Coupling biogeochemical tracers with fish growth reveals physiological and environmental controls on otolith chemistry

Biogeochemical tracers found in the hard parts of organisms are frequently used to answer key ecological questions by linking the organism with the environment. However, the biogeochemical relationship between the environment and the biogenic structure becomes less predictable in higher organisms as physiological processes become more complex. Here, we use the simultaneous combination of biogeochemical tracers and fish growth analyzed with a novel modeling framework to describe physiological and environmental controls on otolith chemistry in an upwelling zone. First, we develop increasingly complex univariate mixed models to describe and partition intrinsic (age effects) and extrinsic (environmental parameters) factors influencing fish growth and otolith element concentrations through time. Second, we use a multivariate mixed model to investigate the directionality and strength between element-to-element and growth relationships and test hypotheses regarding physiological and environmental controls on element assimilation in otoliths. We apply these models to continuous element (Na, Sr, Mg, Ba, Li) and growth increment profiles (monthly resolution over 17yr) derived from otoliths of reef ocean perch (Helicolenus percoides), a wild-caught, site-attached, fully marine fish. With a conceptual model, we hypothesize that otolith traits (elements and growth) driven by environmental conditions will correlate both within an otolith, reflecting the time dependency of growth and element assimilation, and among individuals that experience a similar set of external conditions. We found some elements (Sr:Ca and Na:Ca) are mainly controlled by physiological processes, while other elements (Ba:Ca and Li:Ca) are more environmentally influenced. Within an individual fish, the strength and direction of correlation varies among otolith traits, particularly those under environmental control. Correlations among physiologically regulated elements tend to be stronger than those primarily controlled by environmental drivers. Surprisingly, only Ba:Ca and growth are significantly correlated among individuals. Failure to appropriately account for intrinsic effects (e.g., age) led to inflated estimates of among individual correlations and a depression of within individual correlations. Together, the lack of among-individual correlations of otolith traits in properly formulated models and the biases that can be introduced by not including appropriate intrinsic covariates suggest that caution is needed when assuming multi-elemental signatures are reflective solely of shared environments.