Analysis of neighborhood competition among annual plants: implications of a plant growth model

Abstract

Regression analyses between individual plant growth and indices of local crowding are a popular means to examine the effects of spatially-explicit competition. Low correlations between performance and local crowding are thought to reflect the importance of non-competitive factors in determining plant size. Simulation analyses using an individual-plant, spatially-explicit model of population dynamics suggest that this conclusion is not universally valid. Individual plant growth was prescribed as a function of a growth rate coefficient, growing space, and mass. Regression analysis recovered the known linear relationship between growth and growing space when growth rate coefficients were constant for all plants, but recovered a curvilinear relationship when growth rate coefficients varied randomly among plants. When growing space was approximated by a relative competition index, random variation in growth rate coefficients caused a triangular envelope relationship between growth and neighborhood competition in which there was much variability in growth rates when plants grew without neighbors and little variability in growth rates when plants lost resources to neighbors. The sessile nature of plants ensures that spatially-explicit competition can be important. More attention needs to be given to experiments that document this rather than correlative analyses at a given point in time.