372 



Fishery Bulletin 94(2), 1996 



of large larvae to a predator may have more than 

 compensated for the fact that fewer larvae were cap- 

 tured as prey size increased, between trials. 



I evaluated this proposition by estimating the to- 

 tal mass of prey that were consumed by predators in 

 each of the 23 predation trials conducted by Pepin et 

 al. (1992). I estimated the total number of prey in- 

 gested for each trial in two ways. First, as an esti- 

 mate of the maximum number of prey ingested, I used 

 the difference between the number of prey present 

 at the beginning and end of each trial (Pepin, unpubl. 

 data). Second, I calculated an estimate for the mini- 

 mum number of prey ingested by using Equation 2 

 from Pepin et al. ( 1992). For each trial I estimated 

 the total mass of prey consumed by multiplying the 

 number of prey ingested by the mass of a capelin 

 larva whose length was equal to the mean of prey 

 lengths used in that trial. To calculate the mass of 

 larvae of average length, I used a general allometric 

 length-weight relationship for capelin ( Pepin, 1995 ). 

 The results demonstrate that the maximum total 

 number of prey consumed per predator decreased 

 significantly with increasing prey length between 

 trials (Fig. 1A; F=22.1, df=l,21, P=0.0001, for the 

 pooled data set). Moreover, the total maximum mass 

 of prey consumed increased significantly with in- 

 creasing prey length (Fig. IB; F=52.8, df=l,21, P« 

 0.001; see also Fig. 9 in Pepin et al., 1992). The lat- 



ter result also held within all three categories of prey 

 and predator length ratios. 



It is worth noting that the group of triangles be- 

 low the regression lines (Fig. 1, A and B) represent 

 five trials from an experiment that employed the 

 smallest predators (mean=38.4 mm vs. 50.6 mm for 

 all 18 trials in other experiments; t=3.5, df=21, 

 P=0.002) and the lowest temperatures (mean=11.6"C 

 vs. 13.2"C for all other trials; £=2.6, df=21, P=0.016). 

 It is likely that the low number of prey consumed 

 during those trials (mean=393 vs. 474 for all other 

 trials; t =7.9, df=21,P«0.001) resulted from the small 

 predator size and low temperature, because smaller 

 predators consume fewer prey (Pepin et al., 1992) 

 and activity and ingestion rates are generally reduced 

 at low temperatures. 



The minimum total number of prey consumed also 

 decreased significantly with increasing prey length 

 between trials (F=20.8, df=l,21, P=0.0002, for the 

 pooled data set). The mass of prey consumed also in- 

 creased with prey length within each prey and preda- 

 tor length category when the minimum number of prey 

 consumed were used. When the results from each prey 

 and predator length category were pooled, the relation- 

 ship between prey consumption and prey length was 

 not significant owing to the low prey consumption val- 

 ues for the five trials with anomalous temperatures 

 and predator lengths (see above). There was a highly 



5C 



4! 











' 



578 





17 



; ; 











5 56 



5 78 



600 



Figure 1 



The total mass and number of larval capelin. Mallotus villosus, consumed by threespine stickleback, 

 Gasterostrus cirulfatus. Estimates of the maximum total number (A) (y=918 85 9x) and mass (3)y=-ll.l+4.7x) 

 of prey consumed by sticklebacks versus the mean length of capelin larvae from 23 predation trials conducted 

 by Pepin et al. ( 1992). The categories of prey and predator length ratio were assigned arbitrarily (on the basis 

 of unpubl. data from P. Pepin). Regression equations were based on the pooled data set. 



