Quadratic variation in old-field species richness along gradients of disturbance and nitrogen

We followed species richness, colonizations, extirpations, and turnover for eight years in 16 combinations of disturbance and nitrogen addition applied to a 30-yr-old Minnesota field. Four levels each of disturbance and nitrogen were applied in a completely factorial design, producing 16 treatments. There were four replicates of most treatments and 14 replicates of the four extreme treatments (lowest and highest disturbance and nitrogen). After 5-6 yr, species richness varied quadratically and significantly with disturbance in plots receiving no nitrogen or the lowest level of added nitrogen. Richness increased over time at intermediate disturbance as annuals colonized plots otherwise dominated by perennials. The number of growth forms also varied quadratically with richness, whereas the number of species belonging to each growth form did not, suggesting that high richness was due to coexistence of species with different life histories. The strongest quadratic variation in species richness was associated with significant quadratic variation in colonization with increasing disturbance, and no significant variation in extirpation, suggesting that increased richness at intermediate disturbance was attributable to colonization at the middle of the disturbance gradient, and not to extirpation at either end of the gradient. Richness decreased with increasing nitrogen at all levels of disturbance. Colonization also decreased and extirpation increased with increasing nitrogen availability, suggesting that decreases in richness with increasing nitrogen reflect accelerated extirpation, and not colonization of nitrogen-poor soils. Simultaneous competition experiments showed colonizations in disturbed plots to be related to reduced root competition, and extirpations in fertilized plots to be related to increased shoot competition. Surprisingly, untreated control plots had higher rates of colonization, extirpation, and species turnover than disturbed or fertilized plots. Turnover in control plots was caused mostly by uncommon species, and control plots were always dominated by the perennial grass Schizachyrium scoparium. In contrast, the dominant species of the extreme environmental treatments varied over time in spite of their low levels of richness. The array of responses generated by our factorial experiment helps explain why richness varies with disturbance and fertility in different ways along different natural gradients: patterns and dynamics of richness clearly depend on location within the fertility - Disturbance matrix.