Non-Manipulative Estimates of Competition Coefficients in a Montane Grassland Community

Abstract

1. A method is given for estimating competition coefficients in multispecies grassland communities without manipulation, using the natural turnover of individuals at a fine spatial scale. The method requires a detailed map of the spatial distribution of plants documented at several points in time, and uses non-linear regression of the local density of the plants on their densities in small neighbourhoods at a previous time. 2. The method is tested by generating realizations of two spatiotemporal stochastic processes for which the true parameter values are known. It is shown that non-linear regression successfully recovers the major features of the competition matrices. 3. The method is applied to a montane grassland dominated by four species of grass, for which spatial data on four plots are available for an 11-year period, and for which results of manipulation experiments are also available. 4. The results show that competitive interactions between species are as strong as interactions within species. There are strong asymmetries in the competition coefficients of species pairs, but little sign of species specificity. Intransitivities of the interaction matrix are not evident. The competition coefficients obtained show a good measure of agreement with the results of manipulation experiments that have been carried out on the community. 5. It is argued that non-manipulative methods of the kind described here hold a useful key to understanding interactions in plant communities.