Pepin: The encounter rate of Mallotus villosus witli fish eggs 



211 



based on an evacuation rate of 5Q'7< of eggs per hour. 

 A low predation rate on fish eggs is not entirely unex- 

 pected. The eggs of cunner and cod from our samples 

 had an average diameter of 0.86 and 1.3 mm, whereas 

 the modal length of capelin from our collections was 

 approximately 130 mm, making the eggs -1% of the 

 predator's body length. Based on Paradis et al.'s (1996) 

 analysis, maximum predation rates by fish feeding on 

 ichthyoplankton occur when the prey are 10% of the 

 predator's length, and the minimum predation rates 

 were observed when prey were -0.5-1% of the predator's 

 length. Paradis et al.'s (1996) summary equations indi- 

 cate that an average predation rate of 0.007-0.03 eggs/ 

 h would not be inconsistent with laboratory estimates 

 of predation rates for a range of evacuation times from 

 1 to 4 hours and would be considerably lower than the 

 predation rates estimated in the present study. How- 

 ever, estimates based on a simple foraging model (e.g., 

 Paradis et al., 1999) for a modal capelin swimming at 

 1 body length/s and with a reactive distance of 1 body 

 length yield a mean of 12 eggs/h, and halving these 

 parameter values yields a predation rate of 1.6 eggs/h. 

 These alternative methods of estimating predation rates 

 indicate that capelin are somewhat more likely to feed 

 on fish eggs in the field than laboratory studies would 

 indicate but less likely than would be anticipated from 

 a simple foraging model. The results from this study do 

 not provide an exhaustive estimate of encounter rates 

 between capelin and pelagic fish eggs, but the range 

 of effective encounter rates derived from the different 

 approaches highlights the lack of understanding of this 

 key parameter essential for any model of multispecies 

 interactions. Such a broad range of expected predation 

 (and encounter) rates indicates that in order to develop 

 predictive skills in estimating the potential impact of 

 planktivores on fish eggs, a better understanding of the 

 factors that prompt a predator to feed on a fish egg is 

 needed. Most laboratory studies dealing with predation 

 by fish on fish eggs and larvae have used single prey 

 and few have provided a measure of the probability of 

 attack (Bailey and Houde, 1989; Paradis et al., 1996). 

 Feeding patterns can be affected by the complexity of 

 the available prey community (Kean-Howie et al., 1988; 

 Gotceitas and Brown, 1993; Pepin and Shears, 1995), 

 but methods to relate prey consumption with prey avail- 

 ability in the field are still limited. 



Uncertainty in encounter and predation rates are due 

 partly to errors in the estimation of prey availability 

 and stomach content and partly due to our general lack 

 of knowledge concerning the evacuation rate of fish 

 eggs from predator stomachs. For this study I relied on 

 Hunter and Kimbrell's (1980) observations of evacuation 

 rates for northern anchovy from experiments conducted 

 at 15°C. Water temperatures in Trinity Bay in July 

 2000 showed considerable variation with depth, rang- 

 ing from ~13°C at the surface to -1°C at 100 m, with 

 the overall average temperature over the water column 

 being ~3-4°C. Given that most metabolic processes 

 increase by 1.3-2 times over ten degrees Celsius (e.g., 

 Brett and Groves, 1979), evacuation rates in the pres- 



ent study could have been half of those measured by 

 Hunter and Kimbrell (1980) and thus would imply that 

 our observations of stomach contents could represent 

 70% of the eggs consumed in the last hour rather than 

 the 50% assumed in the analysis. The result would be 

 that estimates of volumes swept by capelin would have 

 to be reduced by -30%, further reducing their potential 

 impact on ichthyoplankton populations. The general un- 

 certainty in our knowledge of evacuation rates for fish 

 eggs from the stomachs of capelin, and from most other 

 planktivorous fish, therefore limits the inferences that 

 can be derived about their impact on ichthyoplankton 

 survival. 



A second source of uncertainty is due to the inher- 

 ent sampling variability associated with the study of 

 elements that form a small fraction of the prey of a 

 predator. This study was based on more than 1000 

 specimens from a small region in order to provide the 

 greatest accuracy possible within the confines of the 

 study area. Stomach sampling is often restricted to 

 fewer specimens sampled over a much broader geo- 

 graphic range. Without a large number of observations 

 from a local environmental setting (i.e., site), estimates 

 of mean numbers of rare prey types would likely be 

 unable to reflect the effect of site-specific differences 

 in environmental conditions on predator feeding and 

 would thereby increase the uncertainty around the 

 estimated effective encounter rates. The situation is 

 well illustrated by the doubling in the frequency of 

 occurrence of fish eggs in the specimens that were ran- 

 domly selected for complete gut analysis versus all the 

 specimens collected for analysis. Further uncertainty 

 comes from the variability in estimates of egg density 

 encountered by capelin — a variability that tended to 

 result in slightly tighter confidence intervals in esti- 

 mates of effective volume swept than the uncertainty 

 due to variability in the number of prey per predator 

 stomach. Variations in egg densities from plankton 

 samples did result in lower estimates of effective vol- 

 ume swept, partly as a result of the rare high densities 

 (typical of highly skewed distribution) that characterize 

 the variability in plankton catches (Power and Moser, 

 1999). Although this study is not intended to provide 

 a general estimate of the effective encounter rate of 

 capelin feeding on fish eggs, the approach provides a 

 useful example of the sampling requirements for the 

 study of feeding on prey that are a minor part of the 

 diet of a predator but which may be greatly affected by 

 the predator's impact on the prey population. 



Greater efforts must be directed at understanding 

 patterns of predation in the field, in relation to prey 

 availability, if the role of planktivorous fish in ichthyo- 

 plankton dynamics is to be understood. The assump- 

 tions in the estimation approach could affect the effec- 

 tive volume swept: estimates of the effective volume 

 swept are based on average prey densities integrated 

 over the water column, and the simplification of the 

 encounter model (Eq. 1) incorporates possible variations 

 in the probability of attack (e.g., due to diurnal varia- 

 tions in visibility of eggs) into the value of K. Pepin et 



