Brodeur and Rugen: Vertical distribution of ichthyoplankton in the northern Gulf of Alaska 



233 



A potential disadvantage to a diurnal ascent is 

 increased susceptibility to visually feeding 

 planktivorous fishes. However, acoustic and trawl 

 survey data suggest that epipelagic fish predators 

 are rare during the spring in this area and the 

 majority of the nekton biomass is found in midwater 

 or near the bottom (Brodeur et al., 1991), well be- 

 low the depth of most larvae. On the other hand, 

 euphausiids, which are possibly the major inverte- 

 brate predator on walleye pollock yolk-sac larvae, 

 undergo a nocturnal ascent to surface water and 

 descend to greater depths during the day in Shelikof 

 Strait (Bailey et al., 1993). If euphausiids were also 

 predators on non-pollock larvae and feed only in the 

 surface layer above the nightime depths of these lar- 

 vae, then a distinct advantage would be conferred 

 upon individuals adopting a reverse diel migration 

 pattern, as has been postulated for copepods (Ohman 

 et al., 1983; Ohman, 1990). Based on field and experi- 

 mental results, it has become increasingly apparent 

 that predators can alter the diel vertical distribution 

 patterns of invertebrate prey (Ohman et al., 1983; 

 Gliwicz, 1986; Bollens and Frost, 1989; Levy, 1990; 

 Neill, 1990; Frost and Bollens, 1992), but evidence for 

 this effect on larval fish as prey is presently lacking. 



Although a variation in depth by time of day was 

 apparent for all species and consistent among spe- 

 cies, it was not substantial enough to be statistically 

 significant in all cases (e.g. G . macrocephalus). This 

 may be due in part to the lack of resolution of our 

 sampling intervals. The smallest average migration 

 that we could detect is -15 m; thus, diel vertical 

 migrations less than that were not likely to be de- 

 tected. Although a daily ambit of 30 m is not excep- 

 tional for larger larvae, it may be excessive for newly 

 hatched individuals. For a study specifically exam- 

 ining the diel vertical distribution of the species 

 considered here, we recommend sampling with a 

 multiple net system every 5 m over the upper 40 m 

 of the water column. Some bias may have also re- 

 sulted from combining tows from different years, 

 weeks, or geographic areas into our four time peri- 

 ods, which was necessitated by the relatively low oc- 

 currence rate and densities of these taxa. However, 

 the remarkably strong and consistent diel differ- 

 ences among the different taxa, despite this intro- 

 duced sampling variability, lend credence to our 

 findings. 



If there was differential migration by size classes 

 of larvae, this condition might also obscure some of 

 our results. The vertical distribution of larval 

 lengths of the dominant species did not show any 

 consistent patterns by time of day. The mean length 

 by depth varied significantly for H. elassodon; 

 smaller larvae were found at greater depths during 



the daytime and at the surface at night. This can- 

 not be explained by visual gear avoidance alone 

 since the nighttime pattern would then be expected 

 to be random rather than exhibit the increasing 

 mean length with depth that we observed. A possible 

 explanation for this pattern might be that larger 

 larvae may migrate a greater distance than smaller 

 larvae, a pattern frequently observed in other fish 

 larvae (Neilson and Perry, 1990). It is also possible 

 that the migration of different size classes is asyn- 

 chronous (Pearre, 1979). However, the available size 

 ranges of the dominant species in our data was not 

 extensive enough to examine diel migration patterns 

 of different size classes. Moreover, caution should be 

 exercised in examining larval length data in mul- 

 tiple net systems. Since larvae shrink upon death 

 (Theilacker, 1980; Hay, 1981) and the likelihood of 

 death may be related to time in net, we may assume 

 that larvae caught in the first (deepest) net may 

 have undergone more shrinkage than those in the 

 last (surface) net. 



In conclusion, this study shows that all the com- 

 mon larvae exhibit a reverse vertical migration pat- 

 tern, opposite to that of the overall dominant spe- 

 cies, walleye pollock. In Auke Bay, an inland 

 embayment in Southeast Alaska (58°22' N) on the 

 eastern side of the Gulf of Alaska, Haldorson et al. 

 ( 1993) found a Type I migration for the numerically 

 dominant osmerid larvae in their sampling and a 

 Type II migration for the five next most abundant 

 taxa {T. chalcogramma, H. elassodon, P. bilineatus, 

 Leuroglossus schmidti, and Agonidae). These au- 

 thors attribute this diel-depth distribution pattern 

 to temperature preferences by each species, al- 

 though their vertical temperature gradients were 

 more pronounced than what we observed in our 

 study. Since most abiotic variables (other than light 

 intensity) and food resources varied little over the 

 depths through which much of the migration oc- 

 curred in Shelikof Strait, we hypothesize that the 

 reverse migration pattern that we documented was 

 either a predator-avoidance mechanism or else an 

 optimization of light levels for feeding. The preva- 

 lence of reverse migration in this and other studies 

 suggests that it may be more common than previ- 

 ously suspected, especially in higher latitude ecosys- 

 tems, and the factors contributing to this phenom- 

 enon merit further investigation. 



Acknowledgments 



The MOCNESS tow collections used in this study 

 were made available by Lew Incze (Bigelow Labo- 

 ratory) and Peter Ortner (Atlantic Oceanographic 



