472 



Fishery Bulletin 88(3). 1990 



ship between egg size and larval size. Size variation in 

 walleye pollock eggs due to annual, seasonal, and geo- 

 graphical factors was examined from eggs caught in 

 ichthyoplankton surveys in the Gulf of Alaska from 

 1981 to 1986. Variation due to female age, length, and 

 condition was studied using eggs taken from spawn- 

 ing females caught at sea and from females held in the 

 laboratory. Effects of egg size on larval size at hatching 

 and at yolksac absorption were investigated in rear- 

 ing experiments, using eggs from captive females and 

 eggs taken in ichthyoplankton surveys. 



Methods 



Field studies 



Walleye pollock eggs were collected in the Gulf of 

 Alaska between 1981 and 1986 with 60-cm bongo nets 

 (0.505-mm mesh), deployed with standard MARMAP 

 procedures (Posgay and Marak 1980). 



Eggs selected for measurement were chosen from 

 archived samples (Table 2) taken in different years 

 (1981-86), months (March-May), and areas (Shelikof 

 Strait, the Semidi Islands to the Shumagin Islands, and 

 west of the Shumagin Islands; Fig. 1). Variation in 

 mean egg size among individual sampling stations was 

 examined by measuring 25-50 eggs from up to 10 sta- 

 tions per year/month/area combination. Eggs at similar 

 stages of development (early blastula to early germ 

 ring) were measured with an ocular micrometer on a 

 dissecting microscope at 40 x magnification. All eggs 

 were preserved in 3-5% buffered formalin. 



Because mean egg sizes were significantly different 

 {F = 4.75, p<0.0005) among sampling stations, nested 

 analysis of variance models with unequal cell sizes were 

 used to test year, month, and area effects, with sta- 

 tions as the nested factor. Variation in egg size among 

 years (within area and month), among months (within 

 year and area), and among areas (holding year and 

 month constant) had to be tested separately, due to the 

 incompleteness of the available dataset, including miss- 

 ing months and areas in some years (Table 2). 



The General Linear Models procedure of SAS-PC 

 (SAS Inst. 1985) was used in this analysis. The F- 

 statistic was derived using the Satterthwaite approx- 

 imation (Gaylor and Hopper 1969 as cited in Sokal and 

 Rohlf 1981), except in one case where the criteria for 

 using this test were not satisfied; in this case a simple 

 approximation test (Sokal and Rohlf 1981) was used. 



Effects of maternal age, size, and condition on egg 

 diameter and dry weight were examined using eggs 

 from females caught by midwater trawl in Shelikof 

 Strait in late March, 1986 (Table 2). Hydrated (fully 

 ripe) eggs were collected from spawning females, and 

 preserved in 5% buffered formalin. For each female, 

 fork length (FL), gonad- and liver-free body weight, 

 and liver weight were determined and otoliths removed 

 for ageing. Body weight and liver weight were used 

 as indices of condition. 



Sixty eggs per female were measured. Because 

 hydrated eggs taken from the ovary are sometimes 

 slightly nonspherical, egg diameters were measured on 

 the axis that fell along the transect to make sure mea- 

 surements were random. Five to ten eggs per fish were 

 subsampled for dry weight; eggs were measured. 



