Testing hypotheses about fecundity, body size and maternal condition in fishes

Recent research suggests that maternal condition positively influences the number of eggs spawned in fishes. These studies commonly choose a priori to use body length rather than weight as an explanatory variable of offspring production, even though weight is usually the better predictor of fecundity. We are concerned that consistent exclusion of body weight as a predictor of egg production inflates the variance in fecundity attributable to maternal condition. By analysing data on three populations of Atlantic cod (Gadus morhua, Gadidae) and 10 populations of brook trout (Salvelinus fontinalis, Salmonidae), we illustrate the need for a statistically defensible method of model selection to distinguish the effects of maternal condition on egg production from the effects of body size alone. Forward stepwise regression and null model analyses reveal how length-based regressions can significantly over-estimate correlations between condition and fecundity, leading us to conclude that the effect of condition on egg productivity may not be as ubiquitous or as biologically important as previously thought. Our work underscores the need for greater statistical clarity in analyses of the effects of maternal condition on reproductive productivity in fishes.