Harvey: Effects of El Nino events on consumption and egg production of Sebastes spp. 



81 



Table 4 



Energy consumption estimates for S. mystinus by a deter- 

 ministic baseline model (parameters given in Table 11 

 and simulations run for relative partial sums of squares 

 (RPSS) analysis. Estimates from the RPSS analysis were 

 determined at three levels of parameter uncertainty, with 

 parameter coefficients of variation (CV) equal to 2, 10, or 

 20%. 



Estimated energy consumption 

 (MJ:mean ±SD) 



Model 



Females 



Males 



Baseline 

 CV = 2% 

 CV = 10% 

 CV = 20% 



285.0 



286.0 ±16.1 

 314.3 ±102.8 



515.1 ±1131.4 



174.6 



175.3 ±10.1 

 183.9 ±57.6 

 209.0 ±131.7 



fishing. However, the El Nino models may have overesti- 

 mated per-recruit consumption because I did not add in 

 direct El Nino related mortality; natural mortality may 

 actually increase during El Nino years, as suggested by 

 anecdotal mass mortality events affecting S. mystinus 

 during the 1982-83 El Nino (Bodkin et al., 1987). 



More dramatic than the effect of El Nino on energy 

 consumption was the effect on egg production. Indi- 

 vidual and per-recruit lifetime fecundity dropped (by 

 roughly 12-19% and 15-23%, respectively) in the El 

 Nino models — an effect that was even more drastic as 

 fishing pressure increased. These declines were dispro- 

 portionate in comparison to changes in long-term energy 

 consumption, which declined by <4% at the individual 

 scale and <7% at the per-recruit scale under even an 

 arduous El Nino regime; and compared to changes in 

 the size of age-30 individuals, which were essentially 

 equal in the baseline and El Nino models. In other 

 words, under a long-term climate regime with El Nino 

 events, total energy demand of females is similar to a 

 baseline regime, and lifetime gross conversion efficiency 

 (growth/consumption) increases, but the conversion ef- 

 ficiency of consumption into reproduction is constrained 

 considerably. That constraint is due largely to delayed 

 maturity, poorer overall fecundity (particularly in El 

 Nino years), and, at the per-recruit scale, the culling 

 effect of natural and fishing mortality. 



Of course, the implications from the models for S. 

 mystinus must be viewed as hypotheses based on a ge- 

 neric Sebastes model. Although the ability of the bioen- 

 ergetics approach to synthesize demographic, physiologi- 

 cal, and environmental data makes it a powerful tool 

 for characterizing dynamic linkages between fish, prey 

 communities, and climate, use of this approach for stud- 

 ies of Sebastes will require additional empirical data. 

 A rich body of information exists for some parameters, 

 such as growth rate, fecundity, and depth distribution 

 (Love et al., 1990; Love et al., 2002). However, many 



relevant data are lacking, notably diet data. Because 

 of seasonal changes in temperature and reproductive 

 state, rockfish energetics are also seasonal. Seasonal 

 diet changes have been observed in several (largely in- 

 shore) species (Love and Ebeling, 1978; Hallacher and 

 Roberts, 1985; Hobson and Chess, 1988; Murie, 1995). 

 Diets may also change with fish size (Love and Ebel- 

 ing, 1978; Murie, 1995). Data that capture the trophic 

 ontogeny of different species would allow a better depic- 

 tion of how energy consumptive patterns of a population 

 change with demographics, particularly given the dis- 

 proportionate demands of younger age classes (Fig. 4). 

 When possible, diet data should be based on weight or 

 volume so that estimates of energy requirements can be 

 readily converted into masses of prey consumed. 



Properly incorporating environmental variability will 

 require information not just on temperature variability, 

 but on how rockfish growth, reproduction, and diet vary 

 under different climate regimes. As discussed previ- 

 ously, El Nino and Pacific Decadal Oscillation events 

 have been shown to affect growth, fecundity, and re- 

 cruitment success of some well-studied species of rock- 

 fish. Little information is available on how these factors 

 are affected by La Nina events, however. Furthermore, 

 climate variability may lead to markedly different prey 

 communities (Brodeur and Pearcy, 1992; Lea et al., 

 1999), resulting in diet shifts about which we currently 

 have little information for most rockfish. Because S. 

 mystinus maintained relatively high energy demand 

 during El Nino years, despite slower growth rates and 

 lower fecundity, the prey quality and quantity during 

 such events is clearly important. 



Ultimately, these models can be expanded to the pop- 

 ulation level to place rockfish in the context of their 

 communities. This approach can elucidate how factors 

 such as fishing, environmental variability, and recruit- 

 ment variability influence the role of rockfish as preda- 

 tors on specific prey taxa, as has been done in bioener- 

 getics models for other predators (Kitchell et al., 1997; 

 Essington et al., 2002; Schindler et al., 2002). Because 

 energy budgets are influenced by fish size and reproduc- 

 tive state, expanding to the population level will require 

 size- or age-structured population models, such as those 

 used in many rockfish stock assessments (e.g., Pacific 

 Fishery Management Council, 2000). Most Sebastes 

 stock assessments to date are for species that live in 

 shelf or slope habitats, whereas the species whose food 

 habits and basic energetic information are best known 

 are inshore species. Therefore, a key part of producing 

 useful bioenergetics analysis at the population level will 

 be to prioritize populations or species assemblages for 

 which bioenergetics models might be most useful, and to 

 identify which type of information (population structure 

 or basic biology and ecology) is lacking. 



Finally, the generic model parameters in this study re- 

 quired information from several species. Interspecies pa- 

 rameter borrowing has been criticized (Ney, 1993), and 

 the results from such models deserve careful appraisal. 

 The sensitivity analysis demonstrates the importance 

 of this issue: with increasing parameter uncertainty. 



