371 



Size-dependent predation risk in 

 larval fishes: mechanistic inferences 

 and levels of analysis 



Douglas F. Bertram* 



Friday Harbor Laboratories, University of Washington 



620 University Road. Friday Harbor. Washington 98250-9299 



In any ecological study, the level of 

 observation, as well as the aggre- 

 gation and analysis of data, affects 

 the inferences that can be drawn. 

 In fisheries science, many studies, 

 covering a wide range of levels of 

 analysis, have examined size-de- 

 pendent processes, particularly 

 predation, as a primary source of 

 mortality for larval fish (reviewed 

 by Bailey and Houde, 1989; Bert- 

 ram and Leggett, 1994; Leggett and 

 Deblois, 1994). A general result 

 from those studies is that mortal- 

 ity rate declines as body size in- 

 creases both across (e.g. Peterson 

 and Wroblewski, 1984) and within 

 species (e.g. van der Veer and 

 Bergman, 1987). However, the use 

 of that general result to forecast 

 survival probabilities at lower lev- 

 els of aggregation and analysis (e.g. 

 for individual larvae) could lead to 

 erroneous predictions (Pepin and 

 Miller, 1993). Several recent stud- 

 ies have questioned the validity of 

 the generality that increasing lar- 

 val size reduces predation risk 

 when analyses were conducted at 

 the level of the individual and when 

 the effects of prey size on vulner- 

 ability to predation were examined 

 independently of age effects (Li tvak 

 and Leggett, 1992; Pepin et al., 

 1992; Bertram and Leggett, 1994). 

 The results of Pepin et al. ( 1992) 

 show that the effect of larval size 

 inversely affects vulnerability to 

 predation, depending on the level 

 at which the information is aggre- 

 gated and analyzed. Contrasting 

 patterns of larval mortality caused 



by predation within and between 

 mesocosmic (2.7 m J circular tanks) 

 trials were observed in the preda- 

 tion of yolk-sac and first-feeding 

 capelin larvae, Mallotus villosus, by 

 threespine sticklebacks, Gaster- 

 osteus aculeatus (Pepin et al., 

 1992). Prey age ranged from 1 to 

 5 d (median post-emergence) and 

 prey size ranged from 5 to 5.8 mm. 

 In each experimental trial, single 

 naive predators were offered a co- 

 hort of 500 similar-age prey for 24 

 h. Comparisons between experi- 

 ments indicated that greater mean 

 size and age of larvae in the meso- 

 cosm reduced mortality due to pre- 

 dation. However, within experi- 

 ments, larger larvae suffered higher 

 mortality than did smaller indi- 

 viduals of the same age (see also 

 Litvak and Leggett, 1992). Resolv- 

 ing the discrepancy between these 

 results is important because the 

 conclusions differ depending upon 

 the level of observation and analy- 

 sis. Large size and age were ben- 

 eficial to larval survival when mean 

 values for each trial were compared 

 over the full range of available prey 

 sizes (i.e. between trial compari- 

 sons). Conversely, large size at age 

 was detrimental to larval survival 

 at a lower level of observation and 

 analysis (i.e. within trials) when a 

 reduced range of prey sizes was 

 examined. Patterns and generali- 

 zations based on observations from 

 higher levels of aggregation (e.g. 

 cohorts, populations, species) and 

 extended ranges of data, although 

 conceptually appealing, may be 



misleading for gaining insights into 

 the mechanistic processes that af- 

 fect larval survival at the level of 

 the individual, where selection acts 

 (Pepin and Miller, 1993). Moreover, 

 to achieve an understanding of the 

 mechanisms that determine over- 

 all survival, information from sev- 

 eral levels of observation must be 

 integrated (Crowder et al., 1992; 

 Pepin and Miller, 1993). In this 

 study I use fundamental foraging 

 theory to examine an underlying 

 mechanism that could have pro- 

 duced both of the observed mortal- 

 ity patterns in Pepin et al.'s ( 1992) 

 results at low and higher levels of 

 analysis, within and between trials, 

 respectively. 



Pepin et al. (1992) argued that 

 the within- versus between-trial 

 differences occurred because preda- 

 tors actively selected larger larvae 

 within trials ("because of greater 

 encounter or attack probabilities 

 due to greater activity or longer 

 perception distance [i.e. the dis- 

 tance that larvae were visible to the 

 predator] relative to smaller prey") 

 but as the average size of prey in- 

 creased between trials, the average 

 predation rate decreased owing to 

 greater avoidance capabilities of 

 the larger, older larvae. In accor- 

 dance with foraging theory (e.g. 

 Stephens and Krebs, 1986), the 

 predators in Pepin et al.'s (1992) 

 study may also have actively se- 

 lected larger larvae because of their 

 higher caloric value and hence their 

 profitability to a predator within 

 trials (prey handling time was 

 likely to be size-independent in the 

 study system; energetic costs of at- 

 tack were also assumed to be inde- 

 pendent of prey size). Furthermore, 

 the consistently high profitability 



Send reprint requests to Department of 

 Biological Sciences. Simon Fraser Univer- 

 sity, Burnabv, B.C., V5A 1S6 Canada. 



Manuscript accepted 11 October 1995. 

 Fishery Bulletin 94:371-373 ( 1996). 



