BECKER and CHEW: PREDATION ON CAPITELLA SPP. 



intermediate faunal zone (i.e., 30-70 m depth) 

 identified for Puget Sound demersal fish assem- 

 blages by Wingert and Miller (1979). All three 

 target species are sufficiently abundant in this 

 depth zone to allow a quantitative food habits 

 analysis to be conducted. Data from both tran- 

 sects were pooled prior to analysis. 



Fishes were collected using a 7.6 m (headrope) 

 otter trawl having a body mesh size of 3.2 cm 

 (stretched) and a cod-end liner mesh size of 0.8 cm 

 (stretched). Trawling was conducted along both 

 transects at a constant vessel speed of approxi- 

 mately 2.5 kn. All positioning was achieved using 

 the LORAN-C navigation system. To assess diel 

 variations in feeding behavior, hauls were made 

 along each transect during four periods of the diel 

 cycle: morning (0900-1030 h), afternoon (1300- 

 1500 h), evening (1900-2030 h), and night (2330- 

 0100 h). Each transect was sampled twice during 

 each time period, yielding eight hauls per tran- 

 sect or a total of 16 hauls for the study. 



At sea, the stomach contents of the target spe- 

 cies were fixed by injecting, using a 50 cc syringe, 

 a 10% solution of buffered formalin into the body 

 cavity of each individual. These fishes were then 

 brought to the laboratory and stored at 4°C. 



Benthic invertebrates along each transect were 

 sampled within 2 days of trawling. Organisms 

 were collected using a 0.1 m^ van Veen bottom 

 grab, sieved through a 1.0 mm mesh screen, fixed 

 with a 10% solution of buffered formalin, and 

 transferred to 70% ethanol for storage. A single 

 grab sample was taken during daytime at each of 

 five sampling points positioned at approximately 

 equal distances along each transect (Fig. 1). 



Laboratory Analysis 



Within 5 days of sampling, the total length (TL) 

 of each fish was measured to the nearest 1.0 mm. 

 The body cavity was then opened and the stomach 

 was removed by severing the esophagus and py- 

 lorus. Stomachs were stored in 70% ethanol prior 

 to analysis. 



For food habits analysis, stomachs were sub- 

 sampled from the total pool of available stomachs. 

 To minimize within-species variation as a result 

 of size-dependent foraging patterns (e.g., Gabriel 

 and Pearcy 1981), only individuals within an 80 

 mm length range were selected for analysis. The 

 ranges used for English sole, Dover sole, and rex 

 sole were 240-320, 200-280, and 210-290 mm TL, 

 respectively. Each length range bracketed the 

 median length observed for each species. 



Identifications of all invertebrates in stomachs 

 and benthic samples were made using a dissect- 

 ing microscope. Sizes of all Capitella spp. were 

 estimated using the width of the fifth setiger (cf. 

 Tsutsumi and Kikuchi 1984). This measurement 

 was used instead of body length because many of 

 these polychaetes were fragmented during grab 

 sampling or ingestion by the fishes. Setiger 

 widths were measured to the nearest 0.1 mm 

 using an ocular micrometer. 



The dietary contribution oi Capitella spp. to the 

 total stomach contents of each target species was 

 estimated using percentages based on numerical 

 proportions. In addition, the total number of prey 

 per stomach (i.e., Capitella spp. plus all other or- 

 ganisms) was used as an index of feeding inten- 

 sity for each species. 



Statistical Analysis 



Nonrandom predation on Capitella spp. (i.e., se- 

 lection) was tested by comparing the numerical 

 proportions of these polychaetes in the stomachs 

 of the fishes with the proportion found in the ben- 

 thos using a 2 X 2 contingency formulation and 

 the chi-square criterion (Pearre 1982). Direction 

 of selection was determined by inspecting the rel- 

 ative proportions of prey in the stomachs and ben- 

 thos. Nonrandom size selection of Capitella spp. 

 was tested by comparing the size distributions of 

 these polychaetes in the stomachs of the fishes 

 with the size distribution found in the benthos 

 using the Mann-Whitney U-test. In both of these 

 analyses, four comparisons (i.e., one for each time 

 period) for each species were made with a single 

 set of benthic observations. Because these four 

 comparisons lacked independence, significance 

 levels were adjusted conservatively using Bonfer- 

 roni's technique (Miller 1981). 



To examine how the foraging patterns of Eng- 

 lish sole differed between habitats where benthic 

 assemblages were dominated by Capitella spp. 

 and habitats where assemblages did not include 

 these polychaetes, the values of feeding intensity 

 (i.e., numbers of prey per stomach) found in the 

 present study were compared with those obtained 

 at six other sites in Puget Sound by Becker 

 (1984). These six sites were located at depths be- 

 tween 12 and 32 m, and fishes were sampled and 

 processed using methods identical to those de- 

 scribed for the present study. Values of feeding 

 intensity were compared during each period of 

 the diel cycle using the Mann-Whitney U-test. 

 Similar analyses could not be conducted for Dover 



473 



