HURRICANE DISTURBANCE AND THE POPULATION-DYNAMICS OF A TROPICAL UNDERSTORY SHRUB - MEGAMATRIX ELASTICITY ANALYSIS

The role of environmental variation in the population dynamics of the tropical understory shrub Ardisia escallonioides was explored using si ngle-environment matrix analysis and dynamic-environment megamatrix an alysis. We investigated how variation in forest-canopy openness caused by hurricanes affected population dynamics. A 7 x 7 patch-transition matrix was created to model forest dynamics based on the historical fr equency and strength of hurricanes in south Florida. Seven 8 x 8 stage -classified matrices for plant population dynamics were derived from f our field populations and two posthurricane years. These populations s panned a strong gradient of forest-canopy openness resulting from hurr icane disturbance caused by Hurricane Andrew (24 August 1992). The pat ch-transition matrix was analyzed for its stable patch distribution. T he seven single-environment matrices were each analyzed for population growth rates and elasticity of stage classes. The 56 x 56 megamatrix was analyzed for population growth rate, reproductive value, stable pa tch-stage distribution, and elasticity of both patches and stages. Pop ulation growth rate of the single-environment matrix models (8 x 8) wa s lowest for the shadiest environment (lambda = 1.0), intermediate but variable for intermediate environments, and highest for the most open environment (lambda = 1.96). The population growth rate (lambda) of t he megamatrix model was 1.71. The stable patch-stage distribution was characterized by a high proportion of seeds in open patches. Reproduct ive value was highest for very large adults in open patches. Megamatri x elasticity analysis revealed several components of the demography th at were not apparent from examination of the single-environment matric es. Although closed-canopy patches were the most common type of enviro nment, elasticity analysis of the megamatrix showed that the highest p atch elasticity was for the most open environment. We found contrastin g predictions for the relative importance of life history stages to po pulation growth rate from single-environment analysis vs. the megamatr ix analysis. In the most common single-environment patch, very large a dults had the highest elasticity. In the megamatrix, small juveniles c ontributed the most elasticity and all stages contributed some elastic ity. For plant species inhabiting temporally varying environments, meg amatrix analysis offers new insight into the importance of both patche s and stage classes.