616 



Fishery Bulletin 93(4), 1995 



rankings that were quite similar to the groundfish 

 abundance rankings estimated during 1990 and were 

 much more similar than those estimated from larval 

 collections only two months earlier in 1991. Addi- 

 tional collections taken earlier in 1991 (late April and 

 early May) showed even less coherence with the adult 

 groundfish community (Brodeur, unpubl. data). Ap- 

 parently, many of the smaller species not vulnerable 

 to the survey trawls leave the plankton and settle to 

 rocky habitats or in nearshore areas during the sum- 

 mer, leaving many of the numerically dominant 

 gadids and pleuronectids to be sampled. We chose to 

 correlate the rankings rather than the actual abun- 

 dance of species among the surveys because small 

 differences in timing of spawning relative to the tim- 

 ing of the survey can have drastic effects on abun- 

 dance owing to mortality and changes in catchability. 

 For example, on the basis of the Methot trawl catches, 

 H. elassodon appear to be much more abundant than 

 T. chalcogramma in our survey area. This may be 

 due to the fact that they hatch out about one month 

 later than pollock (Rugen 2 ) and thus have undergone 

 substantially less larval mortality. Subsequent stud- 

 ies have also indicated that the 1991 year class of T. 

 chalcogramma had very high larval mortality owing 

 to either poor feeding conditions or to advection off 

 the shelf (Bailey et al., 1995), resulting in very low 

 recruitment that year (Bailey et al. 3 ). 



The similarity in the species groupings found with 

 the two methods (Recurrent Group Analysis and 

 TWINSPAN), each of which uses different resolutions 

 of the same data (presence/absence vs. abundance), 

 substantiates the conclusion that certain taxa tend 

 to be associated in our study area. Whether these 

 groupings result from behavioral aggregation by cer- 

 tain species that have been adapted to a particular 

 habitat (Frank and Leggett, 1983) or whether hy- 

 drographic conditions passively transport larvae 

 spawned in the same area to the same nursery area 

 (Richardson et al., 1980; Olivar, 1987; Sabates and 

 Maso, 1990), or some combination of both (Cowen et 

 al., 1993), cannot be determined from our data. How- 

 ever, many of the specimens collected in the Methot 

 net may no longer be considered passive organisms 

 because they can actively swim against currents 

 while seeking out or remaining within favorable habi- 

 tats. Because juvenile fishes respond not only to en- 

 vironmental conditions but also readily respond to 

 the presence of conspecifics and potential predators 

 (Olla et al., in press), several factors can influence 



their distribution patterns in natural conditions. 



The lack of clearly defined boundaries between the 

 station groups that we observed may be characteris- 

 tic of this dynamic environment. In our study area, 

 vigorous mixing and strong currents (Reed and 

 Schumacher, 1986) do not allow formation of well- 

 defined mesoscale physical boundaries (see also Doyle 

 et al., 1995). However, the mid-summer ichthyo- 

 plankton community appears to reflect a large-scale 

 onshore to offshore gradient of environmental charac- 

 teristics that include midwater temperatures. 



The ultimate usefulness of age-0 surveys for pre- 

 dicting year-class strength depends upon the rela- 

 tive mortality pressure occurring after the survey 

 period. Although managers seek information on the 

 relative strength of a year class as early as possible, 

 the accuracy and precision of the index may be low 

 for this life history stage for species that suffer vari- 

 able late juvenile predation losses. For T. chalco- 

 gramma, substantial predation upon juveniles may 

 occur in late summer (Livingston 4 ), which may af- 

 fect the magnitude of subsequent recruitment dur- 

 ing some, if not all, years. A distinct advantage for 

 early or mid-summer surveys of juveniles is that 

 fish at this stage have not developed complicated diel 

 vertical and inshore migrations or complex aggrega- 

 tion and schooling patterns that generally make later 

 stage assessment so difficult (Koeller et al., 1986; 

 God0 et al., 1991; Lough and Potter, 1993; Wilson et 

 al., in press). Determining whether year-class 

 strength for this population is set by mid-summer will 

 require more years of abundance estimates, as well as 

 sampling for several other life history stages in months 

 both before and after the period surveyed in this study 

 (Bailey and Spring, 1992; Bailey et al. 3 ). 



Acknowledgments 



We extend our appreciation to Sarah Hinckley for 

 designing the study and for serving as Chief Scien- 

 tist on the cruise. We thank Jay Clark and Bill Rugen 

 for technical assistance in data analysis, Kathy Mier 

 for statistical assistance, Leslie Lawrence for pro- 

 cessing the temperature data, and Art Kendall, Kevin 

 Bailey, Ann Matarese, Geoff Moser, Bill Rugen, and 

 two anonymous reviewers for comments on earlier 

 drafts of the manuscript. 



3 Bailey, K. M., R. D. Brodeur, and A. B. Hollowed. Cohort sur- 

 vival patterns of walleye pollock, Theragra chalcogramma, in 

 Shelikof Strait, Alaska: a critical factor analysis. Submitted 

 to Fisheries Oceanography. 



4 Livingston, P. A. Groundfish utilization of walleye pollock 

 {Theragra chalcogramma), Pacific herring (Clupea pallasi), 

 and capelin {Mallotus villosus) resources in the Gulf of 

 Alaska. Submitted to Fishery Bulletin. 



