342 



Fishery Bulletin 93(2). 1995 



It is not clear why the amount of shrinkage is re- 

 lated to larval size for some fixatives and not for oth- 

 ers. In this study for example, shrinkage was a con- 

 stant proportion of size for larvae preserved in Z- 

 Fix. However, in Bouin's solution and 5% formalin, 

 smaller larvae shrank more than larger ones. 



The net-treatment experiment showed that shrink- 

 age increased with elapsed time in the net, as in other 

 studies addressing net-capture shrinkage of larval 

 fish (Theilacker, 1980; Hay, 1981; McGurk, 1985). 

 Although our field collections averaged 7.5 min, we 

 included net shrinkage values for up to 20 min in a 

 net for adjusting the size of larvae collected in stan- 

 dard bongo net hauls and MOCNESS tows. 



The histological criteria demonstrated that a sig- 

 nificant number of larval walleye pollock were starv- 

 ing in the Shelikof Strait, Alaska, in April and May, 

 1991. These results agree with the 1991 field study 

 of Bailey et al. 3 who used biochemical criteria to de- 

 termine condition of walleye pollock collected in the 

 same area of Shelikof Strait. They found that more 

 walleye pollock were in poor condition in 1991 than 

 in 1990. Subsequent larval mortality, determined 

 independently by ageing larvae from sequential 

 cruises spanning 2 to 3 weeks, was also higher in 

 1991 than in 1990. 3 Concentrations of copepod nau- 

 plii and invertebrate eggs, the main prey eaten by 

 walleye pollock (Canino et al., 1991), were anoma- 

 lously low throughout Shelikof Strait in 1991, aver- 



aging 6 prey/L, as compared with 38 prey/L in 1990 3 

 and >20 prey/L in earlier years (Incze et al., 1990; 

 Canino et al., 1991). Others have shown that condi- 

 tion of wild larvae is associated with food availabil- 

 ity. In particular, Canino et al. (1991) using a bio- 

 chemical index showed that larval walleye pollock 

 in Shelikof Strait inhabiting areas of sparse prey 

 were in poorer condition and had fewer prey in their 

 guts than their counterparts inhabiting areas of high 

 prey density. Likewise, larval haddock, Melano- 

 grammus aeglefinus, and cod, Gadus morhua, a close 

 relative of walleye pollock, were shown to be in poorer 

 condition in well-mixed areas on Georges Bank than 

 in stratified sites where prey levels were higher 

 (Buckley and Lough, 1987). Kashuba and Mathews 

 (1984) showed that poor histological condition of lar- 

 val shad, Dorosoma spp., correlated with low prey 

 levels and with a subsequent abrupt decline in the 

 population. 



Our results for walleye pollock indicate that the 

 youngest larvae are most vulnerable to starvation. 

 While 29% of the first-feeding walleye pollock (com- 

 bined <5.50 mm SL groups; n=29/99; Table 5) were 

 classified as starving, one week later the number was 

 reduced to 12%, and 2 weeks later it was zero (Table 

 5). Others also have found that starvation of larval 

 fishes in the sea decreases quickly, usually within 1 

 or 2 weeks, as fish larvae mature (O'Connell, 1980; 

 Theilacker, 1986; Robinson and Ware, 1988). Resis- 

 tance to starvation increases after larval fish first 

 feed (Hunter, 1972; Blaxter and Staines, 1971), ac- 

 quire the ability to eat more varied prey (Hunter, 

 1972; Arthur, 1976) and are able to store energy re- 

 serves (Ehrlich, 1974; Fraser, 1989; Hakanson, 1989). 

 We also found patchy areas along the Shelikof Strait 

 sea valley in early May 1991 with large numbers of 

 starving larvae, two to three times the background 

 level. Whether small areas of high mortality affect 

 total recruitment is unknown. 



Despite arguments to the contrary (Sissenwine, 

 1984; Peterman et al., 1988) and a general belief that 

 starvation is not a widespread occurrence in the sea 

 (Heath, 1992), evidence from this study shows that 

 starvation does occur and that it is the young stages 

 of walleye pollock that are vulnerable. The advan- 

 tages of the midgut histological assay, rather than 

 one requiring grading of several tissues, is that it 

 takes less time than does an extensive histological 

 background and is a quantitative rather than a quali- 

 tative measure (Theilacker and Watanabe, 1989). 

 Additionally, rates of starvation-induced mortality 

 may be estimated by using the assay and employing 

 laboratory-determined growth rates to determine size 

 and stage durations. Currently, studies are under- 

 way to correlate changes in the physical environment 



