NOTES -

Limnol. Oceanogr., 34(6), 1989, 1072-1083 0 1989, by the American Society of Limnology and Oceanography, Inc. Zooplanktivorous and variable diel vertical migration in the marine planktonic copepod Calanus paczfzcus

Abstract-As a field test of the hypothesis that tions have generated much debate over the variable diel vertical migration in zooplankton is proximate and ultimate causes of vertical a response to variations in the abundance of vi- sually orienting predators, we sampled juvenile migration behavior, with several hypothe- and adult planktivorous fish inhabiting a tem- ses being proposed to explain its possible perate fjord (Dabob Bay, Washington) concur- adaptive significance (see Kerfoot 1985 for rently with the vertical distributions of a popu- a recent review). from visual lation of adult female Calanus pacificus (Copepoda: Calanoida) known to exhibit season- predators is perhaps the most often hy- ally variable diel vertical migration. Results of pothesized selective pressure responsible for nighttime trawling on seven dates between April the evolution of vertical migration in zoo- 1985 and October 1986 and subsequent stomach plankton (Zaret and Suffern 1976; Pearre content analyses showed that eight of fish, 1979; Kerfoot 1985; Gliwicz 1986a; Lam- comprising 45 different size classes, Fedon adult female C. pacificus; all species of fish showed pert. 1987; Frost 1988). Previous studies marked seasonal and interannual variations in supporting the predation hypothesis can be their abundances. The extent to which the pop- divided into two general categories: theo- ulation of C. pacificus exhibited diel vertical mi- retical arguments based on demographic and gration was determined for each cruise date by energetic models showing that only through calculating the proportion of the population crossing 75-m depth on a diel cycle. A statistically reduced mortality from avoidance of pred- significant regression was obtained between ators can vertical migration confer ‘an ad- strength of diel vertical migration in C. pacificus vantage in population growth to diel mi- and abundance of fish actively preying on the grants (Stich and Lampcrt 198 1; Ohman et copepod. We propose that the timing and mag- nitude of changes in the migration behavior of al. 1983; Lampert 1987; Frost 1988), and C. pacificus in Dabob Bay are directly linked to field studies purporting to show a relation- the year-class strengths of the dominant species ship between the presence or abundance of of planktivorous fish. predators and the extent to which vertical migration is exhibited by planktonic prey (Kozhov 1963; Zaret and Suffer-n 1976; Wil- Observations of diel vertical migration liamson and Magnien 1982; Fancett and have been well documented for aquatic or- Kirnmerer 19 8 5; Gliwicz 198 6a, b; Luecke ganisms in diverse phyla. These observa- 1986). The first approach has placed theoretical limits on adaptive significance of migration behavior, while at the same time providing Acknowledgments This research was supported by U.S. National Sci- a specific hypothesis for further testing. The ence Foundation grant OCE 84-08929 and a College second approach, attempting to provide di- of Ocean and Fishery Sciences Select Program Grant rect field evidence linking prey behavior with from the University of Washington. Contribution 1807 the presence or abundance of predators, has from the School of Oceanography, University of Wash- ington. provided much circumstantial support for We thank D. Thoreson, S. Jonasdottir, C. Mobley, the predation hypothesis but has often suf- R. McQuin, P. Crawford, M. Landsteiner, D. Gun- fered from ambiguous results due to the use derson, K. Fresh, P. McAllister, R. Donnelly, and C. of widely separated field sites (e.g. the com- Simenstad for assistance in various aspects of planning parative lake studies of Zaret and Suffern and carrying out the fieldwork. Analysis of fish stomach contents by J. Cordell and zooplankton samples by G. 1976 and Gliwicz 1986a), that thereby al- Heron are also acknowledged. low variable environmental factors to 1072 Notes

obfuscate the role of predation. In addition, the use of nonquantitative techniques for assessing predators and their feeding habits has too often left open the question of direct DEPTH IN METERS linkage between predator and prey (e.g. Ko- 47O zhov 1963; Williamson and Magnien 1982; 50’ Fancett and Kimmerer 1985; Gliwicz 1986a). In short, although predator avoidance is the most commonly proposed explanation for the adaptive significance of diel vertical migration, field evidence lags behind the theoretical arguments. What is needed is a time series of concurrent, quantitative ob- servations of predator abundances and prey migration under variable conditions of ver- tical migration behavior of the prey. There- 47* fore, the most judicious choice for a field 45’ test would be a single prey species that ex- hibits variable diel vertical migration and would include the measurement, over time, of all environmental parameters (both phys- ical and biological) hypothesized to affect migration behavior. The marine planktonic copepod Calanus paczjicus has been shown to exhibit seasonal and intergenerational variability in diel ver- tical migration off southern California (En- right and Honegger 1977; Koslow and Ota 198 1; Huntley and Brooks 1982) and both 122O 5o” seasonal and interannual variability in Da- Fig. 1. Bathymetry of Dabob Bay and the location bob Bay, Puget Sound, Washington (Runge of the central, deep station (D- 193 m). 198 1; Frost 1988). Frost (1988) has shown that this variation in Dabob Bay is unre- ( 1988) assessing the abundances and feed- lated to those environmental factors pre- ing habits of potential predators of C. pa- viously proposed to provide an energetic or cz$cus. Our purpose was threefold: to quan- mortality-independent demographic ad- titatively estimate the abundance of vantage to diel migrants (namely, food planktivorous fish between seasons and be- availability, in situ growth rate of females, tween years; to identify, from stomach con- and thermal stratification of the water col- tents, those fish actively feeding on adult umn). These results led him to hypothesize female C. paczjicus; and to relate the abun- that diel vertical migration in adult females dance of actively feeding predators to diel of C. paczjkus is a response to visually ori- vertical migration behavior in adult female enting zooplanktivores, most likely plank- C. paczjicus. tivorous fish. Planktonic invertebrate pred- Dabob Bay (Fig. 1) was chosen as the site ators were eliminated from consideration of our study because its circulation regime because all the species potentially preying is favorable to repeated sampling of popu- on adult C. paczjicus in Dabob Bay are non- lations of planktonic organisms (Frost 1988 visual predators, most forage nocturnally, and references therein). Abundances of and many are diel vertical migrators them- planktivorous and the vertical distri- selves (Frost unpubl.). In this note we pre- bution of adult female C. pacz&us were de- sent results of an 18-month field study, un- termined at a central, deep (193 m) station dertaken concurrently with the study of Frost during June, August, and October 1985 and 1074 NC?tes late April-early May, June, August, and Oc- species (except salmonids) then counted, tober 19 8 6. Additionally, zooplankton, but weighed,‘ and measured for length (fork not planktivorous fish, were sampled in late length for those fish possessing forked tails April 1985. and standard length for all other species). Two trawl types were deployed on each Lengths were measured to the nearest 1 cm cruise to sample juvenile and adult plank- for fish > 5 cm, and to the nearest 0.5 cm tivorous fishes. Qualitative hydroacoustic for fish ~5 cm. Squalus acanthias (spiny observations (105 kHz) with upward- and dogfish) and Porichthys notatus (plainfin downward-looking transducers indicated midshipman) were counted and weighed but that some fish vertically migrated, typically not measured for length. Gut cavities of all descending during the day to the very bot- salmonids (Oncorhynchus spp.) were in- tom (i.e. well below the distribution of Cal- jected with a small volume of Formalin (0.5- anus) - a depth which effectively precludes 3.0 ml) to halt digestion; salmonids were these fish from preying on Calanus during then preserved in 10% buffered seawater the day. Other fish remained in the surface Formalin for species identification and layer both day and night; our daytime trawl- measurement (fork length to the nearest ing, however, yielded no catches, due no millimeter) in the laboratory. Of the re- doubt to the well-known daytime avoidance maining nonsalmonid fishes, 1O-20 indi- capabilities of pelagic fish. Thus, we trawled viduals were selected from predetermined only at night and assume that nighttime size classes (see below) for each species and abundances are indicative of potential pred- injected and preserved with Formalin. The ators of Calanus in the surface layer total time between bringing the catch aboard throughout the day. A midwater trawl with and preservation of the fish and their stom- mouth area of 81.0 m2 (9.0 x 9.0 m) and ach contents was usually - 10 min, but nev- stretch-mesh size ranging from 8.9 cm near- er >20. est the mouth to 1.3 cm in the cod end was Abundance estimates for each trawl type used to sample from 50 m to the surface were calculated by averaging the number of and, less frequently, from 25 m to the sur- fish per unit of volume for the two to three face. The depth of the net was monitored tows of each gear type. Because capture ef- with a bathykymograph. The mouth of the ficiencies are assumed to be 100% for both net was kept open by two 230-kg doors rigged trawl types, abundance estimates are con- to the bridles. Tow speed averaged 150 cm servative. Estimates of the number of fish s-l (range: 120-180 cm s-l) and volumes per unit of volume of near-surface water (50 filtered, calculated from mouth area and m to surface) were calculated with a weight- length of tow, ranged from 115,000 to ed average of the mean abundance estimates 449,000 m3. of the two gear types, assuming that the sur- A surface tow net with mouth area of 18.3 face tow net sampled from 3 m to the surface m2 (3.0 x 6.1 m) and mesh size grading and that the midwater trawl sampled from from 8.9 cm nearest the mouth to 0.6 cm 50 m to 3. This latter assumption is based at the cod end was used to sample the sur- on the fact that no fish were caught in in- face 3 m. Two pipes with floats at the top dependent surface tows (3 m to surface) with and weights at the bottom were attached to the midwater trawl, due presumably to the the sides of the net to keep the mouth ver- disturbance of surface waters by the vessel. tical. The surface tow net was kept fully Each species of fish was divided into the open by deployment between two boats (the following size classes for stomach content 20-m RV CZzfird A. Barnes and a 5-m Bos- analyses: 149,50-99, 100-149, L 150 mm. ton Whaler). Average tow speed was 80 cm Salmonids were further divided into 50-74 s-l (range 50-l 00 cm s-l) and volumes fil- and 75-99 mm. The choice of size classes tered ranged from 7,300 to 3 1,700 m3. Two was sometimes based on a previously ob- or three tows were taken with each net on served change of diet with size (e.g. the each cruise. switching from epibenthic to neritic prey at Catches were immediately removed from -50 mm in pink and chum salmon, as the cod end, and all fish were identified to shown by Okada and Taniguchi 197 1 and Notes 1075

Table 1. Abundances of planktivorous fish [No.(104 m3)-I] in near-surface water (50-O m) at the central station in Dabob Bay. Estimates are a weighted average of the two trawl types.

1985 1986 Planktivorous fish Jun AW Ott ApdMay Jun AW Ott Ammodytes hexupterus (Pacific sandlance) 10.6 0.001 0.02 Clupea harengus pallasi (Pacific herring) 0.71 0.41 0.21 0.31 0.11 9.82 Engraulis mordux (northern anchovy) 0.014 0.035 Gasterosteus aculeutus (threespine stickleback) 0.036 0.76 0.023 0.19 0.052 1.58 2.32 Merluccius productus (Pacific hake) 1.53 4.36 1.6 0.54 1.88 13.4 5.35 Microgadus proximus (Pacific tomcod) 0.009 Oncorhynchus gorbuscha (pink salmon) 0.057 Oncorhynchus k&z (chum salmon) 1.23 0.83 0.093 1.91 3.66 0.5 1 0.18 Oncorhynchus kisutch (coho salmon) 0.022 0.91 0ncorhynchu.s tshawytscha (chinook salmon) 0.086 0.067 0.11 2.23 0.14 Porichthys notatus (plainfin midshipman) 0.158 0.158 0.24 0.18 0.66 0.20 Psychrolutes parudoxus (tadpole ) 0.04 Unidentified embiotocid (surfperch) 0.023 Unidentified osmerid (smelt) 0.019 Unidentified salmonid 0.023 0.018 Total 3.01 6.89 2.29 13.8 7.12 18.6 18.1

Simenstad et al. 1980) and at other times tical distribution and abundance of adult was based on the need for keeping the total female C. paczjkus are described elsewhere number of fish stomachs analyzed to a rea- (Frost 1988). sonable number given limited resources. For A combined total of 4,225 fish repre- each cruise date, 5-14 individuals (when senting 14 species were caught during the available) were selected from each size class seven cruises to Dabob Bay. Spiny dogfish of each species of fish for microscopic ex- (S. acanthias) was discounted as a potential amination of stomach contents. The stom- predator of C. paczjkus based on a review ach from the esophagus to just posterior to of the literature (Hart 1973 and references the pylorus was dissected from each fish. therein) and is not included in the catch Prey organisms were sorted, identified, and results. Our catch results show the abun- enumerated under a dissecting microscope. dance of planktivorous fish in Dabob Bay Prey were identified to the lowest possible to vary seasonally, both for individual taxon (usually species) and developmental species and total fish (Table 1). The seasonal stage allowed by the state of digestion. In occurrence of various size classes of the dif- the case of C. pacijcus, the later copepodid ferent species in our catches corresponds well stages and the sex of the adults were almost with published (e.g. Hart 1973; Miller and always identifiable. The actual number of Borton 1980; Garrison and Miller 1982) and stomachs analyzed from any one size class unpublished (Bollens et al. in prep.) ac- and species depended on a somewhat sub- counts of the life histories of these fishes in jective assessment of the interstomach vari- the neritic northeast Pacific. ability of contents at the time of dissection. Considering seasonal variations in abun- When abundance and composition of prey dance of total fish in 1985, the most striking types were relatively consistent among feature was the twofold to threefold greater stomachs, only five individual stomachs abundance in August as compared to either from that size class were analyzed; when June or October. This was due largely to the variability was greater or abundance lower, much greater numbers of Merluccius pro- a many as 14 stomachs were analyzed from ductus in August. Having hatched as larvae each size class. For the purposes of this pa- in late winter and early spring (Bailey and per, we report only on the presence or ab- Yen 1983; Bollens et al. in prep.), this species sence of adult female C. paczjkus in the had by August attained a size great enough stomach contents. to be retained by our nets (30-60 mm). Sim- Sampling and sample analysis for the ver- ilarly, Clupea harengus paZZasiand Gaster- 1076 Notes

osteus aculeatus reached their highest abun- remained abundant during October 1986, dances in our catches in August, due largely in stark contrast to the very small catches to young-of-the-year fish available to our we obtained in October 1985. Interannual nets on this cruise. Juvenile salmonids (On- differences in the abundances of A. hexap- corhynchus spp.), although present on all terus are more difficult to assess given that three cruises in 1985, were most abundant this species seems to be available to our during June and August, a time during which sampling gear (upper 50 m of the water col- these species are known to migrate through umn at night) only in the relatively early Puget Sound on their way to oceanic waters stages of its life history, corresponding to a (Simenstad et al. 1982). season (spring) when we sampled with trawls Striking seasonal variations in the abun- in 1986, but not in 1985. Results from ich- dance of individual species and total plank- thyoplankton surveys, however, showed tivorous fish were observed in 1986 as well. larval A. hexapterus to be present in spring In late April-early May, juvenile Ammody- 1986 but absent during 198 5 (Bollens et al. tes hexapterus was overwhelmingly domi- in prep.). Additionally, juvenile (young-of- nant in our catches. This species spawns the-year) and adult A. hexapterus were early in winter (Trumble 1973) and would caught in our trawls in June and August therefore be expected to produce young-of- 1986, whereas no A. hexapterus individuals the-year that are available to our nets at an were caught in summer or fall 198 5. These earlier date (i.e. spring) than the other species observations suggest that had we trawled for of planktivorous fish in Dabob Bay, most juvenile and adult fishes in May 198 5, of which spawn in late winter-early spring abundances of young-of-the-year A. hex- (Bollens et al. in prep.). In June, abundances apterus would have been lower than we ob- were relatively low for all species, except served in May 1986, i.e. the 1986 year-class larger juveniles of M. productus (age-1 and of A. hexapterus was probably larger than age-II) and juvenile salmonids (especially the 1985 year-class. Oncorhynchus keta). The very much re- A total of 396 fish (comprising 13 species) duced catches of A. hexapterus in June and collected at the central station during the later were likely due to initiation of demer- seven cruises in 198 5 and 1986 were ana- sal behavior in the young-of-the-year ju- lyzed for stomach contents. Data on pres- veniles (Reay 1970 and references therein; ence or absence of adult female C. pacz$cus but see Mater 1966), thereby greatly reduc- were the basis for determining which plank- ing susceptibility to capture by our sampling tivorous fishes actively preyed on C. pacif- gear. The highest abundance of total plank- icus on any single cruise date. Eight species tivorous fish in 1986 was obtained in Au- comprising 45 different size classes were gust, when young-of-the-year M. productus found to have adult female C. paczj?cusin and G. aculeatus were extremely abundant their stomachs. On any one cruise date, a in our catches. Juvenile salmonids, espe- minimum of five and a maximum of 10 cially Oncorhynchus tshawytscha, were also species/size classes were found to be feeding abundant during August. Total abundance on C. paczjicus. The seasonal and interan- of fishes remained high in October, when nual variability is striking (Table 2). young-of-the-year C. harengus pallasi, M. We quantified the degree to which the productus, and G. aculeatus were all abun- population of C. pacz$cus undertakes a nor- dant. mal diel vertical migration (up at night, In comparing years, the seven dominant down during the day), or the “strength” of species in our catches (A. hexapterus, C. ha- migration behavior (herein denoted as v), rengus pallasi, G. aculeatus, M. productus, by calculating the difference between the 0. keta, Oncorhynchus kisutch, and 0. tsha- fraction of the population above a given wytscha) were all more abundant in 1986 depth at midnight and the fraction of the than 1985, due mostly to larger numbers of population above that same depth during young-of-the-year fishes in 1986. Further- midday. This parameter is then a measure more, C. harenguspallasi, G. aculeatus, and of the fraction of the population migrating M. productus young-of-the-year juveniles across a particular depth on a diel cycle. Notes 1077

Table 2. Abundances of planktivorous fish [No.( lo4 m3)-‘1 found to prey on adult female Calanus pacijicus in near-surface water (50-O m) at the central station in Dabob Bay. Estimates are a weighted average of the two trawl types.

1985 1986 Planktivorous fish Jun Aug Ott May Jun Aug Ott Ammodytes hexapterus 50-99 mm 10.6 Clupea harengus pallasi 50-99 mm 0.61 0.29 0.08 5.03 100-149 mm 0.07 0.05 0.19 4.53 1150 mm 0.04 0.07 0.02 0.25 Engraulis mordax 100-149 mm 0.04 Gasterosteus aculeatus 50-99 mm 0.04 0.04 0.82 2.07 Merluccius productus 549 mm 0.04 2.69 3.44 50-99 mm 0.04 0.69 7.37 2.28 100-149 mm 2150 mm 1.38 0.99 1.21 Oncorhynchus keta 549 mm 0.20 50-74 mm 1.10 0.31 75-99 mm 0.34 0.53 2.65 100-149 mm 0.85 0.48 0.09 0.59 0.5 1 2150 mm 0.34 Oncorhynchus kisutch 100-149 mm 0.81 Oncorhynchus tshawytscha 75-99 mm 0.07 loo-149 mm 0.02 0.07 2.24 Total 2.78 6.01 1.71 12.6 4.36 14.5 14.2

Although the maximum depth or amplitude (V= 0.33) 1985, vs. moderate in both early of migration of C. pacz$cus varies season- May (I’= 0.53) and June (V = 0.50) 1986. ally, this species, when migratory, always Calanus paczjkus was strongly migratory in descends to at least 75 m in the daytime late summer (August) of both years (V = (Fig. 2). This depth, therefore, was chosen 0.74 for 1985; V = 0.79 for 1987). In early as the reference depth across which we mea- fall (October), C. pacz$cus vertically mi- sured the proportion of the population un- grated during both years, but the degree of dertaking diel migration. It should be em- migration varied considerably, being great- phasized that we do not present calculations er in 1986 (V= 0.96) than 1985 (V= 0.61). of the amplitude of the vertical migration, The dependence of the strength of the diel but rather the percentage of the population vertical migration exhibited by C. paczjicus undertaking migration across 75-m depth. (V’) on the abundance of planktivorous fish Both seasonal and interannual differences actively feeding on C. paczjicus in the sur- in the strength of diel vertical migration in face 50 m of the water column was exam- adult female C. pacz$cus are apparent. The ined with simple linear regression (Fig. 3). extent to which C. pacificus exhibits vertical The regressions for individual years (1985 migration in the spring and early summer and 1986) suggest a dependence, but the varies between years, being extremely small slopes of these lines do not differ statistically in late April (V’ = 0.05) and small in June from zero. Combining the data for both 1078 Notes

1985 1986

c

0.05

LL som-f 25-26 APRIL

O-

50 -

100 - 0.33 0.50

150-

m IIT25-26 JUNE U S-11 JUNE I 1 E 200- 40 me3 4m’3 ? L Or b 0

50- 100 - 0.74 1 0.79 150- I 150 m-3 T I ,’ 75 mw3 d20-22 AUGUST IL ’ 200 L ::I5-6 AUGUST

0.67 hT0.96 I J 150 m-3 300 m-3 48-9, 14-15 OCTOBER 13- 15 OCTOBER 200 L Notes 1079 years, thus increasing the number of obser- rengus pallasi, G. aculeatus, and M. pro- vations, resulted in a statistically significant ductus) then reach a maximum in late sum- regression. Whether the dependence is ac- mer and may persist into autumn, coinciding tually linear is somewhat uncertain given with the time when the proportion of the the broad confidence intervals for the slope population of adult C. paczjkus undertaking and intercept. diel vertical migration also reaches a max- In general, the proportion of the popu- imum. lation of adult female C. pacijicus under- The strong dependence of our abundance taking diel vertical migration is both sea- estimates on the mesh size of our sampling sonally and interannually variable and is gear begs the question of the possible role related to the abundance of actively feeding that smaller or younger fish might play as planktivorous fish. We propose that the predators of C. pacz$kus. We were not able timing and magnitude of these behavioral to adequately sample late larval and early changes in C. pacijicus are directly linked to juvenile stages of fish (20-40 mm long). the year-class strengths of the various species These stages of pelagic fish are known to of planktivorous fish in Dabob Bay in the avoid plankton nets (Murphy and Clutter following way. If the abundance of fish is 1972) and it has been shown that a gap in thought to be the cause of the moderate mi- sampling capability may exist until these gration behavior of C. pacz$kusin May 1986 fish grow large enough to be caught by com- ( V = 0.5 3) then the large number of young- mercial-sized nets and larger midwater of-the-year A. hexapterus (comprising 77% trawls (Mais 1974; Methot 1986). There- of all planktivorous fish and 84% of all pred- fore, the abundances and feeding habits of atory fish) might be largely responsible. Al- these sized fishes, and thus their potential though the relationship between the abun- predatory impact on adult female C. pacif- dance of planktivorous fish and the presence icus, were impossible for us to assess. Pre- or absence of migration behavior in C. pa- vious studies, however, have shown that c&us in spring is difficult to assess given larval stages of marine fishes only rarely take the lack of data on fish abundance in spring prey as large as adult Calanus (Blaxter 1965; 198 5, there is reason to believe, as discussed Hunter 1980). As a first approximation then, above, that the 1986 year-class of A. hex- fish smaller than those retained by our trawls apterus was much larger than the 198 5 year- (i.e. ~20-40 mm) are assumed not to be class. feeding on adult C. paczjicus. We therefore suggest that the abundance This relationship between the presence or of young-of-the-year A. hexapterus, being abundance of predators and the vertical mi- an early winter spawner and thus producing gration behavior of zooplanktonic prey has juvenile fish of a size capable of preying on been suggested before. Gliwicz (19863) stat- adult C. pacz$kus by spring, determines ed that seasonal differences existed in both whether C. paczjkus is migratory in late the abundance of predators and the vertical April-early May. As summer progresses, migration of several species of zooplankton young-of-the-year of other, later-spawning in tropical Lake Cahora Bassa. Though he species of planktivorous fish (C. harengus claimed that vertical migrations were “most pallasi, G. aculeatus, and M. productus) be- pronounced” in late August-arly Septem- gin to attain a size large enough to feed on ber, he didn’t present data for any other adult C. paczjkus. Additionally, abundances sampling date, thereby precluding compar- of seaward-migrating juvenile salmonids ison of migration behavior with abundance increase from spring to early summer. The of predators. Additionally, his reference to abundances of young-of-the-year juveniles “most pronounced” migration likely refers of the later-spawning nonsalmonids (C. ha- to the amplitude of migration (rather than c Fig. 2. Vertical distribution of adult females of CuIunus pacijicus in Dabob Bay in 1985 and 1986. For each set of dates, day (clear) and night (darkened) vertical distributions are means of four vertical series of samples collected in pairs near noon and midnight on each of two consecutive days. Numbers in italics are values of V, calculated as described in text. 1080 Notes

1985 the proportion of the prey population un- 80 dertaking migration), which he attributes to light (proximate factor) rather than preda- 70 August tors (a possible ultimate cause). . October Narver (1970) concurrently sampled _ n 60 sockeye salmon (Oncorhynchus nerka) ju- V veniles < 1-yr old and the zooplankton of 50 Babine Lake, British Columbia, during June-October. The cladoceran Bosmina 40 June coregoni and the Calanoid copepod Hetero- n I . I. I . I . I .I cope septentrionalis exhibited seasonally 30 1 2 3 4 5 6 7 variable diel vertical migration behavior. Heterocope septentrionalis was nonmigra- tory in June (concentrated near the surface both day and night) but became increasingly 1986 migratory, both in amplitude and propor- 100 - October. tion of the population migrating, during 90 - summer and autumn. Bosmina coregoni showed a reverse diel vertical migration (i.e. August near the surface during the day and at depth 70 V at night), which Zaret (1980) attributed to 60 response to the normal diel vertical migra- June tion of the predatory H. septentrionalis. Al- 50 April/May though the amplitude of the reverse migra- 8040 -.:---“:., tion of B. coregoni was shown by Narver to 30 increase in summer and fall, the proportion 4 6 8 10 12 14 16 of the population exhibiting this behavior appears to have been fairly constant during all sampling periods. The increased ampli- 1985 and 1986 tude of the migrations of both zooplankters 100 - might be simply a result of increased clarity 8 of the lake in summer and fall (as shown by Narver’s Secchi disk data).

n The change from nonmigratory to migra- tory behavior (i.e. the proportion of the

V 60- n population migrating) in the case of H. sep- nO: A n n tentrionalis, however, suggests a more fun- damental change in the environment and 40 - / n resulting behavior than that caused by changing light regime. On the basis of Nar- 1 I 20 ’ ver’s standard net tows, the abundance of 0 10 20 sockeye increased between early July and F mid-August, then declined through Octo- Fig. 3. Linear regressions for the dependence of the ber. Heterocope septentrionalis was a dom- strength of diel vertical migration (V, from Fig. 2) in adult female Calanus paczjicus on the abundance of inant prey of sockeye on all sampling dates predatory fish in the upper 50 m of the water column (between July and October). Although not at night (F, from Table 2) during 1985, 1986, and both discussed by Narver, the increase in abun- years combined. Lines fitted by functional regression dance of sockeye during July and August (Ricker 1973); only the line for both years combined (V = 3.77F + 33.76) has a slope differing significantly corresponds closely to the change in behav- from zero (t,,,,,, = 2.94, P < 0.05; 95% C.L. of slope ior of H. septentrionalis from nonmigratory 0.48, 7.06; 95% C.L. of y-intercept 2.41, 65.11). to migratory. Although other environmen- tal parameters were not measured and other possible factors influencing migration can- Notes 1081

not therefore be ruled out (e.g. thermal strat- Euchaeta elongata above 50 m at night. ification and food availability), Narver’s ob- Luecke ( 1986) presented diel vertical dis- servations suggest the predation hypothesis tributions of Chaoborus jlavicans in Lake as an explanation for initiation and en- Lenore, Washington, and showed a change hanced display of vertical migration behav- from nonmigratory to migratory behavior ior in H. septentrionalis. in this species spanning the period of intro- Similarly, Tessier ( 198 6) presented data duction of cutthroat trout (Salmo clarki for Daphnia catawba in Lake Lacawac, henshawi) into the lake. Similarly, Dorazio Pennsylvania, showing a change in diel ver- et al. ( 1987) reported on a summer shift in tical migration behavior from nonmigra- net zooplankton (especially Daphnia puli- tory in April to migratory in June and July. caria) of Lake Michigan from nonmigratory He also found that recruitment and growth to migratory between 1983 and 1985. Dur- of the dominant planktivorous fish, Perca ing this period visual observations of plank- jlavescens, occurred in spring and summer, tivorous fishes (alewife and bloater chub) respectively, indicating close correspon- suggested increasing numbers of predators, dence of seasonal changes in migration be- although abundances and feeding habits of havior of prey with increasing abundance the fishes apparently were not quantified. of predators. These previous reports corroborate our Kozhov ( 196 3) described seasonal own field observations on the dependence changes in both the amplitude and intensity of variable migration behavior in plankton- of diel vertical migrations in the zooplank- ic on the variation in abundance of ton of . Changes in migration their predators, and while correlation is not amplitude were attributed to changing light necessarily causation, it is suggestive that C. regimes, while changes in intensity of mi- pacz%cusmay be undertaking diel vertical gration were attributed to periods of active migration in an effort to avoid visually ori- feeding by planktivorous fish. The intensity enting predators, i.e. planktivorous fish. This of migration (percent of total population possibility is all the more likely given the undertaking migrations) of copepodid stages inability of any of the energetic or demo- of Epischura baicalensis was very weak in graphic models hypothesizing enhanced spring but pronounced in late summer when growth or reproduction from diel vertical the dominant planktivorous fishes (Cotto- migration to explain the variable nature of comephorus spp., spp., and Co- this behavior by C. paczjicus in Dabob Bay mephorus spp.) migrated offshore and ac- (Frost 1988). That is, both our demographic tively fed in surface waters. Although lacking models (Frost 1988) and our field obser- quantitative data on the seasonal variability vations point to predator avoidance as the of predator abundances, Kozhov’s obser- adaptive significance of diel vertical migra- vations lend support to the predation hy- tion. pothesis as an explanation for seasonal vari- Even if the variable diel vertical migra- ability of diel vertical migration in tion behavior of planktonic prey in these zooplankton. previous reports and our Dabob Bay field Interannual variability in diel vertical mi- study are indeed responses to predators, the gration of zooplankton and its possible re- actual mechanism by which predators trig- lationship to predator abundance have re- ger this response is an open question. Ever ceived even less attention than seasonal since the seminal work of HrbaCek et al. variability. Ohman (1983) related the re- (196 1) and Brooks and Dodson (1965) over verse diel vertical migration of a small cala- two decades ago, studies of predator-prey noid copepod (Pseudocalanus sp.) to the interactions in the plankton have focused normal diel vertical migration of known in- on selective predation as the mechanism by vertebrate predators for a single location and which predators affect their prey. Indeed, season over several years. He found strong such a mechanism is implicit in earlier the- correlation between the proportion of adult oretical analyses of induction of diel migra- female Pseudocalanus sp. below 50 m at tory behavior in zooplankton (Wright et al. night and the number of Sagitta elegans and 1980; Ohman et al. 1983), and observed 1082 Notes

intraspecific genetic differences between mi- BLAXTER, J. H. S. 1965. The feeding of herring larvae gratory and nonmigratory genotypes of and their ecology in relation to feeding. Cal. Coop. Fish. Invest. (CalCOFI) Rep. 10: 79-88. freshwater cladocerans (Weider 1984; Du- BROOKS, J. L., AND S. L. DODSON. 1965. Predation, mont et al. 1985) support the possibility of body size, and composition of the plankton. Sci- such a mechanism. Recently, however, there ence 150: 28-35. has developed a greater appreciation of the DODSON, S. I. 1988. The ecological role of chemical ability of prey organisms to respond phe- stimuli for the zooplankton: Predator avoidance behavior in Daphnia. Limnol. Oceanogr. 33: 143 l- notypically, both morphologically (Have1 1439. 1987) and behaviorally (Petranka et al. 1987; DORAZIO, R. M., J. A. BOWERS, AND J. T. LEHMAN. Kats et al. 1988), to the chemical exudates 1987. Food-web manipulations influence grazer of predators, and such a mechanism for the control of phytoplankton growth rates in Lake Michigan. J. Plankton Res. 9: 891-899. induction of diel vertical migration behav- DUMONT, H. J., Y. GUISE, I. CARELS, AND H. M. ior in zooplankton has been suggested by VERHEYE. 1985. Experimental isolation of pos- Frost (1988) and Dodson (1988). itively and negatively phototactic phenotypes from Our data are consistent with either mech- a natural population of Daphnia magna Straus: A contribution to the genetics of vertical migration. anism of induction, but the rapid rate of Hydrobiologia 126: 121-127. change of migration patterns in 198 5 -none ENRIGHT, J.T., AND H.-W. HONEGGER. 1977. Diurnal in May, weak in June, strong in August, and vertical migration: Adaptive significance and tim- moderate in October- would require very ing. Part 2. Tests of a model: Details of timing. different rates of population growth between Limnol. Oceanogr. 22: 873-886. FANCETT, M.S., AND W.J. KIMMERER. 1985. Vertical the two prey genotypes (migratory genotype migration of the demersal copepod Pseudodiap- r > nonmigratory genotype r between May tomus as a means of predator avoidance. J. Exp. and August; nonmigratory genotype r > mi- Mar. Biol. Ecol. 88: 31-43. gratory genotype r between August and Oc- FROST, B. W. 1988. Variability and possible adaptive tober), perhaps to a degree that excludes significance of diel vertical migration in Calanus pacificus, a planktonic marine copepod. Bull. Mar. such a genotypic change in the population Sci. 43: 675-694. as a reasonable possibility. GARRISON, K. J., AND B. S. MILLER. 1982. Review What is required to more fully test our of the early life history of Puget Sound fishes. Univ. field data against these two possible mech- Wash. Fish. Res. Inst. Rep. FRI-UW-8216. GLIWICZ, M. Z. 1986~. Predation and the evolution anisms of induction is to include rate mea- of vertical migration in zooplankton. Nature 320: surements of both predation and prey re- 746-748. sponses. The necessary data (stomach - 1986b. A lunar cycle in zooplankton. Ecology contents, temperature, and feeding perio- 67: 883-897. dicity of fish for gut evacuation rates and HART, J. L. 1973. Pacific fishes of Canada. Bull. Fish. Res. Bd. Can. 180. thus predation rates; egg production rates, HAVEL, J. E. 1987. Predator-induced defenses: A re- development times, and mortality rates of view, p. 263-278. In W. C. Kerfoot and A. Sih C. pacijicus) are currently being analyzed [eds.], Predation: Direct and indirect impacts on and will allow us to place limits on the rate aquatic communities. New England. HRBACEK, J., M. Dvo~ZAKOVA, V. KORNEK, AND L. at which C. paczjicus could respond at a pop- PROCH.&K~VA. 196 1. Demonstration of the ef- ulation-genetic level to predation from fect of the fish stock on species composition of planktivores. zooplankton and the intensity of metabolism of the whole plankton assemblage. Int. Ver. Theor. Stephen M. Bollens Angew. Limnol. Verh. 18: 162-l 70. Bruce W. Frost HUNTER, J. R. 1980. The feeding behavior and ecol- ogy of marine fish larvae, p. 287-326. Zn J. Bar- School of Oceanography, WB- 10 dach et al. [eds.], Fish behavior and its use in the University of Washington capture and culture of fishes. Int. Center Living Seattle 98 195 Aquat. Resour. Manage. HIJNTLEY, M., AND E. R. BROOKS. 1982. Effects of References age and food availability on diel vertical migration of Calanus pacificus. Mar. Biol. 71: 23-3 1. BAILEY, K. M., AND J. YEN. 1983. Predation by a KATS, L.B.,J. W. PJZTRANKA,AND A. SIH. 1988. An- carnivorous marine copepod, Euchaeta elongata tipredator defenses and the persistence of am- Esterly, on eggs and larvae of the Pacific hake, phibian larvae with fishes. Ecology 69: 1865-l 870. Merluccius productus. J. Plankton Res. 5: 7 l-82. KERFOOT, W. C. 1985. Adaptive value of vertical NCPtes 1083

migration: Comments on the predation hypothesis REAY, P. J. 1970. Synopsis of biological data on North and some alternatives. Contrib. Mar. Sci. Atlantic sandeels of the Ammodytes (A. to- 27(suppl.): 9 l-l 13. bianus, A. dubius, A. americanus, and A. marinus). KOSLOW, J. A., AND A. OTA. 198 1. The ecology of FAO Fish. Synop. 82. l vertical migration of three common zooplankters RICKER, W. E. 1973. Linear regression in fishery re- in the La Jolla Bight, April-August, 1967. Biol. search. J. Fish. Res. Bd. Can. 30: 409434. Oceanogr. 1: 107-l 34. RUNGE, J. A. 198 1. Egg production of Calanus pa- KOZHOV, M. 1963. Lake Baikal and its life. Junk. &jicus Brodsky and its relationship to seasonal LAMPERT, W. H. 1987. Vertical migration of fresh- changes in phytoplankton availability. Ph.D. the- water zooplankton: Indirect effects of vertebrate sis, Univ. Wash., Seattle. 116 p. predators on algal communities, p. 291-299. In SIMENSTAD,~. A., K.L. FRESH,AND E.O.SALO. 1982. W. C. Kerfoot and A. Sih [eds.], Predation: Direct The role of Puget Sound and Washington coastal and indirect impacts on aquatic communities. New estuaries in the life history of Pacific salmon: An England. unappreciated function, p. 343-364. In V. S. Ken- LUECKE, C. 1986. A change in the pattern of vertical nedy [ed.], Estuarine comparisons. Academic. migration of Chaoborus flavicuns after the intro- -, AND OTHERS. 1980. Prey community struc- duction of trout. J. Plankton Res. 8: 649-657. ture and trophic ecology of outmigrating juvenile MACER, P. J. 1966. Sand eels (Ammodytidae) in the chum and pink salmon in Hood Canal. Univ. southwestern North Sea, their biology and fishery. Wash. Fish. Res. Inst. Rep. FRI-UW-8026. Fish. Invest. Ser. 2 24: l-55. STICH, H.-B., AND W. LAMPERT. 198 1. Predator eva- MAIS, K. F. 1974. Pelagic fish surveys in the Cali- sion as an explanation of diurnal vertical migra- fornia Current. Cal. Fish. Bull. 162: l-79. tion by zooplankton. Nature 293: 396-398. METHOT, R. D. 1986. Frame trawl for sampling pe- TESSIER, A. J. 1986. Comparative population regu- lagic juvenile fish. Cal. Coop. Fish. Invest. (Cal- lation of two planktonic Cladocera (Holopedium COFI) Rep. 7: 267-278. gibberum and Daphnia catawba). Ecology 67: 285- MILLER, B. S., AND S. F. BORTON. 1980. Geographical 302. distribution of Puget Sound fishes: Maps and data TRUMBLE, R. J. 1973. Distribution, relative abun- source sheets. V. 1 and 2. Univ. Wash. Fish. Res. dance, and general biology of selected underuti- Inst. Wash. State Sea Grant Publ. lized fishery resources of the eastern North Pacific MURPHY,G. I., AND R.I. CLUTTER. 1972. Sampling Ocean. MS. thesis, Univ. Wash., Seattle. 178 p. anchovy larvae with a plankton purse seine. Fish. WEIDER, L. J. 1984. Spatial heterogeneity of Daphnia Bull. 70: 789-798. genotypes: Vertical migration and habitat parti- NARVER, D. W. 1970. Diel vertical movements and tioning. Limnol. Oceanogr. 29: 225-23 5. feeding of underyearling sockeye salmon and the WILLIAMSON, C. E., AND R. E. MAGNIEN. 1982. Diel limnetic zooplankton in Babine Lake, British Co- vertical migration in Mesocyclops edux: Implica- lumbia. J. Fish. Res. Bd. Can. 27: 28 l-3 16. tions for predation rate estimates. J. Plankton Res. OHMAN, M. D. 1983. The effects of predation and 4: 329-339. resource limitation on the copepod Pseudocalanus WRIGHT, D., W. J. O'BRIEN, AND G. L. VINEYARD. sp. in Dabob Bay, a temperate fjord. Ph.D. thesis, 1980. Adaptive value of vertical migration: A Univ. Wash., Seattle. 249 p. simulation model argument for the predation hy- -, B. W. FROST, AND E. B. COHEN. 1983. Re- pothesis. Am. Sot. Limnol. Oceanogr. Spec. Symp. verse diel vertical migration: An escape from in- 3: 138-147. New England. vertebrate predators. Science 220: 1404-1407. ZARET, T. 1980. Predation and freshwater commu- OKADA, S., AND A. TANIGUCHI. 1971. Size relation- nities. Yale. ship between salmon juveniles in shore waters and AND J. S. SUFFERN. 1976. Vertical migration their prey animals. Bull. Fat. Fish. Hokkaido Univ. in ‘zooplankton as a predator avoidance mecha- 22: 30-36. nism. Limnol. Oceanogr. 21: 804-813. PEARRE, S., JR. 1979. On the adaptive significance of vertical migration. Limnol. Oceanogr. 24: 781- 782. PETRANKA, J.W.,L.B. KATS,AND A. SIH. 1987. Pred- ator-prey interactions among fish and larval am- Submitted: 21 June 1988 phibians: Use of chemical cues to detect predatory Accepted: 17 April 1989 fish. Anim. Behav. 35: 420-425. Revised: 29 June 1989