Growth history and crown vine coverage are principal factors influencing growth and mortality rates of big-leaf mahogany Swietenia macrophylla in Brazil.
Current efforts to model population dynamics of high-value tropical timber species largely assume that individual growth history is unimportant to population dynamics, yet growth autocorrelation is known to adversely affect model predictions. In this study, we analyse a decade of annual census data from a natural population of big-leaf mahogany Swietenia macrophylla King to quantify the strength and duration of growth autocorrelation, and experimentally investigate the role of crown vine coverage, a major predictor of performance. The sample population consisted of 358 trees >10 cm diameter. The relative contributions of predictor variables including diameter, crown vine coverage and growth history to models of growth and mortality were evaluated using Akaike's Information Criterion. Autocorrelation among trees was incorporated into growth models using generalized least squares. We experimentally removed vines from heavy-laden trees to test the strength and persistence of their impact on stem diameter growth. Previous growth explained the highest amount of variation in annual diameter growth; the best-fitting model of autocorrelation was an AR(7) autoregressive model, indicating that autocorrelation persisted throughout the study period. Other factors explaining variation in growth were, in decreasing order of importance: year of measurement, crown vine coverage, diameter, crown illumination and fruit production. The best logistic regression model for mortality retained diameter growth plus crown vine coverage as predictors of risk. The vine release experiment strongly supported these results. Released trees grew faster than control trees but required ≥5 years to approach growth rates of trees with minimal vine coverage. Trees with heavy vine coverage also experienced higher mortality rates. Synthesis and applications. These results indicate that growth autocorrelation is strong, persistent, and an important predictor of future performance; demographic models for tree growth and yield projections can be improved by accounting for growth history. Our results also indicate that targeted silvicultural practices such as vine cutting can increase long-term growth and timber yield. These findings further current understanding of tropical tree growth and survival, and offer improved management tools for sustainable harvest practices for mahogany and similar species.