542 



Fishery Bulletin 92(3). 1994 



with each egg or larva, used to estimate mean depth, 

 is the depth of the net where it was collected. Addi- 

 tionally, observations of lengths of larvae in each net 

 were used to estimate mean lengths. Gaussian sta- 

 tistics, including analysis of variance (ANOVA) and 

 analysis of covariance (ANCOVA), were used to com- 

 pare mean depths of eggs and larvae and stages of eggs, 

 and lengths of larvae with various environmental vari- 

 ables: time of day, temperature, salinity, and depth. 



Temperature and salinity data from vertical Sea- 

 bird conductivity, temperature, and depth (CTD) 

 casts made in conjunction with the MOCNESS tows 

 in 1986 were processed at the Pacific Marine Envi- 

 ronmental Laboratory, Seattle. CTD data were col- 

 lected with Seabird sensors on the MOCNESS in 

 1987 and 1988 and were processed at the Atlantic 

 Oceanographic and Meteorological Laboratory, Mi- 

 ami. CTD data were available at 1-m depth inter- 

 vals. In comparing temperature and density (ex- 

 pressed as o t ) with depths of occurrence of eggs and 

 larvae among tows, the mean and standard devia- 

 tion of temperature and density were calculated 

 within the depth interval of +/- one standard devia- 

 tion of the mean depth of eggs or larvae. Egg mean 

 depth weighted by abundance was regressed on tem- 

 perature. Meteorological data from the Kodiak air- 

 port for the sampling periods in April and May 1986, 

 1987, and 1988 were used to model hourly irradi- 

 ance at depth in the area by using an attenuation 

 coefficient of 0.16 (a mean value of 33 measurements 

 [SD=0.103] made aboard ship in Shelikof Strait be- 

 tween 3 and 23 May 199 1 2 ). 



2 Davis, R. AFSC. Personal commun., April 1992. 



Data from the 36 tows were grouped into nine "se- 

 ries" that were numbered sequentially based on cal- 

 endar day without regard to year. The first time se- 

 ries included eggs sampled on 12-16 April, and the 

 ninth series included larvae sampled on 23-25 May. 

 Tows within a series were taken with the same ra- 

 tionale; i.e. in a fixed location or following a drogue, 

 with similar depth schemes, and close to the same 

 dates (Table 1). Each series was composed of one to 

 eight tows taken in areas of high egg and/or larval 

 concentrations except series six and seven, which 

 were taken primarily for zooplankton studies (Fig. 

 1). Series eight, in mid-May, was taken about seven 

 weeks after peak spawning, in the spawning area. 

 Series four and nine had time series sufficient to in- 

 vestigate diel differences in vertical distribution. 

 Tows during these two series were taken at local 

 midnight (midway between sunset and sunrise), 

 dawn (sun 20° above the horizon), local noon, and 

 dusk (sun 20 above the horizon). 



Results 



Overall densities 



Egg densities were highest in the first series of tows, 

 taken between 12 and 16 April, when mean concen- 

 tration in the water column based on all nets in the 

 tows varied from 12,057 to 34,734 eggs/1,000 m 3 (SD 

 ranged from 5,827 to 30,619). Egg densities were 

 greater than 1,000/1,000 m 3 per tow in series two, 

 and generally decreased as the season progressed 

 (Table 2). Eggs were present during series three to 

 eight through the middle of May, but in reduced con- 



