What predicts the richness of seeder and resprouter species in fire-prone Cape fynbos: Rainfall reliability or vegetation density?

In ecosystems subject to regular canopy fires, woody species have evolved two general strategies of post-fire regeneration. Seeder species are killed by fire and populations regenerate solely by post-fire recruitment from a seed bank. Resprouter species survive fire and regenerate by vegetative regrowth from protected organs. Interestingly, the abundance of these strategies varies along environmental gradients and across regions. Two main hypotheses have been proposed to explain this spatial variation: the gap dependence and the environmental-variability hypotheses. The gap-dependence model predicts that seeders are favoured in sparse vegetation (vegetation gaps allowing effective post-fire recruitment of seedlings), while resprouters are favoured in densely vegetated sites (seedlings being outcompeted by the rapid crown regrowth of resprouters). The environmental-variability model predicts that seeders would prevail in reliable rainfall areas, whereas resprouters would be favoured in areas under highly variable rainfall that are prone to severe dry events (leading to high post-fire seedling mortality). We tested these two models using distribution data, captured at the scale of quarter-degree cells, for seeder and resprouter species of two speciose shrub genera (Aspalathus and Erica) common in fire-prone fynbos ecosystems of the mediterranean-climate part of the Cape Floristic Region. Contrary to the predictions of the gap-dependence model, species number of both resprouters and seeders increased with values of the Normalized Difference Vegetation Index (a widely used surrogate for vegetation density), with a more marked increase for seeders. The predictions of the environmental-variability hypothesis, by contrast, were not refuted by this study. Seeder and resprouter species of both genera showed highest richness in environments with high rainfall reliability. However, with decreasing reliability, seeder numbers dropped more quickly than those of resprouters. We conclude that the environmental-variability model is better able to explain the abundance of woody seeder and resprouter species in Southern Hemisphere fire-prone shrublands (fynbos and kwongan) than the gap-dependence model.