From the individual to the landscape and back: time‐varying effects of climate and herbivory on tree sapling growth at distribution limits

Abstract

As herbivory can modulate climate‐induced shifts in species distribution, disentangling the relative importance of herbivory and climate on plant growth can help to predict and manage future changes in vegetation, such as those occurring at treeline areas. An individual‐based hierarchical Bayesian time‐series model (individual‐based model; IBM) was developed to estimate the time‐varying impact of climate and herbivory on individual pine sapling height growth in woodland and treeline ecosystems of southern Europe during a 16‐year period. The performance of the IBM was compared to a linear mixed‐effects (LME) model to test for potential inferential effects when individual variability is marginalized. Time‐varying models were also compared to constant parameter approaches. Model fitting and posterior predictive checking suggests a better statistical performance of individual‐scale, time‐varying inference. LME modelling overestimated herbivory effects and underestimated environmental stochastic effects, and model validation indeed suggested severe overfitting in the LME model relative to the IBM strategy. These results reveal a potential failure of common aggregation strategies to correctly resolve the effects of climate and herbivory variability at the individual scale. Moreover, ignoring the time‐varying nature of the effects may preclude the correct estimation of the temporal scale of climate and herbivory impacts. In general, we found stronger individual‐ and time‐averaged effects of summer precipitation relative to the effects of herbivory, particularly at the treelines. Also, individual pine responses showed effects of the same sign more consistently in the case of precipitation. This suggests that precipitation is more pervasive at the population level, while herbivory act as a spatially aggregating force through individual‐level damage. Synthesis. Our results suggest that accounting for individual and temporal variability in ecological inference greatly improves the assessment of the relative importance of climate and herbivory on species distribution shifts. Strong effects of precipitation at the treeline could allow tree upward expansion, although increasing ungulate populations and associated browsing damages might limit positive climatic responses of pine saplings in the near future. In this context, the spatially heterogeneous effect exerted by herbivory could result in diverse vegetation structures in ecotones, adding a new dimension to the predictions on climate‐driven vegetation shifts.