Larval traits show temporally consistent constraints, but are decoupled from postsettlement juvenile growth, in an intertidal fish.

Complex life cycles may evolve to dissociate distinct developmental phases in an organism's lifetime. However, genetic or environmental factors may restrict trait independence across life stages, constraining ontogenetic trajectories. Quantifying covariance across life stages and their temporal variability is fundamental in understanding life-history phenotypes and potential distributions and consequences for selection. We studied developmental constraints in an intertidal fish (Bathygobius cocosensis: Gobiidae) with a discrete pelagic larval phase and benthic juvenile phase. We tested whether traits occurring earlier in life affected those expressed later, and whether larval traits were decoupled from postsettlement juvenile traits. Sampling distinct cohorts from three annual breeding seasons afforded tests of temporally variability in trait covariance. From otoliths (fish ear stones), we measured hatch size, larval duration, pelagic growth (larval traits) and early postsettlement growth (juvenile trait) in 124 juvenile B. cocoensis. We used path analyses to model trait relationships with respect to their chronological expression, comparing models among seasons. We also modelled the effect of season and hatch date on each individual trait to quantify their inherent variability. Our path analyses demonstrated a decoupling of larval traits on juvenile growth. Within the larval phase, longer larval durations resulted in greater pelagic growth, and larger size-at-settlement. There was also evidence that larger hatch size might reduce larval durations, but this effect was only marginally significant. Although pelagic and postsettlement growth were decoupled, pelagic growth had postsettlement consequences: individuals with high pelagic growth were among the largest fish at settlement, and remained among the largest early postsettlement. We observed no evidence that trait relationships varied among breeding seasons, but larval duration differed among breeding seasons, and was shorter for larvae hatching later within each season. Overall, we demonstrate mixed support for the expectation that traits in different life stages are independent. While postsettlement growth was decoupled from larval traits, pelagic development had consequences for the size of newly settled juveniles. Temporal consistency in trait covariances implies that genetic and/or environmental factors influencing them were stable over our three-year study. Our work highlights the importance of individual developmental experiences and temporal variability in understanding population distributions of life-history traits.