GABRIEL and PEARCY: FEEDING SELECTIVITY OF DOVER SOLE 



category (Figure 3D), electivity increased among 

 all three size categories of fish. This pattern 

 existed for the large-bodied, tubed polychaetes 

 Pectinaria californiensis (20), Pista cristata (27), 

 Nothria sp. (13), and Maldanidae (11), the large- 

 bodied, tubeless polychaete Glycera capitata (7); 

 two small-bodied, tubeless polychaetes Decamas- 

 tus gracilis (4) and Ninoe gemmea (10); and the 

 aplacophoran molluscs (33). In a fifth pattern (not 

 shown I, electivity of a prey taxon did not change 

 significantly with fish size. The polychaetes Lum- 

 brineris latreilli, Glycinde picta, the Paraonidae, 

 and the Sigalionidae followed this pattern, as did 

 the molluscan genus Macoma. Taxa belonging to 

 each category of electivity patterns were not 

 necessarily identical to taxa belonging to each 

 analogous category of frequency of occurrence 

 patterns, since the index of electivity of a 

 prey taxon was based on proportion of numerical 

 abundances of the prey taxon in the diet, rather 

 than on frequency of occurrence. 



Among polychaetes at SGIO, the most highly 

 selected prey taxa were Nephtys sp. and Tharyx 

 sp. by fish of all sizes (Table 4). No specimens of 

 Travisia foetida or Lumbrineris latreilli were 

 found in the core samples so the values shown in 

 Table 4 were based on pooling of the taxa at the 

 generic level. Once again, all principal molluscan 

 taxa were negatively selected, with the exception 

 of aplacophorans (no significant selection) and 

 pteropods, which were not found in core samples. If 

 values for all cumaceans of genus Campylaspis 

 were pooled, the taxon appeared positively 

 selected; however, no species common to both fish 

 and box core were found. In the case of amphipods, 

 three of the four principal taxa were not rep- 

 resented in box core samples. The fourth, Am- 

 pelisca macrocephala, appeared positively selected 

 based on these samples. 



Merely because a taxon is positively selected 

 does not mean it plays an especially important 

 role in diet. For example, frequency of occurrence, 

 biomass contribution, and numerical abundance 

 of gastropods in diets of fish at SGIO were small 

 compared with other taxa. Yet this taxon was posi- 

 tively selected (Table 5). Conversely, even though a 

 taxon is negatively selected, it may still play an 

 important role in diets. Positive selection may also 

 be an artifact of the environmental sampling de- 

 vice. Crustaceans appear highly selected at SGIO 

 (Table 5), but this may be partly due to the ineffi- 

 ciency of the box corer in sampling motile epifauna 

 relative to infauna. 



A significant positive correlation was found be- 

 tween prey size (weight/length) of 16 prey and the 

 length offish at which that prey began to signifi- 

 cantly increase in frequency ( r = 0.540, P-^0.05) 

 (Figure 4 1. Large prey were consumed by large 

 fish. Prey which decreased in frequency were gen- 

 erally small-bodied polychaetes (Ampharetidae, 

 Nephtys sp., Myriochele heeri, M. oculata), am- 

 phipods, and cumaceans. In the case of the five 

 prey taxa which showed no significant change in 

 frequency in fish of different sizes ( Aplacophora, 

 Glycinde picta, Prionospio sp., Tharyx sp., and 

 Lumbrineris latreilli) some other criteria for in- 

 clusion in diet may have been more important 

 than size. 



When prey frequency increased with increased 

 size of Dover sole at SG29, predator selectivity also 

 appeared to increase. From the iterative chi- 

 square tests described earlier, a fish length was 

 found which divided the total fish size range into 

 (usually) two length intervals of statistically 

 homogeneous prey frequency for each prey taxon. 

 A value of £ was calculated for each interval, and 

 the difference in values between the two intervals 

 was determined for each applicable prey taxon. 

 This difference was then plotted against the fish 



5r- 



20 



e 



bJ 

 M 

 CO 



> 

 Q 

 O 

 CD 



X 



UJ 

 Q 



15 



20 25 30 35 



FISH LENGTH (cm) 



Figure 4.— Body size of prey vs. fish length at which frequency 

 of prey increased significantly Numbers designate prey taxa i see 

 Table 3). 



757 



