Reconciling larval and adult sampling methods to model growth across life-stages.

Individual growth rates are intrinsically related to survival and lifetime reproductive success and hence, are key determinants of population growth. Efforts to quantify age-size relationships are hampered by difficulties in aging individuals in wild populations. In addition, species with complex life-histories often show distinct shifts in growth that cannot be readily accommodated by traditional modelling techniques. Amphibians are often characterized by rapid larval growth, cessation of growth prior to metamorphosis, and resumption of growth in the adult stage. Compounding issues of non-linear growth, amphibian monitoring programs typically sample larval and adult populations using dissimilar methods. Here we present the first multistage growth model that combines disparate data collected across life-history stages. We model the growth of the endangered Reticulated Flatwoods Salamander, Ambystoma bishopi, in a Bayesian framework, that accounts for unknown ages, individual heterogeneity, and reconciles dip-net and drift fence sampling designs. Flatwoods salamanders achieve 60% of growth in the first 3 months of life but can survive for up to 13 years as a terrestrial adult. We find evidence for marked variability in growth rate, the timing and age at metamorphosis, and maximum size, within populations. Average size of metamorphs in a given year appeared strongly dependent on hydroperiod, and differed by >10mm across years with successful recruitment. In contrast, variation in the sizes of emerging metamorphs appeared relatively constant across years. An understanding of growth will contribute to the development of population viability analyses for flatwoods salamanders, will guide management actions, and will ultimately aid the recovery of the species. Our model formulation has broad applicability to amphibians, and likely any stage-structured organism in which homogenous data cannot be collected across life-stages. The tendency to ignore stage-structure or omit non-conforming data in growth analyses can no longer be afforded given the high stakes of management decisions, particularly for endangered or at-risk populations.