A NULL METHOD FOR NEIGHBORHOOD MODELS OF PLANT COMPETITIVE INTERACTIONS

Neighborhood (spatial) models of plant populations have assumed implicitly that competition between adjacent individuals is the primary determinant of observed dynamics. In contrast, non-spatial models now routinely test this assumption against the null hypothesis that individual differences in intrinsic growth rate alone can explain observed patterns. In a widely-used neighborhood model (Pacala and Silander 1990), competition is modelled as a direct effect of neighbors on target plant biomass, which in turn affects plant fecundity. We have developed a statistical null model for Pacala and Silander's (1990) neighborhood model for plant monocultures that explicitly incorporates variation in plant growth rate, and permits a test of the hypothesis that neighborhood competition controls plant ground area occupied (= rosette diameter) and biomass. An additional alternative hypothesis to our model is that neighbors can affect fecundity independently of their effects on plant diameter and biomass. We tested these three hypotheses - (1) no effect of competition (null); (2) neighborhood competition directly affecting plant diameter and biomass, and indirectly affecting fecundity (Pacala and Silander 1990); (3) neighborhood competition directly affecting plant diameter, biomass, and fecundity - by replacing, with appropriate controls, Silander and Pacala's (1985) experiments with the rosette-forming crucifer, Arabidopsis thaliana. Neighborhood competition in dense stands resulted in simultaneous changes in plant diameter, biomass, and fecundity relative to plants grown in the absence of competition. There also were apparent effects of neighbors on plant fecundity independent of their effects on plant size alone. The null model was not supported, but Pacala and Silander's model did not fit the data as well as the model that incorporated direct effects of neighbors on plant fecundity.