Abstract
Our goal was to elucidate the population dynamics of the perennial understory herb Calathea ovandensis in a rain forest in southern Mexico using matrix projection model analysis. We emphasize the magnitude and consequences of spatiotemporal variation in (1) basic demographic parameters (growth, survival, and reproduction) (2) asymptotic demographic properties of a given environment (the asymptotic population growth rate and the associated stable—stage distribution and reproductive values) and (3) demographic sensitivities associated with a given environment (sensitivity and elasticity). We obtained 6 yr (1982—1987) of empirical data from four study plots (differing in substrate, light, and density) from which we used the first 5 yr (1982—1986) to construct 16 plot—year and 1 pooled population projection matrices. This stage—structured population was characterized by a long—lived seed bank, temporally variable seedling recruitment (10—fold variability among years), high mortality of seedlings (>90%), very low mortality of reproductives (usually <10%), fertility that increased markedly with plant size, and the ability of large plants to shrink rather than die under adversity. Within these broad outlines, the magnitudes of transitions representing demographic fates exhibited considerable variation through space and time, some parameters varying much more than others (CV from 22 to 400%). Growth and reproduction were positively correlated across environments. The least variable parameters were seed dormancy and stasis of small reproductives. Observed stage distributions were reasonably close to the stable stage distributions (mean = 86.1% similar). In most plot—years, the stable—stage distribution was dominated by seeds, followed by seedlings, and then small reproductives and the reproductive values increased with size class. Population growth rates, given by the dominant eigenvalue of the matrices, ranged from 0.73 to 1.25. Analysis of the mean dynamics gave λ = 0.97 (using a variety of analytical approaches) and our analysis of the overall pooled dynamics gave a λ = 0.99, indicating that the habitat at the study site favored the persistence of Calathea ovandensis. An el Nino even coincided with the year of the highest population growth rate. Survival, growth, and reproduction varied significantly through space and time, and different plot—years were beneficial to different stages. Most interestingly, stage—specific sensitivity parameters (sensitivity and elasticity) also varied through space and time. Spatiotemporal variability of sensitivity structure has important implications. Determination of stages most "critical" to population dynamics will depend upon knowledge of this variation. Population growth rate was significantly positively correlated with elasticity of seed production, seed germination, and seedling growth. These results indicate that the opportunity for selection on plant characters affecting particular life history stages varies through space and time even if the effect on the single—stage transition probability does not vary. Selection on characters affecting juvenile stages may be stronger in populations of higher growth rates.
