Understanding or predicting the responses of natural populations to climate is an urgent task for ecologists. However, studies linking the temporal dynamics of populations to climate remain limited. In population ecology, studies typically assume that populations respond only to the climate of the most recent growing season or year. However, evidence also exists that certain organisms respond to the climate in a lagged-response fashion. Antecedent effect models are a set of statistical tools that help researchers select the past climatic factors that best predict a response (e.g., survival, reproduction). These models use the evidence provided by the data to select, for example, specific months, or time windows correlated with a response. Thus, antecedent effect models could improve our ability to perform exploratory analysis and to predict the effects of temporal climatic variation on population dynamics. Here, we compare the predictive performance of antecedent effect models against simpler models. We fit three antecedent effect models: (1) weighted mean models (WMM), which weigh the importance of monthly anomalies based on a Gaussian curve, (2) stochastic antecedent models (SAM), which weigh the importance of monthly anomalies using a Dirichlet process, and (3) regularized regressions using the Finnish Horseshoe prior (FHM), which estimate a separate effect size for each monthly anomaly. We compare these approaches to a linear model using a yearly climatic predictor, and a null model with no predictors. We use high resolution demographic data from 34 plant species ranging between 7 and 36 years of length, which reflect the typical temporal grain employed to study the effects of climate on populations. We fit models to the overall asymptotic population growth rate (λ), and to its underlying vital rates (survival, development, and reproduction) to examine the extent to which potential lagged effects occur. Antecedent effect models do not consistently outperform the linear or null models. Average differences in performance are small among models, response variables, and across sample sizes. Surprisingly, ranked in order of decreasing predictive performance, the best approaches are: null model > linear model > FHM > SAM > WMMs. Development is on average easier to predict than survival and reproduction, and these have higher predictive power than the overall rate of growth of the population, λ. Synthesis: In temporal datasets with limited sample size, antecedent effect models are better suited as exploratory tools. These models can help identify which single monthly anomalies drive population dynamics, while controlling the effect of noise on estimates. As such, antecedent effect models can be a useful tool for hypothesis generation, and to increase our understanding of the links between climate and plant population dynamics.
