Pepin The encounter rate of Mallotus villosus with fish eggs 



207 



algorithm failed to converge because of missing observa- 

 tions. Effective volume swept (K) was calculated from 

 the estimated mean number of eggs per stomach (as a 

 measure of encounter rates) and the estimated density 

 of eggs based on plankton samples by re-arranging 

 Equation 2 and assuming that encounters between 

 prey and predators were random over the top 100 m 

 of the water column. I define K as the effective volume 

 swept because it implicitly includes some measure of the 

 probabilities of attack and capture. Values of the esti- 

 mated volume swept were corrected for evacuation rates. 

 Hunter and Kimbrell (1980) found that the evacuation 

 rate of northern anchovy {EngrauUs mordax, Girard 

 1854) feeding on fish eggs at 15°C was -SO'/r/h based 

 on an exponential model (exponential rate of 0.7/h), 

 which is appropriate for microphagic fish (Temming et 

 al., 2002). Based on the duration of our tow and the 

 handling time to process the samples, the observations 

 of stomach content from this study represent 50% of 

 the eggs consumed during the last hour before capture. 

 Hunter and Kimbrell's (1980) estimate of the evacuation 

 rate is approximately twice that obtained in field studies 

 by Arrhenius and Hansson (1994) and Darbyson et al. 

 (2003) for herring (C. harengus L.) feeding primarily on 

 copepods. However, the estimates from the latter two 

 studies were based on total gut fullness and not on the 

 disappearance of specific food items, such as fish eggs. 

 As a result. Hunter and Kimbrell (1980) provided the 

 only current measure of the rate of disappearance of 

 eggs in the stomachs of planktivorous fish. Increases or 

 decreases in evacuation rates because of inaccuracies in 

 our knowledge of the rate of disappearance of eggs from 

 stomach contents will result in corresponding changes 

 in the estimate of the effective volume swept. By using 

 the mean number of eggs per stomach as an estimator 

 of the effective volume swept, the probabilities of attack 

 and capture within the reactive distance are assumed 

 to be 100%. The resulting value of the effective volume 

 swept represents an average over the top 100 m of the 

 water column, the depth range over which we sampled 

 both capelin and zooplankton and over which I assume 

 that the encounters are random. Ideally, one would 

 want to obtain a depth-stratified measure of stomach 

 contents, and prey availability and overlap between prey 

 and predator, such as has been inferred in the Baltic 

 Sea (Kbster and Mbllmann 2000; Koster et al. 2001), 

 but this was not possible in the present study because 

 there were insufficient data on the vertical distribution 

 of capelin. 



Results 



Capelin used in this analysis ranged from 53 to 195 

 mm TL (mean = 122 mm; median = 130 mm). Average 

 zooplankton concentrations ranged from 210 to 3096 

 organisms/m'. In most samples, the copepods Calanus 

 finmarchicus and Pseudocalanus sp. and the larva- 

 cean Fritillaria borealis (Lohmann 1896) represented 

 60-80% of the zooplankton by number, based on samples 



? 0.01 



0.0001 



10000 



1000 



o 



70 



90 110 130 150 170 190 



Length (mm) 





••• Vj* 



• • •* 



70 



90 



110 130 

 Length (mm 



^ — »A« — * 

 150 170 190 



8 12 16 20 



Time of day (h) 



24 



Figure 1 



Weight of stomach contents (upper graph) 

 and number of prey per stomach (center 

 graph) in relation to capelin {Mallotus 

 villosus) length, for specimens with com- 

 plete analysis of contents, and average 

 and standard deviation of the log-trans- 

 formed gut fullness index (GFI) in rela- 

 tion to time of day ( lower graph), based 

 on all fish sampled. Statistics were cal- 

 culated for all samples collected within 

 each 4-hour time interval. 



from bongo nets. In both species of copepods, copepodite 

 stages II and III were most abundant. A finer mesh 

 net would have revealed that Oithoina similis was also 



