FISHERY BULLETIN: VOL. 87, NO. 3, 1989 



hatchling length, and between initial yolk vol- 

 ume and yolk volume at hatching agree with 

 most previous reports based on grouped data 

 (reviewed in Blaxter 1988). The positive rela- 

 tionship between initial yolk volume and age at 

 hatching differs from the generally reported pat- 

 tern of no association. Moreover, the increase in 

 length with age at hatching we observed is con- 

 trary to that reported by Bengtson et al. (1987) 

 who analyzed daily hatching cohorts within fam- 

 ilies. Hatching late, however, appears to occur 

 at the expense of yolk reserves (Fig. 1). This 

 negative relationship was not present in the cor- 

 relations based on family averages (Table 1). 



Neither initial yolk volume, length at hatch- 

 ing, nor yolk volume at hatching was related to 

 posthatching lifespan when evaluated within in- 

 dividuals. This contrasts both with our correla- 

 tions based on family averages and with previous 

 generalizations (from group averages) that large 

 initial egg size leads to hatchlings that survive 

 longer in the absence of food (Blaxter 1988). The 

 only association with posthatching Hfespan that 

 we detected was the direct influence of oil 

 globule volume at hatching. Our data indicate 

 that if all else were held constant, an increase in 

 oil globule volume delays the time to starvation 

 of individual larvae. This prolonged posthatching 

 hfespan should increase the chances of encoun- 

 tering suitable food before irreversible starva- 

 tion and, thus, the chances of survival. Lipids, 

 whether aggi-egated in oil globules or dispersed 

 throughout the yolk, appear to serve primarily 

 as energy reserves (Blaxter 1969). Conservation 

 of oil globules (relative to yolk) has been re- 

 ported for species that have oil globules in their 

 yolk sacs (e.g., May 1971; Bagarinao 1986) al- 

 though oil globules may also serve to regulate 

 buoyancy. 



A variety of evidence argues for advantages of 

 large size at hatching (a positive correlate of 

 initial yolk volume). Larger larvae have greater 

 mouth widths (Shirota 1970), have gi'eater suc- 

 cess in establishing feeding (Knutsen and Tilseth 

 1985), are more effective predators (Bla.xter and 

 Staines 1970; Hunter 1981), consume a gi'eater 

 range of prey sizes (Hunter 1981), and have 

 higher survival when predators are absent 

 (Rosenberg and Haugen 1982; Henrich 1988) or 

 present (Lilleland Lasker 1971; Bailey 1984; 

 Folkvord and Hunter 1986; Purcell et al". 1987). 

 Although many of these studies compared larvae 

 of two or more gi'oups that differed in age as well 

 as size, the results are likely to apply to the finer 

 size differences among contemporaries. Mortal- 



ity has also been reported to be size specific and 

 to be concentrated early in the life cycle in na- 

 tural populations (Crecco et al. 1983; Smith 1985; 

 Rice et al. 1987; Savoy and Crecco 1988). 



The cost of producing large eggs is frequently 

 evaluated under the assumption of a trade-off 

 between egg size and egg number (Svardson 

 1949; Smith and Fretwell 1974). This assumption 

 has led to optimality derivations for balances 

 between size and number of eggs under given 

 conditions. Observed patterns of size and num- 

 ber of eggs among species, populations, or repro- 

 ductive modes are then evaluated in light of the 

 predicted optima (Ware 1975; Sargent et al. 

 1987; Tanasichuk and Ware (1987)). The correla- 

 tions between ELH traits we observed suggest 

 that egg size expresses reciprocity with other 

 ELH traits to which optimahty methods may be 

 applied. For example, a corollary of the direct 

 relationship between length and age at hatching 

 in capelin is that an inverse relationship exists 

 between length at and developmental rate to 

 hatching (the reciprocal of age at hatching). If 

 daily mortality rates are gi-eater in the embry- 

 onic than in the larval period, an extended em- 

 bryonic period is seen as a cost of being large at 

 hatching. 



This optimality approach is inherently deter- 

 ministic. It does not explicitly admit within- 

 population and within-female variation in egg 

 size (Fig. 2) and in other ELH traits we ob- 

 served. Rather, this variation is viewed as fail- 

 ure to achieve the optimal trait value or com- 

 bination of trait values. In contrast, we consider 

 trait variation itself to be adaptive. For capehn, 

 the emergence of larvae from their intertidal 

 incubation sites into the nearshore water is 

 linked to the episodic occurrence of relatively 

 warm water and nearshore turbulence, gener- 

 ated by onshore winds and coastal water mass 

 exchanges (Frank and Leggett 1981a, 1983; Tag- 

 gart and Leggett 1987b). The attainment of a 

 developmental stage from which an embryo can 

 be induced to hatch (e.g., through a sudden rise 

 in temperature and/or turbulence) or its nutri- 

 tional state at the time of emergence, if pre- 

 viously hatched (Frank and Leggett 1982), are 

 important determinants of survival during early 

 pelagic life. However, the intertidal incubation 

 zone is variable for capelin in three fundamental 

 ways. First, temperature directly influences 

 developmental rates of embryos. Local tempera- 

 tures in the intertidal substrate oscillate up to 

 10°C (Frank and Leggett 1981b) with tidal cycle. 

 Thus, there are large microsite differences in 



520 



