Interspecific relationships between growth and survival are critical determinants of tree species diversity maintenance in forests. The trade-offs between growth and survival in co-occurring tree species are believed to arise along a continuum of life-history strategies. For example, co-occurring species range from those that grow slowly and survive well in resource-poor environments to those that grow quickly but have low survival rates in resource-rich environments. However, uncertainties remain regarding how growth-survival trade-offs are related to species traits, tree sizes, or environmental conditions.

We examined how the relationships between species traits and growth–survival relationships shift in response to changes in stem sizes and across census periods with different climate conditions (frequency of strong winds, drought intensity) across 45 co-occurring tree species based on 23 years of growth and survival records in a warm temperate rain forest on Yakushima Island, Japan. We developed hierarchical Bayesian models of relative growth and survival rates, including leaf traits, wood density, and 95-percentile maximum stem diameter as explanatory variables. We tested the relationships between estimated trait-mediated growth–survival relationships and the intensities of climate events during five census periods.

Each trait’s effects on growth–survival relationships differed across the five census periods in response to climate conditions. Interspecific growth–survival relationships affected by a single trait axis for leaves or wood tended to be negative. In contrast, those affected by the maximum stem diameter tended to be positive. Such trends increased with more frequent strong winds or more intense droughts. The single-trait effects on growth–survival relationships were stronger for smaller sizes than for larger sizes. For all traits combined, we found a significant growth–survival trade-off only for small-sized stems in three of five census periods.

Synthesis: Our results indicate that the effect of species traits on the growth–survival relationships depended on tree sizes, the census periods, or both in response to the frequency or intensity of climate events. We argue the importance of incorporating spatial and temporal variations in environmental conditions into long-term data from tree census to predict forest dynamics.