Implications of vulnerability to hurricane damage for long-term survival of tropical tree species: a Bayesian hierarchical analysis

Tropical forests in the Caribbean are often subject to catastrophic disturbances by hurricanes. Despite the high frequency and intensity of hurricanes in this region, their effect on tropical forest dynamics remain poorly understood. In an effort to better understand the importance of hurricanes to a Puerto Rican tropical forest, we employ Bayesian statistical methods to identify factors that may determine the response of individual trees to hurricane damage. These factors include the effect of tree size and taxonomic identity on the vulnerability of trees to wind damage and the effect of wind damage, tree size, and crowding on individual survival following a hurricane event. In this analysis, we use data from the censuses of the 16-ha Luquillo Forest Dynamics Plot (LFDP) in Puerto Rico that include assessments of damage following Hurricane Hugo (Sept. 1989), status (alive or dead) 2–5 years after Hugo, size, neighborhood crowding, and spatial location for each tree. At the species level, the association between life-history traits and structural attributes generate a positive relationship between shade tolerance and resistance to hurricane damage. We focus our analyses on four relatively common tree species in the LFDP that represent a range of life-history strategies: shade intolerant, early successional species Alchornea latifolia and Casearia arborea, and shade tolerant species Dacryodes excelsa and Manilkara bidentata. At the stand level, spatial variation in storm severity is an important driver of individual tree damage, but its direct effects are difficult to separate from other landscape-level factors that interact with hurricane intensity to affect tree survival. In this study we build a hierarchical, spatially explicit Bayesian model that provides a straightforward method for evaluating species-specific susceptibility to hurricane damage and the implications for survival. We apply the method as a tool to quantify stand-level spatial variability in hurricane intensity, independent of species composition and stand age or size structure.