieve an energy savings of 20% (compared with swimming at constant depth) by alternately gliding downward at an angle of about 11° (to the horizontal), then actively swimming upward at an angle near 37°. Resistance to Low Oxygen In areas of the world ocean with surface waters not stressfully warm for skipjack tuna there is always available air-saturated water that over- lies oxygen-depleted strata (Barkley et al. 1978). Therefore, the 4-h exposure period we adopted in this study would seem to include all intervals of low-oxygen exposure that skipjack tuna ever need endure at sea. The data suggest for skipjack tuna a threshold of response to hypoxic stress at about 4.0 mg O2/I (Figure 7); this value is at or below that represen- tative of fishes (Davis 1975). In our experiments, the skipjack tuna's response to low oxygen was an increase in swimming speed; this would seem adaptive in that increased swimming speed initi- ated by hypoxic stress would facilitate return of fish from deep, oxygen-depleted water to air- saturated surface water. The 4-h median tolerance limit to low oxygen was also near 4.0 mg O2/I (Figure 6). This value, in keeping with the skipjack tuna's exceptionally high metabolic rate, is apparently higher than that of any other fish yet investigated (Doudoroff and Shumway 1970). Angular Acceleration and Excess Body Temperature Compared with other studies of fish metabo- lism, our experiments with skipjack tuna involved two unusual elements: 1) The fish were forced, by the relatively small size of the tanks, to swim a curved path, and 2) they probably had core temperatures up to several degrees higher than the temperature of the surrounding water. Weihs (1981) has suggested that our continuous- ly turning fish expended more propulsive energy than they would in swimming a straight path at the same speed. A turning tuna must counter centrifugal forces by "banking" with its pectoral fins to produce a component of lift directed in- wards along the turning radius. Therefore, our results may overestimate the oxygen-uptake rates and perhaps also the lower lethal oxygen concen- tration for skipjack tuna at sea. However, we doubt that the magnitude of the overestimate can be very great, for fish in the large and small respirometers (with radii of typical swimming paths about 2 and 0.8 m) respired at similar rates (Table 1). Furthermore, metabolic rates of fish in our experiments were consistent with those inferred from weight and energy "loss" rates of starved skipjack tuna living in tanks 7.3 m in diameter (Kitchell et al. 1978). Oxygen-uptake rates of our test fish also com- pare well with theoretical estimates of the amount of energy consumed by similarly sized fish swim- ming a straight course at the same speeds (Figure 10). The observed oxygen-uptake relationship (milligrams 02/hour) was extrapolated from to 8.5 L/s for four skipjack tuna ranging in weight from 800 to 3,800 g (dashed lines). (Recall that mean speeds of our fish were between only 0.9 and 2.2 L/s.) Superimposed on the empirical relationship are theoretical projections of energy consumption based on estimates of drag force. Theoretical energy uptake — in keeping with the reasoning of Webb (1975), Sharp and Francis (1976), Sharp and Vlymen (1978), and Dizon and Brill (1979) — was computed according to the fol- lowing rationale: 1. Total power required is the sum of the power required for nonswimming processes (P2) plus power required for thrust (Pi), the latter divided by an estimate of total aerobic efficiency =0.2 (Webb 1975). 2. Power required for nonswimming metabolic processes (the standard metabolic rate of a fasted fish from Brill (1979)), P2 - 1.53 • W^^^ where P2 = power (watts), W = weight (kilograms). Brill's (1979) relation is used despite some doubts about the validity of the exponent because it provides for skipjack tuna the only estimate of P2 independent of our data. 45 3. Thrust power must be equal to drag force multiplied by velocity Cd 10-' iplied by velocity Pi = 0.5 p S U^ where Pi = power (watts), p = water density (1.0234), S = surface area (0.4 L^) where L = length (centimeters), U = velocity (centimeters/second), Cd = drag coefficient. 4. The drag coefficient is estimated using Webb's (1975) formula, as Cd = 10 ( pLuy M / where /j. = water viscosity (0.0096). 5. Assuming an oxycaloric equivalent of 3.4 cal/mg O2 , power in watts can be converted into oxygen uptake in milligrams 02/hour by multi- plying watts by 253. The simple model of energy consumption pre- sented here makes no pretention of precision because no attempt was made to accurately deter- mine either the coefficient of drag or the surface area of the fish. Magnuson and Weininger (1978) and Magnuson (1978) did do that. We have in- cluded their estimates for power consumption of a 40 cm, 1,003 g skipjack tuna in Figure 10. The five triangles are estimates of power consumption based on Magnuson's (1978) determination of drag forces, the points based upon Lighthill's (1969) model of thrust forces (data from Magnuson 1978: table XI). Whether a sophisticated estimate of power consumption or a simple one is employed, the correspondence between the theoretically expected and the empirically derived power con- sumption is good. We take this as additional evidence that our experimental values are rea- sonable estimates of oxygen uptake of skipjack tuna swimming straight courses at sea. Skipjack and other tunas are warm bodied, owing to their high metabolic rates coupled with large thermal inertia (Neill et al. 1976; Stevens and Neill 1978). Thus, our fish undoubtedly were warmer than the water in which they swam. Skipjack tuna used in the laboratory experiments probably had core-temperature excesses on the order of 2°-4° C (cf. Stevens and Fry 1971; Neill et FISHERY BULLETIN: VOL. 79, NO. 1 al. 1976); the just-caught fish, being more active, may have had core temperatures as much as 10° C above ambient water temperature (cf. Stevens and Fry 1971). Interpretation of our results has not been complicated by consideration of the difference between tissue and environmental temperatures, because metabolism of skipjack tuna has been shown to be virtually independent of temperature (Gordon 1968; Chang et al. foot- note 4). CONCLUSION Our findings emphasize the unique evolution- ary position of the skipjack tuna (and, by exten- sion, other tunas) among fishes. The skipjack tuna epitomizes what Stevens and Neill (1978) have termed "energy speculators": forms that "operate to maximize energy gain by gambling large energy expenditures ... on the expectation of proportionately large energy returns." The skipjack tuna's "standard" metabolic rate is two to five times that of typical fishes of similar size. Moreover, the skipjack tuna is relatively ineffi- cient in its use of oxygen and food-energy for swimming (at least at low speeds) and it dies at O2 levels still well above those lethal to other fishes. Clearly, the skipjack tuna's ability to sustain high levels of activity has not been achieved without substantial physiological cost. ACKNOWLEDGMENTS This study was initiated by Reuben Lasker and John Hunter, both of the Southwest Fisheries Center La JoUa Laboratory, National Marine Fisheries Service, NOAA. We are grateful for their early involvement and support. We thank J. R. Brett, Fisheries Research Board of Canada, Nanaimo, British Columbia; E. D. Stevens, University of Guelph, Ontario; and D. Weihs, Technion, Haifa, Israel for reviewing the manuscript. LITERATURE CITED Anonymous. 1965. Oxygen studies in relation to catching tuna. Com- mer. Fish. Rev. 27(12):28-29. Baldwin, W. J. 1970. Oxygenating device for live-bait wells. J. Fish. Res. Board Can. 27:1172-1174. BARKLEY, R. A., W. H. NEILL, AND R. M. GOODING. 1978. Skipjack tuna, Katsuwonus pelamis, habitat based 46 GOODING ET AL.: RESPIRATION RATES AND LOW-OXYGEN TOLERANCE IN SKIPJACK TUNA on temperature and oxygen requirements. Fish. Bull., U.S. 76:653-662. BLACKBURN, M. 1965. Oceanography and the ecology of tunas. Oceanogr. Mar. Biol., Annu. Rev. 3:299-322. BLACKBURN, M., AND F. WILLIAMS. 1975. Distribution and ecology of skipjack tuna, Katsu- wonus pelamis, in an offshore area of the eastern tropical Pacific Ocean. Fish. Bull., U.S. 73:382-411. BRETT, J. R. 1965. The relation of size to rate of oxygen consumption and sustained swimming speed of sockeye salmon {Oncorhynchus nerka). J. Fish. Res. Board Can. 22: 1491-1501. 1971. Energetic responses of salmon to temperature. A study of some thermal relations in the physiology and freshwater ecology of sockeye salmon (Oncorhynchus nerka). Am. Zool. 11:99-113. 1972. The metabolic demand for oxygen in fish, particu- larly salmonids, and a comparison with other verte- brates. Respir Physiol. 14:151-170. Brett, J. R., and N. R. Glass. 1973. Metabolic rates and critical swdmming speeds of sockeye salmon (Oncorhynchus nerka) in relation to size and temperature. J. Fish. Res. Board Can. 30: 379-387. BRILL, R. W. 1979. The effect of body size on the standard metabolic rate of skipjack tuna, Katsuwonus pelamis . Fish. Bull., U.S. 77:494-498. Broadhead, G. C, and I. Barrett. 1964. Some factors affecting the distribution and appar- ent abundance of yellowfin and skipjack tuna in the eastern Pacific Ocean. [In Engl, and Span.] Inter-Am. Trop. Tuna Comm. Bull. 8:417-473. BROCK, V. E. 1954. Some aspects of the biology of the aku, Katsuwonus pelamis, in the Hawaiian Islands. Pac. Sci. 8:94-104. Brown, C. E., and B. S. Muir. 1970. Analysis of ram ventilation of fish gills with appli- cation to skipjack tuna (Katsuwonus pelamis). J. Fish. Res. Board Can. 27:1637-1652. Davis, J. C. 1975. Minimal dissolved oxygen requirements of aquatic life with emphasis on Canadian species: a review. J. Fish. Res. Board Can. 32:2295-2332. DIZON, A. E. 1977. Effect of dissolved oxygen concentration and salin- ity on swimming speed of two species of tunas. Fish. Bull., U.S. 75:649-653. DizoN, A. E., AND R. W. Brill. 1979. Thermoregulation in tunas. Am. Zool. 19:249-265. DiZON, A. E., T. C. BYLES, AND E. D. STEVENS. 1976. Perception of abrupt temperature decrease by re- strained skipjack tuna, Katsuwonus pelamis. J. Therm. Biol. 1:185-187. DiZON, A. E., W. H. NEILL, and J. J. MAGNUSON. 1977. Rapid temperature compensation of volitional swim- ming speeds and lethal temperatures in tropical tunas (Scombridae). Environ. Biol. Fishes 2:83-92. DIZON, A. E., E. D. Stevens, w. H. Neill, and j. J. MAGNUSON. 1974. Sensitivity of restrained skipjack tuna (Katsuwonus pelamis) to abrupt increases in temperature. Comp. Biochem. Physiol. 49A:291-299. DOUDOROFF, P, AND D. L. SHUMWAY. 1970. Dissolved oxygen requirements of freshwater fishes. FAO Fish. Tech. Pap. 86, 291 p. Fry, F E. J. 1957. The aquatic respiration of fish. In M. E. Brown (editor). The physiology of fishes. Vol. 1, p. 1-63. Acad. Press, N.Y. 1971. The effect of environmental factors on the physiol- ogy of fish. In W. S. Hoar and D. J. Randall (editors), Fish physiology, Vol. VI, p. 1-98. Acad. Press, N.Y. Gordon, M. S. 1968. Oxygen consumption of red and white muscles from tuna fishes. Science (Wash., D.C.) 159:87-90. IVERSEN, R. T B. 1967. Response of yellowfin tuna ( Thunnus albacares) to underwater sound. In W N. Tavolga (editor). Marine bio-acoustics. Vol. 2, p. 105-121. Pergamon Press, Oxf J0B,S. V 1955. The oxygen consumption of Salvelinus fontinalis. Univ Toronto Biol. Ser. 61, 39 p. KITCHELL, J. F, J. J. MAGNUSON, AND W H. NEILL. 1977. Estimation of caloric content for fish biomass. En- viron. Biol. Fishes 2:185-188. KITCHELL, J. F., W. H. NEILL, A. E. DiZON, AND J. J. MAGNUSON. 1978. Bioenergetic spectra of skipjack and yellowfin tunas. In G. D. Sharp and A. E. Dizon (editors). The physiological ecology of tunas, p. 357-368. Acad. Press, N.Y LAEVASTU, T, AND H. ROSA, JR. 1963. Distribution and relative abundance of tunas in relation to their environment. FAO Fish. Rep. 6: 1835-1851. LIGHTHILL, M. J. 1969. Hydromechanics of aquatic animal propulsion. Annu. Jlev Fluid Mech. 1:413-466. MAGNUSON, J. J. 1965. Tank facilities for tuna behavior studies. Prog. Fish-Cult. 27:230-233. 1969. Digestion and food consumption by skipjack tuna (Katsuwonus pelamis). Trans. Am. Fish. Soc. 98: 379-392. 1970. Hydrostatic equilibrium of Euthynnus affinis, a pelagic teleost without a gas bladder Copeia 1970: 56-85. 1973. Comparative study of adaptations for continuous swimming and hydrostatic equilibrium of scombroid and xiphoid fishes. Fish. Bull., U.S. 71:337-356. 1978. Locomotion by scombrid fishes: Hydromechanics, morphology, and behavior. In W. S. Hoar and D. J. Randall (editors). Fish physiology. Vol. VII, p. 239-313. Acad. Press, N.Y. MAGNUSON, J. J., AND D. WEININGER. 1978. II. Estimation of minimum sustained speed and associated body drag of scombrids. In G. D. Sharp and A. E. Dizon (editors). The physiological ecology of tunas, p. 293-311. Acad. Press, N.Y. MATSUMOTO, W M. 1975. Distribution, relative abundance, and movement of skipjack tuna, Katsuwonus pelamis, in the Pacific Ocean based on Japanese tuna longline catches, 1964-67. U.S. Dep. Commer, NOAA Tech. Rep. NMFS SSRF-695, 30 p. NAKAMURA, E. L. 1968. Visual acuity of two tunas, Katsuwonus pelamis and Euthynnus affinis. Copeia 1968:41-49. 47 FISHERY BULLETIN: VOL. 79, NO. 1 1972. Development and uses of facilities for studying tuna behavior. In H. E. Winn and B. L. 011a (editors), Behavior of marine animals. Current perspectives in research, Vol. 2, p. 245-277. Plenum Press, N.Y. Nakamura, E. l., and J. H. UcraVAMA. 1966. Length- weight relations of Pacific tunas. In T. A. Manar (editor). Proceedings, Governor's Conference on Central Pacific Fishery Resources, State of Hawaii, p. 197-201. NEILL, W. H., R. K. C. CHANG, AND A. E. DIZON. 1976. Magnitude and ecological implications of thermal inertia in skipjack tuna, Katsuwonus pelamis (Linnaeus). Environ. Biol. Fishes 1:61-80. Peterson, C. H. 1976. Cruising speed during migration of the striped mullet (Mugil cephalus L.): An evolutionary response to predation? Evolution 30:393-396. RAJU, G. 1964. Observations on the food and feeding habits of the oceanic skipjack, Katsuwonus pelamis (Linnaeus) of the Laccadive Sea during the years 1958-59. In Proceed- ings of the Symposium on Scombroid Fishes. Part 2, p. 607-625. Mar. Biol. Assoc. India, Symp. Ser. 1. Rao, G. m. m. 1968. Oxygen consumption of rainbow trout iSalmo gairdneri) in relation to activity and salinity. Can. J. Zool. 46:781-786. RAYNER, M. D., AND M. J. KEENAN. 1967. Role of red and white muscles in the swimming of the skipjack tuna. Nature (Lond.) 214:392-393. Robins, J. R 1952. Further observations on the distribution of striped tuna, Katsuwonus pelamis L., in eastern Australian waters, and its relation to surface temperature. Aust. J. Mar. Freshwater Res. 3:101-110. SHARP, G. D., AND R. C. FRANCIS. 1976. An energetics model for the exploited yellowfin tuna, Thunnus albacares, population in the eastern Pacific Ocean. Fish. Bull., U.S. 74:36-51. SHARP, G. D., AND W. J. VLYMEN III. 1978. The relation between heat generation, conserva- tion, and the swimming energetics of tunas. In G. D. Sharp and A. E. Dizon (editors), The physiological ecol- ogy of tunas, p. 213-232. Acad. Press, N.Y. SlEGEL, S. 1956. Nonparametric statistics for the behavioral statis- tics. McGraw-Hill, N.Y., 312 p. SMIT, H. 1965. Some experiments on the oxygen consumption of goldfish (Carassius auratus L.) in relation to swimming speed. Can. J. Zool. 43:623-633. STEFFEL, S., A. E. DiZON, J. J. MAGNUSON, AND W. H. NEILL. 1976. Temperature discrimination by a captive free- swimming tuna, Euthynnus affinis. Trans. Am. Fish. Soc. 105:588-591. Stevens, E. D. 1972. Some aspects of gas exchange in tuna. J. Exp. Biol. 56:809-823. Stevens, E. D., and F E. J. Fry. 1971. Brain and muscle temperatures in ocean caught and captive skipjack tuna. Comp. Biochem. Physiol. 38A:203-211. Stevens, E. D., and W. H. Neill. 1978. Body temperature relations of tunas, especially skipjack. In W. S. Hoar and D. J. Randall (editors). Fish physiology, Vol. VII, p. 315-359. Acad. Press, N.Y. Strasburg, D. W., and H. S. H. Yuen. 1960. Preliminary results of underwater observations of tuna schools and practical applications of these results. Indo-Pac. Fish. Counc. Proc, 8th Sess., Sect. 3:84-89. TAMURA, T, I. HANYU, and H. NIWA. 1972. Spectral sensitivity and color vision in skipjack tuna and related species. Bull. Jpn. Soc. Sci. Fish. 38:799-802. WEBB, R W. 1975. Hydrodynamics and energetics of fish propulsion. Fish. Res. Board Can., Bull. 190, 158 p. WEIHS, D. 1973a. Mechanically efficient swimming techniques for fish with negative buoyancy. J. Mar Res. 31:194-209. 1973b. Optimal fish cruising speed. Nature (Lond.) 245: 48-50. 1977. Effects of size on sustained swimming speeds of aquatic organisms. In T J. Pedley (editor). Scale effects in animal locomotion, p. 333-338. Acad. Press, N.Y. 1981. Effects of swimming path curvature on the ener- getics of fish motion. Fish. Bull., U.S. 79:171-176. WILLIAMS, F 1970. Sea surface temperature and the distribution and apparent abundance of skipjack (Katsuwonus pelamis) in the eastern Pacific Ocean, 1951-1968. [In Engl, and Span.] Inter-Am. Trop. Tuna Comm. Bull. 15:231-281. WINBERG, G. G. 1960. Rate of metabolism and food requirements of fishes. Fish. Res. Board Can. Transl. Ser. 194, 202 p. YUEN, H.S. H. 1970. Behavior of skipjack tuna. Katsuwonus pelamis, as determined by tracking with ultrasonic devices. J. Fish. Res. Board Can. 27:2071-2079. 48 AN ANALYSIS OF CATCH AND EFFORT DATA FROM THE U.S. RECREATIONAL FISHERY FOR BILLFISHES (ISTIOPHORIDAE) IN THE WESTERN NORTH ATLANTIC OCEAN AND GULF OF MEXICO, I971-78» Grant L. Beardsley and Ramon J. Conser^ ABSTRACT Catch and effort data from the United States recreational fishery for billfishes in the Atlantic Ocean and Gulf of Mexico were examined to evaluate their usefulness in determining trends in abundance. In the Gulf of Mexico, data were recorded from both organized fishing tournaments and from non- competitive fishing. A fishing power model was developed and comparisons made between catch per unit effort from tournament data, nontoumament data, and Japanese longline data. The results indicate that catch and effort statistics for white marlin, Tetrapturus albidus, and sailfish, Istiophorus platypterus, in the Gulf of Mexico appear to be reliable and can be aggregated to provide a means of indexing relative abundance of these species. The model did not appear to be appropriate for blue marlin, Makaira nigricans, however The general trend in catch per unit effort from 1972 to 1978 for sailfish and white marlin in the Gulf of Mexico appears to be downward. Based on catch per unit effort from all fishing areas, there appears to be a single stock of white marlin in the western North Atlantic and Gulf of Mexico. In 1971, the National Marine Fisheries Service's Southeast Fisheries Center initiated research on the billfish stocks of the western North Atlantic Ocean and Gulf of Mexico. The purpose of this research was to develop and evaluate a method of determining changes in relative abundance of bill- fish stocks using catch and effort data from the recreational fishery. This report has been prepared to present a description of this research, evaluate the reliability of the sampling techniques, and make a preliminary determination of the validity of catch and effort data from the recreational fishery as an indicator of changes in relative abundance of billfish populations. THE RECREATIONAL FISHERY The development of the U.S. recreational fishery for billfishes (families Istiophoridae and Xiphiidae) has been reviewed in detail by de Sylva (1974). The first sailfish caught by rod and reel in the Atlantic was probably taken off Miami, Fla., around the turn of the century. After World War 11, increased leisure time and affluence coupled with newer and better fishing gear, vessels, and angling techniques spurred a dramatic expansion of the 'Southeast Fisheries Center Contribution No. 81-20M. ^Southeast Fisheries Center Miami Laboratory, National Marine Fisheries Service, NOAA, 75 Virginia Beach Drive, Miami, FL 33149. Manuscript accepted August 1980. FISHERY BULLETIN: VOL. 79, NO. 1, 1981. fishery geographically as well as to a broader seg- ment of the population. In the Atlantic, anglers now fish for billfishes from almost every state along the eastern coast of the United States as well as from the U.S. Virgin Islands, Puerto Rico, and numerous foreign ports. Species The billfish species in the Atlantic recreational fishery are the sailfish, Istiophorus platypterus, the white marlin, Tetrapturus albidus, the blue marlin, Makaira nigricans, and to a much lesser extent the swordfish, Xiphias gladius, and the longbill spearfish, Tetrapturus pfluegeri. Sailfish, the most commonly occurring species in the catch, is more coastal in its habitat than any of the other species and consequently is available to a greater number of anglers. It is also the smallest in aver- age size, with the possible exception of the longbill spearfish, and generally requires less expensive and sophisticated fishing tackle than is commonly used in fishing for marlins. The two marlins are most abundant in oceanic waters, generally far from the coast of the United States, and fishing for marlins usually requires relatively large vessels and expensive fishing gear. Prior to 1976, recre- ational fishing for swordfish was a specialized type of fishing where the fish was usually sighted be- fore the fishing lines were placed in the water. 49 FISHERY BULLETIN: VOL. 79, NO. 1 Fishing was done in a fairly restricted area off the northeastern United States. In 1976, however, a new method of fishing for swordfish was developed off the southeast Florida coast. This method in- volved drifting baited lines at various depths at night. Fishing success using this technique has been substantially higher than by the earlier method, and swordfish are now available to fishermen all along the Gulf of Mexico and Atlan- tic coasts of the United States, whereas previously the fishery was confined to a relatively small geo- graphical area. Longbill spearfish are rare in the recreational catch. They are believed to be primarily an open ocean species and generally are not common in the areas where recreational fishing takes place. Because of the nature of the fishery for swordfish and the scarcity of longbill spearfish in the recre- ational catch, our study involves only the sailfish and the two marlins, and the following discussions deal only with these species. Fishing Techniques Fishing, using rod and reel, is conducted primarily by trolling dead or artificial baits at speeds ranging from 3 to 15 kn. The baits are fished mainly at the surface, although sometimes baits are rigged to troll down to a meter or more beneath the surface. Generally, three to four lines are fished simultaneously, although as many as eight are occasionally used. In some areas the use of live bait has become increasingly popular. Our study does not include catch and effort data involving the use of live bait. Once a billfish is hooked, the boat operator usu- ally maneuvers the boat so that the effort required by the angler is reduced. Once the fish is brought to the boat it is either gaffed and brought onboard or released alive. More and more frequently, anglers and crews are tagging their fish before releasing them in cooperation with the National Marine Fisheries Service Cooperative Game Fish Tagging Program (formerly the National Marine Fisheries Service-Woods Hole Oceanographic In- stitution Cooperative Game Fish Tagging Pro- gram). Billfishes are not highly desirable as food in the continental United States, although they are utilized to some extent as a smoked product. How- ever, many fishermen are learning that fresh mar- lin, in particular, is an excellent food fish. In Puerto Rico and the Virgin Islands there is a great demand for fresh marlin which commands a high price on the local fresh fish markets. THE LONGLINE FISHERY The high seas longline fishery for tunas was begun in the Atlantic by the Japanese in 1956. Fishing effort increased rapidly, peaking in 1965 when almost 100 million hooks were set and the fishery included almost all waters between lat. 40° N and 40° S. Effort fell off rapidly, however, in response to declining catch rates and increasing costs, and by the early 1970's the Japanese were averaging only about 40 million hooks annually. In the mid-1960's, Taiwan and South Korea en- tered the fishery and by the 1970's theirs were the dominant fleets in the Atlantic. An excellent re- view of the development of the fishery is available in Ueyanagi (1974). The longline fishery in the Atlantic is directed primarily at tunas, and billfishes are incidental catches, although large numbers are caught. From 1956 through 1976, for example, almost 140,000 t of white marlin, blue marlin, and sailfish/ spearfish were caught by longliners in the Atlantic (Table 1). There is some evidence that stocks of white and blue marlin in the North Atlantic and South Atlantic are discrete groups (Mather et al. 1972; Wise and Davis 1973). Longline catch per unit effort (CPUE) values for white and blue mar- lins within these two areas in the 1970's are for the most part considerably below those in the 1960's (Figure 1). Table L — Estimated landings, in metric tons, of blue marlin, white marlin, and sailfish/spearfish by the tuna longline fishery in the Atlantic Ocean, 1956-76. Data from Conser and Beard- sley,' tables 1 and 4, for blue marlin and white marlin, and Conser,^ table 1, for sailfish/spearfish. Blue White Sailfish/ Blue White Sailfish/ Year marlin marlin spearfish Year marim marlin spearfish 1956 6 — 1 1967 2,316 1.421 1.421 1957 92 15 39 1968 3,572 2,458 2.281 1958 722 25 50 1969 3,727 2,538 1.586 1959 847 123 72 1970 4,939 2,916 2,758 1960 1,517 206 160 1971 4,316 2,999 1,710 1961 4.004 713 383 1972 3,047 2,452 1,551 1962 7,414 1,984 602 1973 2,925 2,461 1,298 1963 9.034 2,526 841 1974 2,761 2,958 1,413 1964 7,847 3,634 1,240 1975 3,000 1,987 1,122 1965 6.019 4,847 2,587 1976 1,076 2,062 750 1966 3,713 3,296 2,032 'Conser, R. J. and G. L. Beardsley. 1979. An assessment of the status of stocks of blue marlin, Makaira nigricans, and white marlin, Tetrapturus albidus, in the Atlantic Ocean. Collect. Vol. Sci, Pap. 8(2):461-489. Int. Comm. Conserv. Atl. Tunas, General Mola 17, l^^adrid, Spain, ^Conser, R. 1979. Production model analysis of the sailfish and spear- fish stocks in the Atlantic Ocean. Working paper submitted to the Standing Committee on Research and Statistics. Int. Comm. Conserv Atl. Tunas, General Mola 17. Madrid, Spain. November 1979. 50 BEARDSLEY and CONSER: AN ANALYSIS OF CATCH AND EFFORT DATA 70 60 o o He S 30 > P 20 u u Ik u Id 10 - NORTH ATLANTIC WHITE MARLIN -1.5 2.0 1.0 in * o o z o o o K lU a. X bJ 0.5 £ u 56 58 60 62 SA 66 68 YEAR — I 1 1 1 IQ 70 72 74 76 78 60 r ae o o 50 40 P 30 ui > 5 20 I- bl 10 NORTH ATLANTIC ILUE MARLIN EFFORT CPUE 3 <" o o o o o bt a. 1 0. u 56 58 60 62 64 66 68 YEAR 70 72 74 76 78 3.0 2.5 80 r 70 SOUTH ATLANTIC ABLUE MARLIN U) 60 2.0 'Si O o o o z 50 z o S 1.5 § £ 40 at o u Ik a. Ik u 30 1.0 5 u > H 3 u hi 20 & Ik 0.5 U u 10 - 76 78 Figure l. — Effective effort in millions of hooks and catch per unit effort (CPUE) in numbers offish caught per 1,000 hooks for blue marlin and white marlin in the North and South Atlantic Oceans, 1956-77. Data on the Japanese longline fishery are reported by the Fisheries Agency of Japan, Annual report of effort and catch statistics by area on Japanese tuna longline fisheries, 1962-77. THE SAMPLING PROGRAM Prior to 1970 almost no data are available from the U.S. recreational fishery for billfishes on fish- ing effort, although a considerable amount of in- formation exists on catch. Because of the large number of big-game fishing tournaments taking place throughout much of the range of the bill- fishes, and the fairly continuous and often inten- sive fishing effort at various fishing centers along the U.S. coast, it seemed feasible that catch and effort from the recreational fishery could be used to detect changes in relative abundance from year to year as well as short-term changes in availabil- ity. As a consequence, the Oceanic Game Fish In- vestigations Program was organized at the South- east Fisheries Center Miami Laboratory in 1972 to develop an effective system for the collection of billfish catch and effort data in the Gulf of Mexico, Caribbean Sea, and western North Atlantic 51 FISHERY BULLETIN: VOL. 79, NO. 1 Ocean, and to analyze these data to provide infor- mation on temporal and spatial changes in rela- tive abundance. The program is directed primarily at sailfish, white marlin, and blue marlin, although data on recreational catches of yellowfin and bluefin tunas are also recorded. Swordfish and longbill spearfish are rare in the recreational catch, but some data have been obtained. Much preliminary work had already been ac- complished in the Gulf of Mexico by the Southeast Fisheries Center Panama City Laboratory, NMFS, NOA A, in cooperation with various big-game fish- ing clubs and charterboat associations along the coast. Sampling sites were established and cover- age of fishing effort during the fishing season in the gulf was estimated to be as high as 90% (Nakamura and Rivas 1974). Initial contact with big-game fishing clubs, tournament managers, and others associated with big-game fishing tournaments produced a list of about 40-50 tournaments scheduled throughout the Bahamas, Caribbean, Gulf of Mexico, and along the eastern coast of the United States. Con- tact with various state marine research agencies provided cooperative sampling agreements for tournaments within their states. In addition to tournament sampling, port samplers were stationed in the Gulf of Mexico to maintain day- to-day coverage of nontournament fishing activity at major fishing areas along the coast. At present, coverage includes Port Aransas, Tex.; Grand Isle and South Pass, La.; Orange Beach, Ala.; and Pen- sacola, Destin, and Panama City, Fla. The fishing season in the Gulf of Mexico usually runs from April through October. Data Acquisition Sampling procedures at tournaments are reasonably uniform regardless of locality or sea- son. At the end of each fishing day, program samplers interview the angler or a member of the crew of each boat participating in the tournament. Information on environmental conditions, number and species offish hooked, and other ac- tivities are recorded. At most tournaments all of the participating boats aire located at a single marina, and sampling coverage is usually 100%. Tournament sampling is further simplified in that most tournaments have rigidly controlled fishing hours, and all boats in the tournament fish the same amount of time. After the statistical infor- mation is collected, the sampler obtains biological data from each billfish landed. Daily port sampling is more difficult than tour- nament sampling. Fishing frequently takes place from a variety of locations, the boats return to the dock at different times, and fishing effort is fre- quently not as consistent as during tournaments. Much of the success of daily sampling is attribut- able to the samplers' knowledge of the area and their persistence and resourcefulness in obtaining the data. Sampling Problems There are numerous sampling problems that appear to be unique to the recreational fishery for billfishes. The first is the determination of what constitutes the catch portion of the CPUE ratio. When trolling for billfishes there are three distinct levels of activity that feasibly could be associated with effort and provide an estimate of relative abundance. The first is commonly knov^Ti as a fish "raised." This term refers to the visual observation of a billfish behind the trolling baits whether it ultimately strikes the baits or not. The inherent problem in using this measure is species identifi- cation. In addition, it is apparently not uncommon for a single billfish to be raised more than once during a given day's fishing, occasionally by the same boat. There is also the possibility that two or more billfishes are raised in rapid succession, but the observer may interpret this as a single fish. The second level of activity, and the one used in this study, is fish "hooked." Disadvantages of this criterion are differences in the skills of anglers in hooking fish, and the fewer data obtained since many fish that are raised are not hooked. Its ad- vantages are that identification reliability is con- siderably increased since billfishes almost always jump when hooked and positive identification is usually possible. The third level of activity is a billfish "boated" (or caught and released). The biggest difficulty with this measure of catch is that different tournaments use different categories of line-test; comparing CPUE on 9 kg test line with CPUE on 36 kg test line is not reasonable. Another drawback is that the number of data points avail- able from "boated" fish decreases significantly. The value of this measure is that species identification is no longer a problem. We decided to use fish hooked as our measure of catch, and all sub- sequent references to CPUE in the recreational 52 BEARDSLEY and CONSER: AN ANALYSIS OF CATCH AND EFFORT DATA fishery refer to number of billfishes hooked per hour of trolling. Another problem was the determination of which tournaments were suitable for use in the analysis of any given species. Billfish tournaments may be classified as "all billfish" or restricted to a single species or combination of species. In exam- ining CPUE of white marlin in the Bahamas, for example, it is unreasonable to include data from tournaments that are exclusively blue marlin tournaments, because when fishing for blue mar- lin many boats troll large baits, and white marlin, considerably smaller in average size than blue marlin, either refuse to strike at such baits or are unusually difficult to hook. Accordingly, any anal- ysis of a given species used only data from tourna- ments that were specifically directed at that species or that were designated as "all billfish." An additional sampling problem, encountered in almost any kind of fisheries survey, is reliability of recall. We believe that, in general, the respon- dents are able to recall accurately their fishing activity during the day; however, it may occasion- ally be difficult for the angler or crew to recall each species of billfish hooked if fishing was good and several billfishes were hooked during the day. When possible, more than one member of the fish- ing party was consulted if there was some doubt expressed in the original interview. Tournament and port samplers have received excellent cooper- ation at every level, and most of the anglers and crew members make every effort to assist our data collection activities. Consequently, we do not be- lieve that errors in recall significantly affect the results of our analyses. Sampling Coverage Tournament sampling extends along the east and gulf coasts of the United States from Long Island, N.Y., to Port Isabel, Tex. (Figure 2). Addi- tional tournament sampling has been or is being conducted in the Bahamas, Jamaica, Mexico, Puerto Rico, and the Virgin Islands. Tournaments are scheduled throughout the year to coincide with the presence of seasonal concentrations of bill- fishes. In the Bahamas, for example, the tourna- ment season extends from March through July. In southeast Florida, most tournaments are scheduled from November through January. Most of the tournaments sampled are annual events and occur at approximately the same time each year. Tournament scheduling is also arranged so that Figure 2. — Areas in the western North Atlantic Ocean, Carib- bean Sea, and Gulf of Mexico where the recreational fishery for billfishes is sampled. Primary species available by area are: A — blue marlin, sailfish; B — blue marlin, white marlin, sailfish; C — sailfish; D — sailfish; E — blue marlin; F — blue marlin; G — blue marlin, white marlin, sailfish; H — blue mar- lin; I — white marlin. there are few instances where two or more tour- naments are held at the same time in the same area. Seasonal port sampling on a daily basis is con- ducted in the gulf beginning in April and extend- ing through October. The amount of effort mea- sured and the recorded number offish hooked from daily dock sampling from 1971 through 1978 and from tournament sampling, 1972 through 1978, are shown in Table 2. DATA ANALYSIS— GULF OF MEXICO Methodology There are several areas along the Atlantic and Gulf of Mexico coasts of the United States where recreational and commercial fishermen compete for billfishes. This interaction occurs most fre- quently in the northern Gulf of Mexico where in- tensive recreational fishing for billfishes takes place from a number of ports from Florida to Texas during April through October. During the same 53 FISHERY BULLETIN: VOL. 79, NO. 1 TABLE 2.— Data on effort and catch for billfishes recorded by tournament and dock sampling, 1971-78, in the western North Atlantic and Gulf of Mexico. Hours fished No. fish hooked Year Blue marlir White marlin Sailfish Tournaments 1972 23,090.4 365 170 399 1973 17,864.0 512 233 684 1974 18,473.6 537 368 768 1975 26,858.4 684 1,088 1,664 1976 27,368.4 655 750 1,062 1977 35,333.9 772 781 1,989 1978 40,601.0 801 992 1,752 Dock sampling 1971 11,609.9 266 491 482 1972 13,298.2 306 517 528 1973 7,859.5 176 414 140 1974 10,462.8 290 487 197 1975 8,852.5 196 684 434 1976 8,174.5 177 434 577 1977 8,575.0 225 398 232 1978 12,522.7 251 402 200 period of the year, Japanese longliners fish in the same area for yellowfin and bluefin tunas but fre- quently catch billfishes as well. Detailed and con- sistent catch and effort data are available from both the recreational and longline fisheries in this area over the period 1971-78. These attributes make the northern Gulf of Mexico fishery unique when compared with other billfish fisheries in that more than one type of gear is operating at signifi- cant effort levels in the same time and place, and consistent catch and effort statistics are available from both types of fishing operations for a reason- ably long time series. In this analysis we attempt to: 1) determine the utility and consistency of catch and effort data from the recreational and longline fisheries for indexing changes in abundance of billfish populations, 2) obtain species- and area- specific indices of abundance that incorporate both recreational and longline data, and 3) gain a bet- ter understanding of the dynamics of the fishery by modeling the general characteristics of recre- ational and longline fishing for billfishes. The northern Gulf of Mexico was divided into three areas based on the general distribution of recreational fishing (shaded areas in Figure 3). The easternmost. Panhandle, groups fishing effort from Panama City, Destin, and Pensacola, Fla., and Orange Beach and Mobile, Ala. The center. New Orleans, combines effort from South Pass and Grand Isle, La., and the westernmost, Texas, en- compasses all fishing from Texas. Recreational catch and effort data are acquired in each of these areas by sampling both daily, noncompetitive rec- reational fishing as well as fishing conducted dur- ing organized big-game fishing tournaments. From 1971 through 1978 over 136,000 h of tourna- ment and nontournament fishing for billfishes in these three areas were recorded (Table 3). Tour- nament and dock data were processed and monthly total catch and total effort were compiled by species, area, and type of fishing. CPUE was computed for those months in which 60 h or more of fishing effort had been sampled. The 60-h min- imum effort criterion was chosen by making two series of calculations of the variance of monthly CPUE using various minimum effort criteria, and then subjectively selecting a minimum effort level which provided a balance between the variance and sample size considerations. Using the 60-h minimum effort criterion produced more reliable statistics without causing the sample size to be- come unacceptably small. It represents approxi- 100' 30- 25 Galveston Qi' TEXAS 85" PANHANDLE i< inn -30" «c:i==^ > JSLm 25" Figure 3. — The shaded areas, Panhandle, New Orleans, and Texas, are major recreational fishing areas for billfishes. The larger areas, I, II, III, and IV, are 5° squares from which Japanese longline catch and effort data are available. 54 BEARDSLEY and CONSER: AN ANALYSIS OF CATCH AND EFFORT DATA Table 3. — Number of fishing hours recorded from tournament (toum.) and dock sampling at three major fishing areas for bill- fishes in the Gulf of Mexico, 1971-78. See Figure 3 for location of areas. Panhandle New Orleans Texas Table 4. — Japanese catch (in numbers offish) and effort (in numbers of hooks) from the two 5° areas ( longline areas II and IV in Figure 3) in the northern Gulf of Mexico which coincide with recreational fishing areas, 1971-78. BM = blue marlin, WM = white marlin, and SF = sailfish. Year Tourn. Dock Tourn. Dock Tourn. Dock Year Area II Area IV 1971 143.5 8,2877 4,225.2 355.2 3.322.2 3,380.0 462.0 254.3 Effort (hooks) Catch Effort (hooks) Catch 1972 BM WM SF BM WM ,SF 1973 703 6 8 605 2 , 3,730.0 3,395.0 52 7 963 1974 584.2 5,618.1 1,449.7 1971 413,941 220 1,273 853 227,552 114 2,627 1,402 1975 2.020.0 5,587.1 2,441.3 2,034.3 1,767.8 1,230.1 1972 664,295 181 2,280 571 — — — — 1976 4,279.7 4,619.3 3,552.3 1,762.4 2,314.3 1,792.8 1973 237,092 93 998 204 64.787 28 533 797 1977 6.088.3 5,516.2 5,981.0 2,412.6 4,496.0 646.2 1974 53,632 34 213 42 104,298 120 635 505 1978 6,983.4 7,4108 7.576.1 3,966.2 4,999.4 1,145.7 1975 1976 712,659 2,999,552 149 269 546 3,100 313 220,337 140 1,174 878 850 309,1 18 97 622 937 1977 2,206,500 181 993 272 54,407 8 64 87 mately 10 boat-days of fishing and occurs ; in a 1978 1 ,454,447 100 719 62 36,355 6 90 29 region of the effort distribution where moderate changes in the minimum effort criterion would have Httle effect on the number of months used. The resulting CPUE values by area and type of fishing are displayed in Figure 4 for blue marlin, Figure 5 for white marlin, and Figure 6 for sailfish. Monthly catch and effort statistics by 5° area for the Japanese longline fleet are reported by the Fisheries Agency of Japan in the Annual Report of Effort and Catch Statistics by Area on Japanese Tuna Longline Fisheries for the period 1962-77. Japanese longliners fish all four of the 5° areas that compose the northern Gulf of Mexico (Areas I through ly Figure 3). However, 52% of their fish- ing effort in the northern gulf during 1971-77 oc- curred in area II and 39% in area III. Only 1% occurred in area I and 8% in area IV Since nearly all recreational fishing for billfishes in the north- ern gulf occurs in areas II and ly only longline data from these two areas were used in this com- parative analysis. Although no data on the dis- tribution of longline effort within 5° areas are given in the Japanese annual reports, data supplied by Honma^ on the distribution of catch and effort during 1971-75 show that most longline effort in areas II and IV occurred in the more coastal regions which coincides fairly well with the location of recreational fishing grounds as dis- played in Figure 3. Catch and effort statistics by species for areas II and IV were compiled from the Japanese annual reports for 1971-77 (Table 4). Comparable statistics for 1978 were compiled from the quarterly reports submitted to the Southeast Fisheries Center by Japanese longliners fishing in the U.S. Fishery Conservation Zone. CPUE was then computed for those months in which 2,000 hooks or more of fishing effort occurred. As with ^Misao Honma, Far Seas Fisheries Research Laboratory, Shimizu, Japan, pars, commun, July 1977. the recreational data, using a minimum effort criterion produced more reliable statistics and the number of months accepted was not sensitive to moderate changes in the threshold level. The Japanese longline CPUE values by species and comparable recreational fishing area are also shown in Figures 4-6. Murphy (1960), Rothschild (1977), and others have discussed some of the important aspects in- volved in using longline statistics to estimate changes in abundance. One of the demonstrated functional relationships, which may be pertinent in this analysis, is that the average amount of effective effort per hook is a function of the.amount of "soaking time" the gear is in the water. Al- though the Japanese annual reports do not pro- vide time in the water data, NMFS observers aboard Japanese vessels in the northern gulf re- port a consistency in the time the gear is in the water during recent years (Lopez et al. 1979). Al- though no data are available from earlier years of the analysis period, soaking times tend to remain more or less constant in most tuna longline fisheries and consequently, fishing time can be measured by the number of hooks set (Food and Agriculture Organization 1976). The lack of data on time in the water should, therefore, not con- tribute significantly to any bias in the estimates of relative abundance. Another aspect of the longline data which is also pertinent to this analysis is that sailfish and spearfish catches are combined in the Japanese annual reports. This problem may be minimal in coastal areas, however, since Ueyanagi et al. (1970) demonstrated that sailfish are found primarily in coastal areas and spearfish tend to inhabit more oceanic waters. In this analysis all catches from areas II and IV that were reported as sailfish/spearfish in the annual reports were as- sumed to be sailfish. 55 FISHERY BULLETIN: VOL. 79, NO. 1 FIGURE 4.— Monthly CPUE for blue marlin from the longline fishery and the recreational fishery in the three major fishing areas in the northern Gulf of Mexico, 1971-78. Longline CPUE depicted for Panhandle and New Or- leans were derived from data taken from the 5° square labeled II in Figure 3. CPUE for Texas was taken from the 5° square labeled rV in Figure 3. The first month depicted on the abscissa is March (M) 1971. CPUE is in numbers offish caught per 1,000 hooks in the longline fishery and numbers offish hooked per 100 hours fished in the recreational fishery. The Tournament panel displays CPUE calculated from billfish fishing tour- naments while the panel labeled Dock shows CPUE derived from noncompetitive fishing. PANHANDLE (/> TOURNAMENT . (/) i 111 iiil .i DOtJC M 2.9- ^ 8.8- (/) T LONGLINE TT K l / l ^' lIl M ^ NJSDNJSDHJSDHJSDnJSDnjSDHJSDnJSD NEW ORLEANS TOURNANENT (/> a. o (/) DOCK JL J '■■ U'lr 18 (/» (/> 2.5- e.e- rf LONGLINE h ii ' i 'ln i ' i " ^ NJSDHJSDHJSDnJSDMJSDHJSDNJSONJSD TEXAS (/» TOURNAMENT DOCK lln ltl.1 4.1 ill i OS o • o -18 ^ -8 «/» 10- PANHANDLE TOURNAHENT . ii 1 I ii i 1 i - -20 (/» DOCK Jit Jt^ mil .tTtrl. Jn Jlli M *-nlLe 30 2 LONCLINE wi4i I r " i 1 1 f 'l I I' T 'i ii' V f I 'I T 1 1 n" i I'l T l I r iS njSDHJSDHJSDNJSDHJSDHJSDNJSDNJSD a. (/I a. Figure 5.— Monthly CPUE for white marlin from the longline fishery and the recreational fishery in the three major fishing areas in the northern Gulf of Mexico, 1971-78. Longline CPUE depicted for Panhandle and New Or- leans were derived from data taken from the 5° square labeled II in Figure 3. CPUE for Texas was taken from the 5° labeled IV in Figure 3. The first month depicted on the abscissa is March (M) 1971. CPUE is in num- bers of fish caught per 1,000 hooks in the longline fishery and numbers offish hooked per 100 hours fished in the recreational fishery. The Tournament panel displays CPUE calculated fi-om billfish fishing tour- naments while the panel labeled Dock shows CPUE derived from noncompetitive fishing. NEW ORLEANS (/) TOURNAHENT 1 41 jfl A 10 c/l • DOCK .IIL T .t .1 Ill,t I Ii I T A .rflil' 10 10- (/» LONCLINE I _LJ I r " i 1 1 i " i I \ T \ I i " i'i i r i ' i I rn i'i t i i i t i HJSDHJSDNJSOHJSDNJSDHJSDNJSDNJSD TEXAS o TOURNAHENT i i/t DOCK J d »W*, k -10 (/> 20- I LONCLINE !Tirrl* ii r!! ii l' i I I I I I I I I I n I I 1 n I I I I I I I I I I I I . . MJSDnjSDNJSDHJSDHJSDnjSDHJSDNJSD 1971-1978 (/» 57 FISHERY BULLETIN; VOL. 79, NO. 1 Figure 6.— Monthly CPUE for sailfish from the longline fishery and recreational fishery in the three major fishing areas in the north- em Gulf of Mexico, 1971-78. Longline CPUE depicted for Panhandle and New Orleans were derived from data taken from the 5° square labeled II in Figure 3. CPUE for Texas was taken from the 5° square labeled IV in Figure 3. The first month depicted on the abscissa is March (M) 1971. CPUE is in num- bers of fish caught per 1,000 hooks in the longline fishery. The Tournament panel dis- plays CPUE calculated from billfish fishing tournaments while the panel labeled Dock shows CPUE derived from noncompetitive fishing. PANHANDLE (/) TOURHrtHENT . U» tI T ! In ill Jl jJl le (/I 3- DOCK Jl ttIIIt .t TtT .TTm. .ml. Jk jTIIt ..^tk ■e -2e LONGLINE I TT P i nfM ili l i Tf i rf' i irlij T j HJSDMJSDHJSOMJSONJSDHJSOMJSDMJSO TEXAS TOURHAHENT Jl 1 90 - ««J> and ^',. =(/3.. -/3,s.*) The parameters u^ , a '^^ , and ^ '^^ can be estimated by solving the usual normal equations and esti- mates of relative fishing power and relative popu- lation density can be obtained from [P^^ = expCa.^) andD^ =exp(i3V^)]. To apply the basic catch model to the billfish fishery in the northern gulf, it is necessary to assume that for each type of fishing (i.e., dock, tournament, and longline) catchability is constant throughout the analysis period, there is no in- teraction between catchability and density, and units of effort operate independently. The first two assumptions may be tenuous for this fishery and will be investigated in the analysis. The third as- sumption appears to be reasonable. The basic catch model was used initially to de- termine what relationship existed between catch and effort data from dock and tournament data. Figure 7 presents a flow diagram for the determi- THE BASIC CATCH MODEL DOES NOT ADEQUATELY REPRESENT THE DATA Phase three ^...^'flGN I F I CAN?'***.^^^ Difference in power^^ ^^^^ 5: LEVEL,,,-'"^ YES USE FPOW TO FISHING POWER POOL CATCH AND EFFORT DATA POOL CATCH AND STANDARDIZED EFFORT DATA COMPUTE A SINGLE CPUE WHICH REPRESENTS BOTH TYPES OF FISHING THE MODEL IS APPROPRIATE Figure 7. — Flow diagram for determining the appropriateness of using the basic catch model to represent data from the two different kinds of fishing methods. 59 FISHERY BULLETIN: VOL. 79, NO. 1 nation of the appropriateness of the basic catch model for a single species-area case. Separate analyses were performed for blue marlin, white marlin, and sailfish in each of the three recre- ational areas. Since the model assumes that the catchability coefficients of dock and tournament fishing are proportional [Equation (3)], correlation analysis was performed on the dock and tournament CPUE values (Phase 1, Figure 7) and the model was con- sidered appropriate for estimating fishing power only when the CPUE's were significantly corre- lated at the 5% level. Data used in the correlation analysis were from all months in which dock and tournament fishing met the minimum effort threshold concurrently. The two factor ANOVA model [Equation (2)] was then used to test for significant differences in fish- ing power and density, and Tukey's (1949) test was used to test for significant interaction. The data used in the ANOVA were from all months in the 1971-78 period for which dock and tournament sampling met the minimum effort threshold con- currently and for which CPUE's were >0 for both types of fishing. The positive CPUE constraint was necessary because of the log transformation used in obtaining Equation (2). Because the model re- quires that there be no interaction between power and density, the model was not considered appro- priate when interaction was significant at the 5% level. For all cases in which the model was deemed appropriate and the ANOVA test for difference in power was not significant at the 5% level, the catch and effort data were pooled and a single recre- ational CPUE was calculated for those species- area combinations. Where the model was appro- priate and the power was significantly different, dock sampling was designated as the standard and the relative fishing power of tournament fishing was estimated from Equation (5). The computer program FPOW (Berude and Abramson 1972) was used to estimate the relative fishing power. FPOW solves the normal equations like Equation (5) and corrects for the logarithmic bias using a Taylor series expansion of the estimate about its true value (Laurent 1963). The FPOW program was modified to perform the usual F-test for the sig- nificance of the overall regression and to compute the coefficient of determination. As in the ANOVA test, the data used in the fishing power estimation were from all months for which dock and tourna- ment fishing met the minimum effort threshold concurrently, and for which both CPUE's were >0. For those species-area combinations in which the model adequately represented the recreational data, the entire procedure was then repeated in an analogous manner to compare the recreational and longline data. Results The results of the correlation, ANOVA, and re- gression analyses for blue marlin, white marlin, and sailfish from the Panhandle, New Orleans, and Texas areas are summarized in Table 5. Blue Marlin In the Panhandle area, dock and tournament CPUE data are fairly consistent and it appears that fishing power is greater for dock data than for tournament data. When the dock and tournament data were pooled and compared with the longline data, no correlation was found and interaction between power and density was apparent. In the New Orleans and Texas areas, no significant dif- ference in the power of dock and tournament data was found, but the CPUE's were not correlated and interaction was significant in the New Or- leans data. The blue marlin results generally indicate that the basic catch model does not adequately repre- sent the blue marlin data in the northern Gulf of Mexico. While it may be possible to obtain adequate indices of abundance from recreational or longline data, the two types of fishing appear to be providing very different indices in the same local areas, and it cannot be determined which, if either, provides a valid measure of relative abun- dance. It appears that until the dynamics of the blue marlin fishery are better understood, the use of nominal catch and effort data to index relative abundance may produce inconsistent and mislead- ing results. White Marlin In the Panhandle and New Orleans areas, the CPUE's were well correlated, no significant differ- ence in the power of dock and tournament data was found, and no interaction was apparent. When dock and tournament data were pooled and com- pared with longline data, the CPUE's were well correlated, and a significant difference in power was found, but significant interaction was found in 60 BEARDSLEY and CONSER: AN ANALYSIS OF CATCH AND EFFORT DATA s . 1 ^ — z 6 ^ ^ Qi o^ ^=3 in c cd J= c Bii CO Oh ^^ B 2 CO > E o O a. tu o o E c ^^ o 5) ^= Ifish catc 05 fer g' Cl> o Q •« -^ tr to t« CO "H -S g^ c .5=:^ CO o — o 2 S c 131 ^ c CO S 2 1 :s ,2 ^ ^ c c 2 o :s D, ^ y t8 ca "to .0) CO E 1) h= E 2 ^ ^ bC 5 ^ c o 'c ^ £ t2 ■- CO E 5 £■ cn oj O (0 c Q CO -a z c c ca .5 < o §.§ ts (0 1.^ u. 0) -*-> bo +J 1^ o y cc Q. "3 S c -S cs a, - CO < a> > J2 OcT c .9 Z ° m < ^ a> c =■ S o J? o • M o ^ tii nS -t^ "^ S u -" ._ O) 8 $ > <" E 7 < 3 1 1 Cd CO .J s m o < a. H w 1 o ci t^ in CD "^ 1^ o -^ CO O CD d I I I I CD I '°- I O I i2 CD I " 1 1 1 1 ; I ; 1 1 V I I o • cnmocD-^'ir^cDco ;_C\lcDOr^^'-CDC\l>->-CDO> ococNjooT^cbcbT-d'-dcbci-r^P II II II II II II II II II II II II II II II u> ^ 2 o w s ? t^ tn t S V to : - u.u.u,u.u.u.u.u. u. u. u. u. U- u. u. II II II 11 II II II II II II II eg ^ ^ M II to 9 f^ tn f^ II II into — c>j(0-'~<-coina><-cu fvJCM — ^C«ICNJ'-f->'-(M CD E CO f 9 1= t= S c j^ c CO o CD 0) CJcOOOcOcCJC-'cCJcC-'^ OoooooOo°°oOoOo D-jQQQJjQJjQQ-jQJjQ^ c C c a> CO o CO a CO T3 » B 0) •o a) c CO O u, CD O c O c 5 2 c 5 51= 3 I c Z (2,^ Z c- CO co E E £ 1 1 1 1 1 1 1 1 r 1 1 1 1 1 1 1 n 1 1 r 1 1 1 T i 1 1 n I HJSDNJSDMJSDMJSDNJSDMJSDHJSDHJSD NEW ORLEANS FIGURE 8.— Monthly CPUE for white marlin from the Panhandle and New Orleans areas, 1971-78. CPUE is derived from standardized recreational (both dock and tournament fish- ing) and longline effort. The first month de- picted on the abscissa is March (M) 1971. ^nrnjTJTJTYTUJJ:, 1971-1978 part of the comparable longline effort probably occurred in the western portion of area II where few sailfish are caught. Recreational and longline data were pooled, and the aggregate indices of abundance are shown in Figure 9. The results indicate that the basic catch model is adequate for representing the sailfish data in the northern Gulf of Mexico. In view of these results, it then seems appropri- ate to examine white marlin and sailfish catch data from the Gulf of Mexico to see if any trends in relative abundance are apparent. Figure 8 pre- sents CPUE for white marlin from the Panhandle and New Orleans areas and suggests that a gen- eral decline in white marlin abundance has taken place since 1973. There was a peak in 1975 in both areas (also seen in the yearly average of CPUE shown below) and two good months in 1978 in the Panhandle area, but the overall trend would ap- pear to be downward. Figure 9 presents similar data for sailfish from all three gulf areas. For this species, too, the gen- eral trend in relative abundance, at least in the Panhandle and New Orleans areas, appears to be 62 BEARDSLEY and CONSER: AN ANALYSIS OF CATCH AND EFFORT DATA PANHANDLE 3- ^ 3- ^ 2- ' A H K -dQ o - 1 1 1 1 1 1 1 1 r I ri 1 1 1 1 T 1 1 1 T 1 1 1 Ti 1 1 n I HJSDNJS0NJ8DHJ8DNJ8DHJS0NJ8DNJSD (/> < NEW ORLEANS (LL AND DOCK DATA) tDMJtDNJtDNJtDHJt»NJt»NJtONJtO TEXAS %f% M Figure 9.— Monthly CPUE for sailfish from the Panhandle, New Orleans, and Texas areas, 1971-78. CPUE is derived from standardized recreational (both dock and tournament except for New Orleans) and longline effort. The first month depicted on the abscissa is March (M) 1971. (/I \%1\-\%1% 63 nSHERY BULLETIN: VOL. 79, NO. 1 clearly downward. In the Texas area, 1975 and 1976 were years when CPUE was unusually high, but in 1977 and 1978 CPUE dropped back to levels more consistent with earlier years. DATA ANALYSIS— ALL AREAS Only tournament data are available from the recreational fishery in areas other than the Gulf of Mexico, hence a fishing power analysis similar to the one conducted for the gulf is not presented. It is informative, nevertheless, to examine CPUE data from all sources throughout the western At- lantic and Gulf of Mexico in view of the results of the analysis presented for the gulf. Blue Marlin ■OSn .04- w a. .03- .02- ATLANTIC AND CARIBBEAN GULF OF MEXICO — I . 1 ,1 I I 1 I — 1971 72 73 '74 '75 '76 '77 YEAR '78 FIGURE 10.— CPUE, in number offish hooked per hour of fishing, from the recreational fishery for blue marlin in the two major fishing areas, 1971-78. Data on blue marlin were divided into two areas: the Gulf of Mexico and the Atlantic and Caribbean (Figure 10). Both tournament and dock sampling were combined for the gulf. This division does not necessarily mean we support a separate stock theory for these areas, but merely that the geo- graphical separation of fishing effort indicates that this is a logical division for comparative pur- poses. If, however, trends in CPUE from the two areas are similar, one might conclude that there is at least prima facie evidence of a single stock. Figure 10 shows that trends between the two areas are similar only from 1973 to 1976, which is obvi- ously inconclusive. It should also be noted that there is little fluctuation in CPUE over the time series presented, particularly when compared with white marlin (Figure 11) and sailfish (dis- cussed below). Normally one would expect CPUE for a long-lived species with numerous age groups contributing to the fishery to fluctuate much less than that for a species with a relatively short life span where the impact of a large or small incoming year class would be much greater on the fishery. Although no reliable age and growi;h data are available on blue and white marlins, the Atlantic blue marlin grows to a much larger size than either the white marlin or the sailfish, occasion- ally reaching weights of over 580 kg (Interna- tional Game Fish Association 1979) and would therefore appear to be the longest lived of the three species. The trends in CPUE for the three species appear to conform to the general pattern one might expect based on their presumed relative life span and the length of time they would be expected to contribute to the recreational fishery. .14' O. u 0.8' < UJ K 0.6- U bJ ae 0.4- 0.2^ • .208 • ■217 NORTH CAROLINA* TO NEW JERSEY GULF OF MEXICO BAHAMAS — I — 1971 — T— V2 — I— '72 1 '74 YEAR — I— '75 — I — '76 — f— '77 — 1 — '78 Figure U. — CPUE, in number offish hooked per hour of fishing, from the recreational fishery for white marlin in three major fishing areas, 1971-78. White Marlin Data on catch and effort for white marlin were divided into three areas (Figure 11). Mather et al. (1972) hypothesized that the gulf and Atlantic stocks of white marlin were separate based on tag return data and the distribution of CPUE in the Japanese longline fishery. More recent tag return data, however, indicate that there may be consid- erable mixing of white marlin between the Gulf of Mexico and the Atlantic Ocean. "• There is rather ■■Che-ster C. Buchanan, U.S. Fish and Wildlife Service, An- chorage, AK 99503, pers. commun. June 1977. 64 BEARDSLEY and CONSER: AN ANALYSIS OF CATCH AND EFFORT DATA clear evidence (Mather et al. 1972) that the group of white marlin available to the recreational fishery in the Florida Straits and Bahamas in late winter and early spring (labeled Bahamas in Fig- ure 11) is the same group that concentrates off the northeastern coast of the United States in late summer and early fall (labeled North Carolina to New Jersey in Figure 11). If CPUE from the recre- ational fishery is adequately measuring the rela- tive abundance of white marlin stocks, one would expect a high degree of correlation between CPUE from a single stock from three widely separated areas assuming a constant percentage of the total stock was available in each area, each year. By inspection, it is clear that for the time series avail- 170n 160- U ^ 150- »- O z bi -I ae e o lb I < 180-1 O 170- 160- GULF OF,'-' MEXICO BAHAMAS • -•-^^_-. NORTH CAROLINA TO NEW JERSEY . --• 2 FISH BAHAMAS I——- •-„,___^ GULF 0F» • *~"'^V.|i^ "*^'"*=° NORTH CAROLINA* TO NEW JERSEY —I 1971 — I— '72 — I— '73 1 I '74 '75 YEAR — r- '76 77 FIGURE 12.- -Length frequencies of white marUn from three major fishing areas, 1971-77. able a close relationship appears to exist between CPUE from the three areas sampled. It is also interesting to note that 1975 was a good year in all three areas. Although availability obviously plays an important role in affecting CPUE, it seems unlikely that it is the dominant factor in this case since the three fishing areas are widely separated geographically, and conditions affecting availabil- ity would not likely be optimal in all three areas in the same year Correlation coefficients were calcu- lated for all three areas (5 yr) and for the Gulf of Mexico and the Bahamas (7 yr). The multiple cor- relation coefficient for all three areas was signifi- cant at the 95% level {R = 0.925) and the simple correlation coefficient for the Bahamas and the Gulf of Mexico was significant at the 99% level ( r = 0.865). If we are indeed measuring relative abun- dance, then the similarities in all three sets of data support the hypothesis that the three general fish- ing areas harbor a single stock of white marlin. Size data from 1971 through 1977 separated by sex do not reveal any substantial differences among fish in the three areas (Figure 12). Average size has remained fairly stable over the period with females averaging larger than males for all areas. Earlier size data from the recreational fishery, not differentiated by sex, and with the Atlantic and gulf areas combined, suggest that a moderate reduction in average size has occurred since the late 1950's and early 1960's (Figure 13), but that size may have stabilized since 1970. Sailfish CPUE data for sailfish were separated into three areas (Figure 14). These are the major fish- 190' a 180- z w 5 170- o I ui o 160- 150- • BOTH SEXES COMBINED \ ,.„., . 07) \ '^°*-H74) , (464) y' *^*' (438) ^.^ (542) FEMALE - ' (210) J271>^* (128) (3,,_(.00)_,84)_(122)„3„ ,,34)^. Air ^» • MALE — I r 1955 — I r 1960 1965 YEAR 1970 1975 1977 Figure 13. — Length frequencies of white marlin from the recreational fishery in the western North Atlantic and Gulf of Mexico. The number of specimens measured is shown in parentheses. 65 FISHERY BULLETIN: VOL. 79, NO. 1 •SOn .40' .30' .20' 111 3 a. u J .i2n < < .10- w K U w « .08 .06- .04- .02' PALM BEACH* STUART I —I 1~" 1 1 "*. 1973-74 ■74-75 '75-76 76-77 77-7g 78-79 FLORIDA KEYS* — I 1971 72 — r- '73 — r- '74 75 '76 — I '78 YEAR Figure 14. — CPUE, in number offish hooked per hour of fishing, from the recreational fishery for sailfish, 1971-78. The panel labeled Palm Beach-Stuart is an area along the southeast Florida coast about 40 nautical miles long and fishing is concen- trated at the end of one year and the beginning of the next. sharp decline in the northern gulf in 1978 when CPUE fell to the lowest level of the 8-yr time series. By inspection of Figure 14, it can be seen that there is an inverse relationship between CPUE in the Florida Keys and the Palm Beach-Stuart area except for the 1978 Keys point and the 1978-79 Palm Beach-Stuart point. It is also interesting to note that if we shift the CPUE data from the Florida Keys forward by 1 yr, we obtain a strong positive correlation, significant at the 99% level, between the two areas. Our sampling in the Keys occurred in November and early December and much of the catch consists of very small sailfish, often averaging only 4-7 kg. Our sampling in the Palm Beach-Stuart area was in late December and January and the average size of sailfish in this area during those months was about 14-18 kg. Jolley (1977) concluded as a result of his studies of growth by analysis of dorsal spines that age-2 sailfish averaged about 7 kg and age- 3 sailfish about 14 kg. This approximates the difference in size of sailfish caught at the two areas. We believe that tournament sampling in the Keys is provid- ing a measure of the strength of the incoming year class (age 2) and that this strength is reflected in CPUE from the Palm Beach-Stuart area (mostly age-3 fish) some 12-14 mo later. ing areas for sailfish in the continental United States. Although fishing effort for sailfish is fairly intensive all year long along the southeast Florida coast, we divided the area into two specific loca- tions: the Florida Keys and the Palm Beach-Stuart area. Our sampling at these two areas was concen- trated into two separate time periods and we be- lieve that different groups of fish are exploited in each area. CPUE fluctuates widely in all three areas. One might expect this from a relatively short-lived species where the effect of year-class strength on the fishery is pronounced. In addition, the sailfish is more coastal in its habitat than the marlins and availability may be more strongly influenced by environmental conditions (Jolley 1979). CPUE fell sharply in the gulf in 1977 and again in 1978. This overall decline was strongly influenced by a large decline off the coast of Texas where fishing for sailfish is usually better than in any other area of the gulf (Figure 15). CPUE in 1977 and 1978 was the lowest since we began sampling and 8(F/c below the average of the previous 3 yr. There was also a DISCUSSION Catch and effort statistics for white marlin and sailfish from the northern Gulf of Mexico appear to be reliable. With the exception of cases where sample size was inadequate, the data from three nearly independent sources, i.e., dock, tourna- ment, and longline fishing, were consistent over an 8-yr period. It seems likely that if significant biases were present in the data sources, they would have behaved differently over the time series and inconsistencies would have resulted. This consistency over time provides greater confi- dence in each of the individual data sources and enables the pooling of data to form reliable indices of abundance. Although only tournament data are available from areas outside the gulf, a compari- son of these data for white marlin indicates a con- sistency in trends among areas and suggests that catch and effort statistics are providing a reliable means of indexing abundance. We also believe that they provide some evidence that a single stock of white marlin exists throughout the sampling area. 66 BEARDSLEY and CONSER: AN ANALYSIS OF CATCH AND EFFORT DATA 05- .04- 03- .02- .01- 3 a. o 5 'H u u tt .10- OB- 06- .04- .02- NORTHERN GULF OF MEXICO I I WESTERN GULF OF MEXICO — T— 1971 -r- '72 I '74 YEAR T '75 -r- '76 — P- '77 '78 FIGURE 15. — CPUE, in number offish hooked per hour of fishing, from the recreational fishery for sailfish in the northern and western Gulf of Mexico, 1971-78. Although trends in CPUE for blue marlin in the Gulf of Mexico and in the Atlantic are similar for some years, the detailed analysis of data from the gulf indicates that caution should be exercised in interpreting catch and effort statistics for blue marlin. In the gulf, little agreement or consistency could be found among the three data sources, and it appears that the basic catch model is not appro- priate. Some of the competition effect models dis- cussed by Rothschild (1977), in which catchability decreases as effort increases, may be more appro- priate for these data. Rothschild et al. (1970) also demonstrated that the fishing power of various gear types, relative to one another, can change as a function of stock abundance. Declining blue mar- lin abundance during the analysis period (Conser and Beardsley^; Kikawa and Honma^) may have caused this situation to occur in the northern gulf It must be pointed out, however, that white marlin abundance was declining during the same period, but this change in relative fishing power did not occur. ^Conser, R. J., and G. L. Beardsley. 1979. An assessment of the status of stocks of blue marlin, Makaira nigricans, and white marlin, Tetrapturus albidus, in the Atlantic Ocean. Collect. Vol. Sci. Pap. 8(2):461-489. Int. Comm. Conserv. Atl. Tunas, Gen- eral Mola 17, Madrid, Spain. •^Kikawa, S., and M.Honma. 1979. Status ofwhite and blue marlins caught by the longline fisheries in the North Atlantic Ocean. Collect. Vol. Sci. Pap. 8(2):513-515. Int. Comm. Conserv. Atl. Ttmas, General Mola 17, Madrid, Spain. In the analysis of Gulf of Mexico data, the pro- portionality of catchability over time for the vari- ous data sets was examined using correlation analysis of the CPUE's, but the available data did not allow testing the constancy of catchability over time. The results show that the catchability was proportional for all data sets with white marlin and sailfish. Therefore, if any change in catchabil- ity occurred, it would have been in the same direc- tion for all data sets. This appears unlikely, how- ever, since any change in catchability for the recreational fishery would probably have been an increase due to improvements in gear and equip- ment, but an increase in catchability for the longline fishery is unlikely because the fishery has been targeting more on bluefin tuna in recent years, and joint occurrences of billfishes and the tropical tunas tend to be more frequent than joint occurrences of billfishes and the temperate tunas, i.e., bluefin (Fox''). It appears reasonable, there- fore, to assume that catchability has been constant for white marlin and sailfish, but this assumption would be tenuous for blue marlin in the northern Gulf of Mexico. ACKNOWLEDGMENTS We express our sincere appreciation to all.of the anglers, captains, and crew members that partici- pate in the recreational fishery for billfishes for their patience and courtesy in providing the data that form the basis for this report. We also ac- knowledge with gratitude the support of the many charterboat associations, big-game fishing clubs, and tournament committees throughout our sam- pling area. Many of these groups made special arrangements for our samplers so that they could conduct their interviews. We also acknowledge the cooperation of the Government of the Bahamas and the Commonwealth of Puerto Rico and ex- press our thanks for allowing us to sample in their areas. We particularly thank the Florida Depart- ment of Natural Resources, the Georgia Depart- ment of Natural Resources, and the South Carolina Marine Resources Department for sam- pling tournaments in their respective states and providing us with the data. We thank Dade Thornton for his continuing support and assis- tance through the years. This research was sup- ^Fox, W. W. 1971. Temporal-spatial relationships among tunas and billfishes based on the Japanese longline fishery in the Atlantic Ocean, 1956-65. Univ. Miami Sea Grant Program, Sea Grant Tech. Bull. 12, 78 p. 67 FISHERY BULLETIN: VOL. 79, NO. 1 ported by a number of staff members over the years. We particularly acknowledge Paul Pristas of the Southeast Fisheries Center Panama City Laboratory; Chester Buchanan, presently with the U.S. Fish and Wildlife Service in Anchorage, Alaska; Perry Thompson with the Southeast Fisheries Center Pascagoula Laboratory; and Allyn Lopez, Luis Rivas, and Edwin Scott of the Southeast Fisheries Center Miami Laboratory. Edward Houde, William Lenarz, Eugene Naka- mura, and Gary Sakagawa reviewed the manu- script and we are grateful for their careful and thoughtful comments. Grady Reinert prepared all the charts and graphs. LITERATURE CITED BERUDE, C. L., AND N. J. ABRAMSON. 1972. Relative fishing power, CDC 6600, FORTRAN IV Trans. Am. Fish. Soc. 101:133. BEVERTON, R. J. H., AND S. J. HOLT. 1957. On the dynamics of exploited fish populations. Fish. Invest. Minist. Agric, Fish. Food (G.B.), Ser. 2, 19, 533 p. DE SYLVA, D. R 1974. A review of the world sport fishery for billfishes (Is- tiophoridae and Xiphiidae). In R. S. Shomura and F. Williams (editors), Proceedings of the International Bill- fish Symposium, Part 2, p. 12-33. U.S. Dep. Commer., NOAA Tech. Rep. NMFS SSRF-675. FOOD AND AGRICULTURE ORGANIZATION. 1976. Monitoring offish stock abundance; the use of catch and effort data. FAO Fish. Tech. Pap. 155, 101 p. GULLAND, J. A. 1956. On the fishing effort in English demersal fisheries. Fish. Invest. Minist. Agric, Fish. Food (G.B.), Ser. 2, 20(5), 41 p. INTERNATIONAL GAME FiSH ASSOCIATION. 1979. World record marine fishes. 1979 ed. Int. Game Fish. As.soc., Ft. Lauderdale, Fla., 272 p. JOLLEY, J. W, JR. 1977. The biology and fishery of Atlantic sailfish Is- tiophorus platypterus, from southeast Florida. Fla. Mar. Res. Publ. 28, 31 p. 1979. Fishermen help scientists. Underwater Nat. 11(4):15-18. LAURENT, A. G. 1963. The lognormal distribution and the translation method: description and estimation problems. J. Am. Stat. Assoc. 58:231-235. LOPEZ, A. M.. D. B. MCCLELLAN, A. R. BERTOLINO, AND M. D. LANGE. 1979. The Japanese longline fishery in the Gulf of Mexico, 1978. Mar. Fish. Rev 41(10):23-28. MATHER, F J., Ill, A. C. JONES, AND G. L. BEARDSLEY, jR. 1972. Migration and distribution of white marlin and blue marlin in the Atlantic Ocean. Fish. Bull., U.S. 70:283- 298. MURPHY, G. I. 1960. Estimating abundance from longline catches. J. Fish. Res. Board Can. 17:33-40. NAKAMURA, E. L., AND L. R. RiVAS. 1974. An analysis of the sportfishery for billfishes in the northeastern Gulf of Mexico during 1971. In R. S. Sho- mura and F. Williams (editors), Proceedings of the Inter- national Billfish Symposium, Part 2, p. 269-289. U.S. Dep. Commer., NOAA Tech. Rep. NMFS SSRF-675. ROBSON, D. S. 1966. Estimation ofthe relative fishing power of individual ships. Int. Comm. Northwest Atl. Fish. Res. Bull. 3:5-14. ROTHSCHILD, B. J. 1977. Fishing effort, /n J. A. Gulland (editor). Fish popu- lation djTiamics, p. 96-115. Wiley, N.Y. ROTHSCHILD, B. J., G. C. POWELL, J. JOSEPH, N. J. ABRAMSON, J. A. BUSS, AND R ELDRIDGE. 1970. A survey of the population dynamics of king crab in Alaska with particular reference to the Kodiak area. Alaska Dep. Fish Game Inf Leafl. 147, 149 p. TUKEY, J. W. 1949. One degree of freedom for non-additivity Biome- trics 5:232-242. UEYANAGI, S. 1974. A review of the world commercial fisheries for bill- fishes. In R. S. Shomura and F Williams (editors). Pro- ceedings ofthe International Billfish Symposium, Part 2, p. 1-11. U.S. Dep. Commer., NOAA Tech. Rep. NMFS SSRF-675. UEYANAGI, S., S. KIKAWA, M. UTO, AND Y. NISHIKAWA. 1970. Distribution, spawning, and relative abundance of billfishes in the Atlantic Ocean. [In Jpn., Engl, synop.] Bull. Far Seas Fish. Res. Lab. (Shimizu) 3:15-55. WISE, J. R, AND C. W. DAVIS. 1973. Seasonal distribution of tunas and billfishes in the Atlantic. U.S. Dep. Commer., NOAA Tech. Rep. NMFS SSRF-662, 24 p. 68 WESTERN ATLANTIC HAGFISHES OF THE GENUS EPTATRETUS (MYXINIDAE) WITH DESCRIPTION OF TWO NEW SPECIES Bo Fernholm^ and Carl L. Hubbs^ ABSTRACT Recent trawl collections from the continental slopes of the western North Atlantic have yielded three species of the hagfish genus Eptatretus (treated herein) as well as two or three undescribed species of Myxine. Eptatretus is accepted as the generic name for most of the multibranchiate myxinids including all the Atlantic species; Paramyxine Dean 1904, is restricted to western Pacific species. The documentary material of Paramyxine springeri Bigelow and Schroeder 1952, contains two species, one of which is here described and named Eptatretus minor, new species. The two species are sympatric on the continental slope of the northeastern Gulf of Mexico, but appear to occupy relatively narrow, nonoverlapping depth ranges. Eptatretus multidens, new species, is described from the Carib- bean Sea and Atlantic Ocean off French Guiana. The value of tooth counts and the numbers of slime pores is stressed in systematic studies within Eptatretus. Bigelow and Schroeder (1952) described Paramyxine springeri from three specimens caught in 1951 in the Gulf of Mexico. The only recognized species o^ Paramyxine at that time was P. atami Dean 1904 (now known to be a composite of two Japanese species, Fernholm unpubl. data). Other species oi Paramyxine , later described from Taiwan (Teng 1958; Shen and Tao 1975) have strengthened the distinctiveness of that genus by having the generic character of crowded gill (or branchial) apertures even more pronounced than in the type-species. We redescribe P. springeri and refer it to the genus Eptatretus . We also describe two new species o{ Eptatretus and point out the likely occurrence of at least two additional species of this genus from the midwestern Atlantic. We use the name Epta- tretus for the Atlantic hagfishes with several gill apertures to stress that we believe they represent a phyletic line which is independent of, although similar to, that of the Asian hagfishes of the genus Paramyxine. It is more likely that Eptatretus from the Atlantic represents an offshoot from the west- ern American Eptatretus group than that they are directly related to the Asian Paramyxine. That the 'Roskilde University, Department of Biology and Chemistry, Box 260, DK-4000 Roskilde, Denmark. ^Scripps Institution of Oceanography, University of California, San Diego, La Jolla, CA 92093. Carl L. Hubbs died on 30 June 1979. Manuscnpt accepted July 1980. FISHERY BULLETIN: VOL. 79, NO. 1, 1981. western Atlantic species of Eptatretus seem to be restricted to Central American and adjacent waters indicates that they may have crossed be- tween the American continents from the Pacific prior to the appearance of the isthmus. In the western Atlantic there are no records oi Eptatretus outside those shown in Figure 1. On the European and African side of the ocean the only reported captures are those of .B. profundus, E. hexatrema, and E. octatrema, all from South African waters (Barnard 1923, 1950). All specimens treated herein were taken by bot- tom trawl, a method which usually produces few hagfishes. No doubt an expedition with baited traps would provide vastly more material that could fill in some of the gaps in the material we have at our disposal. However, the U.S. govern- ment research vessels (Springer and BuUis 1956; Bullis and Thompson 1965; Bayer 1969) that have secured most of our material give us data which are relatively homogeneous and complete, and thus yield some information on the hagfish habitats. We show that P. springeri Bigelow and Schroeder includes two species, and describe some new forms of Eptatretus. We also mention here that hagfishes of the genus Myxine have been found in the western Atlantic (Hubbs unpubl. data). They are not yet systematically analyzed, but it is expected that they compose two new species. It thus seems likely that the hagfish fauna 69 FISHERY BULLETIN: VOL. 79, NO. 1 45° 30° 0° — FIGURE 1.- -Distribution oiEptatretus species in the western Atlantic Ocean. Numbers indicate more than one specimen taken. Two inferential records are indicated with question marks. Isobaths in meters. of the western Atlantic is about as rich as it is in Japanese waters (cf. Dean 1904). Anticipating more material of Eptatretus from this area, we have chosen to name only those forms that are represented by four or more specimens in the available material. Many hagfish species have been described from one or a very few specimens and this has caused much confusion, especially since subsequent investigators have not been aware of what may be regarded as normal intra- specific variation of characters in different genera of hagfishes. MATERIALS AND METHODS We have examined material from the following repositories: FMNH — Field Museum of Natural History, Chicago; MCZ — Museum of Comparative Zoology, Harvard University, Cambridge; SIO — Scripps Institution of Oceanography, La Jolla, Calif.; UMML— Rosenstiel School of Marine and Atmospheric Sciences, Miami; USNM — National Museum of Natural History, Washington, D.C. De- tails of the studied material are given under each species. Weight for each preserved specimen was re- corded in grams. The following measurements have been taken on the left side of the specimen: Total length (TL): from extreme tip of snout at midpoint, excluding barbels, to rear margin of fin around tip of tail; moderate stretching may be needed to approximate normal form. This process has been used for other measurements. Trunk length: from front of pharyngocutane- ous aperture to front of cloacal slit. Tail length: from front of cloacal slit to tip of tail fin. 70 FERNHOLM and HUBBS: WESTERN ATLANTIC HAGFISHES OF THE GENUS EPTATRETUS Prebranchial length: from tip of snout to front of anteriormost gill aperture. Branchial length: from front of anteriormost gill aperture to front of pharyngo-cutaneous aper- ture. Preocular length: from tip of snout to center of clear area marking ocular region. Body width: maximum, with body molded into seemingly natural conformation. Body depth: maximum overall, near middle of body, including ventral fin fold (if applicable). Body depth excluding fin fold: same region as in body depth, but excluding fin fold. Body depth over cloaca: over front of cloacal slit. Tail depth: maximum, taken at right angle to local axis, including fins. Length of each of the three barbels: from crease at outer-anterior base. The following counts have been taken: Cusps on multicuspids (Figure 2): outer/inner row of teeth on both sides. Usually the head is cut open ventrally to count the teeth. Unicuspids, outer row (including smaller cusps). Unicuspids, inner row (including smaller cusps). Total sum of all cusps. Prebranchial (tip of snout to front of anterior- most gill aperture) slime pores (left side). Branchial slime pores (left side). Trunk slime pores (left side). Tail slime pores (left side). 7 ^ 11 m •*i •^m^ I^HH^I rswH ^^1 ([ ■i f J ■m ^ ^m ^m Figure 2.— Multicusps of Eptatretus springeri by scanning electron microscopy. A— right outer tooth row, B— right inner tooth row to show pattern of fused teeth 3/2 ( see text) . Multicusps ofE. minor. C —right outer tooth row, D— right inner tooth row to show pattern of fused teeth 3/3. Scale indicated in B is 2 mm for all figures. 71 FISHERY BULLETIN: VOL. 79, NO. 1 Total sum of slime pores on left side. Total sum of gill apertures (not counting the pharyngo-cutaneous aperture if the left pos- teriormost gill aperture opens separately). Many specimens were cut open to determine the number of gill pouches and their position relative to the tongue muscle and branching of aorta. Sex was determined by examining the gonad through a cut in the lateral right side body wall anterior to the cloaca. The usefulness of different counts and mea- surements for systematic studies in Paramyxine and Eptatretus have been discussed (Dean 1904; Bigelow and Schroeder 1952; Strahan 1975). We agree with Dean's (1904) statement that ". . . in the case of myxinoids it is peculiarly necessary to base specific determinations upon the average charac- ters of as great a number of individuals as practica- ble." Dean (1904) and Strahan (1975) stressed the importance of the relative position of the gill aper- tures and body proportions, which we also find useful. Dean (1904) tended to disregard the number of slime pores as a systematic character, but we stress the value of that count and point out that both Bigelow and Schroeder (1952) and Stra- han (1975) arrived at ranges for this character which are far too broad because they include two composite species (P. springeri Bigelow and Schroeder andP atami Dean). This inclusion of an undescribed species {E. minor, see below) in P. springeri also led to the erroneous conclusion (Bigelow and Schroeder 1952) that the number of slime pores increased with length of the animal. As indicated earlier (Dean 1904; Strahan 1975), the gill aperture counts vary slightly within the five- to seven-gilled species of Eptatretus , hut when a larger sample is available, the character is, of course, quite valuable and is easily examined. Tooth counts were considered of little value by Dean (1904) and Strahan (1975). We find, however, as had Regan (1912) in his admirable yet terse synopsis of the multibranchiate myxinids, that this is a very useful character. Especially we find the pattern of fused cusps (Figure 2) in the inner and outer row of teeth to be constant within species in all of the hundreds of specimens of Epta- tretus and Paramyxine we have studied (unpubl. data). The movements of the rasping lingual tooth plates are rather elaborate in hagfishes and com- plicate the terminology used to express positional relationships. In agreement with some previous authors (Dean 1904; Strahan 1975), we have cho- sen to call the row of teeth with larger teeth the outer row and the other row the inner. The fused tooth or multicusp we regard as the anterior in each row, and to designate the pattern of fused cusps or teeth in the multicusps of the outer/ inner row we write 3/2 or 3/3, which are the two patterns found in the Atlantic species of Eptatretus. The shape of the gill apertures, as well as exten- sion and shape of the relatively low ventral fin fold, have been used as systematic characters, but we find it difficult to assess these characters in pre- served specimens. The pattern of fused teeth (3/2 or 3/3), supplemented with counts of gill apertures and slime pores, appears to suffice to distinguish the western Atlantic species of Eptatretus. GENERIC ALLOCATION The generic allocation of the polybranchiate species of hagfishes has been considerably dis- cussed (for summary, see Holly 1933 and Strahan 1975). To us it is obvious that the name with prior- ity, Eptatretus Cloquet, 1819, should be used. As stressed earlier (Bigelow and Schroeder 1948; Adam and Strahan 1963; Hubbs 1963; Strahan 1975), there is no obvious advantage in dividing the genus into subgenera. It has been argued that Paramyxine as a genus should be treated as a junior synonym of Eptatre- tus (Strahan 1975). It is true that tendencies to shortening of the gill aperture area can be found in E. springeri (Bigelow and Schroeder 1952) and in E. burgeri (Strahan 1975), but we believe this rep- resents a convergent trend of development in the Atlantic and Pacific Oceans. We retain the generic name Paramyxine for the western Pacific species. Dean (1904) defined the genus Paramyxine on the basis of a single specimen of P. atami. Now that much more material is available, it is somewhat a matter of choice whether one wants to retain Paramyxine for the Asian species. We choose to do so for the following reasons: 1) Dean's concept of crowdedness of gill apertures has been strengthened by the description of Taiwanese species of Paramyxine (Teng 1958; Shen and Tao 1975), which are more extreme in this character than is the type-species; 2) as a further charac- teristic of the many Asian Paramyxine species, we point to the absence of slime pores in the branchial area (with the exception that in P. atami Dean and P cheni Shen and Tao, there may be a single pair of 72 FERNHOLM and HUBBS: WESTERN ATLANTIC HAGFISHES OF THE GENUS EPTATRETUS slime pores in the branchial area); 3) Paramyxine defined in this way is a geographically, and we believe phylogenetically, distinct group limited to the waters of southeastern Asia. Key to the Western Atlantic Species of Eptatretus la. Three anterior teeth in outer row and two anterior teeth in inner row fused at bases 2 lb. Three anterior teeth in each row fused at bases 4 2a. Gill apertures 6 or 7. Body and head stout (Figures 3,4) 3 2b. Gill apertures 5. Body thin. Head narrow. One specimen, 308 mm (Figure 5). South of Bahama Islands sp. B 3a. Slime pores 84-92. Maximum known length 590 mm. Northeastern Gulf of Mexico .. E. springeri 3b. Slime pores 78. One specimen, 433 mm. North of Bahama Islands sp. A 4a. Gill apertures 7 E. multidens? 4b. Gill apertures 6 (rarely 5) 5 5a. A thin whitish middorsal stripe. Total cusp count of teeth 46-54. Maximum known size 395 mm. Northeastern Gulf of Mexico E. minor 5b. No whitish middorsal stripe. Total cusp count of teeth 52-58. Caribbean Sea and Atlantic Ocean off French Guiana and Haiti 6 6a. Slime pores 75. Total cusp count of teeth 58. One specimen 380 mm. North of Haiti sp. C 6b. Slime pores 87-91. Total cusp count of teeth 52-57. Maximum known length 655 mm. Caribbean Sea and Atlantic Ocean off French Guiana E. multidens CM I ' ' I ' I ' I ' I I I I I I I I I I 1 CM 2 3 4 5 6 7 8 9 10 II 12 13 14 15 FIGURE 3. —Eptatretus springeri (MCZ 39939). 73 FISHERY BULLETIN: VOL. 79, NO. 1 DESCRIPTION OF SPECIES Eptatretus springeri (Bigelow and Schroeder) Figure 3, Table I Paramyxine springeri Bigelow and Schroeder 1952:1-10 (in part; original description; holo- type, 590 mm, and an additional specimen, 505 mm; comparison with P. atami; includes, as do the following references, one 338 mm specimen of E. minor). Teng 1958:5-6 (comparison with other species referred to Paramyxine) . Lindberg and Legeza 1959:23-24 (gill apertures). Stra- han and Honma 1961:323-341 (comparison with P. atami and P. yangi). Adam and Strahan 1963:7 (characters; size; Gulf of Mexico). Lindberg and Legeza 1959:19, 21 (gill apertures). Rass 1971:18 (Gulf of Mexico). Material.— MCZ 37399,^ 1 (505 mm), 29 Sep- 3" Additional material" of Bigelow and Schroeder (1952). La- belled as coming from Oregon station 321; 29''27' N, 87°19' W, 400 m, 28 April 1951. In the original description stated to have come from the same trawl haul as the holotype and so regarded here. tember 1951, 1340-1540 h, Oregon station 489, 27°44 ' N, 85°09 ' W, depth 465 m, bottom tempera- ture 10.3° C, bottom type blue mud; MCZ 39939, 3 (500, 509, 576 mm), 13 March 1955, 0835-1305 h, Oregon station 1282, 29°10 ' N, 88°03 ' W, depth 475 m, bottom temperature 10° C, bottom type gray mud; MCZ 42423, 3 (410, 433, 450 mm), 12 March 1962, Oregon station 4076, 28°33' N, 86°27' W, depth 460 m; SIO 76-248 (formerly UMML 4405), I (542 mm), 18 February 1956, 0805-0835 h, Ore- gon station 1450, 29°17 ' N, 87°41 ' W, depth 440 m; USNM 161512 (holotype), 1 (590 mm), 29 Sep- tember 1951, 1340-1540 h, Oregon station 489, 27°44' N, 85°09' W, depth 465 m, bottom temper- ature 10.3° C, bottom type blue mud; USNM 188210, 1 (522 mm), 23 October 1962, Oregon sta- tion 4005, 29°07.5' N, 88°09' W, depth 550 m; USNM 218396, 2 (513, 526 mm), 4 February 1970, Oregon II station 10899, depth 550 m; USNM 218397, 2 (500, 540 mm), 5 February 1970, Oregon II station 10900, 28°50.2' N, 86°59' W, depth 730 m; USNM 218394, 1 (417 mm), 29 August 1970, Oregon II station 11192, 29°19 ' N, 86°45 ' W, depth 420-460 m; USNM 218395, 1 (418 mm), 1 Sep- tember 1970, Oregon II station 11204, 29°12' N, 87°55'W, depth 550 m. Table l. — Characters of western Atlantic species of Eptatretus with 3/2 cusps (see text) on multicuspids of outer/inner row of teeth. Mean ± SD and range given for multiple specimens. E. springeri 16 specimens E. sp. A E. sp. B Item including holotype Holotype 1 specimen 1 specimen Depth of capture (m) 420-730 465 950 590 Total length, TL (mm) 496±53.9 410-590 590 433 308 Weight (g) 231 ±97.3 102-479 479 182 40 Measurements in thousandths of TL: Preocular length 53±5.6 47-64 61 53 — Prebranchial length 243±13.3 215-268 243 215 250 Branchial length 37±7.9 25-56 39 79 39 Trunk length 568 ±17.5 529-61 1 563 550 568 Tail length 156 ±9.8 134-168 155 157 146 Body width 52±6.8 42-70 70 55 37 Body depth: Including fin fold 81 ±8.1 66-99 99 88 67 Excluding fin fold 77±9.4 62-97 97 85 62 Over cloaca 67±7.5 51-77 69 74 48 Tail depth 81 ±10.3 64-93 85 81 67 Barbel length: First 9 + 1.4 7-11 11 15 10 Second 10±1.8 7-14 7 16 12 Third 15±2.8 10-20 10 16 15 Counts: Teeth: Cusps on multicuspids 3/2 3/2 3/2 3/2 Unicuspids, outer row' 10-11 11 -1- 11 10 + 11 10 + 10 Unicuspids, inner row' 9-11 10-1- 10 10 -1- 10 10 -1- 11 Total sum of cusps 50+1.4 48-52 52 51 51 Slime pores, leftside: Prebranchial 18±1.1 16-19 19 14 18 Branchial 3.2±0.9 2-5 4 4 4 Trunk 54±1.8 52-57 55 48 48 Tail 11±1.1 9-13 13 13 11 Total sum 87±2.6 84-92 91 79 81 Gill apertures' 12±0.4 12-13 6-1-6 7 4-7 54-5 'Left -I- right count for single specimen. 74 FERNHOLM and HUBBS: WESTERN ATLANTIC HAGFISHES OF THE GENVS EPTATRETUS Diagnosis. — An Eptatretus with six (rarely seven) gill apertures. Total cusp count 48-52, with three teeth fused in outer and two in inner row of teeth (Figure 2). Slime pores 84-92. Description. — E. springeri is a large hagfish; the 16 available specimens range from 410 to 590 mm with the mean about 500 mm. The five sexually mature females were 500 mm or longer. Eggs in the most mature female, 526 mm, are about 10 x 36 mm. Only 1 of our 16 specimens had the sixth left gill aperture opening separately in front of the pharyngo-cutaneous aperture (cf. E. burgeri with 10% of the animals showing this condition accord- ing to Dean 1904). Two animals had an extra gill aperture on the right side and one an extra on the left side. Of these three, two had an extra seventh gill pouch on the right side and one had an extra seventh pouch on each side; the extra pouches were all more or less reduced in size. The tongue muscle typically overlies gill pouches 1-3, and the aorta divides between gill pouches 4 and 5. The color of our specimens is dark brownish purplish to very light brown; the eyespots are not plainly visible. Distribution. — Eptatretus springeri has been found only in the northeastern part of the Gulf of Mexico (Figure 1) at depths between 410 and 576 m, but it must be realized, of course, that inciden- tal capture (by trawling) is hardly adequate for our distributional map. The southernmost record is that of the holotype (USNM 161512). Habitat and biology. — Specimens were collected by bottom trawl in March (475 m) and September (465 m) where the habitat temperature was about 10° C. It could not be determined whether the animals were in or above the substrate, which seemed to be composed of blue or gray mud. At least six of the specimens were caught during day- time. Some females contained ripe eggs (21-41 FIGURE 4.— Eptatretus sp. A (MCZ 40370). Scale in centimeters. 75 FISHERY BULLETIN: VOL. 79, NO. 1 mm) in February, March, and September; thus the population presumably spawned throughout the year. Eptatretus species A and B Figures 4 and 5, Table 1 Matena/.— Species A, MCZ 40370, 1 (433 mm), 9 June 1958, 1925-2225 h. Silver Bay station 445, 28°03 ' N, 78°44 ' W, depth 910-950 m, bottom type coral and sand; Species A (inferential; the speci- men cannot be located), 21 June 1958, 0730-1030 h. Silver Bay station 490, 29°49 ' N, 80°11 ' W, 330 m, bottom type green mud. Species B, SIO 76-252 (formerly UMML 31521), 1 (308 mm), 27 Sep- tember 1973, RV Columbus Iselin station 137, 26°07' N, 78°34.1-36.6' W, depth 590-560 m. Two specimens oi Eptatretus (herein provision- ally designated A and B) have been taken at depths of 950 and 590 m, respectively, in the vicin- ity of Grand Bahama Island (Figure 1). The pat- tern of fused teeth is 3/2 in these specimens and the total number of cusps is essentially the same. They exhibit differences large enough that they cannot be convincingly included in E. springeri. There are differences in the numbers of gill pouches and apertures, the relative branchial length, and number of prebranchial, trunk, and total slime pores (Table 1). Although there is vari- ability in the number of gill pouches, and speci- mens having one more or less pouch are found among the six-gilled species oi Eptatretus , it seems unlikely that our five- and seven-gilled specimens from south and north of Grand Bahama Island represent a single species with normally six gill pouches. The body width, the relative depth over the cloaca, and counts of prebranchial slime pores (Table 1) also indicate specific distinction for these two Atlantic Eptatretus specimens. Species A, a mature male, is considerably stout- er than the thin specimen designated as species B and is about as stout as E. springeri. The tail is less flaring and more pointed than that ofE. springeri. The skin is light brown overall wdth the ventral fin fold whitish. Patches of transparent skin overlie the eyes. Trunk and total slime pore counts are outside the range of E. springeri in this seven- gilled specimen (Table 1), but the internal miiiim= ■-•--■*'• --!*> 4 5 Figure 5.— Eptatretus sp. B (SIO 76-252). Object dependent from slit is an egg. Scale in centimeters. 76 FERNHOLM and HUBBS: WESTERN ATLANTIC HAGFISHES OF THE GENUS EPTATRETUS anatomy is similar: the tongue muscle overlies gill pouches 1-3, and the aorta divides between gill pouches 5 and 6. Species B is represented by a five-gilled female having a slender body and narrow head. The skin is light pinkish tan, with the ventral side only slightly lighter; no eyespots are visible. The thin ventral fin fold is white, and extends forward from the cloaca, reaching its maximum height at about the middle of the body, and gradually tapers off toward the posterior part of the branchial region. Several small depressions in the skin, about 0.4 mm in diameter, located mostly in the head region, may be traces of ectoparasitic trematodes. Species B differs from E. springeri in internal anatomy, having a tongue muscle overlying only the first gill pouch and an aorta bifurcating between the second and third gill pouch. Eptatretus profundus (Barnard 1923) is the only described species of Eptatretus having five gill apertures. Unfortu- nately, only the holotj^ie is extant. It was mea- sured by Hubbs (figures in parentheses, below) in the South African Museum (no. 13035) and was found not to differ much in length proportions from species B, but it was clearly stouter: body width (thousandths of total length), 37 (63); body depth including fin fold, 67 (94); body depth excluding fin fold, 62 (91); body depth over cloaca, 48 (68). The South African species yielded a lower total tooth count, 51 (42), but a similar number of slime pores, 81 (84). The difference in tooth count and stoutness indicate that the two specimens probably are not conspecific. Eptatretus species C Figure 6, Table 2 Material.— \5S^M 218400, 1 (380 mm), 13 October 1963, Silver Bay station 5146, 19°55.5' N, 72°00' W, depth 860-910 m. This six-gilled specimen, from off Haiti, is a female with eggs 2-3 mm long, apparently in quite early stages of development. The pattern of fused cusps, 3/3, and several other characters indicate relationship to E. minor (described below). It dif- fers from that species, however, in having a longer tail, shorter branchial length, greater body depth, slightly higher tooth count, and a lower prebran- chial slime-pore count. It is similar toE. multidens (described below), but differs particularly in hav- CM I M ' I M ' I ' 1 23456789 10 II CM 12 13 14 15 FIGURE 6.— Eptatretus sp. C (USNM 218400). 77 FISHERY BULLETIN: VOL. 79, NO. 1 Table 2.— Characters of western Atlantic species of Eptatretus with 3/3 cusps (see text) of multicuspids of outer/inner row of teeth. Mean ± SD and range given for multiple specimens. E. minor n. SD. E. multldens n. sp. E. multldens? 17 specimens Holo- 4 specimens Holo- MCZ USNM Item including holotype type including holotype type 40409 218405 E. sp. Depth of capture (m) 300-400 370 510-770 510 500 365 910 Total length, TL (mm) 330±47.8 223-395 359 526± 125.3 377-655 600 331 364 380 Weight (g) 85±33.5 22-138 107 494 ±302.1 164-757 561 84 128 154 Measurements in thousandths of TL: Preocular length 55±8.8 31-62 59 46±3.1 43-49 43 — 61 59 Prebranchial length 243±14.3 201-259 245 200±8.4 188-207 202 214 236. 237 Branchial length 59±7.6 51-72 71 65±4.0 61-69 62 73 78 47 Trunk length 529±15.1 506-559 522 560±8.0 552-571 560 517 504 526 Tail length 165±14.1 139-183 162 179±8.0 169-188 182 196 181 190 Body width 61 ±10.6 48-78 62 45±5.7 39-50 39 48 77 57 Body depth Including fin fold 94±12.3 71-114 92 104±16.0 80-115 109 97 114 100 Excluding fin fold 89+10.5 71-108 88 102±16.0 78-113 107 — 105 99 Over cloaca 69±9.1 52-79 74 72±8.8 60-81 81 69 82 84 Tail depth 82 + 14.8 53-116 84 76±8.7 66-86 79 97 107 79 Barbel length: First 17±2.8 13-23 16 12+3.4 8-15 15 14 16 13 Second 18±3.3 13-25 22 14+0.6 13-14 13 19 17 14 Third 25+4.7 14-32 30 18 + 2.3 15-20 18 21 24 21 Counts: Teeth: Cusps on multicuspids 3/3 3/3 3/3 3/3 3/3 3/3 3/3 Unicuspids, outer row' 8-11 9-F8 11-12 12+11 11 + 12 12 + 13 12 + 12 Unicuspids, inner row' 8-10 9 + 9 9-11 11 + 11 10 + 11 10+11 11 + 11 Total sum of cusps 50±2.7 46-54 47 55±2.4 52-57 57 56 58 58 Slime pores, leftside: Prebranchial 16±1.0 15-18 15 15±1.0 14-16 16 13 15 13 Branchial 5.0±0.4 4-6 5 5.5±0.6 5-6 5 6 6 4 Trunk 45±2.1 41-48 48 54±1.5 52-55 55 50 47 44 Tail 12.4±0.7 11-14 12 15±0.0 15 15 12 11 14 Total sum 78±2.6 74-82 80 89±1.8 87-91 91 81 79 75 Gill apertures' 11.9±0.5 10-12 6-1-6 12±0.0 12 6 + 6 7 + 7 7 + 7 6 + 6 ' Left + right count for single specimen. ing a low total slime-pore count and also in having shorter trunk and branchial lengths, but a longer prebranchial length. Until more material can be examined, it seems desirable to delay the designa- tion of this specimen as a new species. Species C is light brown with plainly visible lighter patches on the skin overlying the eyes. The tongue muscle overlies the three or four anterior- most gill pouches and the aorta divides between gill pouches 5 and 6. Eptatretus minor, new species Figure 7, Table 2 Paramyxine springeri (in part). — Bigelow and Schroeder 1952:1-10 (the 338 mm long specimen in "additional material" stated to have come from Oregon station 321, which was erroneously listed with lat. 27°27' N; correct latitude is 29°27' N). Springer and Bullis 1956:40 (survey records). Bullis and Thompson 1965:17 (survey records). Holotype: USNM 164119, a female 359 mm long, with eggs 9 mm long, from Oregon station 1009, 24°34' N, 83°34' W, 370 m, 14 April 1954, 0450- 78 0730 h; surface temperature 24.4° C, bottom temperature 11.7° C; bottom material white coral and mud; in 12 m (40-ft) shrimp trawl. Paratypes: USNM 218399, two females, 307 and 334 mm, taken with the holotype. Other matenol: FMNH 59959, 1 (340 mm), 13 April 1954, 1815-2110 h, Oregon station 1006, 24°20' N, 83°20' W, depth 350 m, bottom temperature 10.6° C, bottom type coral and mud; FMNH 65817, 2 (355, 370 mm), 14 October 1959, 1805-2150 h, Ore- gon station 2670, 24°26 ' N, 83°24 ' W, depth 390 m, bottom type mud; MCZ 38707, 1 (313 mm), 19 April 1954, 2115-2300 h, Oregon station 1026, 25°08 ' N, 84° 19 ' W, depth 300 m, bottom temperature 10° C, bottom type sand and gravel; MCZ 40679, 1 (306 mm), 7 June 1959, 1940-2140 h, Silver Bay station 1189, 24°20.5' N, 83°25' W, depth 300 m, bottom type mud and sand; MCZ 51084, 1 (355 mm), 23 November 1963, Oregon station 4529, 24°31' N, 83°26' W, depth 390 m, bottom temperature 10.2° C; SIO 76-251, 3 (223, 223, 310 mm), 7-8 June 1959, 2245-0045 h, Silver Bay station 1190, 24°28' N, 83°34 ' W, depth 330 m, bottom type mud and sand; SIO 76-249, 1 (341 mm), 26 July 1963, Oregon FERNHOLM and HUBBS: WESTERN ATLANTIC HAGFISHES OF THE GENUS EPTATRETUS Figure 7. — Eptatretus minor, holotype (USNM 164119). 79 FISHERY BULLETIN: VOL. 79, NO. 1 Station 4338, 24°18' N, 83°18' W, depth 380 m, bottom temperature 8.9° C; SIO 76-250 (formerly UMML 15042), 2 (385, 395 mm), 29 July 1963, Oregon station 4346, 24°28' N, 83°29' W, depth 380 m; USNM 161513,^ 1 (332 mm), 28 April 1951, 1630-1817 h, Oregon station 321, 29°27' N, 87°19' W, depth 400 m, bottom temperature 10° C; USNM 218398, 1 (358 mm), 23 June 1969, Oregon II sta- tion 10643, 29°30' N, 87°09' W, depth 400 m. Diagnosis.— An Eptatretus with six (rarely five) gill apertures. Total cusp count 46-54, with three teeth fused in both outer and inner rows of teeth. Slime pores 74-82. A thin whitish middorsal stripe. Description. — This is a relatively short and stout species of Eptatretus, maturing at a small size (none of our 17 specimens exceeds 400 mm). Our shortest specimens are two sexually mature males, 223 mm each, and a ripe female, 310 mm, extremely swollen with 12 eggs, each measuring about 10 X 31 mm. An inconspicuous ventral fin fold begins well behind the last gill aperture and extends backward to the cloaca. The tongue mus- cle overlies the first three or four gill pouches and the aorta branches between pouches 5 and 6. Eptatretus minor and E. springeri are sympatric in the northeastern Gulf of Mexico (Figure 1). There are important differences between them. The pattern of fused teeth is 3/3 inE. minor and 3/2 in E. springeri (Figure 2). There may be a difference in the relative trunk lengths, E. springeri being longer. This difference is reflected in the nonoverlapping counts of trunk and total slime pores. The relative length of the branchial region tends to be greater inE. minor. In preserved material £J. minor is usually pale in color, while £. springeri is darker. From the material available it appears that E. minor lives at shallower depths, 300-400 m, than E. springeri, 420-730 m. The thin, light middorsal stripe, evident on most species ofE. minor, and the conspicuously long, laterally protruding barbels may be good field characters for that species. In contrast with other field species of Eptatretus, neither £. springeri nor E. minor has a conspicuous lighter patch of skin overlying the eye. Distribution. — All but 2 of our 17 specimens are from the Dry Tortugas grounds in the archipelago ••"Additional material" of Bigelow and Schroeder (1952) and labelled paratype of Paramyxine springeri. 80 extending westerly from the Florida Keys (Figure 1). The distributional pattern may be due, at least to some extent, to the massive exploratory trawl- ing activities carried out by federal research ves- sels to monitor the population of the royal red shrimp, Pleoticus robustus, in that area. The two records outside this area are the 338 mm specimen described by Bigelow and Schroeder (1952) and one of 358 mm from Oregon II station 10643. These two northernmost records indicate an overlap in distribution of E. minor and E. springeri in the northeasternmost part of the Gulf of Mexico. The depth ranges, however, do not overlap. Habitat and biology. — The bottom temperature in the area of E. minor is about 8.9°-11.5° C, and although two stations list coral as bottom type, they also include mud, where the hagfish probably were caught. One station lists sand and gravel as bottom type which would be less suitable for the animal ifE. minor usually digs into a muddy bot- tom, as does E. burgeri (Fernholm 1974). The specimen from that station might have been caught when swimming above the substrate, as is likely, since the trawl haul {Oregon 1026) was taken during the night when hagfish tend to be most active (Fernholm 1974). An indication of in- creased night activity may also be found in the fact that the only two hauls that took three or more specimens [Oregon 1009 and Silver Bay 1190) were taken at night or in early morning. Some females contained ripe eggs (25-33 mm) in April, June, July, September, October, and November; thus the population presumably spawned throughout the year. Considering the economic importance of the shrimp fishery in the areas where E. minor and E. springeri occur, and recent suggestions that Myxine glutinosa is an active predator on pandalid shrimps (Shelton 1978), a detailed investigation on occurrence and ecology of these hagfishes might be of value. Etymology. — The name minor, small, refers to the small size of mature specimens in our samples of E. minor, as compared with those ofE. springeri. Eptatretus multtdens, new species Figure 8, Table 2 Holotype: USNM 218401, a male, 600 mm long, from Oregon II station 11299, 12°52 ' N, 70°43 ' W, 510 m depth, 23 November 1970. FERNHOLM and HUBBS: WESTERN ATLANTIC HAGFISHES OF THE GENUS EPTATRETUS FIGURE 8.—Eptatretus multidens, holotype (USNM 218401) Paratypes: USNM 218404, 1 (655 mm), 12 May 1969, Oregon II station 10611, 7°13' N, 52°52' W, depth 770 m; USNM 218403, 1 (473 mm), 19 November 1969, Oregon II station 10804, 7°18' N, 52°56 ' W, depth 710-630 m; USNM 218402, 1 (377 mm), 23 November 1970, Oregon II station 11300, depth 550 m. Diagnosis. — A six-gilled Eptatretus with three fused teeth in each row (3/3). Total cusp count 52-57. Slime pores 87-91. No middorsal light stripe. Description. — This is a large, deep-bodied hagfish. The color of the four preserved specimens varies from pale brown to medium brown. The paler specimens may have been bleached during preser- vation, as the wrinkled tail is much darker along the creases. On the paler specimens the eyespots can barely be discerned, while on the darker ani- mals the light skin overlying the eyes is clearly visible. The holotype appears bleached on the right anterior part of the body which renders the eyespot visible only on the left side. A small ven- tral fin fold of noncontrasting brownish color is evident from the middle of the body to the cloaca. The tongue muscle overlies the first two or three gill pouches, and the aorta branches at the level of the sixth posteriormost gill pouch. When E. multidens is compared with E. spring- eri, noticeable differences in addition to the im- portant pattern of fused teeth are the longer tail and branchial area but shorter prebranchial ofE. multidens. These differences in body proportions are reflected in the higher slime-pore counts in tail and branchial areas and the lower prebranchial counts ofE. multidens. If E. multidens is compared with E. minor, to which it may be closely related by having the common pattern of fused teeth, it is found that the differences in body proportions are not pro- nounced. The best definitive characters to sepa- rate these two species seem to be the nonoverlap- ping trunk or total slime-pore count and the more noticeable eyespot in E. multidens. Other than E. springeri and E. minor, the only Atlantic Eptatretus with six gills is the probably endemic E. hexatrema from South Africa. This is a 81 FISHERY BULLETIN: VOL. 79, NO. 1 large species having fused teeth 3/2 and a high number, 91-105, of sUme pores (Strahan 1975), seemingly above the range of the western Atlantic species. Distribution. — Records of this species along the northern coast of South America indicate it to be widespread in both the Caribbean and the Atlan- tic. Etymology. — The name multidens is derived from the Latin mult(us), many, and dens, tooth, in ref- erence to the high tooth count in this species. Eptatretus multidens? (Figure 9). — Two speci- mens, MCZ 40409 (331 mm, 23 August 1957, Ore- gon station 1886, 16°55 ' N, 81°12 ' W, depth 500 m, bottom type gray clay) and USNM 218405 (for- merly Department of Biology, University of Panama, no. 523, 364 mm, 5 July 1972, chartered commercial trawler Canopus, locality between Nicaragua and the Colombian border, depth 365 m), obviously conspecific, are similar to E. multi- dens in tooth count and pattern of fused teeth (3/3). A third record where the specimen cannot be located (MCZ 40218, 16 September 1957, 2100- 2200 h, Oregon station 1945, depth 460-550 m) is inferred to belong to the same species. There are, however, important differences in counts of gill apertures, 7-1-7, instead of 6 + 6 as in E. multi- dens, and the slime pores seem to be fewer (Table 2). Also, the prebranchial, branchial, and tail pro- portions are longer whereas the trunk appears to be shorter. Each of the two specimens has seven pairs of gill pouches. The tongue muscle overlies three or four gill pouches in the USNM specimen and five in the MCZ specimen. In each the aorta branches at the level of the seventh gill pouch. It is our belief that when more specimens be- come available it may be necessary to establish this form as a separate species. Until then it is convenient merely to indicate its relationship to E. multidens. I 2 3 4 5 6 7 8 9 10 II 12 13 14 15 FIGURE 9.— Eptatretus multidens? (MCZ 40409). Scale in centimeters. 82 FERNHOLM and HUBBS: WESTERN ATLANTIC HAGFISHES OF THE GENUS EPTATRETUS ACKNOWLEDGMENTS We are particularly grateful to Charles E. Daw- son, Gulf Coast Research Laboratory Museum; Richard H. Goodyear, University of Panama; Rolf Juhl, Southeast Fisheries Center, Pascagoula; Luis Howell Rivero, Miami; C. Richard Robins, UMML; and Robert Schoknecht, MCZ, for making the hagfish material available for this study. For help with hagfish measurements and valuable discussions on hagfish taxonomy we are greatly indebted to Charmion B. MacMillan. J0rgen Nielsen, Richard H. Rosenblatt, and Robert L. Wisner critically read the manuscript and offered valuable suggestions. Elizabeth N. Shor typed the final manuscript. The Swedish-American Foundation and the Danish Natural Science Research council made it financially possible for the senior author to con- centrate on this work when visiting Scripps In- stitution of Oceanography. All this help is grate- fully acknowledged. LITERATURE CITED ADAM, H., AND R. STRAHAN. 1963. Systematics and geographical distribution of Myxinoids. In A. Brodal and R. Fange (editors). The bi- ology of Myxine, p. 1-8. Universitetsforlaget. Oslo, Nor- way Barnard, K. H. 1923. Diagnoses of new species of marine fishes from South African waters. Ann. S. Afr. Mus. 13:439-445. 1950. A pictorial guide to South African fishes. Marine and freshwater. Bailey Bros. & SwirfenLtd.,Lond.,p. 1-2. Bayer, F. M. 1969. A review of research and exploration in the Carib- bean Sea and adjacent waters. In Symposium on investi- gations and resources of the Caribbean Sea and adjacent regions. FAO Fish. Rep. 71, 1:41-91. BIGELOW, H. B., and W C. SCHROEDER. 1948. Cyclostomes. /n Fishes of the western North Atlan- tic, Part one, p. 29-58. Mem. Sears Res. Found. Mar. Res., Yale Univ. 1. 1952. A new species of the cyclostome genus Paramyxine from the Gulf of Mexico. Breviora 8:1-10. BULLIS, H. R., JR., AND J. R. THOMPSON. 1965. Collections by the exploratory fishing vessels Ore- gon, Silver Bay, Combat and Pelican made during 1956- 1960 in the southwestern North Atlantic. U.S. Fish Wildl. Serv., Spec. Sci. Rep. Fish. 510, 130 p. CLOQUET, H. 1819. Dictionnaire des Sciences Naturelles, Paris 15:134- 136. DEAN, B. 1904. Notes on Japanese myxinoids. A new genus Paramy- xine and a new species Howea okinoseana. Reference also to their eggs. J. Coll. Sci., Imp. Univ. Tokyo 19(2), 23 p. FERNHOLM, B. 1974. Diurnal variations in the behaviour of the hagfish Eptatretus burgeri. Mar. Biol. (Berl.) 27:351-356. Holly, M. 1933. Cyclostomata. Das Tierreich 59, 62 p. HUBBS, C. L. 1963. Cyclostome. Encycl. Br. 6:941-944. LINDBERG, G. U., AND M. I. LEGEZA. 1959. Ryby Yaponskogo morya i sopredel'nykh chastei Okhotskogo i Zheltogo morei. (Fishes of the Sea of Japan and the adjacent areas of the Sea of Okhotsk and the Yellow Sea. Part 1. Amphioxi, Petromyzones, Myxini, Elasmbbranchii, Holocephali.) Izd. Acad. Nauk SSSR. Mosk., Leningrad. (Translated by Isr. Program Sci. Transl., 1967, 198 p.; available U.S. Dep. Commer., Natl. Tech. Inf. Serv., Springfield, Va., as TT 67-51392.) RASS, T S. 1971. Animal life. Fishes. [In Russ.] 4:15-18. Moscow. REGAN, C.T 1912. A sjmopsis of the myxinoids of the genus Heptatretus or Bdellostoma. Ann. Mag. Nat. Hist.,Ser. 8, 9:534-536. SHELTON, R. G. J. 1978. On the feeding of the hagfish Myxine glutinosa in the North Sea. J. Mar. Biol. Assoc. U.K. 58:81-86. SHEN, S. C, AND H. J. TAO. 1975. Systematic studies on the hagfish (Eptatretidae) in the adjacent waters around Taiwan with description of two new species. Chin. Biosci. 11:65-78. Springer, S., and h. r. bullis, Jr. 1956. Collections by the Oregon in the Gulf of Mexico. List of crustaceans, mollusks, andfishes identified from collec- tions made by the exploratory fishing vessel Oregon in the Gulf of Mexico and adjacent seas 1950 through 1955. U.S. Fish Wildl. Serv., Spec. Sci. Rep. Fish. 196, 134 p. STRAHAN, R. 1975. Eptatretus longipinnis, n. sp., a new hagfish (family Eptatretidae) from South Australia, with a key to the 5-7 gilled Eptatretidae. Aust. Zool. 18:137-148. STRAHAN, R., AND Y HONMA. 1961. Variation in Paramyxine, with a redescription oi P. atami Dean and P. springeri Bigelow and Schroeder. Bull. Mus. Comp. Zool, Harv Univ 125:323-342. TENG, H. L. 1958. A new cyclostome from Taiwan. [In Chin.] China Fish. Mon. 66:3-6. 83 k OBSERVATIONS ON DISTRIBUTION AND LIFE HISTORY OF SKIPJACK TUNA, KATSUWONUS PELAMIS, IN AUSTRALIAN WATERS Maurice Blackburn^ and D. L. Serventy^ ABSTRACT Skipjack tuna occur in many areas around Australia, but have been little fished or investigated there because of low commercial demand. Their distribution in Australian coastal waters is not continuous, although suitable temperatures occur in all areas. Abundance in coastal waters is probably highest in the southeast. The southern limit of skipjack tuna range varies seasonally with the 15° C surface isotherm, and that temperature appears to be limiting. Length-frequency polygons for skipjack tuna of southeastern Australia show modes at about 37, 46, 53, and 59 cm fork length in the southern summer. The regression of weight W (grams) upon length L (millimeters) for east coast skipjack tuna is W = 0.000000839 L'^^202 Gonads of coastal skipjack tuna are all immature. The euphausiid Nyctiphanes australis is the principal food in east Australian waters south of latitude 34° S. Small fish such as clupeoids are eaten in some of those areas and are the principal food elsewhere. The skipjack tuna, Katsuwonus pelamis (Lin- naeus), occurs in tropical and warm temperate waters of all oceans. It supports large fisheries in many areas and is considered to have much poten- tial for further exploitation (Gulland 1971). The scientific literature on the species is large (Klawe and Miyake 1967) and growing. Australian con- tributions to that literature have been few, al- though the organized collection of data on Austra- lian skipjack tuna began in 1938. One reason is that the Australian tuna industry has shown little interest in skipjack tuna. It operates principally upon southern bluefin tuna, Thunnus maccoyii. The purpose of this paper is to present unpub- lished biological information on skipjack tuna in Australian waters and relate it to what has been published from the region. Many of the observa- tions were made by us. We do not discuss fishing operations or prospects, except to note briefly here that skipjack tuna have been caught in Australian coastal waters by live-bait fishing, purse seining, mesh netting, and trolling. Most of those catches were incidental to fishing for southern bluefin tuna. In the year 1974-75 the Australian tuna catch was 11,288 t, of which 2,375 t was skipjack tuna and the rest southern bluefin tuna (Anony- mous 1975, 1976). It is our opinion, admittedly '741 Washington Way, Friday Harbor, WA 98250. 227 Everett Street, Nedlands, Western Australia, 6009, Aus- tralia. subjective, that the biomass of skipjack tuna in Australian coastal waters is at least as high as that of southern bluefin tuna. It was much higher than the biomass of southern bluefin tuna off east- ern Tasmania in 1965, according to estimates from aerial surveys (Hynd and Robins 1967). One of us originally proposed the name "striped tuna" for this species in Australia, instead of "skip- jack" which is the usual English vernacular elsewhere (Serventy 1941). The intention was to avoid confusion with another pelagic fish which is sometimes called "skipjack" in Australia, namely Pomatomus saltator. "Striped tuna" has not gained acceptance as an English vernacular outside Aus- tralia, however, and both names are now used in Australia. Therefore we now refer to Katsuwonus pelamis as skipjack. METHODS Our summaries of biological data are based on skipjack tuna taken from 1938 to 1965. All were obtained by hook fishing at the sea surface, except for some Victorian samples which may have been caught near the surface in nets. Thus our biologi- cal observations essentially refer to surface fish. The same applies to other specimens mentioned in literature cited, except those noted to be from Japanese longliners. Length was measured from the tip of the snout to the caudal fin fork (FL), sometimes in millime- Manuscript accepted June 1980. FISHERY BULLETIN: VOL. 79, NO. 1, 1981. 85 FISHERY BULLETIN: VOL. 79. NO. 1 ters and sometimes to the nearest centimeter. Other authors whom we quote measured in the same way. Weight offish was measured ungutted, usually at sea, in pounds and ounces. These weights were converted to grams and rounded to the nearest 50 g. Gonads were weighed fresh to the nearest gram, usually at sea. Statements about mean positions of isotherms are based on charts by Vaux (1970) and Gorshkov (1974). SPATIAL DISTRIBUTION Skipjack tuna have an extensive distribution in the Australian region (Figure 1), but prior to 1938 they had been recorded only off New South Wales. It is now known that they have a continuous range in east Australian coastal waters from Lady El- liott Island to Storm Bay, although the limits may vary seasonally as discussed later. The RV War- reen and RV Stanley Fowler of the Commonwealth Scientific and Industrial Research Organization (CSIRO) established this distribution from trol- ling surveys between 1938 and 1951. Most speci- mens were taken on the continental shelf, many of them close inshore. East coast inshore waters north of Lady Elliott Island lie within the Great Barrier Reef. Skipjack tuna are unknown there, although most of the area is well fished by sports and commercial fishermen (Marshall 1964; Hynd 1968). The War- reen and Stanley Fowler prospected by trolling in northern Australian waters from July to October 1949, traversing much of the coast from Torres Strait to Broome. Skipjack tuna were found on banks near the edge of the Australian continental shelf to the south of Timor, but nowhere else, and no other records exist from northern coastal waters. Thus the distribution appears to be quite limited in coastal waters around northeast and northern Australia (Figure 1). This is not true of the adjacent oceanic waters, however. Japanese longline vessels began to fish for tuna in the area of Figure 1 about 1950. They did not seek skipjack tuna but took them incidentally. Figure 1 shows the general areas in which those vessels took any skipjack tuna in the years 1964-67, as established by Matsumoto (1975). Evidently skipjack tuna occur to some extent almost everywhere in ocean waters east, north, and west of the Australian mainland and New Guinea. Skipjack tuna were first recorded in Papua New Guinea waters be- tween 1948 and 1950 by the Australian RV Fair- wind, in localities shown in Figure 1 (Munro 1958). Japanese longlining began there at about the same time. About 1969 vessels of the Japanese live-bait fishery began taking skipjack tuna north and northeast of New Guinea, and a similar fishery was later established by nationals of Papua New Guinea within the same area (Kasa- hara 1977; Lewis and Smith 1977). Figure 1 shows the general area of the Japanese live-bait fishery in 1973, as reported by Kasahara (1977). Skipjack tuna were first recorded off the west coast of Australia in 1945. By 1951 an apparently continuous range from Onslow to Albany had been established, mostly by trolling surveys of the War- reen. Later surveys, made by the Department of Fisheries and Wildlife of Western Australia, ex- tended this range to Broome (Robins 1975). Skip- jack tuna in Western Australia appear to be less abundant than in southeastern Australia, at least on the continental shelf. Most were taken on the outer part of the shelf or just beyond the shelf edge. On the southern coast of Australia east of Al- bany, skipjack tuna were first found by the CSIRO RV Derwent Hunter in 1953, near the edge of the shelf in the eastern part of the Great Australian Bight. Soviet workers extended the known range to the western part of the Bight, again in waters near the shelf edge (Shuntov 1969). In 1978 the CSIRO RV Courageous found skipjack tuna be- tween the western end of the Bight and Albany (Maxwell^). East of the Bight, skipjack tuna have been taken by South Australian tuna fishermen almost to Kangaroo Island, from the shelf edge to near the coast (Olsen ; Williams ). Thus there is probably a continuous distribution along the southern coast of Australia to Kangaroo Island (Figure 1). We know of no certain skipjack tuna occurrences between Kangaroo Island and Australian east coast waters. The most westerly of the east coast records are Lakes Entrance in eastern Victoria, the mouth of the Tamar River in northeast Tas- mania (Scott 1975) and Storm Bay in southeast Tasmania (Figure 1). The apparent gap in skipjack tuna distribution to the west of those places is not readily explained. The skipjack tuna food or- ^J. G. Maxwell, Research Scientist, Division of Fisheries and Oceanography, CSIRO, Cronulla, 2230, Aust., pers. commun. May 1978. •'A. M. Olsen, Director of Fisheries Research, Department of Agriculture and Fisheries, Adelaide, 5000, Aust., pers. commun. May 1978. ^K. F. Williams, Experimental Officer, Division of Fisheries and Oceanography, CSIRO, Cronulla, 2230, Aust., pers. com- mun. May 1978. 86 BLACKBURN AND SERVENTY: DISTRIBUTION AND LIFE HISTORY OF SKIPJACK TUNA 110° 120° 130° 140° 150° 160° 110° 120° 130° 140° 150° 160° Figure l. — Australia and vicinity, showing areas of skipjack investigations and fisheries, edge of continental shelf (dotted line), mean February and August positions of the 15° C surface isotherm, and the following localities mentioned in the text: 1 — Lady Elliott Island, 2 — Sydney, 3 — Lakes Entrance, 4 — Fumeaux Group, 5 — Tamar River, 6 — Storm Bay, 7 — Portland, 8 — Kangaroo Island, 9 — Albany, 10 — Cape Leeuwin, 11 — Fremantle, 12 — Shark Bay, 13 — Onslow, 14 — Port Hedland, 15 — Broome. ganism Nyctiphanes australis, mentioned later, is abundant. Skipjack tuna can tolerate tempera- tures down to 15° C as shown below, and all waters in the area of no skipjack records have mean sur- face temperatures over 15° C in some months in most years. Aerial sightings of presumed skipjack tuna have been made near Portland and off west- ern Tasmania (Williams footnote 5). Nevertheless, we think the gap is real, at least in central and western Bass Strait. We and our colleagues have done much trolling in those areas in various sea- sons and years without catching any skipjack tuna. Southern bluefin tuna are likewise absent or very rare in central and western Bass Strait, al- though quite plentiful east and west of that region (Serventy 1956). Hynd and Robins (1967) showed that surface temperatures in the western ap- proaches to Bass Strait are under 15° C in parts of the summer because of upwelling, and perhaps as cold as that all summer in some years. This might restrict the distribution of skipjack tuna in waters east of Kangaroo Island. Hynd and Robins (1967) discussed the possibility of a similar effect upon southern bluefin tuna. Another possibility is that Bass Strait is too turbid for skipjack tuna, since it is shallow, receives several rivers, and is well mixed by waves and tides. Very little is known about effects of turbidity on tunas, however. 87 FISHERY BULLETIN: VOL. 79. NO. 1 SEASONAL DISTRIBUTION The range of skipjack tuna in eastern coastal waters is subject to seasonal variation. Table 1 shows where specimens have been taken in those waters in each month, in investigations made by or in cooperation with CSIRO, vdth data of all years combined. The southern limit of the range is Table l. — Records of captured skipjack tuna (X) and sightings of skipjack tuna (S) in Australian east coast waters by months. Sources are Robins (1952), Hynd (1968), and unpublished data from the Commonwealth Scientific and Industrial Research Organization for the period 1938-78. Waters between lat. 33° and 24° S were completely covered only in May, July, and September, and not covered at all in November and December. Waters between lat. 33° and 44° S were completely covered in all months. Lat. S Jan. Feb. Mar. Apr. May June July Aug. Sept. Oct. Nov. Dec. 24° -25° X 8 X 25° -26° s 26° -27° 27° -28° X s 28° -29° 29° -30° X X 30° -31° X X X X X 31° -32° X X X X X 32° -33° X X X X X X X X 33° -34° X X X X X X X X X 34° -35° X X X X X X X X X X X X 35° -36° X X X X X X X X X X X 36° -37° X X X X X X X X X 37° -38° X X X X X X X X 38° -39° X X 39° -40° X X X X X X 40° -41° X X X X 41° -42° X X X X 42° -43° X X X X 43° -44° X X farthest south in February-March and farthest north in July-August, and this variation is explic- able. All positions of the limit lie in the well- surveyed waters south of Sydney (approximately lat. 34° S), and they generally agree with positions of limiting sea surface isotherms as was shown by Robins (1952). Figure 1 shows the mean February and August positions of the 15° C surface isotherm, the one closest to the lowest tempera- ture at which any skipjack tuna have been caught (14.7° C, by Robins). The isotherm positions agree fairly well with the observed limits of skipjack tuna range in the same months, considering that no observations were made south of lat. 44° S. The lower limiting temperature for skipjack in abun- dance is about 16° C according to Robins (1952). The mean positions of that surface isotherm (about 2° of latitude north of the 15° C surface isotherm in each month) agree almost exactly with the skipjack tuna range limits in Table 1. From 1938 to 1942 skipjack tuna were hardly ever found south of lat. 43° S and seldom found south of lat. 42° S. They were fairly numerous between lat. 43° and 44° S in 1951, when temperatures were unusually high (Robins 1952). Hynd and Robins (1967) reported aerial sightings of a few schools of presumed skipjack tuna off the southern tip of Tasmania where surface temperatures were prob- ably about 13° C. Neither the species nor the tem- perature was confirmed, however. The occurrence of skipjack tuna at tempera- tures down to 15° C is of special interest, because the species has not been found in such cool waters in other parts of the world. It has been recorded at temperatures dowTi to 17° C in the eastern Pacific (Williams 1970) and 18° C near Japan (Uda 1957). Dizon et al. (1977) exposed four captive Hawaiian skipjack tuna to gradually decreasing water temperatures with the followdng results. Three fish stopped feeding at 17° C and died at 16° C; corresponding temperatures for the fourth fish were 15° and 14° C. It is difficult to recognize or hypothesize any seasonal change in the northern limit of skipjack tuna range in eastern coastal waters, especially in view of the incomplete vessel coverage north of Sydney (Table 1). The northern limit of any occur- rence (excluding offshore data from Japanese longliners) is between lat. 24° and 25° S, and skip- jack tuna were found there in February and June. If data from sightings are accepted there is evi- dence of skipjack tuna between lat. 24° and 26° S in various months from February to October. It would not be surprising if skipjack tuna occurred in those waters to some extent in all months. Mean monthly surface temperatures are maximal be- tween 27° and 28° C, whereas skipjack tuna can tolerate 30° to 32° C (WiUiams 1970; Dizon et al. 1977). On the other hand, the northern limit of the range of skipjack tuna in abundance could be south of the limit of total range and could vary with season, as Robins (1952) claimed. He put the northern limit of the main area of occurrence at about lat. 30° S in August-September and lat. 38° S in February, corresponding to the positions of the 19° C surface isotherm in those months. Robins considered that temperatures about 19° C were limiting for the main occurrence of skipjack tuna, 18° C limiting for occurrence in abundance, and 20.5° C limiting for any occurrence, at the warm end of the distribution along the east coast. The 88 BLACKBURN AND SERVENTY: DISTRIBUTION AND LIFE HISTORY OF SKIPJACK TUNA data in Table 1 are insufficient to support or deny Robins' conclusions as far as the actual skipjack tuna distribution is concerned. We note however that skipjack tuna are often abundant in other parts of the world at temperatures much above those mentioned (Blackburn 1965; Williams 1970). They have been found plentifully at 24° C off New South Wales (Williams footnote 5). Of course, abundance may reflect other conditions, as well as the distributions of temperature most suitable for adults. The occurrence of skipjack tuna in coastal waters north of Sydney, and its possible connec- tions with offshore distributions to the east and northeast, should be further investigated. Our knowledge of seasonal distribution in other coastal waters of Australia is incomplete. Off Western Australia skipjack tuna have been taken as follows, including records by Robins (1975): Broome to Port Hedland, July, August, and Oc- tober; Port Hedland to Shark Bay, January, April-June, August, September, and November; Shark Bay to Cape Leeuwin, February, March, and June-August; off Albany, May-July. Our rec- ords for the area south of Timor are for September and October. Australian and Soviet records be- tween Albany and Kangaroo Island are all for the period December-May. All the areas just men- tioned and others in northern Australia where no skipjack tuna have been found are warm enough for some skipjack tuna to occur all year (i.e., over 15° C at the sea surface. Figure 1). In the Japanese longline fishing area east of Queensland there may be some seasonal change in abundance of skipjack (Matsumoto 1975), but the pattern is not clear. In the similar area west of Western Australia the abundance appears to be low at all seasons. In surface waters of Papua New Guinea, according to Lewis and Smith (1977), there is no obvious seasonal change. LENGTH AND WEIGHT Length measurements of about 4,500 east coast skipjack tuna were made in CSIRO investigations to the end of 1965. The observed range was 30-65 cm FL north of Sydney, 35-66 cm FL in mainland waters south of Sydney, and 35-66 cm FL off Tas- mania. We have no length data for Great Aus- tralian Bight or South Australian fish except those of Shuntov (1969), which were 48-52 cm FL. Larger skipjack tuna to about 80 cm have been taken off New South Wales and South Australia in recent years (Williams footnote 5). Robins (1975) measured about 300 skipjack tuna from Western Australia, which were 29-78 cm FL. Our earlier measurements from the same area fall in that range. For Papua New Guinea a range of 35-62 cm FL was reported by Kearney et al. (1972). All these skipjack tuna were taken very near the sea sur- face. Barkley et al. (1978) hypothesized that large skipjack tuna require lower temperatures than small skipjack tuna and are therefore more abun- dant in the upper thermocline than at the sea surface, in the tropical Pacific. Figure 2 summarizes most of the east coast data in length-frequency polygons for various periods. Most of these sets of measurements are rather small in number, even when combined for certain months and years as in some of the polygons. Data for the Southern Hemisphere winter (polygons A, H, and M) show modes at about 34, 44, and 51 cm. Polygons for the southern summer (C, E, F, G, and J) have modes at about 37, 46, 53, and 59 cm. The first three modes for the summer are close to the three for the winter and are shown as I, II, and III, respectively, for each season. Modes at similar sizes in other polygons are labelled in the same way. This labelling does not imply that the modes represent successive age-groups a year apart, or that the absolute age is knov^n for any mode, be- cause such conclusions could not be drawn with confidence from these scattered data. If the modes do represent successive age-groups, the mean grov^h rate of east coast skipjack tuna must be about 6-10 cm/yr for fish between 35 and 60 cm. Most published estimates of skipjack tuna growth rate in that range of length are higher, as dis- cussed by Shomura (1966), Joseph and Calkins (1969), and Chi and Yang (1973) for Hawaii, Japan, the eastern Pacific, and Taiwan. The range of an- nual growth increment in those studies was 11-27 cm, with many values near 15 cm, and some of those estimates were obtained from tagging. On the other hand, Batts (1972) estimated 8-9 cm/yr for skipjack tuna >40 cm from North Carolina, from annuli in cross sections of dorsal spines. Kearney^ referred to an estimate of 7 cm/yr for Papua New Guinea skipjack tuna, based on tag- ging, but gave no details. Skipjack tuna of modal group I were obtained only from 1938 to 1941 in east coast waters. They may have been particularly abundant then, or there may have been some difference in trolling ^Kearney, R. E. 1978. Some hypotheses on skipjack {Kat- suwonus pelamis) in the Pacific Ocean. South Pac. Comm., Noumea, New Caledonia, Occas. Pap. 7, 23 p. 89 15 - 10 - 5 - 10 - 5 - 10 r 5 10 5 A May-July 1939,1940. N.S.W. 145 I II B Aug-0ct1940 ^ 1941 N.S.W. 130. FISHERY BULLETIN: VOL. 79, NO. 1 20r H Aug-Sept 1950 a N.S.W. 462. 15- 10- 5 - ol 1 L C Nov- Jan 1939/40 '"A N.S.W. 190. 1940/41,1941/42 " D Mar-May 1941,1942 o) (0 ^* c 0) u o Q. lOr 5 20 n 15 - 10 5 15 r 10 E Dec 1965 F Dec-Jan 1963,1964 G Jan 1954 20 30 40 50 cm 20 " 1 Oct-Nov 1950 III A NSW. 109. 15 - \ 10 - \ 5 n - 1 1 iiy \, , Tas.1865. ""Or j peb-Mar 1951 5- 20 I 15 N.S.W. 107. Vic. 321. 10 5 ■J K Apr-May 1951 _L _L ^^^ L Apr-May 1951 10 5 20r M July-Aug 1951 N.S.W. 201. ""S 10 70 20 30 40 50 cm Tas. 282. Tas. 176. " N.S.W. 152. N.S.W. 201. 60 70 Figure 2. — Percentage fork length (centimeters) frequency polygons of skipjack from New South Wales (mostly south of Sydney), Victoria (Lakes Entrance), and Tasmania (Fumeaux Group to Storm Bay). The data are smoothed by a moving average of three. The number of specimens is given after the locality abbreviation in each polygon. Polygons A to D, E to G, and H to M show data for 1939- 42, 1954-65, and 1950-51, respectively. The order of the polygons in each of those groups is based on the month or months in which the data were taken. Roman numerals are for identification of modes, and do not necessarily indicate ages. 90 BLACKBURN AND SERVENTY: DISTRIBUTION AND LIFE HISTORY OF SKIPJACK TUNA method (e.g., speed, lure type, lure size) between that period and later. Robins (1975) took similar fish in the period April-June off Western Aus- tralia, mostly by purse seining but occasionally by trolling. Weight as well as length was measured for 607 east coast skipjack tuna. The following significant linear regression was found between the common logarithms of the variables: log W = -6.0762 + 3.5202 (logL) where W is weight (grams) and L is fork length (millimeters). The coefficient of determination, i? -, is 0.856. Standard errors of the first and second constants in the equation are 0.1595 and 0.0586, respectively. The equation is equivalent to: W = 0.000000839 L 3.5202 The range of L in the data used is 410-645 mm. The heaviest fish weighed 5.67 kg. The 95% confidence limits of the regression co- efficient are 3.4048 and 3.6356, calculated from the standard error. Other published regressions of skipjack tuna weight on length for large samples ( >200) indicate regression coefficients from 3.2164 to 3.67. Those samples were taken in the eastern, central, and northwestern Pacific (Nakamura and Uchiyama 1966, and references there) and off North Carolina (Batts 1972). Standard errors of the coefficients were not published in most cases. The standard error can be calculated from data of Hennemuth (1959), for a regression coefficient of weight on length for 1,280 skipjack tuna from the eastern Pacific (combined areas). The coefficient was 3.336, with 95% confidence limits 3.296 and 3.376. Thus the east Australian and eastern Pacific regressions are significantly different at the 5% level of probability. The meaning of this difference is not clear. The two groups of skipjack tuna probably belong to different populations (Fujino 1972; Sharp 1978). However, Hennemuth (1959) found regression coefficients from 3.144 to 3.555 in different areas of the eastern Pacific north of the Equator, a region considered to contain only one skipjack tuna population (Fujino 1972; Sharp 1978), and some of those coefficients were signifi- cantly different. SEXUAL CONDITION The gonads of 418 east coast skipjack tuna were weighed. Ovary weights ranged from 4 to 30 g. Most ovaries were white to pink. Discrete small ova were visible to the naked eye in some ovaries, but large yolked ova were not observed. Ovaries of a reddish flaccid appearance, which might have been spent, were seen occasionally from April to August in New South Wales and Tasmanian waters. Testes were small, weighing mostly 1-2 g with a maximum of 13 g. Milt could sometimes be expressed from them by pressing. Similar obser- vations were made on a small number of skipjack tuna from the west and northwest coasts of Aus- tralia, except that no gonads were weighed. A specimen taken off Fremantle in July was possi- bly spent. No gonad data are available from South Australia. Orange (1961) compared ovaries of skipjack and yellowfin tunas by means of a "gonad index" equal to 3 8 (gonad weight) / (fish length ) 10 with gonad weight in grams and fish length (fork length) in millimeters. This is a ratio between gonad weight and an estimate offish weight. The estimate is not accurate for skipjack tuna, since weight increases with fork length to some power slightly higher than 3 in that species, as noted earlier. However Orange also compared gonad in- dices with the appearance of the ovaries and ova, and found that only indices over 30 indicated ap- proaching sexual maturity in skipjack tuna. Naganuma (1979) made similar comparisons which indicated that spawning skipjack tuna have gonad indices of 80 or higher, measured on Orange's scale. Thus calibrated, the index has some utility, and it has been employed by other skipjack tuna investigators. None of the gonad indices in our east Australian material reached 30 (Table 2); only 2, out of 224, were slightly over 20. Thus no females appeared to be mature on the basis of gonad index, confirming the observations on the gonads themselves. Yet virtually all these skipjack tuna were at or over the size at which first sexual maturity has been found in other Pacific waters, i.e., about 45 cm (Kearney et al. 1972; Blackburn and Williams 1975; Naganuma 1979). It is clear from these observations that skipjack tuna do not spawn to any significant extent in east Australian coastal waters, and there is no evi- dence that they spawn in any Australian coastal waters; nevertheless, they do spawn in the 91 FISHERY BULLETIN: VOL. 79, NO. 1 Table 2.— Length range, sex ratio, and female gonad indices (see text) for three groups of Australian east coast skipjack tuna. Fork Sex Gonad index length ratio Area Period (cm) (F/M) Range Mean New South Wales, Nov.-Dec. 44-57 47/30 5.6-20.7 12.3 south of Sydney 1941 Tasmania Mar.-May 1942 47-61 166/154 3.4-17.8 10.1 New South Wales, June 51-65 11/10 8.6-21.5 10.7 north of Sydney. 1941, and southern 1942 Queensland offshore tropical waters. Ueyanagi (1970) showed that skipjack tuna larvae occur between November and February in the Coral Sea east of tropical Queensland, north and east of New Guinea, and in ocean waters west of northwest Australia. From May to August the larvae are scarcer in Coral Sea and New Guinea waters than in the preceding period, and possibly so off north- west Australia. Gonad indices are higher in Coral Sea and New Guinea waters in the southern sum- mer than in winter (Naganuma 1979). Thus the skipjack tuna spawning season in waters near Australia is probably the southern summer, and the principal spawning areas seem to be in offshore waters northeast and northwest of the continent. FOOD Observations were made on stomach contents of 660 skipjack tuna from east coast waters and 30 from west and northwest coast waters (Table 3). Euphausiids were mostly Nyctiphanes australis although Thysanoessa gregaria was occasionally observed. Also included with euphausiids were several stomachs which contained a red liquid. This liquid was often found together with euphausiids, never with any other food, and was certainly a product of the digestion of euphausiids. The main point of interest in Table 3 is the proportion of stomachs with euphausiids. Evi- dently euphausiids are almost the sole food of skip- jack tuna in Tasmania and the principal food in southern New South Wales, but a small component of diet in the other sampled areas. Small pelagic fish are a large food item in all areas except Tas- mania. Cephalopods are a minor item in all areas. Table 3 does not include data on east coast skipjack tuna from Robins (1952) because they are non quantitative, but his findings were similar, as follows. Euphausiids were the principal food in Tks- mania and New South Wales waters south of Syd- ney. North of Sydney the principal food was fish, especially the young of pilchard, Sardinops neopilchardus , and anchovy, Engraulis australis. Park and Williams^ found the following in stomachs of skipjack tuna taken near Sydney: fish larvae, mainly pilchard; A^. australis; brachyuran and decapod larvae; copepods; and squid. These changes in diet by area appear to reflect the kinds of small nekton and large zooplankton that are available to skipjack tuna in coastal waters. Nyctiphanes australis is the principal coastal euphausiid in the southeast Australian region. Its range along the east coast is from lat. 31° S to the southern end of Tasmania (Sheard 1953). It is abundant off Victoria and Tasmania (including all of Bass Strait) and also off southern New South Wales, but not common in waters north of Sydney (Blackburn 1980). The species is un- recorded off Western Australia, although it occurs in South Australian waters. Off eastern Tas- 'Park, J. S., and K. Williams. 1977. A study of the relation- ship between the composition of food organisms of skipjack tuna Katsuwonus pelamis and the abundance and species composition of the plankton in the waters adjacent to Cronulla, New South Wales, Australia. Unpubl. manuscr Commonwealth Scien- tific and Industrial Research Organization, Division of Fisheries and Oceanography, Cronulla, 2230, Aust. Table 3.— Foods of skipjack tuna collected in Australian coastal waters, by numbers of stomachs in which they occurred. Nil means empty stomachs. Fish Other Area Nil Euphausiids remains Pilchard' Mackerel fish Squid lyiixed Other^ Total New South Wales, north of Sydney 16 1 7 12 1 1 6 44 New South Wales, Sydney and south 104 "77 10 2 4 M 1 ^5 204 Tasmania 82 322 1 '1 6 412 Western and northwestern Australia 14 6 83 1 '6 30 ^Sardinops neopilchardus . 'Scomber australasicus . ^Yellow liquid, except for two stomachs from Tasmania which contained salps. 'Including three stomachs which also contained hyperiid amphipods. ^Bellows fish (Macrorhamphosidae). ^Euphausiids plus fish or squid. Fish included Scomberesox forsteri , Trachums sp. and Macrorhamphosidae. 'Euphausiids plus squid. 'Flying fish (Exocoetidae), juvenile Gonorhynchus greyi, Harengula sp. and anchovy (probably Engraulis australis). 'Fish plus crustaceans or cephalopods or pteropods. Fish included Myctophidae. Cephalopods included squid and paper nautilus. 92 BLACKBURN AND SERVENTY: DISTRIBUTION AND LIFE HISTORY OF SKIPJACK TUNA mania, where skipjack tuna occur only in summer and autumn, N. australis is probably the only abundant food organism available to them. Pil- chards are not common in that area (Blackburn 1950b). Mackerel, Scomber australasicus, have been recorded from Tasmania, but were not found there by us and are probably rare. Anchovies are common in Tasmania, but their occurrence in summer and autumn is mostly in inlets which skipjack tuna do not enter (Blackburn 1950a). Jack mackerel, Trachurus declivis, are also abun- dant, but mostly occur closer inshore than skip- jack tuna (Hynd and Robins 1967). DISCUSSION The need for further information of various kinds on Australian skipjack tuna has been shown. It would be particularly interesting to know if the apparent discontinuities in distribu- tion are real, and if so, what causes them. The probable determinants of skipjack tuna distribu- tion in surface waters are temperature, food, and turbidity (Blackburn 1965, 1969). Dissolved oxy- gen concentration can be an additional limiting environmental property in the vertical plane, since skipjack tuna are stressed at concentrations below about 2.8 ml/1 (Dizon 1977; Sharp 1978). Concentrations at 100 m in waters near Australia are higher than 3.0 ml/1, however, except in an area off the west coast of West Irian (Reid et al. 1978). Temperatures required by skipjack tuna larvae may be higher than those preferred by adults, causing spawning adults to seek warmer waters than those not spawning (Blackburn and Williams 1975). It has been shown that all surface waters around Australia are warm enough (>15° C) for adult skipjack tuna in the warm season, and most waters warm enough at all seasons. The ab- sence or rarity of skipjack tuna in some of those areas probably indicates that suitable food or- ganisms are scarce, that the waters are too turbid for the fish to find food, or that the vertical dis- tribution of temperature is not such as to force the fish to the surface. ACKNOWLEDGMENTS Many valued colleagues helped with the field work of this study They included J. G. Clark, D. Connolly, R. J. Downie, A. Flett, S. Fowler, and G. E Whitley, all now deceased. We are grateful to others for personal communications mentioned in the text. Kevin Williams provided some of the rec- ords in Table 1. Ian Munro was helpful with nomenclature and records of some food organisms. The regression of weight on length was calculated by Dennis Reid. We received useful comments from J. S. Hynd, G. I. Murphy, K. F Williams, and anonymous reviewers. The first author acknowl- edges support received from CSIRO as a visting scientist in the Division of Fisheries and Oceanog- raphy, which facilitated this work. LITERATURE CITED ANONYMOUS. 1975. Tuna season will be closely watched. Aust. Fish. 34(10):4-8. 1976. Drop in value and quantity of 1974-75 catch. Aust. Fish. 35(8):2. barkley, r. a.,w. h. neill, and r. m. Gooding. 1978. Skipjack tuna, Katsuwonus pelamis, habitat based on temperature and oxygen requirements. Fish. Bull., U.S. 76:653-662. BATTS, B. S. 1972. Age and growth of the skipjack tuna, Katsuwonus pelamis (Linnaeus), in North Carolina waters. Chesapeake Sci. 13:237-244. Blackburn, M. 1950a. A biological study of the anchovy, Engraulis aus- tralis (White), in Australian waters. Aust. J. Mar. Freshwater Res. 1:3-84. 1950b. Studies on the age, growth, and life history of the pilchard, Sardinops neopilchardus (Steindachner), in southern and western Australia. Aust. J. Mar. Freshwa- ter Res. 1:221-258. 1965. Oceanography and the ecology of tunas. Oceanogr. Mar. Biol., Annu. Rev. 3:299-322. 1969. Conditions related to upwelling which determine distribution of tropical tunas off western Baja California. U.S. Fish Wildl. Serv, Fish. Bull. 68:147-176. 1980. Observations on the distribution of Nyctiphanes australis Sars (Crustacea, Euphausiidae) in Australian waters. Aust. CSIRO Div Fish. Oceanogr. Rep. 119, 10 p. BLACKBURN, M., AND F WILLIAMS. 1975. Distribution and ecology of skipjack tuna, Kat- suwonus pelamis , in an offshore area of the eastern tropi- cal Pacific Ocean. Fish. Bull., U.S. 73:382-411. Cm, K.-S., AND R.-T. YANG. 1973. Age and growth of skipjack tuna in the waters around the southern part of Ikiwan. Acta Oceanogr. Taiwan. 3:199-222. DiZON, A. E. 1977. Effect of dissolved oxygen concentration and salinity on swimming speed of two species of tunas. Fish. Bull., U.S. 75:649-653. DizoN, A. E., W. H. Neill, and J. J. Magnuson. 1977. Rapid temperature compensation of volitional swimming speeds and lethal temperatures in tropical tunas (Scombridae). Environ. Biol. Fishes 2:83-92. FUJINO, K. 1972. Range of the skipjack tuna subpopulation in the western Pacific Ocean. In K. Sugawara (editor), The 93 FISHERY BULLETIN: VOL. 79, NO. 1 Kuroshio II, proceedings, Second Symposium on the Re- sults of the Cooperative Study of the Kuroshio and Adja- cent Regions, p. 373-384. Saikon PubL Co.. Tokyo. GORSHKOV, S. G. 1974. Atlas okeanov (Atlas of the oceans), Tikhii okean (Pacific Ocean). [In Russ.] Pergamon Press, Oxf , 302 p. 1976. Atlas of the oceans. Pergamon Press, Oxf, 87 p. (Engl, transl. of index and introd. of Atlas okeanov.) GULLAND, J. A. 1971. The fish resources of the ocean. Fishing News (Books) Ltd., West Byfleet, Engl., 255 p. HENNEMUTH, R. C. 1959. Additional information on the length-weight rela- tionship of skipjack tuna from the eastern tropical Pacific Ocean. [In Engl, and Span.] Inter- Am. Trop. Tuna Comm. Bull. 4:25-37. HYND.J. S. 1968. Report on a survey for yellowfin tuna, Thunnus alba- cares ( Bonnaterre), in Queensland waters. Aust. CSIRO Div Fish. Oceanogr. Tech. Pap. 26, 40 p. HYND, J. S., AND J. R ROBINS. 1967. Tasmanian tuna survey report of first operational period. Aust. CSIRO Div Fish. Oceanogr. Tech. Pap. 22, 55 p. JOSEPH, J., AND T. R Calkins. 1969. Population dynamics of the skipjack tuna (Kat- suwonus pelamis) of the eastern Pacific Ocean. [In Engl, and Span.] Inter-Am. Trop. Tuna Comm. Bull. 13:7-273. KASAHARA, K. 1977. Trends in tropical Pacific skipjack fishery. [In Jpn.] Suisan Sekai 26(3);28-33 (Engl, summ., Aust. Fish. 37(2):6-7, 1978). KEARNEY, R. E., A. D. LEWIS, AND B. R. SMITH. 1972. Cruise report Tagula 71-1. Survey of skipjack tuna and bait resources in Papua New Guinea waters. Papua New Guinea Dep. Agric. Stock Fish. Res. Bull. 8, 145 p. KLAWE, W. L., and M. P MIYAKE. 1967. An annotated bibliography on the biology and fishery of the skipjack tuna, Katsuwonus pelamis, of the Pacific Ocean. [In Engl, and Span.] Inter-Am. Trop. Tuna Comm. Bull. 12:139-363. LEWIS, A. D., AND B. R. SMITH. 1977. Successful year for PNG's fishery. Aust. Fish. 36(10):11-13. Marshall, T. C. 1964. Fishes of the Great Barrier Reef and coastal waters of Queensland. Angus and Robertson, Sydney, 566 p. ]VlATSUMOTO,W. 1975. Distribution, relative abundance, and movement of skipjack tuna, Katsuwonus pelamis, in the Pacific Ocean based on Japanese tuna longline catches, 1964- 67. NOAA Tech. Rep. NMFS SSRF-695, 30 p. MUNRO, I. S. R. 1958. The fishes ofthe New Guinea region. Territ. Papua New Guinea, Fish. Bull. 1:97-369. (Reprinted from Papua New Guinea Agric. J. 10(4):97-369.) Naganuma, a. 1979. On the spawning activities of skipjack tuna in the western Pacific Ocean. [In Jpn., Engl, summ.] Bull. Tohoku Reg. Fish. Res. Lab. 40:1-13. NAKAMURA, E. L., AND J. H. UCHIYAMA. 1966. Length-weight relations of Pacific tunas. In T. A. Manar (editor). Proceedings, Governor's Conference on Central Pacific Fishery Resources, State of Hawaii, p. 197-201. 94 ORANGE, C. J. 1961. Spawning of yellowfin tuna and skipjack in the east- em tropical Pacific, as inferred from studies of gonad development. [In Engl, and Span.] Inter-Am. Trop. Tuna Comm. Bull. 5:459-526. REID, J. L., E. BRINTON, a. FLEMINGER, E. L. VENRICK, AND J. A. MCGOWAN. 1978. Ocean circulation and marine life. In H. Charnock and G. Deacon (editors), Advances in oceanography, p. 65-130. Plenum Publ. Corp., N.Y ROBINS, J. P 1952. Further observations on the distribution of striped tuna, Katsuwonus pelamis L., in eastern Australian waters, and its relation to surface temperature. Aust. J. Mar. Freshwater Res. 3:101-110. 1975. Tuna survey in waters off the Western Australian coast during the period August 1973 to August 1974. West. Aust. Dep. Fish. Wildl. Rep. 18, 85 p. SCOTT, E. O. g. 1975. Observations on some Tasmanian fishes: Part XXI. Pap. Proc. R. Soc. Tasmania 109:127-173. SERVENTY, D. L. 1941. The Australian tunas. Aust. Counc. Sci. Ind. Res., Pam. 104, 48 p. 1956. The southern bluefin tuna, Thunnus thynnus mac- coy ii (Castelnau), in Australian waters. Aust. J. Mar. Freshwater Res. 7:1-43. SHARP, G. D. 1978. Behavioral and physiological properties of tunas and their effects on vulnerability to fishing gear In G. D. Sharp and A. E. Dizon (editors). The physiological ecology of tunas, p. 397-449. Acad. Press, N.Y. SHEARD, K. 1953. Taxonomy, distribution and development of the Euphausiacea (Crustacea). B.A.N.Z. Antarct. Res. Exped. 1929-1931 Rep., Ser. B, 8:1-72. SHOMURA, R. S. 1966. Age and growth studies of four species of tunas in the PacificOcean. In T. A. Manar(editor), Proceedings, Gov- ernor's Conference on Central Pacific Fishery Resources, State of Hawaii, p. 203-219. SHUNTOV, V R 1969. Some features of the ecology of pelagic fishes in the Great Australian Bight. [In Russ.] Vopr. Ikhtiol. 9:995-1005 (Engl, transl., Rrobl. Ichthyol. 9:801-809, 1969). UDA, M. 1957. A consideration on the long years trend of the fisheries fluctuation in relation to sea conditions. Bull. Jpn. Soc. Sci. Fish. 23:368-372. UEYANAGI, S. 1970. Distribution and relative abundance of larval skip- jack tuna {Katsuwonus pelamis) in the western Pacific Ocean. In J. C. Marr (editor). The Kuroshio, a sym- posium on the Japan Current, p. 395-398. East-West Center Press, Honolulu. VAUX, D. 1970. Surface temperature and salinity for Australian waters, 1961-65. Aust. CSIRO Div Fish. Oceanogr. Atlas 1, 198 p. WILLIAMS, F 1970. Sea surface temperature and the distribution and apparent abundance of skipjack (Katsuwonus pelamis) in the eastern Pacific Ocean, 1951-1968. [In Engl, and Span.] Inter-Am. Trop. Tuna Comm. Bull. 15:231-281. A METHOD FOR GROWTH CURVE COMPARISONS Russell F. Kappenman^ ABSTRACT Suppose one has a sample of pairs of age and length measurements from each of two or more populations offish. The mathematical forms of the growth curves associated with the populations are assumed to be specified but each form contains at least one unknown parameter Presented in this paper is a data analytic approach to the problem of deciding which, if any, of the populations have essentially the same growth curve and which have different ones. A common problem in fisheries research is that of comparing two or more growth curves. This prob- lem arises whenever investigators gather data for the purpose of trying to determine whether or not different species or the sexes of a given species of fish grow at different rates, or for the purpose of assessing growth variation of a species from envi- ronment to environment, area to area, or stratum to stratum in which it is found. Since the models generally used for the age- length relationships (e.g., von Bertalanffy, Laird- Gompertz, logistic, etc.) are most often nonlinear in the unknown parameters and cannot be linear- ized by transformations of the variates, the usual techniques for comparing regression equations are not applicable. Up to this point little has been done in the way of development of quantitative methods for determining whether or not unknown growth curves do in fact differ. Thus the investiga- tor can often do little more than visually examine plots of age-length data for samples from the various populations being compared and arrive at some rather subjective conclusions. An exception is a paper by Allen (1976) which treats the special case where each of the growth curves being compared belongs to the von Berta- lanffy family. There are, of course, numerous instances where the von Bertalanffy model is not appropriate and the Allen procedure does not apply if it is not. Further, even if this model is appropriate, some severe assumptions need to be made in order to apply the analysis. These include: 1) the equality of the scale parameters for all curves being compared, 2) the true value of the common scale parameter being exactly equal to its 'Northwest and Alaska Fisheries Center, National Marine Fisheries Service, NOAA, 2725 Montlake Boulevard East, Seattle, WA 98112. estimated value, and 3) the usual normality, independence, and equality of variance assump- tions for the error term. The first assumption quite clearly biases the procedure in favor of the null hypothesis of equality of the growth curves, while validity of the second seems to be too much to hope for. Gallucci and Quinn (1979) also discuss the growth curve comparison problem for the von Bertalanffy case. They essentially reparameterize the model and test the hypothesis of equality of one of the new parameters for all curves being compared, assuming, apparently, that the other two have the same value for all of the curves. The comments in the preceding paragraph also apply to these authors' work. The purpose of this paper is to point out how some predictive sample reuse techniques, described in a recent paper by Geisser and Eddy (1979), can be adapted and applied to a growth curve compari- son problem where two or more populations are being studied, the growth curves associated with each of the populations are unknown and are to be compared, and a sample of age-length data is available from each population. The problem then is to use the data to decide which, if any, of the population growth curves are the same and which are different. We will assume that the growth curves associ- ated with each of the populations are specified except for the values of unknown parameters. These specifications often would be made by plotting the sets of age-length data, fitting vari- ous possible models suggested by the data plots, and selecting the models which best fit the data. The growth curves can, but need not, belong to the von Bertalanffy family. In fact, they can belong to any family. Thus, in essence, we are Manuscript accepted July 1980. FISHERY BULLETIN: VOL. 79, NO. 1, 198L 95 FISHERY BULLETIN: VOL. 79, NO. 1 considering a much more general and widely applicable problem than that discussed by Allen (1976) and by Gallucci and Quinn (1979). For the technique presented here, the only assumption that is made is that the forms of the population growth curves can be specified. No parameters are assumed to be known or equal, and no distri- butional assumptions are made. The solution, described in the following sec- tions, of our growth curve comparison problem is not obtained via a classical statistical hypothesis testing approach. That is, we do not formulate an appropriate null hypothesis and derive a crite- rion which dictates when and only when it should be rejected. Instead, in the spirit of Geisser and Eddy (1979), we formulate various possible mod- els and give a data analytic approach to selecting the one model preferred by the data. A difficulty, which plagues classical hypothesis testing, does not exist for the approach described here. It is the necessity of specifying a signifi- cance level. Historically, significance levels such as 0.10, 0.05, and 0.01 have routinely been used for tests without objective justification. Yet the choice of a significance level affects the conclu- sions arrived at. For example, it is quite possible to find that a hypothesis can be rejected at the 0.05 level, but not at the 0.01 level. Further, the choice of a significance level affects the probabil- ity of rejecting a false hypothesis. Lowering the significance level usually lowers the probability of rejecting a false hypothesis. Requiring one to specify a significance level presumes that one has a sound basis for controlling the probability of rejecting one of two possible hypotheses when it is true, that one can objectively assign a signifi- cance level which controls this probability, and that controlling this probability is more crucial than controlling the probability of not rejecting the hypothesis when it is false. The contention here is that for many, if not most, scientific investigations, the consequences of rejecting one hypothesis when it is true are no more serious than rejecting the other when it is true. That is, often an investigator has no reason to favor either hypothesis, but merely wants to know which one is more reasonable, given the data that has been collected. THE TWO POPULATIONS CASE Suppose that two populations of fish are being studied. For example, the first population might 96 consist of all fish of a given species inhabiting one area while the second might consist of all fish of this species inhabiting a different area. Suppose we are interested in comparing the growth curves associated with the two populations. We consider two possible models, say Mi and M2. The model Mi specifies that the growth curves are the same, while under M2 the two growth curves differ. Let X and y represent, respectively, the age and length of a fish. Then we rewrite Mi and M2 as Mi: y=fix;6) + e (1) no matter which of the two popula- tions fish belongs to. M2: y = fiix;di) + e (2) if the fish belongs to the first population. y = f2ix;d2) + e (3) if the fish belongs to the second population. Here fix; 6), fiix; di), and fiix; 62) are each functions of x. 6, di, and 62 are each vectors of unknown parameters and f, fi , and /2 are speci- fied except for the values of elements of 6, 61 , and 02. Essentially, f, fi, and [2 represent three dif- ferent growth curves which are specified except for the values of unknowm parameters present in each. The function f represents the expected length of a fish whose age is x, assuming equal growth curves for the two populations, while /"i and f2 are, respectively, the expected lengths for fish of age x from the first and second populations, assuming the growth curves differ. As usual, e represents the unknown, random error term. We now give a data analytic approach for selecting one of the two possible models Mi or M2 . The data used to make the selection are pairs of age-length measurements for samples of fish from each of the two populations. Let {xii,yu), ixi2,yi2),---,ixin,yin) represent a sample of n pairs of observations of x and y from the first population and (^21, ^21), (^22, y22),--, ix2m, y^m) represent a sample of m pairs of obser- vations of X and y from the second population. These n + m pairs of observations are the data gathered by the investigator and we want to use these data to select either Mi or M2 . Assume, for the moment, that M2 is correct. Forj - 1, 2,...,n, let ^k^) represent the vector of KAPPENMAN: A METHOD FOR GROWTH CURVE COMPARISONS (least squares, say) estimates of the elements of 01 found by taking the relationship between x and y to be given by Equation (2) and using the data (xii, yii), U12, yi2),- --Axiij -d, yuj-i)), ixnj + i),yuj + i)),...,(xin, yin), that is, all of the n pairs of observations of x and y from the first population except for the jth pair. Set n D21 = 1 [yij - fiixij; dt(j))f. Note that the second term inside the brackets is the predicted length for the jth fish in the sample from the first population, assuming M2 is correct. The observed length of this fish is the first term inside the brackets. Thus D21 is the sum of the squares of the differences between the observed and predicted fish lengths for the fish in the first population sample, for model M2 . Similarly, for 7 = 1, 2,...,m, let ^2(J) represent the vector of (least squares, say) estimates of the elements of 62 found by taking the relationship between x and y to be given by Equation (3) and using the data (X21, ^21), (^22, y22),---,ix2ij-i), y2{j-i)), (x2ij +1) , y2ij +1)), ,ix2m, y2m), i.e., all of the m pairs of observations of x and y from the second population except for the Jth pair. Set n ■D22 = 2 {y2j - f2ix2j; d2{j))f 7=1 and D2 = D21 + D 22. The quantity D22 has an interpretation similar to that given to D21. Putting these two together, we see that D2 represents the sum of the squares of the differences between the observed fish lengths and the predicted fish lengths for all n + m fish in the samples, under model M2 . Next, assume that Mi is correct, pool the data, and consider the n + m pairs (jCn, yn), (xi2, yi2),. .-, iXm, yin), iX2i , y2i), (^22 > y22),---, iX2m, y2m)- Let {xj,yj) represent thejth of these n + m pairs, for 7 = l,...,n + m. Further, forj - 1,2,..., n + m, let d^j), represent the vector of (least squares, say) estimates of the elements of 6 obtained by taking the relationship between x and y to be given by Equation (1) and using the data (xiji), (X2,y2),---,ixj-i,yj~i), ixj + i,yj+i),..., ixn +m , yn +m ), that is, all 7z + m pairs of observa- tions of X and y from the first and second popula- tions except for the jth pair. The sum of the squares of the differences between the observed and predicted fish lengths for all n -I- m fish in the samples, under Mi , is n + m ^1 - S [yj - fixj; eij^)?. J =1 Our rule for selecting either of Mi or M2 can be simply stated as follows. Select Mi if Z)i^D2, otherwise select M2. This rule is a very natural and objective one. It is based on whether the data (i.e., the observed fish lengths) are better predicted by one growth curve or two. If the sum of squares of the differences between observed and predicted lengths under Mi does not exceed the sum of squares of the differences between observed and predicted lengths under M2 (i.e., Di^D2), the data are better predicted by one growth curve than by two and Mi should be selected. Otherwise, they are better predicted by two distinct growth curves and M2 should be selected. AN EXAMPLE To illustrate the procedure described in the previous section, we consider an example. The numbers given in the first two columns of Table 1 are the ages and corresponding lengths of 15 fish taken from the first of two populations, while the numbers in the first two columns of Table 2 are the ages and corresponding lengths of 14 fish taken from the second population. These data are hypothetical. In fact they were generated by a computer. We want to use these two sets of data to decide which of two models, Mi or M2, is preferred, where under Mi the growth curves for the two populations are the same, and under M2 the two populations have different growth curves. Among several growth ciirves, including the von Bertalanffy, Laird-Gompertz, and logistic ones, the best fit, for both data sets as well as the combined data set, was provided by the logistic. The average length of a fish whose age is x, for a logistic growth curve, is f{x;a,b,c) = a 1+e ■(6x + c) (4) where a, b, and c are unknown parameters. Thus, we take Mi to specify that the average length of a fish whose age is x is Equation (4) no 97 FISHERY BULLETIN: VOL. 79, NO. 1 TABLE 1.— Ages, lengths, parameter estimates, and predicted lengths, under models M, and M2, for 15 Population I fish. Least squares estimates Predicted Least squares estimates Predicted Age Length of a,, ti,, c, under M2 length under M2 of a, b, c under M, length under M, 1 4.0 55.27, 0.36, -2.70 4.9 55.67, 0.35, -2.62 5.2 2 8.3 54.76, 0.38, -2.80 6.2 55.44,0.36, -2.66 6.9 3 5.6 55.86, 0.34, -2.56 9.9 55.91,0.34. -2.56 10.0 5 16.0 55.11,0.37, -2.73 15.9 55.57, 0.35, -2.63 16.5 6 20.9 55.13,0.37, -2.74 20.3 55.56, 35, -2.64 20.8 7 27.7 55.40,0.36, -2.77 24.8 55.70,0.35. -2.65 25.4 8 31.9 55.43, 0.36, -2.73 29.9 55.70, 0.35, -2.64 30.4 9 32.9 54.62,0.38, -2.76 35.7 55.28, 0.36, -2.65 35.6 10 40.2 55.24,0.36, -2.71 39.4 55.62, 0.35, -2.63 39.5 11 39.9 54.88.0.38, -2.81 44.0 55.37, 0.36, -2.67 43.7 12 45.6 55.15,0.37, -2.74 46.4 55.56, 0.36, -2.64 46.4 13 49.6 54.96,0.36, -2.72 48.6 55.51,0.35, -2.63 48.7 14 53.0 54.32,0.37, -2.71 50.0 55.26, 0.35. -2.63 50.4 16 54.2 54.23, 0.37, -2.74 52.2 55.27, 0.36, -2.64 52.8 17 51.5 56.97, 0.35, -2.68 54.8 56.25,0.35, -2.62 54.2 TABLE 2.- —Ages, lengths, parameter estimates. and predicted lengths, under models Mi and M2 , for 14 Population II fish. Least squares estimates Predicted Least squares estimates Predicted Age Length of 82, £)2, C2 under M2 length under Mj of a, b, c under M, length under M, 1 6.5 55.65, 0.35, -2.60 5.3 55.43, 0.36, -2.66 5.0 2 6.4 56.09, 0.34, -2.52 7.7 55.62,0.35, -2.62 7.1 3 8.0 56.24, 0.33, -2.48 10.4 55.69. 0.35, -2.60 9.7 4 14.1 55.74, 0.35, -2.60 12.9 55.47, 0.36, -2.66 12.5 5 17.6 55.84, 0.35, -2.59 16.8 55.53,0.36, -2.66 16.3 6 21.8 55.90, 0.35. -2.58 21.1 55.58, 0.36, -2.65 20.7 8 31.4 56.05,0.34, -2.56 30.6 55.65, 0.35, -2.64 30.4 9 34.4 55.48, 0.35, -2.59 35.9 55.46,0.36, -2.64 35.4 11 43.0 55.78, 0.35, -2.57 43.5 55.54,0.36, -2.64 43.3 13 49.1 55.83,0.34, -2.56 48.8 55.54, 0.35. -2.63 48.7 14 50.4 55.92, 0.35, -2.57 50.7 55.58, 0.35, -2.64 50.6 15 52.3 55.78,0.35, -2.56 52.0 55.50, 0.35, -2.64 52.0 16 54.3 55.25,0.35, -2.57 52.7 55.24, 0.36, -2.64 52.8 17 53.0 56.53, 0.34, -2.55 54.4 55.78, 0.35, -2.63 53.9 matter which population it belongs to. On the other hand, we take M2 to specify that the aver- age length of a fish whose age is x is Equation (4) with a, b, and c replaced, respectively, by a/, bi, and a, if the fish belongs to population i, for i = 1, 2. Note that in the notation of the previous section d is the vector whose elements are a, b, and c, while 6i is the vector whose elements are at , bi , and a , for i = 1, 2. Also n -15 and m = 14, for this example. The ith row, or threesome, for i = 1,...,15, of the third column of Table 1 is the set of least squares estimates of ai , 61 , and Ci obtained by assuming M2 to be correct and using all of the age-length data pairs in Table 1, except for the ith pair, to estimate oi , 61 , and Ci . For example, when the data point (8, 31.9) is ignored, the least squares estimates of ai , 61 , and ci are, respec- tively, 55.43, 0.36, and -2.73. The fourth column of Table 1 gives the predicted lengths for each of the first population fish, assuming M2 is correct. That is, the ith element in this column is am) I -I- g-^bi(i)Xi + Ci(i)) (5) where xi is the ith element of the first column and ai(i), biH), and cm) represent the ith three- some of the third column. The ith row, or threesome, for i = 1,...,15, of the fifth column of Table 1 is the set of least squares estimates of a, 6, and c obtained by assuming Mi to be correct and using all of the age-length data pairs in Tables 1 and 2, except for the ith pair in Table 1, to estimate a,b, and c. The last column of Table 1 gives the predicted lengths for each of the first population fish, assuming Mi is correct. The ith element of this column is (5) after dm), hm), and cm) have been replaced by dii), b{i), and cn), where the latter threesome is the ith row of column five. The discussion of columns three, four, five, and six of Table 2 is completely analagous to that given in the preceding two paragraphs for these 98 KAPPENMAN: A METHOD FOR GROWTH CURVE COMPARISONS columns of Table 1. Thus, essentially, the fourth and sixth columns of Table 2 give the predicted lengths of the second population sampled fish for models M2 and Mi , respectively. In the notation of the previous section, D21 is the sum of the squares of the differences between the corresponding elements of columns two and four of Table 1. We find that Dai = 87.31, for this example. Similarly, D22 is the sum of the squares of the differences between the corresponding elements of columns two and four of Table 2 and we find that D22 = 19.39. Further, D2 = D21 + D22 = 106.70. Finally, Di is the sum of the squares of the differences between the correspond- ing elements of columns two and six of Tables 1 and 2. We find that Di = 86.77 and since Di < D2 , the model. Mi , of equal growth curves for the two populations is the one best supported by the data. For this example, the length of the 7th fish from population i was taken to be Yu = a + €, 1+e -ibxij +c) ij for i = 1, 2, where Xij is the age of the 7th fish from population i, a = 55, b = 0.35, c = -2.55 and the e^y's were each normal random variates with mean zero and standard deviation equal to two. The normal variates were generated using the algorithm of Box and Muller (1958). Thus, in essence, we generated both data sets using the same growrth curve and the procedure described in the previous section made the correct selection. MORE THAN TWO POPULATIONS The procedure used to compare the growth curves for two populations is easily extended to the case where the grow^th curves for three or more populations are to be compared. As before, we begin by formulating all possible or plausible models. The number of possible models increases considerably as the number of populations being studied increases. For example, if there are three populations, there are five possible models, say Mi,...,M5. Here Mi specifies that all three growth curves are the same. M2 specifies that the growth curves for the first two populations are the same but they differ from that for the third population. M3 specifies that the first and third populations have the same growth curve but the second population's growth ciirve is different. M4 specifies that the first population's growth curve differs from those for the second and third popu- lations but the latter two are the same. Finally, M5 specifies that all three grov^h curves differ. Once again we assume that the forms of the common and distinct growth curves are specified for each model, but each contains one or more unknown parameters. Once the models have been formulated, the problem is to use data to select one of them as being most plausible. The data consist of samples of pairs of age-length measurements from the populations being studied. For each model, we compute the sum of the squares of the differences between observed and predicted lengths for all of the fish in the samples, where the predicted lengths are computed by assuming the model is correct. The model selected as most appropriate, by the data, is the one that corresponds to the smallest sum of squares. In order to compute the sums of squares, we must calculate a predicted length for each fish in the samples, under each model. For a given fish and a given model, the fish's predicted length is calculated by noting the population from which it came and grouping together all data points from this population and the populations, if any, whose growi;h curves Eire asserted, by the given model, to be equal to the grov^h curve for the fish's popula- tion. The fish's age and length measurements are then eliminated from the group of data points and the remaining data points in the group are used to estimate the unknown parameters in the asserted common grovid;h curve. Once these estimates are obtained, unknown parameters in the asserted common growth curve are replaced by their esti- mates, and the fish's age is substituted into the result to obtain the fish's predicted length. As an example, consider the data given in Table 3. These data represent the ages and corre- sponding lengths of 20 fish taken from each of three populations. Once again, the data have been generated by a computer. Our goal is to use the data to select one of five possible models, Ml , . . . ,M5 , where the M,'s are delineated in the first paragraph of this section. For this example, a growth curve of the form y =a{l 6 0' e ■) + e (6) fits each data set better than the von Bertalanffy, Laird-Gompertz, and logistic growi;h curves. The first term on the right hand side of Equation (6) is 99 essentially a constant times an extreme value for minima distribution function. This growth curve does not appear to have been used in the litera- ture as yet. But it probably should be considered as a possible model whenever the other three are tried, as I have found cases where it fits real data better than the others. Table 3.— Ages and lengths of 20 fish from each of the three populations. Population 1 Population II Population III Age length length length 1 4.7 5.0 5.3 2 5.4 7.5 6.1 3 10.0 14.2 10.8 4 12.6 11.0 13.5 5 19.0 17.8 20.0 6 16.0 17.5 17.2 7 19.0 19.2 20.1 8 22.3 27.9 23.5 9 28.1 27.7 29.2 10 27.2 34.1 28.2 11 38.0 32.7 38.8 12 41.5 40.8 41.8 13 42.1 48.1 41.7 14 49.9 51.5 48.9 15 53.3 53.3 51.4 16 57.0 56.3 54.3 17 56.3 55.9 52.8 18 58.3 57.9 54.1 19 59.6 60.2 55.0 20 59.9 57.9 55.1 Because Equation (6) fits each data set so well, one takes each of the common and distinct growth curves for each model to be in the form of Equa- tion (6). Then for each model one calculates a predicted length for each of the 60 fish in the samples and a sum of squares of the differences between observed and predicted lengths. For the models Mi,...,M5, these sums of squares, are respectively, 334.27, 298.58, 346.45, 331.58, and 312.94. Since the second of these is the smallest, the model, M2, which asserts that the growth curves for the first two populations are the same but that for the third population is different is the selected one. Each of the fish lengths for this example was calculated using Equation (6), where j' represents length and x represents age. Once again, e was taken to be a normal random variate with mean zero and a standard deviation of 2. For each of the fish in the first two of the three samples, a = 60, 6 = 0.10, and c = 0.20. For the 20 fish in the third sample, a = 55, 6 = 0.12, and c = 0.20. Thus the first two data sets were generated using the same growth curve, but the third data set was gener- ated using a different growth curve. Our procedure made the correct selection. FISHERY BULLETIN: VOL. 79, NO. 1 SOME CONCLUDING REMARKS It is, in general, not feasible to attempt to carry out the calculations required for our growth curve comparison procedure by hand or with a desk calculator. This is because nonlinear regres- sion analyses are usually required and there are many of them. However, the computations are easily programmed for a computer. For many growth studies, rather massive amounts of data are gathered. If the amount of data available is excessively large, computer time and costs may become prohibitive. It is natural to ask whether the number of computa- tions required can be reduced by doing away with the process of eliminating a data point from a data set before estimating parameters. Indeed, if this could be done, the number of least squares analyses needed would be drastically reduced. Unfortunately, however, it cannot be done. For it can be shown that if it is done, the selected model will always be the one which asserts different growth curves for all populations. Often though, when there is a large amount of data, each age in the samples is common to many fish. In this case, a possible procedure is to work with the data points consisting of ages and aver- age lengths, thus reducing the number of data points considerably. However, if the numbers of lengths used to calculate the average lengths vary widely from age to age, then it seems sensi- ble to use weighted sums of squares of differences between observed and predicted lengths, and weighted least squares estimates of parameters, with the weights, in each case, being the numbers of lengths used to calculate the average lengths. The idea is that the larger the number of observa- tions used to calculate an average, the closer the average should be to the true growth curve ordi- nate and, thus, the more weight that should be assigned to it. This modification of the present procedure was used in Boehlert and Kappenman (1980). The dangers of extrapolation, after regression analyses, are well known. Thus, the practice of obtaining a predicted value for the dependent variable for a subject whose independent variable value lies outside the range of independent var- iable values used to carry out the regression analysis, is generally discouraged. There may be instances where extrapolation will bias our com- parison procedure away from one or more models. The easiest way of checking to see if it does, in 100 KAPPENMAN: A METHOD FOR GROWTH CURVE COMPARISONS any given situation, is to examine the observed and predicted length differences for each model. Differences, corresponding to youngest or oldest fish, being excessively large for one or more models might be an indication that extrapolation is biasing the procedure. This difficulty did not appear in the examples used in this paper, but it is possible to imagine rare cases where it could be a problem. If this problem does arise, it is easily remedied. One can always eliminate from a sum of squares of differences between observed and predicted lengths those differences whose predic- ted lengths are obtained by extrapolation. If this is done, the sums of squares in the model selection criterion should be replaced by averages of squares of differences. For each of the examples given in this paper, all of the specified growth curves were taken to be of the same form. This is not necessary. Any growth curve can be given any form. For exam- ple, in the two population case, the common growth curve, for the model of equality of growi;h curves, can have a mathematical form which is different from the forms of the growth curves specified under the model of different growth curves. And, in fact, the latter two forms can be different from each other. Thus it is possible to handle the case where Mi specifies equal growth curves and the common growth curve belongs to, say, the logistic family, while M2 specifies differ- ent growth curves and the curves belong to, say, the Laird-Gompertz family or one belongs to the Laird-Gompertz family and the other belongs to the generalized extreme value for minima family. Finally, it should be pointed out that although this paper has been concerned solely with growth curve comparisons, the procedure described here can be applied to the general problem of compar- ing regression equations. The regression equa- tions of interest can be either linear or nonlinear functions of the unknown parameters. Where they are nonlinear is of particular interest since such comparisons have apparently not been dis- cussed in the literature. LITERATURE CITED ALLEN, R. L. 1976. Method for comparing fish growth curves. N.Z. J. Mar. Freshwater Res. 10:687-692. BOEHLERT, G. W, AND R. F. KAPPENMAN. 1980. Latitudinal growth variation in the genus Sebastes from the Northeast Pacific Ocean. Mar. Ecol. Prog. Sen 3:1-10. Box, G. E. P, AND M. E. MULLER. 1958. A note on the generation of random normal devi- ates. Ann. Math. Stat. 29:610-611. GALLUCCI, V F, AND T. J. QUINN II. 1979. Reparameterizing, fitting, and testing a simple growth model. Trans. Am. Fish. Soc. 108:14-25. GEISSER, S., and W E EDDY. 1979. A predictive approach to model selection. J. Am. Stat. Assoc. 74:153-160. 101 CURRENT KNOWLEDGE OF LARVAE OF SCULPINS (PISCES: COTTIDAE AND ALLIES) IN NORTHEAST PACIFIC GENERA WITH NOTES ON INTERGENERIC RELATIONSHIPS' Sally L. Richardson^ ABSTRACT Current knowledge of cottid larvae in northeast Pacific genera is summarized. Larvae are known for representatives of 25 of the 40 genera reported from Baja California to the Aleutian Islands although two genera, Gymnocanthus and Icelus, are represented only by species which live in other areas as adults. Included are illustrations of larvae of 29 species representing the 25 genera plus one potentially new northeast Pacific genus, identified only as "Cottoid Type A." The larvae exhibit a wide diversity of form. Based on shared larval characters, including spine patterns, body shape, and pigmentation, 6 phenetically derived groups of genera are apparent within the 25 genera for which representatives are considered: 1) Artedius, Clinocottus, Oligocottus, Orthonopias; 2) Paricelinus, Triglops, Icelus, Chitonotus, Icelinus; 3) Dasycottus, Psychrolutes, Gilbertidia, IMalacocottus, "Cottoid Type A"; 4) Scorpaenichthys, Hemilepidotus; 5) Blepsias, Nautichthys; 6) Leptocottus , Cottus. Six genera do not fit with any group: Enophrys, Gymnocanthus , Myoxocephalus, Radulinus. Rhamphocottus, Hemitripterus . If these preliminary larval groupings reflect relationship, as evidence indicates, they tend to support a number of previously implied relationships wdthin the cottids, but there are some important differences. These include the distinctiveness of the Artedius (Group 1) line; the separation of Artedius and Icelus, once considered closely related; the relationship of Paricelinus, generally considered a primitive and rather distinct form, with other members of Group 2; the apparent relationship of Icelus to other genera in Group 2 and its questionable placement in a separate family; the distinctiveness of Radulinus, previously considered to be related to Chitonotus and Icelinus. The Cottidae, which in this paper are considered broadly to include sculpinlike fishes of Cottidae, Icelidae, Cottocomephoridae, Comephoridae, Nor- manichthyidae, Cottunculidae, and Psychrolut- idae of the suborder Cottoidei of Greenwood et al. (1966), comprises a diverse group of temperate and boreal fishes. Nelson (1976) estimated that the group may contain over 350 species, three-fourths marine, in about 86 genera. They are generally coastal fishes inhabiting all oceans but the Indian. Greatest species diversity occurs in the North Pacific. The systematics of the group are not well understood (Quast 1965; Nelson 1976). Until recently, larvae of relatively few cottids had been described. They were a difficult group to identify in ichthyoplankton collections, particu- larly in the northeast Pacific where 40 genera are reported to occur between Baja California and the 'This paper was presented at the Second International Sym- posium on The Early Life History of Fish (sponsored by ICES, lAO, ICNAF, lAOB, SCOR) held at Woods Hole, Mass., 2-5 April 1979. An abstract of the paper appeared in the sjrmposium publication. ^Gulf Coast Research Laboratory, East Beach Drive, Ocean Springs, MS 39564. Aleutian Islands (Table 1). With the recent work by Richardson and Washington (1980), larvae are now known for representatives of 25 of these 40 cottid genera, although two genera, Icelus and Gymnocanthus, are represented only by larvae of species that live in other areas as adults. The purpose of this paper is twofold. It presents for the first time a summary of important cottid larval characters (those characters occurring only during the larval period and most useful in identi- fying and distinguishing species) based on the larvae of these 25 northeast Pacific genera. (Lar- vae of these genera that are known for species inhabiting other areas as adults are also con- sidered.) This knowledge, which is a necessary prerequisite for systematic studies using larvae, is presented to provide a foundation to which future work on cottid larvae can be compared and upon which it can be expanded as more larvae become known. The paper also presents a preliminary examination of generic groupings within these northeast Pacific cottid genera based on shared larval characters, i.e., similarity. These phenetic groupings, even though preliminary, are helpful Manuscript accepted August 1980. FISHERY BULLETIN: VOL. 79, NO. 1, 1981. 103 FISHERY BULLETIN: VOL. 79, NO. 1 Table l. — List of cottid genera occuring in the northeast Pacific Ocean between Baja California and the Aleutian Islands based on Howe and Richardson (text footnote 3) with a summary of illustrations (accessible to author) of larvae known for those genera world- wide. Lengths (millimetersl of larvae are reported as they appeared in the literature: NL = notochord length; SL = standard length; TL = total length; mm = no length definition was given. Genus and species Reference Sizes illustrated Artediellus — Artedius harringtoni A. lateralis Artedius Type 2 Ascelichthys — Asemichttiys — Blepsias cirrhosus Chitonotus pugetensis Clinocottus acuticeps C. analis C. recalvus Cottus asper Dasycottus setiger Enophrys bison E. bubalis^ E. lilljeborgi^ Eurymen — Gilbertidia sigalutes Gymr)ocanthus lierzensteini^ G. tricuspis^ G. ventralis^ Hemilepidotus gilbert i^ H. hemilepidotus H. jordani H. papilio H. spinosus H. zapus Hemitripterus americanus ' H. villosus Icelinus spp.' Icelus bicornis ' Jordania — Leiocottus — Leptocottus armatus Malacocottus ?M. zonurus Type 1 MyoxocephaJus aenaeus ' M. octodecemspinosus^ M. polyacanthocephalus M. quadricornis^ (marine form) Ricfiardson and Washington 1980 Budd 1940 Marliave 1975 Richardson and Washington 1980 Blackburn 1973 Marliave 1975 Richardson and Washington 1980 IVIisitano 1980 Richardson and Washington 1980 Eigenmann 1892 Budd 1940 Morris 1951 Stein 1972 Richardson and Washington 1980 Blackburn 1973 Blackburn 1973 Marliave 1975 Misitano 1978 Richardson and Washington 1980 Cunningham 1891^ Mcintosh and Masterman 1897^ Ehrenbaum 1904^ Ehrenbaum 1905-9 Russell 1976^ Bruun 1925" Rass 1949^ Russell 1976' Blackburn 1973 Marliave 1975 Kyushin 1970 Koefoed 1907 Rass 1949 Khan 1972 Ehrenbaum 1905-9 Gorbunova 1964 Hattori 1964 Gorbunova 1964 Peden 1978 Richardson and Washington 1980 Gorbunova 1964 Peden 1978 Gorbunova 1964 Follett 1952 Peden 1978 Richardson and Washington 1980 Peden 1978 Warfel and Merriman 1944 Khan 1972 Fuiman 1976 Kyushin 1968 Okiyama and Sando 1976 Richardson and Washington 1980 Ehrenbaum 1905-9'° Rass 1949 Jones 1962 Blackburn 1973 Marliave 1975 White 1977 Richardson and Washington 1980 Richardson and Bond" Richardson unpubl. data Perlmutter 1939 Khan 1972 Lund and Marcy 1975 Golton and Marak 1969 Khan 1972 Blackburn 1973 Zvjagina 1963 Khan 1972 Khan and Faber 1974 3.0, 4.7, 6.9 mm NL. 7.3, 9.3, 13.6 mm SL 4.1 mmSL 4,8, 11, 14mmTL 3.0. 4.7. 6.0 mm NL, 7.2, 9.9, 1 1 .8 mm SL 12.2 mm SL 10, 14, 19, 15.5 mm TL 3.0, 6.3 mm NL, 8.5, 11.5, 15.4, 16.6 mm SL 3.0,4.8 mm SL 3.7, 3.9, 6.9 mm NL, 7.6, 10.4, 13.8, 16.5 mm SL ca. 4 mm ca. 4 mm 4.6, 5.0, 7.6, 8.3, 9.9, 10.8, 18.0, 24.3 mm TL 5.5,9.0, 10.8 mm TL 5.2 mm NL, 8.2, 9.9 mm SL 7.4 mm SL 7.5 mm SL lOmmTL 5.0, 5.4, 5.8. 6,7, 7.1 , 7.6 mm SL 4.8, 7.0 mm NL, 9.1 mm SL 5.7 mm Larva (size not given) 5.8, 10, 11 mm 5.8, 10, 11 mm 4.5,5.7,6.4,9.5 mm 5.6, 6.8, 8.7 mm TL 6.8 mm 4.08, 4.2, 5.7, 7.0 mm 7.9,9.5 mm SL 7, 13, 15, 25, 34 mm TL 5.79, 6.59, 7.55 mm 10.7, 12.7, 15.5 mm 9 mm 12.2, 13.9, 15.9 mm TL 15, 18 mm 7.5, 11.4, 17.5 mm 7.1, 11.6, 19.2,24.8,32.5 mm 7.25, 10.5 mm ca. 20 mm SL 5.8,5.9,9.1 mmNL, 10.7, 11.5, 19.0 mm SL 6.4, 10.7, 13.0 mm ca. 20 mm SL 10.7, 13.7 mm 12,21 mmSL ca. 20 mm SL 5.0, 6.6, 8.9 mmNL, 11.0, 11.8, 19.0 mm SL ca. 20 mm SL ca. 12 mm 11.7,^4.5, 18.8 mm TL 12.6, 15.5, 20 mm TL 14.78, 15.57, 16.52 mm SL 11.6, 14.4, 17.4,20.0 mm 3.3, 8.6 mm NL, 10.9, 13.8, 15.2, 16.5, 12.5, 16.6 mm SL 25 mm 12.3 mm ca. 4 mm 7.6,8.3, 12.0 mm SL 8, 12, 13mm TL 4.9 mmNL 5.1,8.1 mmNL, 11.1 mm SL 7.0,9.8, 14.2, 24.0 mm SL 6.6, 7.0, 8.8, 9.8, 10.4, 14.2, 24.0 mm SL 6 mm 5.0,7.1, 9.7, 11.8 mm TL 5.4, 6.1,6.8,7.5, 8.5, 9.2 mm TL 6.8,8.5, 10.5, 15.2 mm TL 7.0,9.5, 10.7, 12.5, 14.5 mm TL 7.7, 10.7 mm SL 12.3, 12.8, 13.6, 14.5, 16.2, 32 mm 12.8, 14.5. 17.0 mm TL 12.8, 14.4, 17.0 mm TL 104 RICHARDSON: CURRENT KNOWLEDGE OF SCULPIN LARVAE Table l. — Continued. Genus and species Reference Sizes illustrated M. scorpius ' Nautichthys oculofasciatus Oligocottus maculosus O. snyderi Orthonopias triads Paricelinus hopliticus Phallocottus — Porocottus — Psychrolutes paradoxus Radulinus asprellus R. boleoides Rhamphocottus richardsoni Scorpaenichthys marmoratus Sigmistes — Stelgistrum — Sternias — Sttegicottus — Synchirus — Thyrisicus — Triglops murrayi^ T. pingeli^ Triglops sp. Zesticelus — Mcintosh and Prince 1890 Mcintosh and Masterman 1897'^ Ehrenbaum 1904'^ Ehrenbaum 1905-9 Koefoed 1907 Rass 1949 Bigelow and Schroeder 1953 Khan 1972 Russell 1976 Blackburn 1973 Marliave 1975 Richardson and Washington 1980 Stein 1972 Stem 1973 Stem 1972 Bolin 1941 Richardson and Washington 1980 Blackburn 1973 Marliave 1975 Richardson and Washington 1980 Richardson and Washington 1980 Blackburn 1973 Marliave 1975 Richardson and Washington 1980 OGonnell 1953 Richardson and Washington 1980 Khan 1972 Ehrenbaum 1905-9 Koefoed 1907 Rass 1949 Blackburn 1973 Richardson and Washington 1980 Larva (size not given) Larva (size not given) 8.24, 8 16, 10. 18 mm 8.2. 10, 18 mm 9.5 mm 7 9.9.3 mm 8.2. 10, 18 mm 7.6, 8.5, 10 4, 14.0. 17.4 mm TL 7.5.9.5, 10, 14 mm 7.5, 13 mm SL 9.5. 13, 17. 26 mm TL 11 7mmNL. 166mm SL 4.6-5.2.6.0,6.6. 9.2 mm TL 4.5-5.2, 6.0, 6.6, 9.2 mm TL 4.5-4.75, 5.5 mm TL ca. 3-4 mm SL 5.6, 6.2 mm NL, 13.8, 18.6, 25.6 mm SL 10.3 mm SL 10.5. 13. 14. 13mm TL 4.7, 7.9. 9.6, 10.9 mm NL, 12.6, 14.4 mm SL 8.7 mm NL 6 7, 10 mm SL 10, 11.5, 15 mm TL 8.4 mm NL, 10.6, 1 1 .7 mm SL 5.85,6.26, 10, 17,30,48 mm 5.3, 7.5, 8.6 mm NL, 8.7. 10.4, 13.8 mm SL 8.4, 11.6, 18.9, 23.4 mm TL 18 mm 13,16.5,22 mm 10 mm 8.3, 12mmSL 6.9 mm NL, 15.4 mm SL ' Not northeast Pacific species. ^As Cottus bubalis. ^As Taurulus bubalis. *As Cottus lilljeborgi. ^As Acanthocottus lilljeborgi. ^As Taurulus lilljeborgi. 'Probably Myoxocephalus scorpius (Laroche text footnote 6). ^Species occurs in northeast Pacific but larvae described from other areas. ^Not identified below genus level, '°As Centridermichthys hamatus. "Text footnote 5. '^As Cottus scorpius. in reducing taxonomic problems. The potential usefulness of the larval groups in providing in- sights into systematic relationships and evolu- tionary trends within this difficult group of fishes is also discussed. The use of larval forms of fishes to elucidate systematic relationships has been demonstrated in a number of groups, e.g., ceratioids (Bertlesen 1941), myctophids (Moser and Ahlstrom 1970, 1972, 1974), gonostomatids (Ahlstrom 1974), scombroids (Okiyama and Ueyanagi 1978), and serranids (Kendall 1979). LARVAL CHARACTERS OF COTTIDS pigmentation. Although meristic characters may be of prime utility in identifying cottid larvae, they persist in adults and are not considered truly larval. Meristic characters for northeast Pacific cottids have been discussed by Richardson and Washington (1980) and Howe and Richardson.^ The purpose of this summary is to point out the kinds of larval characters that are useful for identification and the spectrum in which those characters may be exhibited since cottid larvae manifest a wide diversity of form. This summary is based on only the representa- tives of northeast Pacific cottid genera listed in This is a summary of important larval charac- ters in cottids, i.e., those characters occurring only during the larval period which can be of most use in distinguishing species. These characters in- clude preopercular spine pattern, body shape, and ^Howe, K. D., and S. L. Richardson. 1978. Taxonomic review and meristic variation in marine sculpins (Osteichthys: Cottidae) of the northeast Pacific Ocean. Final rep., NOAA NMFS Contract No. 03-78-MO2-120, 1 January 1978 to 30 September 1978, 142 p. North vilest and Alaska Fisheries Center, National Marine Fisheries Service, NOAA, 2725 Montlake Boulevard East, Seattle, WA 98112. 105 FISHERY BULLETIN: VOL, 79, NO. 1 Table 1. Species from areas outside the northeast Pacific are included when larvae are know^n be- cause of the taxonomic information their larvae may provide. Generic level designations are used throughout the text for continuity and emphasis although larvae of all species (number of species based on the taxonomic status summary by Howe and Richardson footnote 3) in a genus may not be known. In some cases the genera are monotypic (Chitonotus, Dasycottus, Gilbertidia, Leptocottus, Orthonopias, Paricelinus, Rhamphocottus , Scor- paenichthys) and thus larval characters of the genus may readily be defined. At least some developmental stages are known for all six species of Hemilepidotus, providing good generic level definition. In some cases larvae of a few, but not all species v^dthin a genus are known [Artedius (3 species out of 7); Gymnocanthus (3 of 6); Hemitrip- terus (2 of 2 or 3); Myoxocephalus (5 of 18); Oligocottus (2 of 4); Radulinus (2 of 5); Triglops (3 of 9)]. In those instances, constancy of larval characters among species provides good indica- tions of generic level definition. Larvae oflcelinus spp. have only been described at the generic level as none of the eight species have yet been distin- guished. For some genera, larvae are known for only one of a few species: Blepsias (1 of 2), Cottus (1 of 2 brackish water species), Icelus (1 of 13), Malacocottus (1 of 5), Nautichthys (1 of 3), Psychrolutes (1 of 2). In those, generic level defini- tion may not be as precise; however, larvae of all species appear rather distinctive and thus may be good representatives of their genera. In the follow- ing summary those genera which provide the best examples of patterns are listed in parentheses. Principal preopercular spines typically (18 of 25 genera) number 4 {Scorpaenichthys, Icelinus, Leptocottus, Enophrys) and may vary in degree of development. Modifications of this basic pattern may occur (Myoxocephalus, 1 Malacocottus) in which four main spines are present with one or two auxiliary spines. Another pattern consists of multiple preopercular spines, usually small, num- bering up to ca. 25 {Artedius, Clinocottus). Some- times only one spine is present (Rhamphocottus) or none (Psychrolutes, Gilbertidia). Spines in other regions of the head (particularly parietal- nuchal, postocular, posttemporal-supracleithral, opercular) may also be important. General body shape can range from rather stubby and deep (Artedius, Enophrys) to moder- ately slender and elongate (Icelinus, Triglops) to globose (? Malacocottus) . The snout can be quite rounded (Scorpaenichthys, Hemilepidotus) or pointed (Icelinus, Chitonotus). Snout to anus length can be rather short, <40% SL (standard length) (Dasycottus), to moderately long, >60% SL (Rhamphocottus), although this can change with development. The gut may appear tightly compacted (Dasycottus) or be distinctively coiled (Cottus). The hindgut may trail somewhat below the body (Artedius, Clinocottus). Unusual gut diverticula may be present (Artedius, Clinocot- tus). Pectoral fins may be noticeably elongated (Nautichthys) or fanlike early in development ( Myoxocephalus ) . Melanistic pigment patterns range from rela- tively unpigmented to heavily pigmented. Pig- ment may be variously present or absent over the head, snout, cheek, jaws, cleithral base, throat. Pigment over the dorsolateral surface of the gut may vary in intensity, ventrolateral extent, and pattern (e.g., bars, Leptocottus; distinct round melanophores, Enophrys). In some species the entire gut region is pigmented (Paricelinus). The ventral midline of the gut may have a distinct line of melanophores (Co^^us, some Myoxocephalus) or be unpigmented (Scorpaenichthys). The nape may be distinctively pigmented (Artedius, Enophrys). The lateral body surface above the gut may be unpigmented (Chitonotus), have dorsolateral pig- ment not extending to the gut (Radulinus) or be entirely pigmented (Scorpaenichthys). In the tail region posterior to the anus, pigment may be absent (some Triglops, Dasycottus), present along only the ventral midline (Artedius, Chitonotus), present along only the ventral and dorsal midlines [Gymnocanthus , small (<8 mm) Hemilepidotus], or present on the lateral body surface, sometimes in combination with a ventral midline series (Scorpaenichthys, Radulinus, Blepsias). Num- ber, spacing, position, and shape of ventral mid- line melanophores are important as is the pos- terior extent of lateral pigment. Melanophores may variously appear along the caudal fin base (Paricelinus, Chitonotus). Pectoral fins are gen- erally unpigmented, but some species have heav- ily pigmented fins (Psychrolutes, Gilbertidia) or a pigment band along the fin margin (Nautichthys). LARVAL COTTID GROUPS Within the 25 cottid genera considered, 6 groups of genera are apparent based on shared larval characters, i.e., similarity , and 6 genera do not fit into any group (Table 2). Characters within each 106 RICHARDSON; CURRENT KNOWLEDGE OF SCULPIN LARVAE Table 2. — Groupings of 25 cottid genera reported to occur in the northeast Pacific Ocean between Baja California and the Aleutian Islands based on shared larval characters. Group characteristics were based on representative species for which larvae are known, as listed in Table 1. Also included in the groupings is an unidentified larval type, "Cottoid Type A" of Richardson and Washington (1980) which may represent a new genus. Group General characteristics Genera Ungrouped genera Multiple preopercular spines, rounded snout, stubby shape, slightly trailing gut, sometimes with gut protrusions or diverticula Four preopercular spines, pointed snout, moderately slender, postanal pigment when present usually restricted to ventral midline Four principal preopercular spines or none, rounded snout, often globose shape with loose skin, pigmented pectoral fins Four preopercular spines, rounded snout, relatively deep bodied, ca. 4-5 mm NL at hatching, postanal pigment dorsally, ventrally. and laterally Four preopercular spines not pronounced, rounded snout, relatively slender, post- anal pigment dorsally, ventrally, laterally, probably >7 mm NL at hatching, pectoral fins unpigmented or with pigment band near margin Four preopercular spines, rounded snout, relatively slender, no additional head spines, postanal pigment restricted to ventral midline Enophrys. Gymnocanthus, Myoxocephalus. Radulinus, Rhamphocottus. Hemitripterus Artedius, Clinocottus. Oligocottus. Orthonopias Paricelinus , Triglops , Icelus , Chitonotus , Icelinus Dasycottus. Psychrolutes, Gilbertidia, '7 Malacocottus . Cottoid Type A (new genus?) Scorpaenichthys , Hemilepidotus Blepsias , Nautichthys Leptocottus, Cottus group and of each ungrouped genus are summa- rized to facilitate recognition and minimize tax- onomic and identification problems involving cottid larvae. These groupings are based on com- plete developmental series to the extent available, but only representative figures illustrating one point on a developmental continuum are pre- sented (Figures 1-9). The groupings are necessar- ily preliminary because not all species in all genera are known as larvae. The groups described below are not arranged in any particular order. Generic designations are used as discussed in the previous section. Group 1 This is the tightest group among the 25 genera. Included are Artedius, Clinocottus, Oligocottus , and tentatively Orthonopias (Figures 1, 2). The unique multiple preopercular spine pattern dis- tinguishes it from all other groups or genera. [Although a complete series of Orthonopias has not been described and the spine pattern is un- knov^Ti, small larvae (Figure 2) are very similar to Artedius in form and pigment characteristics and are tentatively included in this group.] The stubby body shape, rounded snout, and somewhat trailing gut Eire remarkably consistent within the group. Presently, identification to genus based on larval characters is still difficult and in need of better definition. Characters used to distinguish species (besides fin ray counts) include: number, spacing, and shape of ventral midline melano- phores; intensity of gut pigmentation; presence of unusual gut diverticula; total number of preoper- cular spines and position of largest spines; num- ber of spines (e.g., none, two, cluster) in the parietal and posttemporal-supracleithral regions; presence or absence of pigment on the nape or head. Although the multiple preopercular spine pat- tern persists through the larval period, adults have four preopercular spines with the lower three reduced and the upper variously modified. Rem- nants of the larval serrations have been observed only in adult A. notospilotus (Howe'*). It is unclear which four spines of the larvae persist in adults. Group 2 This is also a rather cohesive group (Figure 3) consisting of slender forms with pointed snouts and four prominent preopercular spines [Paricel- inus, Triglops, Icelus (tentatively), Chitonotus, Icelinus]. This general body shape is remarkably similar among genera and is not found in any other genera considered. All have a relatively short snout to anus distance. Postanal ventral midline pigment is usually present (absent in one species of Triglops) with some additional melano- phores along the caudal fin base. Dorsal midline pigment is usually absent except for a few spots in some Icelinus and possibly a row in some late stage Triglops. Generic differences include degree of gut pigmentation (e.g., darkest in Paricelinus and some Triglops), number and position of ven- tral midline melanophores, and degree of head spination (e.g., postocular spines in Paricelinus and Triglops). "K. D. Howe, Ph.D. candidate. Department of Fisheries and Wildlife, Oregon State University, Corvallis, OR 97331, pers. commun. September 1978. 107 FISHERY BULLETIN: VOL. 79, NO. 1 FIGURE 1.— Larvae of A) Artedius harringtoni (7.3 mm SL), B) Artedius Type 2 (9.9 mm SL), C) Clinocottus acuticeps (7.6 mm SL), D) Oligocottus maculosus (9.2 mm TL) (A-C, Richardson and Washington 1980; D, Stein 1973). 108 RICHARDSON: CURRENT KNOWLEDGE OF SCULPIN LARVAE Figure 2. — Larvae of A) Artedius harringtoni (3.0 mm NL) and B) Orthonopias triads ( = 4 mm) (A, Richardson and Washington 1980; B, Bolin 1941). The tentative placement of Icelus with this group is of interest as it has been considered to constitute a distinct family, the Icelidae (Jordan 1923; Greenwood et al. 1966) based on the presence of scales in adults. Although larvae of Icelus are known to the author only from descriptions in the literature (Table 1), they strongly resemble other members of this group in form and preopercular spine pattern. Inclusion of Paricelinus is also of interest as it has been considered to be a rather distinct and primitive form (Bolin 1947). It pos- sesses five pelvic soft rays, the ancestral condition, whereas the number of soft rays is reduced to three or two in other members of the group. Group 3 Group 3 (Figure 4) consists of the "psychrolutid" cottids [Dasycottus, Psychrolutes, Gilbertidia, ? Malacocottus ^ "Cottoid Type A" (new genus?)] ^Identification of larvae is tentative pending resolution of taxonomic problems of adults at the generic level [see Howe £md Richardson footnote 3 and also Richardson, S. L., and C. E. Bond. 1978. Two unusual cottoid fishes from the northeast Pacific. Unpubl. manuscr., 6 p. + 25 figs. (Available from senior author.) (Paper presented at the American Society of Ichthyolo- gists and Herpetologists, 1978.)] often considered a separate family (Nelson 1976). ["Cottoid Type A" may possibly be Psychrolutes phrictus but positive identification awaits addi- tional specimens — see Discussion by Richardson and Washington (1980). If it is P. phrictus, larval evidence indicates that the species is incorrectly placed and that a new northeast Pacific genus of cottid is in need of description.] This group is not as cohesive as the two previous groups. The most distinctive character of Group 3 is the pattern of pigmentation of the pectoral fin, a pattern not found in any of the other genera considered. In all, at least the basal portion of the fin develops pigment with the entire fin pigmented in Psychro- lutes, Gilbertidia, and small (<9 mm SL) ?Mala- cocottus. Pigment on small (<8 mm SL) Dasycot- tus is restricted to the inside surface of the pectoral fin base but later it develops distally on the outer surface. (The pigment band near the margin of the elongated pectoral fin of Nautichthys is a very different pattern.) Only Dasycottus and IMala- cocottus develop four preopercular spines, the latter genus with an accessory spine at the base of the second spine. All but the more slender Dasy- cottus have relatively rounded snouts and deep bodies. Both ? Malacocottus and "Cottoid Type A" 109 FISHERY BULLETIN: VOL. 79, NO. 1 >[^-^^:m?^>>^,. Figure 3. — Larvae of A) Paricelinus hopliticus (13.8 mm SL), B) Triglops sp. (15.4 mm SL),C) Icelus bicornis (25 mm), D) Chitonotus pugetensis (11.5 mm SL), E) Icelinus sp. (13.8 mm SL) (A, B, D, E, Richardson and Washington 1980; C, Ehrenbaum 1905-9). 110 RICHARDSON: CURRENT KNOWLEDGE OF SCULPIN LARVAE Figure 4. — Larvae of A) Dasycottus setiger (8 mm NL), B) Psychrolutes paradoxus (transforming, 13 mm SL), O Gilbertidia sigalutes (13 mm SL), D) ? Malacocottus zonurus (10.4 mm SL), E) Cottoid Type A (9.8 mm SL) (A-D, original illustrations; E, Richardson and Washington 1980). Ill FISHERY BULLETIN: VOL. 79, NO. 1 develop a pronounced globose appearance with an outer bubble of skin. Small larvae of all forms have head and gut pigment and "Cottoid Type A" also has lateral pigment posterior to the anus. Some lateral pigment develops later in all but Dasycottus. "Cottoid Type A" has unique "thumb- tack" prickles covering the belly region. The pelvic fins of 1 Malacocottus and "Cottoid Type A" often appear to be withdraw^n into pockets of skin. Group 4 Group 4 (Figure 5) includes larvae with four preopercular spines, conspicuously rounded snouts, and relatively deep bodies with rather heavy pigmentation, except at the smallest sizes (Scorpaenichthys, Hemilepidotus) . Differences between genera include the longer gut and the preanal fin fold of Scorpaenichthys and the in- creased head spination oi Hemilepidotus (parietal, nuchal, postocular, posttemporal-supracleithral). Pigmentation is generally heavier in Scorpae- nichthys than in Hemilepidotus of comparable size. It is initially concentrated along the dorsal and ventral midlines, particularly in Hemilepi- dotus, filling in laterally with development. Lar- vae of both are neustonic. The two genera in this group are certainly more A1 H* jf-^? 5K*^ ---.-•'*■* ^^it Q^y "=1^^. f:-{-3^,—^.J..jZf..'JJ-'-~'^- J ^b) Figure 5. — Larvae of A) Scorpaenichthys marmoratus (8.7 mm SL), B) Hemilepidotus spinosus (ILO mm SL), C) H. hemilepidotus (10.7 mm SL) (A-C, Richardson and Washington 1980). 112 RICHARDSON: CURRENT KNOWLEDGE OF SCULPIN LARVAE similar to each other than to any other cottids considered, although Scorpaenichthys was given familial status in the past (Jordan 1923; Tarenets 1941). Group 5 This group consists of two elongated, slender- bodied genera, Blepsias and Nautichthys (Figure 6). Both hatch at a relatively large size, >7 mm NL (notochord length). Both have rounded snouts, relatively heavy pigmentation, and four preoper- cular spines that never become pronounced and sharp. Nautichthys (at least A^. oculofasciatus) develops greatly elongated pectoral fins soon after hatching, each of which develops a pigment band near its distal margin. The genera Blepsias and Nautichthys have been placed in a separate family, Blepsiidae, in the past (Jordan 1923). Group 6 Group 6 (Figure 7) contains Leptocottus and Cottus (as based on the brackish water species C. asper). They share several characters, including the relatively slender body, rounded snout, four preopercular spines, absence of other head spines, ventral midline pigment along gut, and postanal pigment restricted to ventral midline. Both hatch and transform at similar sizes, ca. 3-4 mm NL and ca. 10-12 mm SL, respectively. Leptocottus has a unique gut pigment pattern of bars, and Cottus has a distinctively coiled gut. Ungrouped Genera Enophrys (Figure 8) has four pronounced pre- opercular spines, rounded snout, deep stubby shape, pigmented nape, and postanal pigment only along the ventral midline. It has a postocular spine and opercular spines, and a preanal fin fold. Melanophores over the gut are distinctively round in shape and densely concentrated. This suite of larval characters is not shared by any other genus. Enophrys bears some resemblance to Group 6 (Leptocottus-Cottus) but differs too much to be part of it. The deep body, bulging gut, and pig- mented nape somewhat resemble Group 1 (Ar- tedius et al.) but spine patterns differ drastically. Larvae of E. bubalis and E. lilljeborgi from the North Atlantic (Table 1) are extremely similar to E. bison from the northeast Pacific. Gymnocanthus (Figure 8) apparently never de- velops pronounced preopercular spines, according to the literature (Table 1). Larvae of G. tricuspis ^^*^ Figure 6. — Larvae of A) Blepsias cirrhosus (11 mm NL) and B) Nautichthys oculofasciatus (11.7 mm NL) (A, original illustration; B, Richardson and Washington 1980). 113 FISHERY BULLETIN: VOL. 79, NO. 1 • --^ -,-^ FIGURE?.— Larvae of A) Lep« FIGURE 9.— Larvae of A) Radulinus asprellus (10.9 mm NL), B) Rhamphocottus richardsoni (8.4 mm NL), C) Hemitripterus villosus (17.4 mm SL) (A, B, Richardson and Washington 1980; C, Okiyama and Sando 1976). unpigmented area on the body above the abdom- inal cavity. Small larvae also resemble Scor- paenichthys except that they have a distinct lateral midline series of melanophores and soon develop a pointed snout and more slender body. Rhamphocottus (Figure 9) is one of the most aberrant cottid forms. It develops only one pre- opercular spine, an unusual snout, a deep body, heavy pigmentation, and a preanal fin fold. At small sizes, ca. 6-7 mm NL, it bears some resem- blance to Scorpaenichthys in pigmentation and shape, but it has a longer gut, more pigment ventrally along the head and gut, and a pigmented preanal fin fold. By 8-9 mm SL, the distinct shape of Rhamphocottus is obvious and spinelike prickles develop over the body. The single species has been considered to represent a separate fam- ily (Jordan and Evermann 1898; Jordan 1923; Taranets 1941). Hemitripterus (Figure 9) is also a heavily pig- mented and distinct form. Based on the literature (Table 1), it has four preopercular spines, a moder- 116 RICHARDSON: CURRENT KNOWLEDGE OF SCULPIN LARVAE ately pointed snout, relatively deep body, and long gut. Larvae are quite large, ca. 12-14 mm NL, at hatching. The larvae of H. americanus from the Atlantic and H. villosus from Japan (Table 1) are very similar. The heavy body pigment is a char- acter shared with a number of apparently un- related cottid genera. The genus was considered to constitute a separate family, Hemitripteridae, by Jordan (1923) and Taranets (1941). DISCUSSION The present state of cottid systematics is con- fused and the group and its allies are in need of intensive study (Nelson 1976; Howe and Richard- son footnote 3). Family limits are not well defined (compare, e.g., Jordan 1923; Berg 1940; Taranets 1941; Greenwood et al. 1966; Bailey et al. 1970; Nelson 1976). Some genera still need revision and potential new species remain to be described (e.g.. Nelson 1977; Richardson and Washington 1980; Howe and Richardson footnote 3). Studies of intergeneric relationships have been few (e.g., Regan 1913; Taranets 1941; Bolin 1947; Watanabe 1960) and these had many disagreements (Table 3). Jordan and Evermann's (1898:1879-1800) com- ment of North American Cottidae still has merit, "The family is an extremely varied one which cannot be readily throvvoi into subordinate groups. Almost every species has an individuality of its own " Because of the confused state of cottid system- atics it seems reasonable to consider whether this preliminary summary of 25 genera of cottid larvae may provide insight into intergeneric relation- ships within the group. Whether these pheneti- cally derived larval groupings are indicative of relationships among cottid genera depends on whether the groups actually possess a set of shared, derived characters. Determination of de- rived states of larval characters is difficult when dealing with such a diverse group as the cottids and their allies because the larvae of many species are still unknown and complete developmental series have not been described for many other species. Such determinations are further hindered by the confused state of adult cottid systematics. Although an in depth analysis of derived charac- ter states is beyond the scope of this study, consideration of several factors allows discussion of the potential significance of, and possible rela- tionships within, at least some of the larval cottid groups described here. Larval characters such as spine patterns, relative body form, and pigmenta- tion have been used to demonstrate or clarify systematic relationships in other groups of fishes, e.g., scorpelarchids (Johnson 1974), gonostomatids (Ahlstrom 1974), myctophids (Moser and Ahl- strom 1974), myctophiforms (Okiyama 1974), ma- rine teleosts in general (Ahlstrom and Moser 1976), bothids (Futch 1977), scombroids (Okiyama and Ueyanagi 1978), serranids (Kendall 1979). In these studies, similarity of larval form has been in remarkable agreement with relationships implied from adult characters. Although larval characters have not been used previously as indicators of relationship (i.e., based on synapomorphies) among cottids, it seems highly probable that at least some of these characters would be as useful in cottids as in other groups of fishes. In addition, if the cottids were derived from an ancestral stock related to the Scorpaenidae, the most generalized group in the Order Scorpaeniformes (Gill 1889; Taranets 1941; Bolin 1947) and if Bolin's (1947) ancestral cottid form is valid and Scorpaenichthys closely resembles the primitive condition, then primitive or derived states of at least some larval characters of cottids can be postulated. Primitive states of larval characters may include four strong preopercular spines, relatively deep but compact body, compact gut, lack of gut diverticula, posses- sion of a preanal fin fold, rounded snout, relatively short pectoral fin. Derived character states may include reduced size and/or numbers of preoper- cular spines or modification of the basic pattern of four, slender or globose body, trailing gut, pres- ence of gut diverticula, no preanal fin fold, semi- pointed or pointed snout, elongated pectoral fin. Pigment patterns are more difficult to evaluate as presumably they may possibly reflect responses to habitat or may be more easily modified genetical- ly than other morphological characters. This idea has been generally discredited in other groups where larval characters have been used to imply relationships (e.g., Ahlstrom 1974; Moser and Ahlstrom 1974; Kendall 1979) as pigment patterns have substantiated findings based on other char- acters. Recent experiments on larvae of the zebra- fish, Brachydanio rerio, (Milos and Dingle 1978) have demonstrated constancy in numerical regu- lation of melanophores which indicates larval pigment patterns may not be as plastic as once thought. Among the cottids, Scorpaenichthys is heavily pigmented but Enophrys, also considered to be a relatively primitive form (Sandercock and Wilimovsky 1968), is not. Heavy pigmentation 117 FISHERY BULLETIN: VOL. 79, NO. 1 seems to be related to a neustonic habitat in some (e.g., Scorpaenichthys , Hemilepidotus) but not others (e.g., Radulinus). Relative constancy of pigment pattern (such as presence or absence of lateral pigment posterior to the anus) within a group used in conjunction with other characters, however, may provide additional evidence for within-group relationships. If this line of reasoning and these assumptions are valid, then certain trends seem apparent which may be indicative of relationships. Group 1 (Artedius et al.) appears to be a natural group sharing a number of derived characters not pres- ent in any other group or genus (i.e., multiple preopercular spines, somewhat trailing gut, un- usual gut diverticula, or at least bulging guts). A preanal fin fold is apparently absent and pig- ment pattern is relatively constant. The grouping agrees with findings of Taranets (1941), in part, and Bolin (1947), who considered the genera to be closely related (Table 3). It seems to be a rather specialized group as Bolin (1947) implied, and, based on the distinctiveness of larval characters, may warrant consideration at possibly the sub- familial level. Group 2 iParicelinus et al.) shares the derived slender body form with pointed snout, and also possesses relative constancy of pigmentation, i.e., no lateral pigment. Relationships among at least some of the genera in this group have been implied previously (Table 3). The distinctiveness of larval form within this group suggests a separate line- age; this group may warrant possible subfamilial status. In Group 3, all but Dasycottus share a highly modified larval form tending in degrees toward globose. The constancy of the pigmented pectoral fin is unique among all groups or genera con- sidered. With the possible exception of Dasycottus, the genera appear to bear at least some relation- ship to each other. Group 4 is the most generalized in that a num- ber of primitive character states are exhibited and relationships cannot be assessed on given present Table 3. — Intergeneric relationships of cottids as interpreted by A = Regan (1913), B = Taranets (1941), C = Bolin (1947), and D = Watanabe (1960). Included are only those 25 northeast Pacific genera for which larvae are known and discussed in this paper. Parentheses indicate a more distant relationship. 3 <1> Genus a S 3 O C g o o o o c o o o o <.1 2, c ra P 3 € & 1 tn R- O P F •a ^ i^ .£ 7 mm NL) of lar- vae in Group 6 represents another specialization indicative of relationship; such a large size is not known in any of the other genera except Hemitrip- terus whose larvae are ca. 12-14 mm NL at hatching. Group 6 genera share a somewhat elongate form and constancy of pigmentation, i.e., lack of lateral pigment, although these characters alone do not provide strong evidence of relation- ship. That the six ungrouped genera did not share a set of derived characters suggests that they bear no close relationship with one another. In summary, this preliminary examination of larval characters within 25 genera of cottids has provided some new insights into cottid systemat- ics. Larval evidence seems to support current concepts of generic limits in most instances (e.g., Enophrys, Hemitripterus , Hemilepidotus) and has indicated a potentially new northeast Pacific genus represented by "Cottoid Type A." Larval characters offer support for the distinctiveness of some genera (e.g., Rhamphocottus) and strong relationships among others (e.g., the Artedius group). Some of the larval groupings discussed here tend to support previously implied relation- ships within the cottids (compare Tables 2 and 3) but some important differences seem apparent [e.g., the distinctiveness of Group 1 demonstrated herein; the separation o{ Artedius and Icelus, once considered closely related (Jordan 1923); the rela- tionship of Paricelinus , generally considered a primitive and rather distinct form (Bolin 1947; Sandercock and Wilimovsky 1968), with other members of Group 2; the apparent relationship of Icelus to other genera in Group 2 and its question- able placement in a separate family (Jordan 1923; Greenwood et al. 1966); the distinctiveness of Radulinus , previously considered related to Chi- tonotus and Icelinus (Bolin 1947)]. Because of the wide diversity of form among cottid larvae, they offer great potential for helping to clarify relation- ships and evolutionary trends within this difficult group of fishes. However, larvae of many species remain to be described (rearing may be the best approach), generic limits of larval characters must be defined, and developmental sequences including osteology need to be examined before that potential can be fully realized. ACKNOWLEDGMENTS Many who helped make this paper possible were acknowledged by Richardson and Washington (1980). In addition, larvae for illustration were provided as follows: Dasycottus setiger, J. R. Dunn (Northwest and Alaska Fisheries Center, Nation- al Marine Fisheries Service, NCAA); Psychrolutes paradoxus, Gilbertidia sigalutes, J. Blackburn (Alaska Department of Fish and Game); IMala- cocottus zonurus, P. Wagner and G. Mueller (Uni- versity of Alaska); Blepsias cirrhosus, Myoxo- cephalus polyacanthocephalus , A. Lamb (Pacific Environment Institute, British Columbia) and C. Moffett (Bellingham, Wash.). B. Washington illustrated these specimens and provided tech- nical assistance. N. Y. Khan granted permission to reproduce a figure of Gymnocanthus tricuspis from his dissertation. J. L. Laroche (Oregon State University) provided information on preopercular spines and pigmentation on Myoxocephalus . Con- versations on cottid systematics with K. Howe (Oregon State University) were particularly in- formative and stimulating. K. Howe and B. Wash- ington read the manuscript and made helpful comments. LITERATURE CITED AHLSTROM, E. H. 1974. The diverse patterns of metamorphosis in gonosto- matid fishes - an aid to classification. In J. H. S. Blaxter (editor), The early life history offish, p. 659-674. Springer- Verlag,N.Y. AHLSTROM, E. H., AND H. G. MOSER. 1976. Eggs and larvae of fishes and their role in systematic investigations and in fisheries. Rev. Trav Inst. Peches Marit. 40:379-398. BAILEY, R. M., J. E. FITCH, E. S. HERALD, E. A. LACHNER, C. C. LINDSEY, C. R. ROBINS, AND W. B. SCOTT. 1970. A list of common and scientific names of fishes from the United States and Canada. 3d ed. Am. Fish. Soc. Spec. Publ. 6, 149 p. BERG, L. S. 1940. A classification offishes, both recent and fossil. Tr. Zool. Inst. Akad. Nauk SSSR 5(2):87-517. BERTELSEN, E. 1941. The Ceratioid fishes. Ontogeny, taxonomy, distribu- tion and biology Dana-Rep., Carlsberg Found. 39, 276 p. BIGELOW, H. B., AND W. C. SCHROEDER. 1953. Fishes of the Gulf of Maine. First Revision. U.S. Fish Wildl. Serv., Fish. Bull. 53, 577 p. BLACKBURN, J. E. 1973. A survey of the abundance, distribution, and factors affecting distribution of ichthyoplankton in Skagit Bay. M.S. Thesis, Univ Washington, Seattle, 136 p. BOLIN, R. L. 1941. Embryonic and early larval stages of the cottid 119 FISHERY BULLETIN: VOL. 79, NO. 1 fish Orthonopias triads Starks and Mann. Stanford Ichthyol. Bull. 2:73-82. 1947. The evolution of the marine Cottidae of California with a discussion of the genus as a systematic category. Stanford Ichthyol. Bull. 3:153-168. BRUUN, A. F. 1925. On the development and distribution of the Norway bullhead iCottus lilljeborgi Collett). Cons. Perm. Int. Explor. Mer, Publ. Circonstance 88, 15 p. BUDD, P L. 1940. Development of the eggs and early larvae of six California fishes. Calif. Div Fish Game, Fish Bull. 56, 50 p. COLTON, J. B., JR., AND R. R. MARAK. 1969. Guide for identifying the common planktonic fish eggs and larvae of continental shelf waters, Cape Sable to Block Island. Bur. Commer. Fish., Biol. Lab., Woods Hole, Mass., Lab. Ref 69-9, 43 p. CUNNINGHAM, J. T. 1891. On some larval stages of fishes. J. Mar. Biol. Assoc. U.K., New Ser., 2:68-74. EHRENBAUM, E. 1904. Eier und Larven von Fischen der deutschen Bucht. III. Fische mit festsitzenden Eiem. Wiss. Meeresunters. Komm. wiss. Unters. Dtsch. Meere Kiel Biol. Anst. Helgol. 3:127-200. 1905-9. Eier und Larven von Fischen. Nordisches Plankton, Zool. Teil, 1:1-413. EIGENMANN, C. H. 1892. The fishes of San Diego, California. Proc. U.S. Natl. Mus. 15:123-178. FOLLETT, W. I. 1952. Annotated list of fishes obtained by the California Academy of Sciences during six cruises of the U.S.S. Mulberry conducted by the United States Navy off central California in 1949 and 1950. Proc. Calif. Acad. Sci., Ser. 4, 27:399-432. FUIMAN, L. A. 1976. Notes on the early development of the sea raven, Hemitripterus americanus. Fish. Bull., U.S. 74:467-470. FUTCH, C. R. 1977. Larvae of Trichopsetta ventralis (Pisces: Bothiidae), with comments on intergeneric relationships within the Bothidae. Bull. Mar. Sci. 27:740-757. GILL, T 1889. On the classification of the mail-cheeked fishes. Proc. U.S. Natl. Mus. 11:567-592. GORBUNOVA, N. N. 1964. Razmnozhenie i razvitie polucheshuiykh bychkov (Cottidae, Pisces). (Breeding and development of hemilep- idotine sculpins (Cottidae, Pisces).) Tr. Inst. Okeanol. Akad. Nauk SSSR 73:235-252. [Translated from Russ. In T. S. Rass (editor). Fishes of the Pacific and Indian Oceans, biology and distribution, p. 249-266. Isr. Program Sci. TransL, 1966, 266 p.; avail. U.S. Dep. Commer., Natl. Tech. Inf Serv., Springifield, Va., as TT 65-50120.] GREENWOOD, P H., D. E. ROSEN, S. H. WEITZMAN, AND G. S. Myers. 1966. Phyletic studies of teleostean fishes, with a pro- visional classification of living forms. Bull. Am. Mus. Nat. Hist. 131:341-455. HATTORI, S. 1964. Studies on fish larvae in the Kuroshio and adjacent waters. [In Jpn., Engl, synop.] Bull. Tokai Reg. Fish. Res. Lab. 40, 158 p. JOHNSON, R. K. 1974. A revision of the alepisauroid family Scopelarchidae (Pisces: Myctophiformes) . Fieldiana, Zool. 66, 249 p. JONES, A. C. 1962. The biology of the euryhaline fish Leptocottus armatus armatus Girard (Cottidae). Univ. Calif Publ. Zool. 67:321-368. JORDAN, D. S. 1923. A classification of fishes including families and genera as far as known. Stanford Univ. Publ., Univ. Ser, Biol. Sci. 3:79-243. JORDAN, D. S., AND B. W. EVERMANN. 1898. The fishes of North and Middle America. Bull. U.S. Natl. Mus. 47:1241-2183. KENDALL, A. W, JR. 1979. Morphological comparisons of North American sea bass larvae (Pisces: Serranidae). U.S. Dep. Commer, NOAA Tech. Rep. NMFS CIRC-428, 50 p. KHAN, N. Y. 1972. Comparative morphology and ecology of the pelagic larvae of nine Cottidae (Pisces) of the northwest Atlantic and St. Lawrence drainage. Ph.D. Thesis, Univ. Ottawa, 234 p. KHAN, N. Y, AND D. J. FABER. 1974. A comparison of larvae of the deepwater and fourhom sculpin, Myoxocephalus quadricornis L. from North America. 1. Morphological development. In J. H. S. Blaxter (editor), The early life history of fish, p. 703-712. Spnnger-Verlag, N.Y. KOEFOED, E. 1907. Poissons. In L. P. R. d'Orleans, Croisiere ocean- ographique accomplie a bord de la Belgica dans la Mer du Gronland 1905, p. 485-499. Charles Bulens, Brussels. KYUSHIN, K. 1968. The embryonic development and larval stages of Hemitripterus villosus (Pallas). Bull. Fac. Fish., Hokkaido Univ 18:277-289. 1970. Embryonic development and larvae of Gymno- canthus herzensteini Jordan and Starks. Jpn. J. Ichthyol. 17:74-79. LUND, W. A., JR., AND B. C. MARCY, JR. 1975. Early development of the grubby, Myoxocephalus aenaeus (Mitchill). Biol. Bull. (Woods Hole) 149:373-383. MARLIAVE, J. B. 1975. The behavioral transformation from the planktonic larval stage of some marine fishes reared in the labora- tory. Ph.D. Thesis, Univ. British Columbia, Vancouver, 231 p. MCINTOSH, W C, AND A. T MASTERMAN. 1897. The life-histories of the British marine food-fishes. C. J. Clay, Lond., 516 p. MCINTOSH, W C, AND E. E. PRINCE. 1890. On the development and life-histories of the tele- ostean food- and other fishes. Trans. R. Soc. Edinb. 35:665-946. MILOS, N., AND A. D. DINGLE. 1978. Dynamics of pigment pattern formation in the zebra- fish, Brachydanio rerio. II. Liability of lateral line stripe formation and regulation of pattern defects. J. Exp. Zool. 205:217-224. MISITANO, D. A. 1978. Description of laboratory- reared larvae of the buf- falo sculpin, Enophrys bison Girard (Pisces: Cottidae). Copeia 1978:635-642. 1980. A record of internal fertilization in the roughback 120 RICHARDSON: CURRENT KNOWLEDGE OF SCULPIN LARVAE sculpin Chitinotus pugetensis with descriptive notes of their early larvae. Copeia 1980:162-165. MORRIS, R. W. 1951. Early development of the cottid fish Clinocottus recalvus (Greeley). Calif Fish Game 37:281-300. MOSER, H. G., AND E. H. AHLSTROM. 1970. Development of lantemfishes (family Myctophidae) in the California Current. Part 1. Species with narrow- eyed larvae. Bull. Los Ang. Cty. Mus. Nat. Hist. Sci. 7, 145 p. 1972. Development of the lantemfi.sh, Scopelopsis multi- punctatus Brauer 1906, with a discussion of its phylo- genetic position in the family Myctophidae and its role in a proposed mechanism for the evolution of photophore patterns in lantemfishes. Fish. Bull., U.S. 70:541-564. 1974. Role of larval stages in systematic investigations of marine teleosts: The Myctophidae, a case study. Fish. Bull., U.S. 72:391-413. Nelson, J. s. 1976. Fishes of the world. Wiley, N.Y., 416 p. 1977. Fishes of the southern hemisphere genus Neophryn- ichthys (Scorpaeniformes: Cottoidei), with descriptions of two new species from New Zealand and Macquarie Island. J. R. Soc. N.Z. 7:485-511. O'CONNELL, C. P. 1953. The life history of the cabezon Scorpaenichthys marmoratus (Ayres). Calif Dep. Fish Game, Fish Bull. 93, 76 p. OKIYAMA, M. 1974. The larval taxonomy of the primitive myctophiform fishes. In J. H. S. Blaxter (editor), The early life history offish, p. 609-621. Springer- Verlag, N.Y. OKIYAMA, M., AND H. SANDO. 1976. Early life history of the sea raven, Hemitripterus villosus, (Hemitripterinae, Cottidae) in the Japan Sea. [In Jpn., Engl, abstr.] Bull. Jpn. Sea Reg. Fish. Res. Lab. 27:1-10. OKIYAMA, M., AND S. UEYANAGI. 1978. Interrelationships of scombroid fishes: an aspect from larval morphology. Bull. Far Seas Fish. Res. Lab. (Shimizu) 16:103-113. PEDEN, A. E. 1978. A systematic revision of the hemilepidotine fishes (Cottidae). Syesis 11:11-49. PERLMUTTER, A. 1939. Section I. An ecological survey of young fish and eggs identified from tow-net collections. In A biological survey of the salt waters of Long Island, 193§, Part 11, p. 11-71. N.Y. Conserv. Dep., Suppl. 28th Anou. Rep., 1938, Salt-water Surv. 15. QUAST.J. C. 1965. Osteological characteristics and affinities of the hexagrammid fishes, with a synopsis. Proc. Calif Acad. Sci., Sen 4, 31:563-600. RASS, T S. 1949. The composition of fish fauna of the Barents Sea and the systematical characters of the fish eggs and larvae. [In Russ.] Tr. Vses. Nauchno-Issled. Inst. Morsk. Rybn. Khoz. Okeanogr. 17:7-65. REGAN, C. T 1913. The osteology and classification of the teleostean fishes of the order Scleroparei. Ann. Mag. Nat. Hist., Ser. 8, 11:169-184. RICHARDSON, S. L., AND B. B. WASHINGTON. 1980. Guide to identification of some sculpin (Cottidae) larvae from marine and brackish waters off Oregon and adjacent areas of the northeast Pacific. U.S. Dep. Commer., NOAA Tech. Rep. NMFS CIRC-430, 56 p. Russell, F S. 1976. The eggs and planktonic stages of British marine fishes. Acad. Press, Lond., 524 p. SANDERCOCK, F K., AND N. J. WILIMOVSKY. 1968. Revision of the cottid genus Enophrys. Copeia 1968:832-853. STEIN, R. 1972. Identification of some larval Pacific cottids. M.S. Thesis, Humboldt State Coll., Areata, Calif, 41 p. 1973. Description of laboratory-reared larvae of Oligocot- tus maculosus Girard (Pisces: Cottidae). Copeia 1973: 373-377. taranets, a. Y 1941. K klassifikatsii i proiskhozhdenifbychkovsemeystva Cottidae. (On the classification and origin of the family Cottidae.) Izvestiya Akad. Nauk SSSR. Otd. Biol. 1941 (3):427-447. [Translated from Russ., Inst. Fish., Univ. B.C., Mus. Contrib. 5, 28 p.] Warfel, h. e., and D. MERRIMAN. 1944. The spawning habits, eggs, and larvae of the sea raven, Hemitripterus americanus, in southern New England. Copeia 1944:197-205. WATANABE, M. 1960. Fauna Japonica Cottidae (Pisces). Tokyo News Service, 218 p. WHITE, W. A. 1977. Taxonomic composition, abundance, distribution and seasonality of fish eggs and larvae in Newport Bay, California. M.S. Thesis, California State Univ., Fuller- ton, 107 p. ZVJAGINA, O. A. 1963. Materialy po razmnozheniyu i razvituju ryb morya Laptevykh 2. Ledovitomorskaya rogatka, i 3. Aziatskaya Koryushka. (Materials on the reproduction and develop- ment of fish of the Laptev Sea 2. Arctic sculpin and 3. Asiatic smelt.) [In Russ., Engl, summ.] Tr. Inst. Okeanol. 62:3-12. [Engl, transl. in Lib. Nat. Mus. Can., Ottawa.] 121 GROWTH AND AGE STRUCTURE OF LARVAL ATLANTIC HERRING, CLUPEA HARENGUS HARENGUS, IN THE SHEEPSCOT RIVER ESTUARY, MAINE, AS DETERMINED BY DAILY GROWTH INCREMENTS IN OTOLITHS^ David W. Townsend^ and Joseph J. Graham'' ABSTRACT Larval Atlantic herring, Clupea harengus harengus, were sampled in the Sheepscot River estuary, Maine, using both towed, and buoyed and anchored plankton nets from October 1978 to March 1979, to determine growth rates and age structure. Larval densities and length-frequency distributions, de- termined from the buoyed and anchored net samples, and the ages of larvae captured in the towed nets, as determined by daily growth increments in the otoliths, showed that there were at least two normally distributed age-groups of larvae which entered the estuary in November and December The two groups hatched during early October and late November, and each appeared in the estuary when about 4 weeks old. Each of the two age-groups of larvae experienced a reduction in growth rate during the latter half of January and early February. The older of the two groups grew approximately 2.1 mm per week from October to early January and from late February to early March. These older larvae grew little if any, during the midwinter period. The younger of the two groups of larvae showed a similar reduction in growth rate during midwinter and grew about 2.0 mm per week before and about 1.5 nmi per week after this period. Research on larval Atlantic herring, Clupea harengus harengus Linnaeus, has been conducted extensively in the western North Atlantic in re- cent years. This has resulted in numerous ac- counts of the abundance and distribution of the larvae, as well as estimates of the generalized growth rates. The growth of Atlantic herring lar- vae in the Gulf of Maine-Bay of Fundy areas has been reported by Tibbo et al. (1958), Tibbo and Legare (1960), Das (1968, 1972), Sameoto (1972), Graham et al. (1972), and Boyar et al. (1973). These workers used the length-frequency method to determine average growth rates of the larvae. Various studies on the seasonal abundance and size distribution of Atlantic herring larvae have shown that in some years there may be more than one mode in the length-frequency distribution for a particular time and geographical area (Tibbo et al. 1958; Tibbo and Legare 1960; Das 1968, 1972; Graham et al. 1972; Boyar et al. 1973; Graham in press), indicating multiple spawnings. These poly- modal length -frequency distributions of Atlantic herring larvae complicate growth rate estimates 'Ira C. Darling Center Contribution No. 149. ^Department of Oceanography, University of Maine at Orono, Ira C. Darling Center, Walpole, ME 04573." ^Department of Marine Resources, Fisheries Research Labo- ratory, West Boothbay Harbor, ME 04575. since an individual sample may not represent a single homogeneous group of larvae. A relatively new technique for studying the growth of larval fishes was introduced by Pan- nella (1971, 1974). He observed daily growth incre- ments in the otoliths of some tropical and low- temperature adult fishes. Brothers et al. (1976) and Struhsaker and Uchiyama (1976) verified the daily nature of these growth increments in several species of larval fishes. Subsequently, others ap- plied this technique to age and growth studies (Ralston 1976; Taubert and Coble 1977; Barkman 1978). Rosenberg and Lough"* used otoliths to age larval Atlantic herring from Georges Bank. The purpose of our study was to use the otolith aging technique to investigate the growth of Atlantic herring larvae in the Sheepscot River estuary of Maine and to examine the age structure of the larvae entering the estuary. METHODS Larval herring were sampled in the Sheepscot River estuary of Maine using both towed, and "Rosenberg, A. S., and R. G. Lough. 1977. A preliminary report on the age and growth of larval herring ( Clupea harengus L.) from daily growth increments in otoliths. 1977/L:26. Manuscript accepted August 1980. FISHERY BULLETIN: VOL. 79, NO. 1, 1981. ICES CM. 123 FISHERY BULLETIN. VOL. 79, NO. 1 buoyed and anchored plankton nets. Shaw^ has shown that when used at night there is a little, if any, difference between these gear with regard to catch rates or larval fish avoidance. Only the sam- ples collected by the towed nets were used for otolith analysis. These towed net samples were collected at night on seven occasions from 24 October 1978 to 6 March 1979 using aim, 0.75 mm mesh plankton net. One daytime sample (10 Jan- uary) was taken with a 61 cm bongo net with 0.505 mm mesh nets on each side. Buoyed and anchored net samples were collected from 5 October 1978 to 27 February 1979 as part of the regular larval Atlantic herring monitoring program conducted by the Maine Department of Marine Resources. The buoyed and anchored nets consisted of six lines of nets fished at four stations in the estuarine channel. Each line had four 0.5 m, 0.75 mm mesh nets, with a digital flowmeter mounted in each net. The nets were set at dusk and retrieved at dawn each sampling date, and fished approxi- mately one semidiurnal tidal cycle. The buoyed and anchored net samples were preserved in 5% Formalin^ and length-frequency distributions and catch rates for larval Atlantic herring deter- mined. The characteristics and performance of the buoyed and anchored nets were reported by Gra- ham and Venno (1968), Graham and Davis (1971), and Graham (1972). The larvae from the towed samples were not preserved, but were sorted within 2 h of collection, placed in plastic Petri dishes, and frozen fresh at -18° C for future otolith analysis. The samples were later thawed and each fish measured to the nearest 0.5 mm. Figure 1 shows that the frozen larvae shrink an average of about 1-2 mm more than Formalin preserved larvae. The sagittae, or largest otoliths, from both sides of the head were teased onto a microscope slide under a binocular microscope. The otoliths were mounted in Per- mount and covered with a glass coverslip. The numbers of daily growth increments in one of each pair of sagittae were counted under a compound microscope at 1,000 x magnification. The incre- ments were counted twice and their mean number computed. Only those otoliths in which there was a difference between counts of 5% or less were used in the analysis. These data were used to esti- LiJ o a: UJ PRESERVED 5% Formalin N = 233 x = 370 FROZEN N=2I9 7 = 35.8 u o UJ Ql '■^Richard F. Shaw, Ph.D. candidate, Department of Ocean- ography, University of Maine at Orono, Ira C. Darling Center, Walpole, ME 04573, pers. commun. May 1980. "Reference to trade names does not imply endorsement by the National Marine Fisheries Service, NOAA. 10 8 6 4 2 12 10 8 6 4 2 LENGTH millimeters (TL ) Figure l. — Length-frequency comparison of 5% Formalin pre- served versus frozen Atlantic herring larvae captured with 61 cm bongo nets on 4 April 1980 in the Damariscotta River estuary, Maine. Larvae from the starboard net were preserved in For- malin and larvae from the port net were frozen. Measurements were performed 2 mo later. A modified t-test (Snedecor and Cochran 1967) showed that the two means were significantly different (P< 0.01). mate daily grovv^h rates and age composition of the larvae. The daily grovd;h increments in larval Atlantic herring otoliths show up very clearly, and only 17 of the 317 larvae examined had oto- liths with increments too faint to be counted accurately. RESULTS Larval Age Structure Changes in modal lengths of larvae during autumn and winter indicate that groups of lar- vae entered the estuary and subsequently lost their identities through differential mortality and growth, since the larvae are not known to depart the estuary once established there (Graham et al. 1972). Length-frequency data from the buoyed and anchored net samples (Figure 2) showed a tri- modal length distribution (range 6-30 mm) for larvae present in the Sheepscot River estuary on 19 October. A large group of smaller larvae entered on 2 November when only a trace of the 124 TOWNSEND and GRAHAM: GROWTH AND AGE STRUCTURE OF LARVAL HERRING liJ < UJ O q: LiJ OCT. 5 1 RANGE N = I7 1 1 1 5 10 OCT. 19 20 life 30 h 40 50 N-335 1 1 N=4I2 15 10 JAN. 3 1 , 5 1 M J_^N=n8 ^ N=225 10 20 30 40 50 LARVAL LENGTH mm (S L) Figure 2. — Length-frequencies of Atlantic herring larvae cap- tured at night in the buoyed and anchored nets in the Sheepscot River estuary, Maine, 5 October 1978 to 27 February 1979. larger larvae from the 19 October population remained. By 16 November there was only one major size mode of larvae in the estuary. During December a second size mode of smaller larvae began to appear and in late December and early January at least two modes were present. By 27 February the two modes were no longer distin- guishable. The appearance of the larval group in December was also evidenced by a leveling off of the decline in larval catch rate for that month (Figure 3). The catch rates then increased in Jan- uary as the second group became established in the estuary. Assuming the otolith growth increments are deposited with a daily periodicity, increment counts can be used to estimate larval age and growth rates. Figure 4 shows the age structure of the Atlantic herring larvae captured with the towed net in the estuary. The larger and presum- ably older larvae represented in the 19 October sample of the buoyed and anchored nets (Figure 2) were not detected in the towed net samples; there appeared to be only one major age-group of larvae occupying the estuary through November. The addition of the second age-group of larvae was readily apparent by 10 January and resulted in a shift of the age-frequencies to the reader's right. This second and younger group of larvae was pres- ent until the last sampling date (6 March), when the remnant larvae from the first group were detected as a negative skewness in the age-fre- quency histogram. The bottom panel of Figure 4 shows the bimodal rO e 4 - A o o N=4874 ^ 3 -z. - J \ CATCH J \ \ r-^ LARVAL ill 1 1 1 1 1 1 II 1 5 19 2 16 4 13 21 3 15 29 8 27 OCT NOV. DEC. JAN. FEB. Figure 3. — Catch rates of Atlantic herring larvae for the buoyed and anchored nets in the Sheepscot River estuary, Maine, based on flowmeter readings. 125 FISHERY BULLETIN: VOL. 79, NO. 1 RANGE OCT. 24 • • N = 4 N=29 RANGE NOV 29 N = 6 o < K ^n Z UJ 2b o ?n a: lij lb Q. 10 5 15 10 5 15 10 5 SEPT. TOTAL OCT. NOV. 5 12 19 26 3 10 17 24; SEPT. OCT 14 21 28 5 NOV. Figure 4. — Age distributions of Atlantic herring larvae caught in the towed nets as determined by daily growth increments in the otoliths. The ages are represented as the times of first daily growth increment formation. The distributions for the individual sample dates and the total for all the sample dates are given. The dates on the abscissa correspond to the time of formation of the first otolith daily growth increment. These dates are determined by subtracting the number of days, represented by the number of otolith increments, from the date of capture. Two modes are apparent in the total histogram. age distribution of larvae from all the towed sam- ples. This bimodal age-frequency distribution was analyzed graphically by the method of Harding (1949) (Figure 5). The total age-frequencies were plotted on probability paper as cumulative per- centages, and were found to fall on a sigmoidal curve having a point of inflection on the 32% vertical. The fitted line EF in Figure 5 is the resul- tant of the two straight lines AB and CD which were found by assuming the data to be bimodally distributed (see Harding 1949 for a more com- plete discussion). Line AB represents those larvae (about 32% of the total) in which the first otolith groMd;h increment was deposited before 30 Octo- ber. This date is approximately the point of inflec- tion of the plotted data. Line CD represents the second, younger group of larvae (about 68% of the total). The two straight lines, AB and CD, indicate that the two groups of Atlantic herring larvae which entered the estuary were normally distrib- uted with respect to age. The mean dates on which the first otolith growth increments were laid down for each age-group were determined by the inter- section of lines AB and CD with the vertical at 50%. The standard deviations of these dates were estimated from the points where the two lines intersect the verticals at 15.87% and 84.13%, the standard deviation being half this distance on the Y-axis. The mean dates on which the first daily grovd;h increments were laid down were approxi- DEC. JAN. X 1- 1 k N L = 300 n — 1 q: I 1- _i 1 1- ^^ IJ- 12 19 26 3 10 Li_ o DEC. JAN. ^ 3 26 O '9 g 12 5 Is > 21 i 1^ 7 24 H O 17 O 10 26 5 GROUP 2 MEAN DATE -^ NOV 23 SO = 19 WEEKS. GROUP I MEAN DATE =^ OCT 10 SD = i8 WEEKS 001 01 I 2 5 10 1 20 30 405060 70 80| 90 95 98 99 998 1587 84 13 CUMULATIVE PERCENTAGE (N =300) Figure 5. — Probability plot of the total bimodal age distribu- tion of Atlantic herring larvae depicted in Figure 4. The dots are the dates on which the first otolith daily growth increment was formed in each larva and are plotted as cumulative percentages. The circles are the cumulative percentages for each of the 2 age- groups using the inflection point of the sigmoidal curve as a dividing point for the two groups. See text for explanation. 126 TOWNSEND and GRAHAM: GROWTH AND AGE STRUCTURE OF LARVAL HERRING mately 10 October for the first group of larvae and 23 November for the second. This analysis indi- cates, along with the length-frequency data (Fig- ure 2) and the larval catch rates (Figure 3), that there were at least two peaks in spawning effort along the Maine coast, separated by about 6 wk, and that the two groups of larvae entered the estuary at separate times. The times of hatching for each group of larvae can be approximated by assuming that the first otolith daily growth incre- ment is formed at the time of yolk-sac absorption (Rosenberg and Lough footnote 4) and allowing about 5 d for yolk-sac absorption at 10° C (Blaxter and Hempel 1966). The two broods of larvae which entered the estuary in November and December were probably hatched, therefore, in early October and mid-November, and were probably spawned in late September and early November. Larval Growth Rates Growth of the larvae was examined separately for each of the two major age-groups which en- tered the estuary. The first group included those in which the first otolith daily growth increment was laid down before 30 October, which is the dividing point between the two age distributions discussed above. The second group of larvae in- cluded those in which the first daily growth incre- ment was laid down on or after 30 October. Both age-groups of larvae experienced approxi- mately a 2-3 wk period of retarded growth. The changes in growi;h rate appeared as breaks in the plotted data in Figures 6 and 7. Figure 2 showed also that modal lengths increased only slightly, if at all, from 29 January to 27 February. The first major group of larvae to enter the estuary showed retarded growth (Figure 6) beginning at a length of about 35 mm and about 80-100 d after 10 Octo- ber, the mean date on which the first otolith daily growth ring was formed (Figure 5). Thus, this period of retarded growrth began during the latter half of January and continued until early Febru- ary. The second major group to enter the estuary showed retarded grovvi;h (Figure 7) beginning at a length of about 26 mm, and 50-60 d after 23 No- vember, the mean date of the first otolith daily growth ring for group 2 (Figure 5). This period of retarded growth also began during the latter half of January and continued until early February. It appears, then, that these two groups of larvae, which differed in age by about 6 wk and in length by about 9 mm, experienced similar reductions in their growth rates during the same period in late January and early February. Apparently the en- vironment at this time was not conducive to their growth. Assuming that growth was interrupted during late January and early February for each of the 2 age-groups of larvae, regression lines were calcu- lated for those larvae caught before the interrup- tion in grovvi:h and for those caught after. The larvae caught 30 January were therefore not in- cluded (Figures 6, 7). The slopes and elevations of the two regression lines for each age-group were compared using the ^-test described by Zar (1974: 228-230). There was no significant difference be- tween slopes for group 1 but the elevations differed significantly (P<0.01). The two regression lines for group 2 differed significantly in slope (P < 0.05) and in elevation (P<0.01). Group 1 larvae, then, grew about 2.1 mm/wk before and after the inter- rupted grovvi;h period. Group 2 larvae grew ap- proximately 2.0 mm/wk before this period and about 1.5 mm/wk after. DISCUSSION Previous workers have reported polymodal length-frequency distributions of Atlantic herring larvae in the Gulf of Maine-Bay of Fundy areas (Tibbo et al. 1958; Das 1968; Graham et al. 1972; Boyar et al. 1973). Graham et al. (1972) detected two broods of Atlantic herring larvae during Sep- tember 1964 in the Boothbay area of the western Gulf of Maine, which includes the Sheepscot River estuary. The two broods were indicated by length- frequency modes of 9 and 13 mm. They reported that in 1965 only a single brood was detected in the area initially and that a second group of smaller larvae appeared in November. They suggested that the variations in lengths of the larvae might be attributed to the location of the Boothbay area within a coastal zone of transition in hatching times. Atlantic herring larvae hatch earlier in the eastern coastal Gulf of Maine than in the west, and may be carried westward and into the Booth- bay area by coastal currents (Graham 1970; Gra- ham et al. 1972). This may explain the variation in modal sizes on 19 October when the buoyed and anchored nets (Figure 2) captured larvae recently hatched (<10 mm) and others obviously older. The two groups of larvae captured in November and December (Figures 2, 4) perhaps also drifted along the coast before entering the estuary, since each group was about 4 wk old when first sampled. 127 FISHERY BULLETIN: VOL. 79, NO. 1 E E X h- LJ 60 p 55 50 45 40 35 30 25 20 15 10 N = 90 A a X 24 OCT (4) 8 NOV (29) 29 NOV (4) 19 DEC (14) 10 JAN (2) 30 JAN (6) 6 MAR (31) Y« 10.59 + 0.29X Y= -7.42 +0.36 X I I I J L J L -I L I I I I 1 L 45 40 35 30 25 -20 E E LU 10 20 30 40 50 60 70 80 90 100 1 10 120 130 140 150 160 170 180 NUMBER OF OTOLITH GROWTH INCREMENTS 190 Figure 6. — Growth of the first group of Atlantic herring larvae which entered the Sheepscot River estuary, 1978-79. This includes all larvae from the towed net samples in which the first otolith daily growth increment was formed before 30 October. The plotted symbols indicate the collection date and the numbers in parentheses indicate sample size. Regression lines were calculated for the samples collected before and after the winter period of interrupted growth and the 30 January samples were therefore not included. The growth rates of Atlantic herring larvae in the Sheepscot estuary as determined using daily growth increments in the otoliths were about 2 mm/wk, excluding the winter period of retarded growth. The growth rate of group 2 larvae, how- ever, was lower after this period. Our estimates of larval growth rates, excluding the retarded growth period, are comparable with autumn and spring values reported by other workers (Table 1). Rosenberg and Lough (footnote 4) used daily growth increments in the otoliths to study the growth of Georges Bank herring larvae. The lar- vae were from a short sampling period (1-18 Octo- ber 1976), but the authors estimated the growth rate to be about 2.4 mm/wk. This October growth Table L — Published growth rate estimates for fall spawned Atlantic herring larvae in the northwest Atlantic. Growth rate (mm/wk) Method used Period studied Source 1.7 Mean lengths First 150 days Tibtx) et al. 1958 <1 Mean lengths Oct. -June Sameoto 1971, 1972 1-2 Mean lengths Nov.- Mar, Boyar et al. 1973 1.4-1.8 Mean lengths Sept-Dec. Graham et al. 1972 2 Modal lengths Sept. and Oct. Das 1968. 1972 <1 Modal lengths Winter Das 1968, 1972 1.5 Modal lengths April Das 1968. 1972 2 Modal lengths May Das 1968, 1972 2.4 Larval otoliths 1-18 Oct. Rosenberg and Lough 1977 rate estimate is greater than our fall and spring estimates, possibly the result of different water temperatures. The relatively wide range in esti- 128 TOWNSEND and GRAHAM: GROWTH AND AGE STRUCTURE OF LARVAL HERRING N = 210 E E X h- o -z. LU A 29 NOV (2) 45 19 DEC (7) X 10 JAN (43) 40 -h 30 JAN (23) • 6 MAR (135) 35 30 25 20 15 10 Y=I0.I6+0.28X •jfi^**» • Y=II.29+0.2IX ^l^/T'^*- n40 35 -- iG 30 E E X 25 ^ 'Z. UJ _l 20 15 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 NUMBER OF OTOLITH GROWTH INCREMENTS Figure 7. — Growth of the second group of Atlantic herring larvae which entered the Sheepscot River estuary, 1978-79. This includes all larvae from the towed net samples in which the first otolith daily growth increment was formed on or after 30 October. The plotted symbols indicate the collection date and the numbers in parentheses indicate sample size. Regression lines were calculated for the samples collected before and after the winter period of interrupted growth and the 30 January samples were therefore not included. mates reported by others were perhaps developed from samples containing more than 1 age-group. Such combinations are difficult to distinguish in length-frequency data as pointed out by Das (1968, 1972) and Graham et al. (1972). Graham (in press) reported that from one to as many as four broods of larvae entered the Sheepscot estuary annually in recent years. The winter retardation of larval growth rate observed in the Sheepscot estuary (Figures 6, 7) was more brief than the general slowing down of growth throughout the winter reported by others (Tibbo et al. 1958; Das 1968, 1972; Graham et al. 1972; Boyar et al. 1973). The duration of slowed growth lasted only 2 or 3 wk, from the latter half of January to early February, and occurred approxi- mately when the second group of larvae was abun- dant in the estuary, but the exact cause of the retarded growth is not clear. Midwinter in general has been shown to be a period of stress for larval herring. Chenoweth (1970) showed that the rela- tive condition of Atlantic herring larvae was poorest in February 1965, 1966, and 1967 and in January 1968 and that these periods of low condi- tion factors coincided with the high mortalities reported by Graham and Davis (1971). Midwinter is a time when food densities are lowest (Sherman and Honey'), when water temperatures approach the lethal limit (Graham and Davis 1971) and when the feeding activity of the larvae is lowest (Sherman and Honey 1971). Any or all of these factors may have contributed to the period of re- tarded growth of larvae in our study. In conclusion, it appears that the ages of larval herring determined by the otolith growth incre- ^Sherman, K., and K. Honey. 1970. Seasonal succession of the food of larval herring in a coastal nursery area. ICNAF Res. Doc. 70/72. 129 FISHERY BULLETIN: VOL. 79, NO. 1 ments are in good agreement with the observed progression of length-frequencies M^ith time, and that such precise age determinations hold much potential for further work on the dynamics of larval fishes. However, it would be advisable to investigate further, under controlled laboratory conditions, the factors controlling the grow^th of Atlantic herring larvae and otolith growth incre- ment deposition. ACKNOWLEDGMENTS We would like to thank Ron Aho, Mike Dunn, Gilbert Jaeger, David Hodges, and Richard Shaw for sampling assistance. We also thank B. J. McAlice and H. H. DeWitt for their helpful suggestions. LITERATURE CITED Barkman, R. C. 1978. The use of otolith growth rings to age young Atlantic silversides, Menidia menidia. Trans. Am. Fish. See. 107:790-792. BLAXTER, J. H. S., AND G. HEMPEL. 1966. Utilization of yolk by herring larvae. J. Mar Biol. Assoc. U.K. 46:219-234. BOYAR, H. C, R. R. MARAK, F. E. PERKINS, AND R. A. CLIFFORD. 1973. Seasonal distribution and growth of larval herring (Clupea harengus L.) in the Georges Bank-Gulf of Maine area from 1962 to 1970. J. Cons. 35:36-51. BROTHERS, E. B., C. B. MATHEWS, AND R. LASKER. 1976. Daily growth increments in otoliths from larval and adult fishes. Fish. Bull., U.S. 74:1-8. CHENOWETH, S. B. 1970. Seasonal variations in condition of larval herring in Boothbay area of the Maine coast. J. Fish. Res. Board Can. 27:1875-1879. DAS, N. 1968. Spawning, distribution, survival, and growth of larval herring ( C/upea harengus L.) in relation to hydro- graphic conditions in the Bay of Fundy Fish. Res. Board Can., Tech. Rep. 88, 162 p. 1972. Growth of larval herring iClupea harengus) in the Bay of Fundy and Gulf of Maine area. J. Fish. Res. Board Can. 29:573-575. Graham, J. j. 1970. Coastal currents ofthe western Gulf of Maine. Int. Comm. Northwest Atl. Fish., Res. Bull. 7:19-31. 1972. Retention of larval herring within the Sheepscot estuary of Maine. Fish. Bull., U.S. 70:299-305. In press. Monitoring winter mortality and spring abun- dance of larval herring, Clupea harengus L., along coastal Maine (1964-1977). Rapp. P.-V. Reun. Cons. Int. Explor Men Graham, J. j., S. B. Chenoweth, and C. W. Davis. 1972. Abundance, distribution, movements, and lengths of larval herring along the western coast of the Gulf of Maine. Fish. Bull., U.S. 70:307-321. Graham, J. J., and c. w. Davis. 1971. Estimates of mortality and year-class strength of larval herring in western Maine, 1964-1967. Rapp. P.-V. Reun. Cons. Int. Explor Mer 160:147-152. Graham, j. j., and R m. w. Venno. 1968. Sampling larval herring from tidewaters with buoyed and anchored nets. J. Fish. Res. Board Can. 25:1169-1179. Harding, j. P. 1949. The use of probability paper for the graphical anal- ysis of polymodal frequency distributions. J. Mar Biol. Assoc. U.K. 28:141-153. PANNELLA, G. 1971. Fish otoliths: daily growth layers and periodical patterns. Science (Wash., D.C.) 173:1124-1127. 1974. Otolith growth patterns: an aid in age determina- tion in temperate and tropical fishes. In T B. Bagenal (editor). Proceedings of an International Symposium on the Ageing of Fish, p. 28-39. Unwin Brothers, Surrey Engl. Ralston, S. 1976. Age determination of a tropical reef butterflyfish utilizing daily growth rings of otoliths. Fish. Bull., U.S. 74:990-994. Sameoto, D. D. 1972. Distribution of herring (Clupea harengus) larvae along the southern coast of Nova Scotia with observa- tions on their growth and condition factor. J. Fish. Res. Board Can. 29:507-515. Snedecor, G. W, and W. G. Cochran. 1967. Statistical methods. 6th ed. Iowa State Univ Press, Ames, 593 p. Sherman, K., and K. a. Honey. 1971. Seasonal variations in the food of larval herring in coastal waters of central Maine. Rapp. P.-V. Reun. Cons. Int. Explor Mer 160:121-124. Struhsaker, P, and j. H. UCHIYAMA. 1976. Age and growth of the nehu, Stolephorus purpureas (Pisces: Engraulidae), from the Hawaiian Islands as indicated by daily growth increments of sagittae. Fish. Bull, U.S. 74:9-17. Taubert, B. D., and D. W. Coble. 1977. Daily rings in otoliths of three species of Lepomis and Tilapia mossambica. J. Fish. Res. Board Can. 34:332-340. TiBBO, S. N., AND J. E. Henri Legar^. 1960. Further study of larval herring i Clupea harengus L.) in the Bay of Fundy and Gulf of Maine. J. Fish. Res. Board Can. 17:933-942. TIBBO, S. N., J. E. Henri legar£, l. W. Scattergood, and R. F Temple. 1958. On the occurrence and distribution of larval herring (Clupea harengus L.) in the Bay of Fundy and the Gulf of Maine. J. Fish. Res. Board Can. 15:1451-1469. ZAR.J. H. 1974. Biostatistical analysis. Prentice-Hall, Englewood Cliffs, N.J., 620 p. 130 FEEDING SELECTIVITY OF SCHOOLS OF NORTHERN ANCHOVY, ENGRAULIS MORDAX, IN THE SOUTHERN CALIFORNIA BIGHT J. Anthony Koslow^ ABSTRACT Direct field measurements of the feeding of five schools of northern anchovy over four sets of conditions indicate consistent size-selective feeding on the dominant zooplankton taxa. At low-to- moderate prey concentrations (10-40 mg carbon per cubic meter), the schools consumed 35-50% of the total zooplankton biomass and >90% of the largest zooplankters present. The schools' feeding was a positive function of prey size primarily. The density of particular prey items did not significantly affect feeding selectivity. The northern anchovy fed preferentially upon a particular species in only one instance. No significant difference was found in the selectivity of two northern anchovy schools composed primarily of late 0-group and Il-group fish, respectively, that were feeding under similar feeding conditions. At prey concentrations of 10-40 mg carbon per cubic meter, the degree of selectivity was inversely related to the size of the largest prey available. The prey size at which consumption is predicted to be 100% was proportional to the size of the largest prey. Field studies have demonstrated that planktiv- orous fish can control zooplankton community structure in oligotrophic lakes and stocked fish ponds by selectively feeding upon the larger, more visible prey organisms (see Gliwicz and Prejs 1977 and Dodson 1979 for a critical discus- sion of this work). However, while highly produc- tive regions in the world's oceans typically sup- port large populations of schooling, planktivorous fish, the impact of these fish populations upon marine zooplankton communities is not known. Taking as an example the estimated consump- tion of zooplankton by the northern anchovy, Engraulis mordax, in the Southern California Bight, it becomes clear that marine fish popula- tions may have considerable impact on the zoo- plankton in the system. The prey consumption of the northern anchovy may be calculated based upon data for the biomass of the population, its annual reproduction and growth, and assump- tions concerning its metabolic efficiency. The results of this calculation can then be compared with estimates of zooplankton production in the region. In the Southern California Bight, the spawning biomass of the northern anchovy in the mid- 1960's to early 1970's averaged between 1.32 and 2.35 X 10^ t over a 40 x 10^ km^ area or 1.34-2.25 g C/m^ (calculated from Smith 1972). Engraulis mordax spawns approximately 20 times annually and produces 389 eggs/g wet weight at each spawning (Hunter and Goldberg 1980). Averaged over the year, this is equivalent to a daily produc- tion rate of 0.43%, based upon a dry weight per egg of 0.030 mg (Hunter and Leong^) or 0.20 mg wet weight (assuming a 15% wet weight:dry weight conversion): [(389 x 20 x 0.2 x 10"^)/ 365] x 100 = 0.43. The growth rate of the northern anchovy past the first year of life is negligible, approximately 0.08% /d from the end of the first year to the end of the third year (calculated from Sakagawa and Kimura 1976). The total daily production of the adult northern anchovy is thus approximately 0.43% + 0.08% = 0.51%. Assuming a 10-30% efficiency of food conversion (Paloheimo and Dickie 1966 and references therein; Jones and Hislop 1978; Lane et al. 1979), mature northern anchovy consmne 1.7-5.1% of body weight daily: 0.51 X 1/0.30 = 1.7; 0.51 x 1/0.10 = 5.1. [The food consumption rate for 0-group northern anchovy is considerably greater since the daily growth rate during the first year is about 6.1% (calcu- lated from Sakagawa and Kimura 1976).] The ma- ture northern anchovy stock therefore consumes 'Scripps Institution of Oceanography A-008, University of California, San Diego, La Jolla, CA 92093; present address: Department of Oceanography, Dalhousie University, Halifax, Nova Scotia B3H 4J1, Canada. ^Hunter, J. R., and R. Leong. The spawning energetics of female northern anchovy, Engraulis mordax. Unpubl. manuscr. Southwest Fisheries Center La Jolla Laboratory, National Marine Fisheries Service, NOAA, PO. Box 271, La Jolla, CA 92038. Manuscript accepted August 1980. FISHERY BULLETIN: VOL. 79, NO. 1, 1981. 131 FISHERY BULLETIN: VOL. 79, NO. 1 0.02-0.11 g C/m^ per d. Primary production in the Bight averages 0.5-1.0 g C/m^ per d (Eppley et al. 1979). The northern anchovy feeds primarily upon zooplankton. Assuming a 20% conversion of primary to secondary production, the spawning population ofE. mordax consumes 10 to >100% of secondary production in the Southern California Bight. This figure appears to give the right order of magnitude, since >80% of acoustic targets from pelagic surveys in this area are estimated to be northern anchovy schools (Mais 1974). It should be noted that the Southern California Bight appears to be an area of net zooplankton consumption, since zooplankton densities are typically greater in the California Current north of the Bight than in the Bight itself (Reid 1962). Presumably the zooplankton are being consumed as they are carried into the area. The northern anchovy also supplements its diet by filter feed- ing on ph3rtoplankton. If the feeding of the northern anchovy is size selective, its impact on the community of zoo- plankton could be considerable. Furthermore, the northern anchovy population is, relatively speaking, not all that large. The population of the Peruvian anchovetta, which is predominant- ly phytophagous but is a zooplankton feeder in certain areas and at certain stages of its life history (Rojas de Mendiola 1971), is estimated to have been an order of magnitude more densely concentrated (Walsh et al. 1980). Most quantitative studies of the feeding selec- tivity of marine fish have been conducted in the laboratory with small groups of fish (Leong and O'Connell 1969; O'Connell 1972; Durbin and Durbin 1975). This permits only a crude approx- imation of their impact upon marine systems, where these fish populations are predominantly found in massive shoals. For example, approxi- mately 90% of the biomass of the northern anchovy population is found in schools >25 t (calculated from Hewdtt et al. 1976). While Eggers (1976) modelled the energetics of planktivorous fish schools using the extensive literature on the feeding of individual fish, there is no experimen- tal data on the feeding of fish schools to test such models. Without better data on the feeding of schooling fish on the zooplankton, contemporary models of marine zooplankton community d5rnam- ics have perforce concentrated upon interactions among the lower trophic levels (Steele 1974; Steele and Frost 1977). I report here the results of in situ measure- ments of the feeding selectivity of schools of E. mordax in the nearshore waters of the Southern California Bight. These represent the first direct quantitative field measurements of the feeding of schools of planktivorous fish. METHODS A vessel with side-scanning sonar and echo sounder was used to track and determine the dimensions of large (25-200 t), near-surface schools of northern anchovy. A school was consid- ered appropriate for study when 1) the school was near the surface and of sufficient size (>50 m along the axis perpendicular to the school's movement) that plankton samples could unequiv- ocally be taken in its wake, 2) the school did not show signs of being disturbed by the ship's pres- ence, and 3) the school was either directly observed to be feeding (October 1976) or its gen- eral configuration and movement were consistent with feeding behavior. It was assumed that when feeding, a school would either form an amorphous "ball" (Radakov 1973) or that its long axis would be normal to its axis of motion (Weihs 1973), and that the school's velocity would not exceed sever- al body lengths per second. When a school was selected for study, a cruci- form drogue with surface buoy was dropped into its center. The school's movements in relation to the drogue were monitored for 10-25 min, during which time the school usually moved 100 to several hundred meters from the drogue. A weighted buoy was then placed over the school. Thus a transect was established, over which the school had passed while presvunably feeding. The school's physical dimensions and swimming speed relative to the water could be determined using the ship's sonar, echo sounder, and by timing the school's movement between the two buoys. In general, the sampling regimen consisted of taking two replicate samples with zooplankton nets first in the wake of the school between the buoys and then in "control" areas either in front of or several hundred meters to the side of the school. The nets were lowered obliquely from the surface to the average depth of the school (as determined by echo sounder), towed at that depth for 2 min, and then hauled to the surface (total length of tow about 100 m). A 0.5 m diameter plankton net (102 /um mesh) with a TSK^ flow- meter was towed in a harness with a 1.0 m 132 KOSLOW: FEEDING SELECTIVITY OF NORTHERN ANCHOVY SCHOOLS diameter (505 /u.m mesh) net with a digital meter. Zooplankton samples were obtained successfully on four cruises conducted in the spring, summer, and fall of 1975-76. All sampling was conducted in daylight hours. Samples of the northern anchovy from the schools were taken by a com- mercial purse seiner on all but one cruise (April 1976) for positive species identification, analysis of their size composition, and examination of gut contents. The sampling scheme varied slightly on sever- al of the cruises: 1) The data from April 1976 represent the results of four replicate tows taken in the wake of the school and three control tows, rather than the two replicates taken for each set of tows on other cruises. 2) On the last cruise of October 1976, the concentration of plankton was measured before and after a school passed through a single patch of water. The control tows were taken first, directly in front of the school; the second set was obtained after the school had passed through the same area. 3) On the first cruise of August 1975, the 0.5 m diameter net was used alone. However, no large zooplankters were found in the samples from this cruise, and those collected were well within the net's range of maximal efficiency, as determined by comparison of catches from this net and the larger, 1 m net on subsequent cruises. Analysis of the plankton samples consisted primarily of determining the size-frequency compo- sition of the zooplankton in the wake of the school as compared with its composition in control tows. I selected for analysis dominant species from the major taxa of zooplankton occurring in the sam- ples (i.e., copepods, chaetognaths, cladocerans, and larvaceans (Table 1)). Species were also se- lected on the basis of size, so that representatives of the smallest and largest commonly occurring zooplankters in each set of samples were enumer- ated. Following Cassie (1968), aliquot size was determined to count 20-50 organisms/size cate- gory; size categories with actual counts <10 were lumped with the adjacent size category. Copepods were enumerated by life history stages, other organisms by body length. To facilitate compari- son, results were converted whenever possible to micrograms carbon (/xg C) using conversions obtained from the literature for Calanus (Mullin and Brooks 1976), microcopepods (Landry 1976, Table l. — Plankton biomass and genera enumerated from control samples of plankton tows taken around northern anchovy schools in the nearshore zone of the Southern California Bight. Genera enumerated Biomass Cruise (mgC/m^) Microzooplankton Macrozooplankton Aug. 1975 '41 Acartia Paracalanus None Sagitta («3 mm) Oikopleura 8 Mar. 1976 ^663 Evadne None 9 Mar. ^639 Evadne None Apr. '33 Acartia Calanus Sagitta (0-3 mm) Sagitta ( * 1 2 mm) Oct. ^10 Acartia Combined Paracalanus Clausocalanus- Ctenocalanus Harpacticoids and cyclopoids None Moderate. ^High. ^Low. 1978; Bartram et al.^), and Sagitta (Reeve 1970; Sameoto 1971). The total zooplankton biomass in the plankton samples was determined from dis- placement volumes of the samples taken with the 0.5 m diameter net; these values were converted to milligrams carbon per cubic meter (mg C/m^) (Wiebe et al. 1975). RESULTS Characteristics of Northern Anchovy Schools The estimated biomass of the five schools studied ranged from 25 to 200 t (Table 2). The length of the schools (the dimension normal to the school's motion) varied by a factor of 4 (55- 200 m). The breadth of the schools (the dimension parallel to the school's axis of motion) was gen- erally less than their length and varied by less than a factor of 2 (30-55 m). The breadth of a feeding school, as a function of the number offish from front to back, is critical to the degree the school depletes the plankton. The lesser variabil- ity in the breadth of the schools may result from behavioral regulation of this parameter, which determines the relative difference in feeding con- ditions encountered from front to back of the school. However, these data are inadequate to "Reference to trade names does not imply endorsement by the National Marine Fisheries Service, NOAA. "Bartram, W. C, D. M. Checkley, and J. F. Heinbokel. 1976. Further use of a deep tank in the study of the planktonic food chain. IMR Rep., IMR Ref 76-7, p. 157-166. Institute of Marine Resources A-018, University of California, San Diego, La Jolla, CA 92093. 133 FISHERY BULLETIN: VOL. 79, NO. 1 Table 2. — Characteristics of northern anchovy schools examined for feeding selectivity: overall school biomass, physical dimensions, velocity and prey consumption, number measured, and mean and range of standard lengths oi EngrauUs mordax within the schools. a b c d e f 9 Dimensions (m) Plankton Body wt. Standard length (mm) Biomass (t wet wt.) Velocity (m/s) consumption (mgC/m^) Cruise Length Front-bacl< Vertical (%) n x±95% C.L. Range Aug. 1975 ^42 200 30 6 — 20.5 — 20 115.20: ►5.02 96-139 8 Mar. 1976 ^25 35 35 "26 — — — 55 122,84±29.17 83-148 9 Mar. ^50 55 "55 20 0.29 — — 16 98.25±6.37 84-127 Apr. '125 100 40 40 0.38 14.85 1.0 — - — Oct. ^200 ^200 "40 20 — 3.47 — 54 99.48: t1.52 90-144 'Percentage body-weight consumed/h = [(volume swept clear/h)(food removed/unit volume)/(school biomass)] x 100 = [(7r(0.5b)(0.5d)(e x m^)(f x 10-3)/(a x 10« x 0.05)] x 100. ^Estimated from physical dimensions, density of 1.5 kg/m^ (Hewitt etal. 1976), and assuming actual shape to be cylinder (August 1975) or oblate spheroid (April 1976). ^Fisherman's estimate based on purse seining of school. "Back-calculated from tonnage, physical dimensions, and assumptions of average school density = 1.5 kg/m^ and of a cylindrical (October 1976) or oblate spheroidal (March 1976) school shape. ^Estimate. evaluate relations between feeding conditions and school size and configuration. Samples of the northern anchovy were obtained from four of the schools. Two of the schools were composed of northern anchovies that had com- pleted approximately 1 year's growth (98-99 mm SL); the other schools were composed of predomi- nantly I-group (115 mm SL) and Il-group (123 mm SL) northern anchovies (Table 2; Sakagawa and Kimura 1976). Since the schools composed of the largest and smallest fish were sampled under the same feeding conditions (i.e., the spring diatom bloom dominated by cladocerans of March 1976; Tables 1, 3), the feeding selectivity of large and small northern anchovies can be compared. Table 3. — Ivlev's Electivity Index (E) as computed from the frequency of size classes of Evadne spp. examined in northern anchovy stomach contents and plankton tows taken in vicinity of the two schools sampled on 8 and 9 March 1976. Body Stomach samples (n = 10) Propor- Plankton tows Replicate Propor- E = length Total tion in sample tion in (r-p)l Date (Mm) count ration (r) counts tows (p) (r+p) 8 Mar. 200-299 9 0.036 44; 30 0.10 -0.47 300-399 75 .30 145; 101 .33 -.05 400-499 42 .17 132; 59 .26 -.21 500-599 70 .28 83; 53 .18 .22 600-699 37 .15 57; 33 .12 .11 700-899 15 .060 2; 2 .0054 .83 Total 248 741 9 Mar. 200-299 4 .03 15; 14 .14 -.64 300-399 10 .08 19; 13 .16 -.35 400-499 30 .23 24; 15 .19 .09 500-599 35 .27 42; 24 .32 -.10 600-699 33 .25 23; 11 .17 .20 >700 19 .15 4;0 .02 .76 Total 131 204 Feeding of Northern Anchovy Schools The impact of the schools' feeding could be clearly determined from the plankton samples during all sampling periods except those of March 1976. In all 40 size categories of prey enumerated from August 1975 and April and October 1976 cruises, the median concentration was less in tows taken in the wake of the school than in control tows (Table 4). These data were analyzed as the fraction consumed [= 1 - (density of organisms in wake of school)/(density in controls)] as a function of the prey organisms' body size. In computing regressions to analyze the feeding selectivity of the northern anchovy schools, it was often not clear either by eye, through analysis of residuals, or from the significance level of the regression whether a linear or curvilinear rela- tionship best fit the data (Figure 1). In these instances, two regressions were performed: a lin- ear regression and a regression in which the independent variable (i.e., prey body size) was loge-transformed. In computing the linear regres- sions (Figure 1), data points are excluded past the first size class at which the school has effectively consumed all the plankton (i.e., when consump- tion is >90%). An arcsine transformation was not performed, although it has been recommended for regressions performed on data expressed as frac- tions or percentages (Sokal and Rohlf 1969). The arcsine transformation did not significantly affect the form of the regressions presented below and only slightly enhanced their significance level. The data are therefore presented untransformed (Figure 1). There was consistently a significant positive relationship between the fraction consumed and the size of the anchovy's prey on the cruises (Figure 1), despite the diversity of prey items within each cruise and the considerable differ- ences in the composition and density of the zoo- plankton between cruises (Tables 1, 4). The frac- tion consumed ranged from 10-30% for the smallest organisms enumerated to 95-100% for 134 KOSLOW: FEEDING SELECTIVITY OF NORTHERN ANCHOVY SCHOOLS Table 4. — Sample counts, density (numbers and biomass per cubic meter), and fraction consumed of prey items examined from control plankton tows and tows taken in the wake of northern anchovy schools, August 1975 and April and October 1976. Control tows Behind-school tows Repli- Median Repli- Median Frac- Body weight cate Abundance (no./m^) biomass cate Abundance (no./m^) biomass tion sample (Acg sample (mQ con- Prey organisms Size class (MgC) counts Median Range C/m3) counts Median Range C/m3) sumed August; Euterpina Total 0.10 128:87 13,300 9,830-16,700 1.330 36,45 5.430 5.370-5.500 543 059 acuUfrons Nauplii Acartia tonsa Total 0.21 275: 302 35,000 34,100-35,900 7.360 135: 197 22,100 20,600-23.500 4.630 37 Nauplii Paracalanus CIII-VI 070 34:53 209 1 78-240 165 9: 12 56.1 55.0-57.3 443 .73 parvus A tonsa CIII-VI 1.35 136:209 828 711-944 1.120 16:31 122 97.7-148 166 85 Sagitta spp 500-999 ixm 0.19 33:11 111 49.7-172 21.1 12; 4 46.2 19.1-733 8 88 58 1,000-1.499 Acm 0.62 264: 145 1,020 655-1,380 631 67:31 279 148-409 173 .73 1,500-1,999 /um 1.38 32:20 129 90.4-167 178 2; 1 8.50 4.80-12.2 11.7 93 2,000-2.999 /xm 3.42 14:12 31.9 27.2-36.6 109 0:0 .00-00 1.00 •3,000 fxm 6.08 9:12 50.6 47.0-54.2 308 0:0 .00-00 1 00 Oikopleura spp. <200 fim 64:22 217 99.4-334 27:22 135 105-165 .38 (Trunk length) 200-299 /im 58:51 267 230-303 25: 18 119 85.9-153 .55 300-399 fim 15; 10 61.8 45.2-78.4 3:6 23.5 18.3-28.6 •62 •400 Aim 6: 10 38.3 31.4-45.2 1:0 3.10 .00-6.10 92 April: A. tonsa CI-IV 1.1 42:38: 26 91:34: 20 424: 279: 19.7 19.4-33.6 21.6 21: 15 13.6 13.1-14.1 15.0 .31 cv 2.3 17.6 14.9-72.7 40.5 21:20 15.8 14.1-17.5 36.4 .10 CVIcJ 3.2 144 84.9-339 461 172: 130 115 114-116 368 .20 114 CVI9 4.25 211:294: 134 152 99.8-169 646 204: 85 106 74.5-137 451 30 Calanus CI 1.4 32: 203: 25.6 14.1-105 35.8 20:13 12.4 11.4-13.4 174 .51 pacificus 19 Cll 26 31: 173: 26 44:68: 30 481:466: 24.8 19.4-89.5 64.5 14:5 689 4.38-9.40 17.9 .72 cm 4.9 35.2 22.3-35.2 172 31; 14 16.6 12.3-20.8 81.1 52 CIV 14 30.9 27.8-46.6 433 137: 137 12.1 11.1-14.8 169 .61 297 164; 147 cv 27 222: 136: 114 10.7 9.00-21.5 289 12; 16; 21:34 1.69 .97-2.60 45.6 84 CVId' 35 55:35: 46 4.31 2.32-5.33 151 1;0; 2;1 .08 .00-0.16 2.80 .98 CVI 9 ' 68 45:25: 27 2.53 1.66-4.36 172 1;0; 1;11 .08 .00-0.84 5.44 97 Sagitta spp. 1.0-1.9 mm 0.96 29:19: 54 43: 26: 40 33: 22: 23 16:21: 9 16: 16: 9 21; 19; 23.2 9.8-40.2 22.3 16:35 20.7 10.7-30.7 19.9 .11 2 0-2.9 mm 3.22 29.8 13.4-34.4 96.0 31: 18 18.3 15.8-20.8 58.9 .39 3.0-3.9 mm 7.13 17.1 11.4-26.4 122 7:7 5.42 4.70-6.13 386 68 4.0-4.9 mm 12.9 12.8 6.70-109 165 11:3 5.01 2.63-7.39 64.6 61 5.0-6.9 mm 25.5 8.28 670-12.8 211 9: 10 7.40 6.04-8.76 189 .11 7.0-7.9 mm' 43 1.26 0.94-2.03 54.2 13; 10; .76 .15-1.08 32.7 .40 10 6:2 8.0-8.9 mm' 58 30:23: 10 1.52 0.94-2.91 882 15: 12: 13: 19 1.26 1.02-1.45 73.1 .17 9.0-9.9 mm' 76 34; 47: 20 3.11 1.87-3.29 236 20: 12: 18:34 1.52 1.30-2.60 116 51 10.0-10.9 mm' 95 47: 56: 35 3.71 3.28-4.55 352 25:6: 14:23 1.43 .65-2.03 136 61 a1 1.0 mm' 119 35:45: 21 298 1 .97-3.39 355 9: 1: 3; 7 039 .11-0 73 464 .87 October: Harpacticoids 100-299 Aim 009 46:34 585 476-694 52.6 70:43 523 415-631 47.1 .11 and cyclopoids 300-499 Mm 0.28 12:7 140 98 0-181 39.2 9: 11 93.6 81.1-106 26.2 .33 A. tonsa ^ Cl-ll 0.29 92:71; 62:52 404 286-557 117 12: 16: 12:21 57.4 45.1-81.0 16.6 .86 CIII-VI 1.33 35:23: 22: 12 135 66-211 180 0:5; 0: 1 5.45 .00-18 7.25 96 Paracalanus- Cl-ll 0.14 205: 184 1,190 1,003-1,424 167 262:216 868 764-984 122 .27 Clausocalanus- 182:259 260: 202 Ctenocalanus^ CIII-VI 0.54 48: 64: 92:89 418 286-509 226 15:21: 12; 12 556 45.4-75.7 30.0 87 'Based on CalCOFI net tows. ^Sample counts for A tonsa and combined Paracalanus-Clausocalanus-Ctenocalanus are presented in pairs, representing replicate counts from tfie same tows. 135 FISHERY BULLETIN; VOL. 79. NO. 1 COPEPOD CEPHALOTHORAX LENGTH (>jm) 200 600 800 H-H — I \ Sagitta TOTAL LENGTH (mm) 1.0 15 2.0 2 5 30 H h 0.3 C(/ig) 7 p jg) PER ORGANISM (A) 6.0 I 00 0.90 80 - 0.70 - cn 0.60 h z o o 2 0.50 g H < 0.40 a: 0.30 0.20 0.10 0.00 (B) X y^ / X /y= 0.0017 +0,11 r2= 94 p < 05 1 1 X Oikopleura spp 1 1 100 200 300 400 500 BODY LENGTH (>jm) PER ORGANISM COPEPOD CEPHALOTHORAX LENGTH (pm) 1000 1500 2000 2400 CEPHALOTHORAX LENGTH (pm) 300 400 500 600 Sagitta TOTAL LENGTH (mm) 2 040 6 80 10 11.0 C Ug ) 0.9 1 A 1 A 1 ^ /I 1 1 / / ^ 0.8 - /-y = 034LN(x) +0.97 / r2=0 76 Q LlJ Z) in z 0.7 0.6 / / p < 05 o o / Ay= 0.59X + 30 2 o 0.5 - // r^ = 0.55 1- 0.4 / . P <.I0 0.3 0.2 : i A MICROCOPEPODS - - 0.1 ■I 1 1 .1 1 1 1 1 1 x/ 1,1 1 (D) / ^^•''"'-y = 12 LN(x) + 0.46 09C — / /"^ r^= 0.69 _ / / / < . / p <; .025 / 0.80 - / \y = 00I2X + 053 / / / r2=0 77 ^ y^ / Q 70 4 / / p-= 025 / ^ ^ LJ J / /^ 5 1 / /^ _) in 0.60 4 1. / tr _ 2 O 1 / '^ ^ii^' 0036X + 027 o y / y^ r2=0 35 z 050 y X * P < 05 - o / y^ 1- — ^ 040 • /• _ tr u. . 0.30 0.20 > ^ • Sagitta spp • A Acortio lonso X Calanus pocificus 0.10 1 1 1 1 1 1 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 C (pg) PER ORGANISM 20 40 60 80 100 120 C (>ig) PER ORGANISM 136 KOSLOW; FEEDING SELECTIVITY OF NORTHERN ANCHOVY SCHOOLS l.uvj 1 1 1 1 1 (E) 0.80 - D / 0.60 - / _ 040 / X UJ Q / ^y = 0023X -1 13 z 0.20 _ °/ r2=0 83 > / p< 01 h- / D > / H 0.00 - / — O a / UJ / _J / UJ -020 - / D _ > / UJ / _I / > -040 - /■ - * — ( / UJ D MARCH 8, 1976 -O60 — _ 1 -0.80 MARCH 9, 1976 - -1.00 1 1 1 1 1 250 350 450 550 650 750 BODY LENGTH (pm) Figure L — Fraction consumed of zooplankton prey items by northern anchovy schools as a function of prey body size ( A-D). Linear regressions were performed on all data sets, excluding from the regression data points past the first size class at which the school consumed >90% of prey. A curvilinear fit to the data based on a regression using a loge-transformation of values on the abscissa is shown where a curvilinear relationship provides as good or better fit. The regressions in Panels A-C include the data from each species found in the panel. In Panel D, the linear and curvilinear regressions in the upper left are based on data for Calanus pacificus alone; the solid lower regression is based upon data for Sagitta spp. and Acartia tonsa; the dashed oblique line is based upon the total data set for the sampling period. Dashed horizontal lines represent the fraction consumed of the total available zooplankton biomass. The significance level for each regression (which is also the significance level for each regression coefficient) is indicated. Panels: A and B — August 1975; C— October 1976; D— April 1976; E— March 1976; Ivlev's Electivity Index (E) as a function of the pigmented body length of Evadne spp. the largest zooplankters (Table 4, Figure 1). Over- all, the northern anchovy schools consumed 35- 50% of the total available zooplankton in the areas sampled (Figure 1). On the summer cruise (August 1975), a diverse assemblage of small plankton was present. Prey consumption appeared to be a function primarily of their size (expressed as their weight in micro- grams carbon) (Figure lA). A single regression, whether linear or curvilinear, adequately de- scribes the northern anchovy's feeding selectivity for such morphologically dissimilar organisms as copepod nauplii, Euterpina acutifrons and Para- calanus parvus; small calanoid copepodites, Acar- tia tonsa and P. parvus; and small chaetognaths, Sagitta spp. There is no indication of species preference. Nor is there evidence that prey density significantly influenced the rate at which they were consumed. While the relative concentration of similar-sized prey items varied widely, they were consumed at equivalent rates [Table 4, compare the density and consumption of copepod nauplii and Sagitta spp. (0.5-1.0 mm), P. parvus CIII-VI and Sagitta spp. (1.0-1.5 mm), and A. tonsa CIII-VI and Sagitta spp. (1.5-2.0 mm)]. The data for the northern anchovy's feeding on the larvacean, Oikopleura spp., from the August cruise was not directly comparable to data for the other species sampled at this time because their body length cannot be converted to a carbon value for the whole organism (including "house"). How- ever, the northern anchovy's feeding on Oiko- pleura appeared to be a linear function of prey size (Figure IB). The autumn cruise (October 1976) was charac- terized by a low density of zooplankton — the standing crop was a factor of 3-4 less than that encountered during the April 1976 and August 1975 cruises — entirely dominated by small zoo- plankton as in the August 1975 cruise (Tables 1, 4). Again the northern anchovy's feeding selectiv- ity was positively related to prey size (Figure IC). A curvilinear relationship here provides a better fit to the data. Only on the cruise of April 1976 were both large and small zooplankton present; however, the zoo- plankton density was comparable to that found on the summer cruise (August 1975) (Table 1, com- pare the range of the prey sizes; Table 5, compare Figure ID with Figure lA, C). As in the summer Table 5. — Maximum prey size and its density, prey size at 100% consumption by northern anchovy schools as predicted from linear regressions, and slope of linear regressions of the school's feeding selectivity of prey size. Apri 1976 Sagitta spp. August Calanus and October Item 1975 pacificus Acartia tonsa 1976 Maximum prey size (mqC) 6.08 68.00 119.00 1.33 Concentration of largest prey item (MgC/m^) 300 172 345 180 Predicted prey size (mqC) at 100% consumption 1.68 39.20 203.00 1.19 Slope of regression of feeding selec- tivity 0.31 0.012 0.0036 0.59 137 FISHERY BULLETIN: VOL. 79, NO. 1 cruise (August 1975), the schoors feeding on both the small copepods, A. tonsa, and the chaetognath, Sagitta spp., increased as a single function of prey weight (Figure ID). However, the school appeared to select the larger copepod, Calanus pacificus, over the other prey examined from this cruise. To test for the differential feeding selectivity for C. pacificus, a single linear regression was per- formed through the pooled data for prey consump- tion from the sampling period (Figure ID) (Quade 1967). All data points for C. pacificus and only 3 of 14 data points for the consumption of Sagitta spp. and A. tonsa lie above this regression line, indicat- ing a significantly heterogeneous distribution of the data (X2 = 11.57; P< 0.01). For the two sets of samples collected from two different schools on consecutive days during the March 1976 cruise, no significant differences were found between the control tows and those taken in the wake of the school either in the displacement volumes or in the plankton's size-frequency com- position. This cruise was undertaken during an intense spring diatom bloom; ambient plankton concentrations were a factor of 20 greater than during any other cruise. At these plankton densi- ties, northern anchovies could fill their stomachs [ca. 5% of body weight (Rojas de Mendiola and Ochoa 1973) for Engraulis ringens or 0.05 g C] in approximately 40 min by simple filtration (fil- tering rate per individual northern anchovy = 2 1/min (Leong and O'Connell 1969)). Thus, the schools, or some part of them, may have ceased feeding. Furthermore, at these high densities of plankton, the schools would have to ingest far more material than on the previous cruises to consume a detectable fraction of the plankton. The fish stomachs examined from both schools were full, but the data on the dimensions and biomass of the schools indicate they were not significantly more densely packed. The schools had thus appar- ently been feeding, at least intermittently, but under these conditions, their feeding selectivity could not be determined from the plankton sam- ples alone. To analyze the northern anchovy's feeding dur- ing this cruise, I compared the size-frequency composition of prey in the stomach contents with that found in the zooplankton tows (Table 3). The data were analyzed using Ivlev's Electivity Index: E = (r - p)l{r +p), where r = the proportion the prey item represents in the diet and p = the proportion the prey represents in the plankton samples (Ivlev 1961). 138 The feeding of the northern anchovy on the cladocerans (predominantly Evadne nordmanni), which dominated the plankton during the spring bloom was a linear function of prey size (Figure IE). No significant difference was found between the electivity of the two schools sampled under similar conditions of food density and composition on this cruise, although northern anchovies from the school of 8 March 1976 were the largest and those sampled on the following day were the smallest encountered during the study. Their difference in mean length {x = 122.84 and 98.25, respectively) indicates the schools were composed predominantly of Il-group and 0-group fish, respectively, with a difference in mean weight of approximately a factor of 2 (calculated from Saka- gawa and Kimura 1976). But this difference in size apparently did not lead to a significant difference in their feeding selectivity under the sampling conditions. Comparison of Feeding Selectivity Between Cruises The three northern anchovy schools studied on the cruises of August 1975 and April and October 1976 each consumed approximately 100% of the largest prey available and a small fraction of the smaller prey. However, because the size distribu- tion of available prey varied greatly between cruises, prey items that were almost entirely removed from the water when only small prey were available (e.g., the later copepodite stages of small copepods, such as A. tonsa or P. parvus, encountered during August 1975 or October 1976) were virtually ignored when larger prey were present (e.g., on the cruise of April 1976, compare Figure lA, C with Figure ID). The prey size at which the northern anchovy school's consumption was approximately 100% on these three cruises (which may be defined as the intersection of the linear regressions with the line y = 1) varied by more than a factor of 100 (Table 5; Figure lA, C, D). Furthermore, while the school's feeding selectivity was consistently a positive function of the prey's size, the slopes of the linear regressions from the cruises of August 1975 and April and October 1976 also varied by more than two orders of magnitude (Table 5). Both factors are related to the size range of prey available to the anchovy on these cruises, which varied to a similar degree. There is a positive relation between the prey KOSLOW: FEEDING SELECTIVITY OF NORTHERN ANCHOVY SCHOOLS size at which the anchovy schoors consumption is about 100%^ and the size of the largest plankters enumerated (Figure 2A). The northern anchovy apparently adjusts