Maximum penalized likelihood estimation in semiparametric mark-recapture-recovery models

摘要: We discuss the semiparametric modeling of mark-recapture-recovery data where temporal and/or individual variation model parameters is explained via covariates. Typically, in such analyses a fixed (or mixed) effects parametric specified for relationship between and covariates interest. In this paper, we use penalized splines, to allow considerably more flexible functional forms. Corresponding models can be fitted numerical maximum likelihood estimation, employing cross-validation choose smoothing data-driven way. Our contribution builds on extends existing literature, providing unified inferential framework open populations, interest typically lies estimation survival probabilities. The approach applied two real datasets, corresponding gray herons (Ardea cinerea), probability as function environmental condition (a time-varying global covariate), Soay sheep (Ovis aries), weight individual-specific covariate). proposed compared standard (logistic) regression new interesting underlying dynamics are observed both cases.