COMPLEX DYNAMICS IN EXPERIMENTAL POPULATIONS OF AN ANNUAL PLANT, CARDAMINE PENSYLVANICA

To study the numerical dynamics of plant populations, 12 experimental populations of an annual greenhouse weed, Cardamine pensylvanica, were maintained for 15 generations in controlled-environment growth chambe rs by growing plants in an array of pots and allowing seed for the nex t generation to disperse into an adjacent array of fresh pots. Discret e generations were enforced by harvesting mature plants after seed dis persal, but germination, recruitment, competition, and dispersal occur red naturally. The numerical dynamics of the experimental populations cycled from high to low density with a period of four to five generati ons, as indicated by negative autocorrelations in population size at l ags of two and three generations. Demographic data collected during th e experiment indicate that population density affected plant growth an d seed set. Independent estimates of low-density recruitment were also high enough to predict complex population dynamics from simple models of direct density-dependent population regulation. However, simple po pulation models fit to the time series data predicted stable dynamics. Similar models including time-lagged density dependence qualitatively reproduced the dynamics of the experimental populations. Delayed feed back through maternal effects or interacting herbivores or pathogens m ay be possible causes of the observed dynamics. This suggests that alt hough plant population dynamics may be stabilized by direct density de pendence, delayed density dependence could destabilize dynamics.