Wilson; Effects of year and region on abundance and size of Theragra chalcogramma 



825 



that the Kodiak region is sometimes a relatively impor- 

 tant nursery area despite its position upstream of the 

 main spawning area. 



Materials and methods 



The data selected for this study comprise the longest 

 time series available from the GOA where the collection 

 method was consistent (Bailey and Spring, 1992; Spring 

 and Bailey*). They were collected in late summer (August- 

 October) 1985-88 by the NMFS (Fig. 1, Table 1). The pur- 

 pose of these sui^veys was to obtain a coarse picture of the 

 distribution and size of age-0 pollock in the western GOA; 

 These surveys were the first to target age-0 pollock in this 

 area and, as such, were exploratory. 



The area surveyed was the continental shelf and bays 

 around Kodiak Island and as far southwest as time 

 allowed (Fig. 1). The Alaska Coastal Current (ACC) flows 

 southwestward through the area but circulation is com- 

 plicated by sea valleys, islands, and other topographic 

 complexity (Lagerloef 1983; Reed and Schumacher, 1986; 

 Stabeno et al., 1995). The largest sea valley forms Shelikof 

 Strait and it forms a natural division within the western 

 GOA between eastern and western regions. The Kodiak 

 Island Archipelago is a prominent feature of the eastern 

 region. Smaller islands characterize the western region, 

 which has a relatively broad shelf 



Sampling was accomplished by trawling on echo layers. 

 Echo sign was monitored along predetermined transects. 

 No effort was made to include acoustic data because a pre- 

 liminary study found no relation between echo sign and 

 age-0 catches (Bailey and Spring, 1992). Samples were col- 

 lected only during the day, so that the problem of diel ver- 

 tical fish migi-ation was avoided. 



Echo sign believed to be age-0 pollock was sampled by 

 using an 18.6-m high, opening shrimp trawl with steel 

 V-doors (1.5 m x 2.1 m, 568 kg) attached by 18-m bridles. 

 The trawl was made of 3.2-cm stretched-mesh, nylon net- 

 ting, and a 3-mm mesh liner was inserted into the codend. 

 For bottom samples, a 16.8-m tickler chain was used and the 

 footrope of the net was held above a bottom contact chain by 

 30-cm chain lengths. For midwater tows, a 363-kg weight 

 was attached to each lower wing to help hold the trawl 

 mouth open, and the chain on the footrope was removed. A 

 netsonde system (BEN-MAR or Furuno) was used to posi- 

 tion the trawl vertically at the desired depth. It was not 

 possible to open and close the net at depth. Towing speed 

 and duration averaged about 4.5 km per h and 10-15 min, 

 respectively. At each sampling location, sea surface temper- 

 ature was measured with a bucket thermometer. 



All taxa caught were identified, enumerated, and 

 weighed. At this time of year, age-0 pollock are readily 

 distinguished from older individuals by a clear separa- 

 tion in size (Bailey et al., 1996b; Brodeur and Wilson, 

 1996a; author's unpubl. data). Fork length was measured 

 to the nearest centimeter. All age-0 pollock were measured 

 except those in large catches, when about 100 randomly 

 selected fish were measured. Length data were missing for 

 five samples, each of which comprised only a few fish. 



The effect of year and region was examined first on 

 age-0 abundance, and then on age-0 length. Finally, year- 

 and region-specific abundance and size estimates were 

 compared with recruitment. 



Examination of effects of year and region on age-0 abun- 

 dance was accomplished in five steps. The first step was 

 to standardize the catch to account for differences in the 

 volume filtered between tows. Volume filtered for each tow 

 was calculated as the trawl mouth area times distance 

 fished. The mouth area was assumed to be 37 m- for hauls 

 on the bottom (Wathne, 1977) or 28 m- for hauls in midwa- 

 ter (Wilson et al., 1996). No adjustment was made for dif- 

 ferences in the depth sampled because most (92%) samples 

 were collected at depths where net dimensions are rela- 

 tively stable (Wilson et al., 1996). Distance fished was the 

 straight-line distance between two geogi'aphic points along 

 which the net was fished at depth. Dividing each age-0 

 catch by the volume filtered produced fish density (fish per 

 m'^), which was used to indicate abundance. Volume was 

 not converted to sea surface area because of uncertainty 

 about the depth and area represented by each sample. By 

 using density, the assumption of a constant depth and area 

 is explicit, and different values can easily be applied post- 

 hoc if an estimate of absolute abundance is desired. 



The second step was to adjust for mortality during the 

 3-5 week survey period. An instantaneous daily mortality 

 rate of 0.014 (Hollowed et al, 1996) was used to adjust each 

 density to the overall median sui-vey date (3 September). 



The third step was to account for 22 tows throughout 

 the survey area that were apparently replicates of other 

 tows. Samples were judged to be pseudoreplicates if they 

 were consecutive, were the same type (bottom or midwa- 

 ter), and were within one nautical mile and 50 m depth of 

 each other. The pseudoreplicates were averaged together 

 to give 384 samples from the original 406 trawl hauls. 



The fourth step was to gi-oup fish densities by geo- 

 graphic area by using the three regions defined by Bailey 

 and Spring (19921. A minor adjustment to this stratifi- 

 cation scheme involved moving the delineation between 

 the Unimak and Shumagin regions farther west to better 

 reflect annual differences in sampling effort (Fig. 1). The 

 Unimak Island region was sui-veyed only during 1987-88. 

 Thus, samples from the area upstream of the main spawn- 



