Journal of Plankton Research Vol.9 no. I pp , 1987

Transcription

1 Journal of Plankton Research Vol.9 no. I pp. 95-, 987 Vertical distributions, diel and ontogenetic vertical migrations and net avoidance of leptocephali of Anguilla and other common species in the Sargasso Sea Martin Castonguay and James D.McCleave Department of Zoology and Migratory Fish Research Institute, University of Maine, Orono, ME 9, USA ^Present address: Department of Fisheries and Oceans, Institut Maurice Lamontagne, 85 route de la Mer, CP, Mont Joli, PQ G5H 3Z, Canada Abstract. - and night-time vertical distributions and their ontogenetic changes in Anguilla leptocephali and other common species of leptocephali were determined and compared during five cruises in the Sargasso Sea using an opening closing -m ring net to sample discrete depth strata between m and 35 m deep. No difference in vertical distribution was ever found between Anguilla mstrata (American eel) and A. anguilla (European eel). Anguilla leptocephali <5 mm long did not exhibit a diel vertical migration, as they were distributed between 5 m and 3 m both by day and by night. The vertical distribution of these small leptocephali is probably roughly representative of the depth distribution of adult spawning. Anguilla fc5 mm did perform a diel vertical migration. Anguilla of the length range mm mostly occurred between m and 5 m by day and between 5 m and m by night. Anguilla mm were found deeper than Anguilla < mm by day, between 5 m and 75 m, and mostly between 3 m and 7 m by night. Nemichthys scolopaceus <8 mm did not perform a vertical migration, as they occurred between the surface and m both by day and by night. Larger N. scolopaceus exhibited a diel vertical migration, as most were caught below m by day and above m by night. There was a distinct vertical migration in Derichthys serpenrinus. Larvae of this species mostly occurred between 5 m and 5 m by day and between 3 m and 7 m by night. The other common species of leptocephali (Serrivomer beani, S. brevidentatus, Ariosoma balearicum, Conger oceanicus, Avocettina infans, Nessorhamphus ingolfianus, Anarchias yoshiae, Gymnothorax sp., and Kaupichthys hyoproroides) were mostly found in the upper m of the ocean at night. The ascent of leptocephali seemed to occur rapidly, mostly in the hour following sunset, and the descent to occur more gradually. However, interpretation of these data was difficult because of net avoidance. A high index of visual avoidance by day for most leptocephali, calculated from the ring net series, indicated that most leptocephali were efficient at seeing and escaping the net during daytime, but visual avoidance by day was not important in small Anguilla. Introduction Diel vertical migration of zooplankton usually occurs from deeper waters during the day to shallower waters at night. It mobilizes a far greater biomass than any other regular movement of animals (Longhurst, 97). Although diel vertical migration has been known for more than 5 years, there is still no general agreement as to its adaptive significance, perhaps because no single explanation can account for its adaptive value in all species and in different life stages of a species. Eel larvae (including Anguilliformes, Elopiformes and Notacanthiformes) are an important component of the biomass of the Sargasso Sea. These larvae, referred to as leptocephali, share common morphological characteristics: they are almost transparent and highly laterally compressed; their heads are small; and they possess fang-like teeth. They are mostly epipelagic (Keller, 97) and spend several months in the plankton before metamorphosing into juveniles (Smith, 979). Downloaded from at Fisheries and Oceans on December 8, 5 IRL Press Limited, Oxford, England 95

2 M.Castonguay and J.D.McCkave Schmidt (95), in his famous work on the breeding places of the European eel (Anguilla anguilla) and American eel (Anguilla rostrata), provided little information on vertical distribution. He stated only that larvae 7 5 mm long were found between 75 and 3 m deep, whereas 5 mm larvae were distributed between the surface and 5 m. Schoth and Tesch (98) later determined that Anguilla larvae were concentrated between 5 and m deep by day and between 5 and m deep by night. Subsequently Schoth and Tesch (98) argued that Anguilla leptocephali occur deeper during the day and shallower at night as their body lengths increase. Older leptocephali of the European eel (I- and II-group) have been captured about m deep at night and between 5 and m deep in the daytime (Tesch, 98; Kracht and Tesch, 98). Changes in vertical distribution with age have been termed 'ontogenetic vertical migration' (Russell, 97). True ontogenetic vertical migration should involve both changes in the mean depth and the limits of the migration ranges (Pearre, 979). Here day- and night-time vertical distribution of Anguilla leptocephali were determined and compared and the ontogenetic changes in vertical distribution analyzed. Information on the vertical distribution of leptocephali other than Anguilla is even more sparse. Leptocephali are generally restricted to the upper m of the ocean (Bengtson, 973; Tighe, 975; Keller, 97; Schoth and Tesch, 98). Diel vertical migration could not be demonstrated for many species of leptocephali in these studies because of small daytime catches. Here the day- and night-time vertical distributions of Nemichthys scolopaceus (Nemichthyidae) and Derichthys serpentinus (Derichthyidae) are compared and the ontogenetic changes in vertical distribution are analyzed. The night-time vertical distributions of the following species are compared: Serrivomer beani (Serrivomeridae), S. brevidentatus (Serrivomeridae), Ariosoma balearicum (Congridae), Conger oceanicus (Congridae), Avocettina infans (Nemichthyidae), Nessorhamphus ingolfianus (Derichthyidae), Anarchiasyoshiae (Muraenidae), Gymnothorax sp. (Muraenidae) and Kaupichthys hyoproroidcs (Xenocongridae). Because of small daytime catches in these species, only limited inferences can be drawn on their vertical migrations. Differences in catch rates of invertebrate zooplankton and ichthyoplankton between day and night reported by many authors have been attributed to differential day/night net avoidance (i.e. greater daytime avoidance) (e.g. Clutter and Ankaru, 98; Lenarz, 97; Wiebe et ai, 98). Moreover, night to day ratios of catch of fish larvae have been shown to increase with increasing size of the larvae (Ahlstrom, 95). The problem of net avoidance has long been recognized (see review by Clutter and Ankaru, 98). Theoretical quantitative models have been published in an attempt to evaluate total net avoidance (Barkley, 9, 97; Laval, 97). In this study day- and night-time densities are compared as a means to estimate the visual net avoidance by day among species of leptocephali. A comparison of visual avoidance by day in Anguilla with respect to length is also presented. Application of the quantitative models is not possible. Light is recognized as the dominant stimulus controlling the vertical migrations of fish (Blaxter, 975). The accuracy of the synchronization between the vertical migration of leptocephali and sunrise and sunset is important. Indeed, in order to describe a depth profile of the plankton by pooling samples collected at different times of a given light period, one assumes that vertical migration occurs only at dawn and dusk. In this Downloaded from at Fisheries and Oceans on December 8, 5 9

3 Anguilla and other leptoccpbali in the Sargasso Sea study we present evidence on the time of vertical migration in relation to dawn and dusk in different species of leptocephali. Salinity and temperature have been reported to constrain the vertical movements of organisms in some circumstances (see review by Longhurst, 97). We examine briefly the vertical distribution of Anguilla leptocephali in relation to salinity and temperature. Materials and methods The day- and night-time vertical distributions of leptocephali were studied during five cruises in the Sargasso Sea: February- March 983 (Cape Florida 833), 3 March - April 983 (CF835), July -3 August 98 (Columbus Iselin 88), 9 September-9 October 98 (CI8), and -3 March 985 (CF853). A series of discrete depth samples during daytime and a similar series during nighttime at the same location are hereafter referred to as a depth profile. Two depth profiles were made at different locations during the first 983 cruise and the 98 cruises. The two depth profiles of the second 983 cruise were at the same location. One deptii profile was made in 985. The depth profiles (Figure ) were intended to be in areas of abundance of Anguilla larvae, as determined from Isaacs Kidd Midwater Trawl (IKMT) catches. Seven depth profiles were made in the Sargasso Sea, and one depth profile from each of the two 98 cruises was made in the Gulf Stream for comparison. Samples were taken with a Sea Gear -m diameter ring net fully lined with 5 /tin or 5 /tm Nitex netting and with a 3. m mouth area. The smaller mesh was used in 983 and 985 because of the small size of the Anguilla leptocephali ( mm) at that time of year. In 98 the greater length of Anguilla (8 mm) permitted use of larger mesh, although part of CI88-B was fished with the smaller mesh because of damage to the larger mesh net. The ring net was equipped with a choker, a belly band, and a General Oceanics model M-DT double-trip mechanism, which by opening and closing the net at depth, permitted sampling discrete depth strata. A calibrated Rigosha model 53B flow meter allowed an estimate of the water filtered by the net. A Communication Associates, Inc. (CAI) model CDS pressure-sensitive ultrasonic transmitter attached to a depressor just below the net or to the -m ring itself sent a signal to a CAI model CS- or CN-5 hydrophone towed from the ship. The signal was fed into a CAI model CR- receiver and a CAI model CI- decoder, which transformed the period between signals into a depth display. The decoder had been calibrated previously by entering the parameters of the regression of transmitter depth on signal period. Data for the regression were obtained by lowering the transmitter on a CTD (conductivity-temperature-depth) profiler. For each tow the closed ring net was lowered to the desired depth stratum. A first messenger was then sent, which opened the net. The depth of the net was monitored closely to sample the depth stratum obliquely. At the end of fishing the net was closed by sending a second messenger. The towing speed was 75 ± cm s" (mean ± SD). In 983 depth profiles consisted of samples taken in random order of seven, nonoverlapping, discrete, 5-m depth strata between the surface and 35 m for each light period. The 5 tows made in 983 filtered an average of 79 m 3 (SD = 35 m ) of water; six tows of CF833-A were missed because of a storm. Average tow duration was ± min. As the 983 sampling had shown that Anguilla leptocephali longer than 5 mm were not present in the upper m by day or below 5 m by night, Downloaded from at Fisheries and Oceans on December 8, 5 97

4 M.Castonguay and J.D.McCleave I I I I L. 33 CI8I-A (-5 Oct), CI8O8-A "(8-9Jul) - 3 \ #CF853 CI88-B (l-mar) (8- Aug) CF835 # CI8I-B* (5-9 Apr) (- Oct) -CF83O3-A CF83O3- B! ( - IFeb) (- FebT Fig.. Locations and dales of the depth profiles. Profiles CI8O8-A and CI8-A were located in the Gulf Stream, while the rest of the profiles were in the southwestern Sargasso Sea. Profiles CF835-A and CF835-B were at the same location. the sampling scheme was modified in 98. The day part of a depth profile involved samples taken in random order of eight, discrete, 5-m depth strata between m and 3 m. The strata sampled in random order at night were as follows: 3 m, 3-5 m, 5-7 m, 7- m, -5 m. The amount of water filtered and the average tow duration in 98 were ± m 3 and 8 ± 9 min (n = ) respectively. In 985 the 983 sampling plan was repeated with the exception that the deepest stratum, 3 35 m, was deleted. Unfortunately failure of the double-trip mechanism also prevented obtaining a daytime sample of the 5 3 m stratum. The net filtered 3 ± 385 m 3 of water and fished for 7 ± 3 min (n = ). Profiles CF83O3-A and CI88-A are not considered further because their sample sizes were too small. time sampling never started before sunrise and always ended before sunset. The night-time sampling never started <9 min after sunset and always finished at least 3 min before sunrise. CTD casts, which provided vertical profiles of salinity, temperature and density between the surface and 5 m, were usually accomplished both at dawn and dusk and always at one or the other. Expendable bathythermograph (XBT) casts were made in the middle of the day and in the middle of the night. The time of ascent and descent of leptocephali in the water column in relation to Downloaded from at Fisheries and Oceans on December 8, 5 IPS

5 Anguilla and other leptocephali in the Sargasso Sea sunset and sunrise, respectively, were assessed in 98 by fishing a series of three or four 3 min KMT tows around dawn or dusk between 5 and m at one location. The IKMT was fully lined with.8 mm Nitex netting and had a mouth area of 8.8 m. The samples were preserved in 7.5 % (v/v) seawater formalin that was changed after h. Leptocephali were sorted from the samples on board the ship during the 98 cruises. Approximately half the samples were re-sorted ashore; no additional leptocephali were found. Only initial sorting was possible on board during the 983 and 985 cruises, because of the small size of the Anguilla leptocephali. These samples were re-sorted under a dissecting microscope ashore. The leptocephali were then measured and identified to species if possible using Smith (979) as a guide. Data on vertical distribution of plankton are often presented without statistical treatment because the data do not meet assumptions for parametric analysis. We used a nonparametric statistical method that has recently been adapted by J.D.McCleave et al., (submitted for publication) for plankton data. When the expected values were > 5 (Zar, 98), data were analyzed with a nonstandard chi-square goodness of fit test of the null hypothesis that the day- or night-time density of leptocephali with respect to depth was uniform. A nonstandard chi-square test for heterogeneity tested the null hypothesis that the vertical distribution of leptocephali was similar between species, or between day and night, or at different locations, or for different length classes. The expected value for each depth stratum was obtained by multiplying the observed pooled density of larvae (total number of larvae caught in all the depth strata divided by total amount of water filtered), by the amount of water filtered in that depth stratum. Expected values of chi-square tests are normally derived from theory and total n. Our chi-square tests are nonstandard only in the sense that expected values are also derived from observations on an additional independent variable, the amount of water filtered (J.D.McCleave et al., submitted for publication). The lower daytime catches reported for most species in this study were caused presumably by greater visual net avoidance by day resulting from greater light intensity in the water column. We estimated this component of avoidance by the following calculations. First, the day- and night-time densities for a given species were calculated for each profile using these equations: DD = DCI{DL x W) () ND = NC/(NL X W) () where DD and ND are the densities of a given species per m by day and by night, respectively (i.e. the number of leptocephali found under m of ocean surface); DC and NC are the total day- and night-time catches of a profile; DL and NL are the mean tow lengths (m) by day and by night of the tows that fished within the day- or night-time depth range of that species; and W is the mean width of the net mouth {W = \f 3.). Second, day- and night-time densities were compared by calculating: Downloaded from at Fisheries and Oceans on December 8, 5 A = (ND - DD)/ND (3) where A is the index of visual avoidance by day. The maximum value of A is one. If A =.5, then 5% of the animals caught at night presumably avoided the net during the day. 99

6 M.Castonguay and J.D.McCleave 5 DAY o-j-o NIGHT DAY NIGHT 5 5 "i 3 jf 35 a. u 5 -h o-- o-- - X CF835-A DENSITY (catch/io m 8 «rot«r) Fig.. Vertical dimribution of density (catch ~ m~ 3 of water filtered) of AnguilLa by day and by night for the winter/spring depth profiles. Actual catches are also shown. Profile CF83O3-A is not shown because its sample size is too small (n = 5). Results Vertical distribution and vertical migration of Anguilla leptocephali No difference between the vertical distribution of Anguilla rostrata and A. anguilla was found for any profile (chi-square tests for distribution of the two species were pooled) (Table IA). The day- and night-time vertical distributions of Anguilla were always non-uniform for the five profiles in which the sample size was sufficiently large (chisquare tests, all P <.) (Table IB). For each of the four profiles of the winter/spring cruises the largest concentration of Anguilla occurred between and 5 m by day and between 5 and m by night (Figure ); nevertheless there were differences among profiles especially during daytime. Almost all the larvae of profile CF835-A were captured in the -5 m stratum by day, whereas some larvae also occurred between 5 and m during daytime sampling of CF835-B and between 5 and m during daytime sampling of CF83O5. - and night-time vertical distributions of Anguilla from the winter/spring cruises (Figure ) were always significantly different from each other for the three profiles in which the sample size was large enough (chi-square for heterogeneity, all P <.) (Table IC). time distributions of profiles CF835-A, CF835-B and CF853 were all significantly different from one another (chi-square for heterogenei- Downloaded from at Fisheries and Oceans on December 8, 5

7 Anguilla and other leptocephali in the Sargasso Sea Table I. Chi-square tests for (A) differences in vertical distribution between the two species of Anguilla, (B) nonuniformity of vertical distribution, (O differences in vertical distribution between day and night, (D) differences in vertical distribution between profiles (profiles not shown had sample sizes too small for test). Part Comparison Chi-square test Profile Time d.f. X P A B C D A. rostrata versus A. anguilla Uniformity with depth versus night CF83O5-A versus CF83O5-B CF835-A versus CF853 CF83O5-B versus CF853 Heterogeneity Goodness of fit Heterogeneity Heterogeneity CF835-A CF83O5-A CF835-B CF83O5-B CI8O8-B CI8O8-B CI8-B CF853 CF853 CF83O5-A CF83O5-A CF83O5-B CF83O5-B CI88-B CI8O8-B CI8-B CF853 CF85O3 CF83O5-A CF83O5-B CF85O >.9 >.5 >.9 >. >. >.5 >.5 >.5 >.9 <. <. <. <. <. <. <. <. <. <. <. <. <. >.5 <.5 >.9 <. >. ty, all P <.5) (Table ID). By contrast, no significant differences were observed among the night-time distributions of the three profiles (all P >.5). There was essentially no overlap between day- and night-time vertical distributions of larger Anguilla during the summer/fall cruises, as larvae were almost entirely between 3 and m by night and between 5 and 75 m by day (Figure 3). Profiles CI8O8-B and CI8 were significantly different from each other with respect to their night-time distributions (chi-square for heterogeneity, x =., d.f., P <.). A larger proportion of Anguilla was caught between 3 and 5 m at night in profile CI8 than in profile CI88-B. Because of low catches the daytime distributions of these profiles could not be compared statistically. There was no evidence of diel vertical migration in Anguilla larvae < 5 mm total length; in contrast larger larvae performed a distinct vertical migration (Figure ). The vertical range of occurrence of larvae <5 mm long, 5 35 m, was greater than for any larger size class of larvae, both by day and by night. Anguilla leptocephali of the size range mm were concentrated between and 5 m by day and between 5 and m by night (Figure ), from which the amplitude of the diel vertical migration is inferred to be about 5 m. The Downloaded from at Fisheries and Oceans on December 8, 5

8 M.Castonguay and J.D.McCleave E X a. ui a mn DAY f- o-p 9I ^ 8C -3.pi NIGHT CI88-B, o IS DAY DENSITY (catch /* m s water) r (- fl n NIGHT JH F CI8I Fig. 3. Vertical distribution of density (catch * m~ 3 of water filtered) of Anguilla for the summer/fall profiles. Actual catches are also shown. The results of CI8O8-A and CI8O8-B could not be pooled because the depth strata of the two profiles were different. Only C8O8-B is shown; sample size of CI8O8-A is too small (n = 3). Because the sample size of CI8-A was also small (n = ), the results of profiles CI8-A and C8-B are pooled and arereferredto as CI8. Note the difference in the density scale of CI88-B and CI8. o "g 3 j5 35 a. o IO DAY NIHT DAY NIGHT 8 C mm 8 DAY PERCENTAGE OF TOTAL DENSITY o-- o - o-o, NIGHT Fig.. Vertical distribution of density (expressed as percentage of total density in each length class) of Anguilla by day and by night for different length classes. Actual catches are also shown. Data for larvae < mm long are from the winter/spring cruises, while data for larvae mm are from the summer/fall cruises. Downloaded from at Fisheries and Oceans on December 8, mm length class was also present in the 5- m depth stratum by day, where the mm length class was absent. The daytime distributions of these two size classes were significantly different from each other (chi-square for heterogeneity, x =.85, d.f., P <.), while the night-time distributions were not (x = 5., d.f., P >.5). The presence of eight larvae of the mm length class

9 Anguilia and other leptocephali in the Sargasso Sea 9 TEMPERATURE (*C) SALINITY (X.) DENSITY (SI8MA-T) SALINITY (%.) TEMPERATURE <*C) Fig. 5. Vertical distribution of temperature ( Q, salinity (%o) and density (sigma-t) between the surface and 5 m from a CTD profile of CF853. between and 5 m at night may be explained by their small mean length (. mm) compared to the larger mean length (7.5 mm) of the larvae of the same length class caught between 5 and m, as these small larvae may have not yet achieved a diel vertical migration. While larvae of the length range mm occurred between 5 and m by day, larger larvae were caught deeper by day, between 5 and 75 m. There was no evidence of change in daytime distributions among length classes. 9.9 mm, mm and ^ mm (Figure ). Maximal night-time concentrations of larvae ^ mm occurred between 5 m and 7 m. In contrast to the daytime distributions, the night-time distributions apparently remained about the same once Anguilla reached 5 mm (Figure ). Under the assumption that the day- and night-time centers Downloaded from at Fisheries and Oceans on December 8, 5 3

10 M.Castongnay and J.D.McCleave r SO 5 *l OAY NIGHT DAY NIGHT OAY NIGHT 3 lie 7 H 3 I l ISO 3 35 O o- - - I 3 - *o I? 799mm «a - e 5 ' ec -3) PERCENTAGE OF TOTAL DENSITY ^His o- ^^H !mm Fig.. Vertical distribution of density (expressed as percentage of total density in each length class) of N. scolopaceus by day and by night for different length classes. Actual catches are also shown. Data are from the winter/spring cruises and the summer/fall cruises. of distribution of Anguilla larvae ^ mm are and m respectively, an amplitude of vertical migration of about m was estimated. This value is clearly larger than in Anguilla < mm. Small Anguilla leptocephali from the winter/spring cruises encountered temperature differences of C during their diel vertical migrations. Larger Anguilla from the summer/fall cruises underwent temperature changes of about C during their diel vertical migrations, from about C by day to 7 C by night. The salinity during all cruises was about 3.5%o with less than.5%o change between the depths of day- and night-time distributions. Vertical profiles of temperature, salinity and density from CF853 are presented as an example (Figure 5) for comparison with Figure and Figure (upper panel). Vertical distribution of other leptocephali N. scolopaceus was the most abundant leptocephalus species of the overall catch. Leptocephali ofn. scolopaceus < mm total length were concentrated in the upper 5 m both by day and by night (Figure ). Also, little evidence of vertical migration was found in larvae of the length range mm. However, larvae >8 mm appeared to perform a diel vertical migration as they were mostly caught in the upper m by night and below m by day (Figure ). Such interpretations on the basis of small daytime catches remain tentative. The vertical distribution of larvae < mm was different from that of larger N. scolopaceus, both by day and by night (chi-square for heterogeneity, d.f., all P < Downloaded from at Fisheries and Oceans on December 8, 5

11 Anguilla and other leptocepbali in the Sargasso Sea 3 SO TO 5 5 ITS 5 5 T5 3 3 SO TO DAY NIGHT DAY NIGHT DAY F 8 8 PERCENTAGE OF TOTAL DENSITY SO 8 Fig. 7. VerticaJ distribution of density (expressed as percentage of total density in each length class) of D. serpentinus by day and by night for different length classes. Actual catches are also shown. Data are from the summer/fall cruises only, as no D. serperuinus were caught in the winter/spring cruises..; day, x =.9; night, x = 9.). By contrast, no changes were observed between the night-time distributions of larvae mm long and of larger larvae (x =., d.f., P >.5). However, the night-time distributions of larvae of the length class mm and of larger larvae were different (x = 7.5, d.f., P <.). This analysis was not carried further because of small night-time catches in larger size classes. A trend to deeper daytime distribution with increasing length of N. scolopaceus was also observed (Figure ), but small sample sizes prevented a statistical comparison. A distinct vertical migration was observed in every size class of D. serpentinus, with larvae mostly occurring between 5 and 5 m by day and between 3 and 7 m by night (Figure 7). If the centers of vertical distribution of D. serpentinus by day and by night were and 5 m, respectively, the amplitude of the diel vertical migration is about 9 m. At night larvae of the length range. 9.9 mm were concentrated between 5 and 7 m, whereas a larger proportion of the larger larvae occurred between 3 and 5 m. -time distributions were different both between larvae > mm total length and larger ones (x =.58, d.f., P <.) and between larvae.-9.9 mm long and larger ones (x =.3, d.f., P <.5). However, the night-time distribution of the mm length category was not significantly different from that of larger larvae (x =.7, d.f., P >.5). Small catches prevented statistical comparisons of daytime distributions among size classes. Other common species of leptocephali from the winter/spring cruises were mostly restricted to the upper m of the ocean at night (Figure 8). Interestingly, while most 5. beani were caught between 5 and m at night, S. brevidentatus were mostly found in the top 5 m. A. balearicum was distributed quite evenly between the two Downloaded from at Fisheries and Oceans on December 8, 5 5

12 M.Castonguay and J.D.McCleave o DAY. 5 5 SO "J 3 ^ 35. U 5 5 SO JOO x*n 3 Is a - I - -I o-f -\z - - NIGHT MMM, DAY NIGHT. ^^^^ l i t P 'r oio J \r ] + + O-ll o -^^ n lu- r Jj + A /»* oio DENSITY (catch/io m s wot«r) DAY NIGHT i ^ z... v".~ 8 Fig. 8. Vertical distribution of density (catch ~ m~' of water filtered) of common species of leptocephali (namely Serrivomerbeani, S. brevidentams, Ariosoma balearicum, Conger oceaiucus, and Anarchias yoshiae) by day aod by night for the winter/spring profiles. Actual catches are also shown. Note the difference in the density scale between the upper and the lower panels. Low numbers (n <) of each of these species were also caught in the summer/fall ciuises but are not reported (except for A. balearicum in Table V). upper depth strata at night, whereas most C. oceanicus and A. yoshiae were captured in the 5 m depth stratum (Figure 8). The night-time vertical distributions of these leptocephali from the winter/spring cruises were nonuniform for the species in which the sample size was sufficiently large (chi-square tests, d.f., all P <.; 5. beam, x = 5.77; 5. brevidentatus, y =.8; A. balearicum, = 77.5). Because of low catches, little can be inferred about daytime distributions of this group or the next group of species. Other common species of leptocephali from the summer/fall cruises were also restricted to the upper m of the water column at night (Figure 9).Most A. infans were caught in the upper 5 m. K. hyoproroides was evenly distributed between the two upper depth strata, while most N. ingolfianus and Gymnothorax sp. occurred in the 5 m depth stratum. Net avoidance The indexes of visual avoidance by day (A) of A. rostrata and A. anguilla were positive in eight of cases (Table IT), but this is not a significant result under the null hypothesis of no relative avoidance (one-tailed sign test, P >.). The mean lengths of dayand night-time catches were significantly different in only two of the comparisons possible, but in one comparison the mean length by day was the larger while the opposite was true in the other comparison (Table II). The index of visual avoidance by Downloaded from at Fisheries and Oceans on December 8, 5

13 Anguilla and other leptocephali in the Sargasso Sea too DAY o-ls -- NIGHT A. imfwm ^M o-^hii * * " " ' DAY NIGHT 3 Is ill I I o-^hio. " \] 3 L - - I I I DENSITY (cotch/ m'wottr) Fig. 9. Vertical distribution of density (catch " m" of water filtered) of common species of leptocephali (namely Avoccttina infani, Gymnothoraz sp., Kaupichihys hyoproroides, and Nessorhamphus ingolfianus) by day and by night for the summer/fall profiles. Actual catches are also shown. Because profile CI8O8-A (in which the depth strata were as in the winter/spring profiles) is included with the other summer/fall profiles, the latter profiles have their -3 m and 3-5 m depth strata as well as their 5-7 m and 7- m depth strata pooled. Table D. Index of visual avoidance by day (A), mean length of day catches and mean length of night catches of A. rostrwa and A. anguilla for different profiles. Profile CF833-B CF835-A CF83O5-B CF85O3 CI8O8-B CI8-B A. rostrata A Mean length day (mm). (3) 7. (8) 7. (9) 8.9 (35) 38. (3) 8. (8) Mean length night (mm) 7. () 9. (8).8 (3). () 3. (5) 7. () ', ".'..7 A. anguilla A _ Mean length day (mm). ().8 () 8. (8) 8.8 () - Mean length night (mm) _ 7. () 8.7() 7.9 (5) 9. () 38.5 () Sample sizes are given in parentheses, (-tests: * indicates that the mean length by night is significantly different from the mean length by day, P <.5; ** indicates P <.. Downloaded from at Fisheries and Oceans on December 8, 5 day in Anguilla is independent of body length as A did not vary in any consistent manner among length classes (Table HI). The indexes of visual avoidance by day of leptocephali other than Anguilla from the winter/spring cruises were high but did not seem to be related to mean lengths (Table IV). 7

14 M.Castonguay and J.D.McCteave Table m. Index of visual avoidance by day of Anguilla with respect to length for the winter/spring profiles (species pooled). Profile Length classes (mm) CF83O3-B CF83O5-A CF83O5-B CF85O No significant differences among length classes were found (anova, F f = 3.8 with and 3 d.f., P >.5). Table IV. Index of visual avoidance by day (A) and mean length of leptocephali. W/S = winter/spring profiles combined; S/F = summer/fall profiles combined. Species Anguilla anguilla Anguilla rostrala Anguilla combined Anguilla combined Ariosoma balearicum Conger oceanicus Derichthys serpenlinus Nemichthys scolopaceus Serrivomcr beard Serrivomer brevidentatus Time W/S W/S W/S S/F W/S W/S S/F W/S W/S W/S A Mean length (mm) Sample size Table V. Density (catch *m 3 of water filtered) and number of leptocephali caught between 5 m and m in relation to sunrise and sunset. Data of tows were pooled. Tows were accomplished during the summer/fall cruises. Time of end of tow Sunset Sunset + h Sunset + h Sunrise h Sunrise - h Sunrise Sunrise + h No. of tows Species Anguilla.9 ().7 (8).8 (9).85 (5).5 () Nemichthys scolopaceus < 8 mm long a 8 mm long.57 (3).3 ().53 (8).8 (7).9 (9).5 (9).55 (3).9 ().83 (3).8 () 5. (3) 3.5 () Derichthys serpentinus. (5).8 (5).3().5 (9) Ariosoma balearicum.5 (). (5).8 (7).5 (7).5 ().() Downloaded from at Fisheries and Oceans on December 8, 5 Time of ascent and descent The ascent of Anguilla occurred mostly later than h after sunset. Anguilla leptocephali from the summer/fall cruises were present between 5 and m only at night (Figure 3). Accordingly, the first tow of each time of ascent series, which ended at sunset, did not catch any Anguilla (Table V). The second tow of each series, which occurred 8

15 Anguilla and other leptocepbali in the Sargasso Sea during dusk, caught a few Anguilla, but most were captured in the last tows of the time of ascent series, which ended h after sunset. The time of descent was more gradual, beginning well before sunrise and ending around sunrise. Our findings from the depth profiles indicate that N. scolopaceus < 8 mm do not perform a diel vertical migration, while larger specimens do (Figure ). However, the time of ascent and descent data (Table V) provide slight evidence that this size group does exhibit a vertical migration. Larger specimens seem to ascend well after sunset, though the sample size is small, and to descend about sunrise. D. serpentinus was present in the upper m only at night (Figure 7). Its ascent in the water column took place after dusk (Table V)- Its descent apparently occurred only after sunrise, as specimens were still caught in tows that ended at sunrise. Three of seven A. balearicum caught by day were in the upper m (Figure 8). However, according to the patterns of time of ascent and descent, for which we have a more adequate sample size, this species does perform a diel vertical migration (Table V). Discussion The American eel (A. rostrata) and the European eel (A. anguilla) are so closely related that some authors dispute that they represent two distinct species (Tucker, 959; Williams and Koehn, 98; Williams et al., 98). Considering this closeness, it is not surprising that no difference in vertical distribution between A. rostrata and A. anguilla was found in this study or by Schoth and Tesch (98). However, differences may occur in older leptocephali that would be involved in the geographic separation of the postlarval stages of the two species. Indeed, vertical migrations through a variable current field may affect horizontal displacement of pelagic organisms (Wroblewski, 98; Rothlisberg et al., 983). Differences in vertical migration between the two Anguilla species could partly explain how A. rostrata detrains from the Gulf Stream to invade the North American coast, while A. anguilla presumably stays in the Stream on its way to Europe. The absence of vertical migration and the great vertical range of occurrence observed in Anguilla larvae < 5 mm indicate that newly hatched larvae do not possess the motile capabilities that allow older larvae to control their vertical position. Leptocephali obtained from artificial maturation experiments on Japanese eels (Anguilla japonica) (Yamamoto and Yamauchi, 97; Research group on eel reproduction, 978) and European eels (Bezdenezhnykh et al., 983) hatched 38-5 h and 5 - h following fertilization, respectively. The hatching time for the American eel is unknown, but is probably similar. Leptocephali of A. japonica reached the length of 5 mm about 5 days after hatching (Yamamoto and Yamauchi, 97; Yamauchi et al., 97). The Anguilla larvae < 5 mm in this study were therefore probably spawned no more than 7 days prior to capture. The swimming capabilities of Anguilla larvae develop very early, as larvae s 5 mm vertically migrate. The depth distribution of Anguilla larvae <5 mm would represent the depth distribution of adult spawning under this assumption. First, no vertical mixing took place between spawning and capture. Considering that most larvae <5 mm occurred below the mixed layer, whose depth varied between 5 m and m, little vertical mixing pro- Downloaded from at Fisheries and Oceans on December 8, 5 9

16 M.Castonguay and J.D.McCleave bably occurred. Second, buoyancy did not substantially affect depth distribution of eggs and larvae < 5 mm long. This condition may not have been fulfilled, as pelagic fish eggs in general (Coombs, 98) and Japanese eel eggs in particular (Research group on eel reproduction, 978) are positively buoyant. Bezdenezhnykh et al., (983) also determined that newly-hatched European eel leptocephali are positively buoyant. Because Anguilla eggs and larvae probably are positively buoyant in their natural environment, the depth distribution of newly-hatched larvae may not be directly representative of the depth distribution of adult spawning. Nevertheless depth of spawning can be roughly delimited. Spawning probably does not occur in the top 5 m of the ocean, as no larvae < 5 mm were ever caught there. The much lower catch of larvae < 5 mm in the 3-35 m depth stratum compared to the 5 3 m stratum argues for a spawning mostly taking place in the upper 3 m of the water column. Schoth and Tesch (98) caught no Anguilla <5 mm in tows between 3 and 5 m deep, but they only caught a few () such small leptocephali above 3 m in their series. However, no lower limit to spawning can be fixed as the rising speed of buoyant eel eggs is unknown. The night-time vertical distribution of Anguilla does not change much once the leptocephali reach 5 mm long. Most larvae of the length range mm were found in the 5 m depth stratum at night, and most larvae 5: mm occurred in the 5 7 m stratum at night. There may however be a slight trend toward an increase in the proportion of larvae in the upper 5 m by night with increasing length. In contrast to the night-time distribution, the daytime vertical distribution of Anguilla changes substantially during growth. Larvae mm long were concentrated between and 5 m by day, while larvae. 9.9 mm long also occurred between 5 and m by day. Larger larvae from the summer/fall cruises were distributed more widely and generally occurred deeper by day (5 75 m). These changes in vertical migration with age appear to fulfill Pearre's (979) criteria for a true ontogenetic vertical migration. On the basis of linear regressions of depth of capture on body length, Schoth and Tesch (98) reported that with increasing size larvae are found deeper by day and slightly shallower by night, whereas we found no changes in night-time distribution. The phenomenon of increasing amplitude of diel vertical migration with increasing size has also been reported in other studies on fish larvae (Seliverstov, 97; Hunter and Sanchez, 97; Smith etal., 978; Loeb, 979; Grave, 98). Significant differences were observed in Anguilla among daytime vertical distributions of the winter/spring profiles and among night-time vertical distributions of the summer/fall cruises. The daytime differences among the winter/spring profiles were presumably not due to differences in length distribution of the larvae. Indeed the profile with the greatest mean length of larvae, CF853, was also the one in which the larvae were on average found at the shallowest depth, which is not in accordance with the ontogenetic trend. Differences in cloud cover were probably not responsible either for daytime differences, as the sun was visible most of the time during the three depth profiles. No explanation other than sampling variability can be offered in this case. Interestingly, different moon phases between CF835 (last quarter) and CF853 (new moon) did not affect the night-time distributions of the winter/spring profiles. The night-time differences between the two summer/fall profiles, in which a larger proportion of Anguilla larvae were caught between 3 and 5 m during CI8 than during CI88-B, may have been caused by a difference in cloud cover. Although a Downloaded from at Fisheries and Oceans on December 8, 5

17 AnguiBa and other leptocephali in the Sargasso Sea full moon prevailed during both profiles, the cloud cover was much greater during CI8. The greater mean length of the larvae captured during CI8 (3. mm) than during CI88-B (3. mm) may also partly explain the difference. The time of ascent and descent data indicated that the upward and downward movements of Anguilla in relation to sunset and sunrise, respectively, are spread over a period longer than h. This is a potential source of bias in the sampling for vertical distribution. If the first day- or night-time tow occurred while the animals were still vertically migrating, then it would have sampled neither a day- nor a night-time distribution, but an intermediate crepuscular distribution. However, this is not a problem in this study, as the time of day or night at which the tows were made did not seem to affect the catch. For example, the largest night-time concentrations of Anguilla during the winter/spring profiles were always in the 5 m depth stratum, regardless of whether fishing occurred immediately after dusk or just before dawn. In addition, although the minimum intervals between sunset and start of fishing and between end of fishing and sunrise were 9 and 3 min respectively, these intervals were much longer in most cases. We were justified in not sampling the top m by day during the summer/fall cruises, as no Anguilla leptocephali were captured in the 5 m depth stratum. Depth profiles of N. scolopaceus revealed evidence of a diel vertical migration only in specimens > 8 mm. However, time of ascent and descent data provide some evidence of a vertical migration in larvae <8 mm as well. But net avoidance should be considered in the interpretation of time of ascent and descent. Avoidance may be partly responsible for the observed patterns of increase and decrease in the catch of leptocephali. For example, avoidance may have caused the small catch of N. scolopaceus in the tows that ended at sunset or soon thereafter, and may have caused the slight decline in catch at sunrise. Considering our estimate of A =.73 in this species, it seems likely that avoidance is responsible for the discrepancy in vertical migration of larvae <8 mm. Bengtson (973) presented some evidence for vertical migration in N. scolopaceus (5 m by day and upper 5 m by night) on the basis of nondiscrete trawls. However, he also reported results of discrete trawls that showed no evidence of vertical migration. He did not separate N. scolopaceus by size class. D. serpentinus exhibited a distinct diel vertical migration in every size class. As in Anguilla, the absence of D. serpentinus larvae in the 5 m stratum by day indicated that no larvae were likely to be found above 5 m. Time of ascent and descent data and depth profile data for this species are in agreement as they both argue for the existence of a vertical migration. The sample size of D. serpentinus that Keller (97) obtained was too small to substantiate the existence of a vertical migration. Although profile data on A. balearicum could not show evidence of vertical migration due to small daytime catches, time of ascent and descent data argue for it. Keller (97) found that unlike most leptocephali, A. balearicum is present in the upper 5 m both by day and by night. Considering that we obtained a high value of A (.89) for this species, it is quite possible that avoidance rather than vertical migration was responsible for the apparent pattern of ascent and descent. No evidence of vertical migration in S. beani and S. brevidentatus could be determined in this study or by Tighe (975). Despite low daytime catches, data on A. yoshiae, K. hyoproroides, and N. ingolfianus argue for the existence of a diel vertical migration in these species. This is in agree- Downloaded from at Fisheries and Oceans on December 8, 5

18 M.Castonguay and J.D.McCleave ment with Keller (97). She also found indications of a vertical migration in Gymnothorax, but our data on this genus do not support her findings. Two cautions are necessary, beyond net avoidance, in interpreting the results of time of ascent and descent tows. An apparently more gradual descent than ascent in some species may be due to the descent series continuing to capture specimens migrating down from shallower depths. The data also contain variability from pooling different locations. Temperature and salinity did not appear to constrain vertical movements of leptocephali. During cruise CI8O8, we hypothesized that the upward movement of Anguilla at night was impeded by a thermocline that was then occurring around 5 m. Data from pairs of tows fished above and below that thermocline tended to support the hypothesis, but during cruise CI8, Anguilla leptocephali crossed the thermocline, which was found deeper, around m. Marine biologists consider net avoidance to be a major problem. Possible avoidance sheds doubt on any estimate of abundance of mobile organisms. It must be taken into account in the comparison of day- and night-time catches of visually orienting animals. In this study, we attempted to evaluate the component of avoidance responsible for differences in abundance estimates between day- and night-time. A high index of visual avoidance by day was calculated for most species of leptocephali, indicating that leptocephali are efficient at seeing the net and escaping before capture. However the low overall A indexes found for A. rostrata and A. anguilla indicate that visual avoidance by day was not important for these species in these samples. This discrepancy between Anguilla and the other species of leptocephali may originate in the fact that Anguilla had the smallest overall mean length of all the species of leptocephali studied. Had the mean lengths of Anguilla been larger, the conclusions regarding avoidance might have been different. The unimportance of visual avoidance by day in Anguilla probably explains the absence of differences in mean length between most day- and night-time samples of Anguilla and the absence of variation in A among size classes of Anguilla. It is important to realize that the avoidance estimates obtained in this study are relative ones, with the night catch as the reference. Visual avoidance by day is probably only part of total avoidance, i.e. that superimposed on other components of avoidance present by night as well as by day. Towed nets may be preceded by acceleration fronts and pressure waves that are detectable (Clutter and Ankaru, 98). Wiebe et al. (98) proposed that bioluminescence in response to the net is the principal cause of avoidance. These findings constitute the first detailed report to be published on the vertical migrations of the poorly known community of larval eels inhabiting the Sargasso Sea. The question of the adaptive significance of their vertical migrations comes naturally to mind. It could be investigated by comparing relevant parameters (e.g. rate of predation, diel rhythm of feeding) between a vertically migrating species, such as Anguilla, and a species that does not vertically migrate, such as S. beard. Information on the locomotory capabilities of leptocephali would shed light on the avoidance phenomenon. Downloaded from at Fisheries and Oceans on December 8, 5 Acknowledgements We thank R.C.Kleckner, E.J.Turner, D.Mullen.L.Castonguay, E.Shelden and D.Brown for their ideas and assistance. Numerous volunteers assisted with field work. This

19 Anguiila and other leptocephali in tbe Sargasso Sea research was funded through grants from the National Science Foundation (grant number OCE 8839) and the National Geographic Society (grant number 9-8) to J.D.M. Awards from the Natural Science and Engineering Research Council of Canada and le Fonds FCAC provided M.C. with financial support, and awards from the Leopold Schepp Foundation and the Eppley Foundation for Research provided J.D.M. with financial support. References Ahlstrom.E.H. (95) Distribution and abundance of egg and larval populations of the Pacific sardine. Fish. Bull.. US, 5, 83-. Barkley,R.A. (9) The theoretical effectiveness of towed-net samplers as related to sampler size and to swimming speed of organisms. J. Cons. Perm. Int. Explor. Mer, 9, 57. Barkley.R.A. (97) Selectivity of towed net samplers. Fish. Bull., US, 7, Bengtson.D.A. (973) Vertical distribution and life history of the fish family Nemichthyidae in the Bermuda Ocean Acre. M.S. Thesis, University of Rhode Island, Kingston, 7 pp. Bezdenezhnykh.V.A., Prokhorchik.G.A., Petrikov.A.M., Petukhov.V.B. and Plyuta.M.V. (983) Obtaining larvae of Anguiila anguiila L. (Pisces, Anguillidae) under experimental conditions. Dovtah BioL Sci., 8, BlaxterJ.H.S. (975) The role of light in the vertical migration of fish a review. In Evans,G.C., Bainbridge,R. and Rackham.O. (eds), Light as an Ecological Factor:. Blaclrwell Scientific Publications, Oxford, pp Clutter.R.I. and Ankaru.M. (98) Avoidance of samplers. In Tranter,D.J. (ed.), UNESCO Monogr. Oceanogr. Methodol., Vol., Zooplankton Sampling. Pan I, Reviews on Zooplanhon Sampling Methods. Imprimerie Rolland, Paris, pp Coombs,S.H. (98) A density-gradient column for determining the specific gravity of fish eggs, with particular reference to eggs of the mackerel Scomber scombrus. Mar. Biol., 3, -. Grave,H. (98) Food and feeding of mackerel larvae and early juveniles in the North Sea. Rapp. P -v. Rtun., Cons. Int. Explor. Mer, 78, Hunter.J.R. and Sanchez.G. (97) Diel changes in swim bladder inflation of the larvae of the northern anchovy, Engraulis mordax. Fish. Bull., US, 7, Keller.A.A. (97) Systematics, vertical distribution, and life history of Anguilliformcs leptocephali in the Bermuda Ocean Acre. M.S. Thesis, University of Rhode Island, Kingston, 5 pp. Kracht.R. and Tesch,F.-W. (98) Progress report on the eel expedition of R.V. 'Anton Dohrn' and R.V. 'Friedrich Heincke' to the Sargasso Sea 979. Env. Biol. Fish.,, Laval,P. (97) Un modele matnematique de l'evitement d'un filet a plancton, son application pratique, et sa verification indirecte en recourant au parasitisroe de l'amphipode hyperide VMlia armata Bovallius. J. Exp. Mar. Biol. Ecol.,, Lenarz.W.H. (97) Dependence of catch rates on size offish larvae. Rapp. P.-v. Rtuns., Cons. Int. Explor. Mer.,, Loeb.V.J. (979) Vertical distribution and development of larval fishes in the North Pacific Central Gyre during summer. Fish. Bull., US, TJ, Longhurst.A.R. (97) Vertical migration. In Cushing.D.H. and Walsh.J.J. (eds), Ecology of the Seas. Saunders Company, Philadelphia, pp Pearre.S. Jr (979) Problems of detection and interpretation of vertical migration. J. Plankton Res.,,9-. Research Group on Eel Reproduction (978) Preliminary studies on the induction of spawning in common eels. Acta Tool. Sinica,, Rothlisberg.P.C, Church J. A. and Forbes,A.M.G. (983) Modelling the advection of vertically migrating shrimp larvae. J. Mar. Res.,, Russell,F.S. (97) The vertical distribution of plankton in the sea. Biol. Rev.,, 3-. Schmidt.J. (95) The breeding places of the eel. Arum. Rep. Smithson. Inst., 9, Schoth.M. and Tesch,F.-W. (98) Spatial distribution of -group eel larvae (Anguiila sp.) caught in the Sargasso Sea in 979. Helgolander Meeresunters., 35, Schoth.M. and Tesch,F.-W. (98) The vertical distribution of small -group Anguiila larvae in the Sargasso Sea with reference to other Anguilliform leptocephali. Meeresforschung, 3, Seliverstov.A.S. (97) Vertical migrations of larvae of the Atlanto-Scandian herring (Clupea harengus). In Blaxter^.H.S. (ed.), The Early Life History of Fish. Springer Verlag, Berlin, pp Downloaded from at Fisheries and Oceans on December 8, 5 3