THE DIVERSITY–DISTURBANCE RELATIONSHIP: IS IT GENERALLY STRONG AND PEAKED?

Abstract

The contemporary literature accepts that disturbance strongly influences pat- terns of species diversity, and that the relationship is peaked, with a maximum at inter- mediate levels of disturbance. We tested this hypothesis using a compilation of published species diversity-disturbance relationships that were gleaned from a literature search of papers published from 1985 through 1996 and from references therein. We identified 116 species richness-, 53 diversity-, and 28 evenness-disturbance relationships in the literature, which we grouped according to shape of relationship (nonsignificant, peaked, negative monotonic, positive monotonic, or U-shaped). We tested the relationships between the strength and shapes of these relationships and attributes of the community, disturbance, and sampling and study design. Nonsignificant relationships were the most common, com- prising 35% of richness, 28% of diversity, and 50% of evenness studies. Peaked responses were reported in only 16% of richness, 19% of diversity, and 11% of evenness cases. Explained variation in the three measures of diversity was variable among studies but averaged -50%. It was higher when few samples and few disturbance levels were examined and when organisms within the samples were not exhaustively censused, suggesting that procedural artifact contributes to these relationships. Explained variation was also higher in studies in which disturbance was measured as a gradient of time passed since the last disturbance (mean r2 = 61%), vs. studies of spatial variation in richness (mean r2 = 42%). Peaked richness relationships had the greatest odds of being observed when sampled area and actual evapotranspiration were small, when disturbances were natural rather than an- thropogenic in origin, and when few disturbance levels were examined. Thus, on average, diversity-disturbance relationships do not have consistently high r2 and are not as consis- tently peaked as the contemporary consensus would suggest.