PLANT COMPETITION ALONG SOIL-MOISTURE GRADIENTS - A FIELD EXPERIMENT WITH THE DESERT ANNUAL STIPA-CAPENSIS

1 The interaction between natural and experimental gradients of produc tivity on competition intensity was tested by neighbour removal experi ments. Water is the main limiting factor in this system and experiment al gradients of productivity were obtained using a series of watering treatments. 2 Competition was determined as the total effect of neighb ours on per-capita seed production of Stipa capensis. Both absolute an d relative competition intensity were calculated for three types of ha bitats, during two successive years, and under different watering trea tments in each habitat-year combination. 3 Both measures were positive ly correlated with productivity, but absolute competition intensity wa s more sensitive to changes in productivity than relative competition intensity. Natural gradients of productivity appeared to have stronger effects on competition intensity than experimental gradients, but thi s was largely due to their wider range. In those cases where the range s of the two types of gradients were similar, experimental gradients h ad a stronger effect on competition intensity than natural gradients. 4 Patterns of spatial variation in competition intensity were correlat ed with standing crop under all watering conditions. However, slopes o f the regression equations obtained for the various watering treatment s were not homogeneous. This indicates that per-gram effects of standi ng crop on competition intensity may fluctuate from year to year, depe nding on rainfall conditions. Regression models constructed to test th e relationships between standing crop and competition intensity-over d ifferent years, habitat types and watering treatments accounted for 88 % of the variation in absolute competition intensity and 83% of the va riation in relative competition intensity. 5 The overall results of th is study are consistent with the hypothesis that plant competition inc reases along productivity gradients. The results also indicate that pa tterns of variation in competition intensity along productivity gradie nts may be influenced by the type of the gradient along which competit ion is measured (natural vs. experimental), its range, and the way com petition intensity is defined.