A structural equation model analysis of postfire plant diversity in California shrublands.

This study investigates patterns of plant diversity following wildfires in fire-prone shrublands of California, seeks to understand those patterns in terms of both local and landscape factors, and considers the implications for fire management. Ninety study sites were established following extensive wildfires in 1993, and 1000-m(2) plots were used to sample a variety of parameters. Data on community responses were collected for five years following fire. Structural equation modeling (SEM) was used to relate plant species richness to plant abundance, fire severity, abiotic conditions, within-plot heterogeneity, stand age, and position in the landscape. Temporal dynamics of average richness response was also modeled. Richness was highest in the first year following fire, indicating postfire enhancement of diversity. A general decline in richness over time was detected, with year-to-year variation attributable to annual variations in precipitation. Peak richness in the landscape was found where (1) plant abundance was moderately high, (2) within-plot heterogeneity was high, (3) soils were moderately low in nitrogen, high in sand content, and with high rock cover, (4) fire severity was low, and (5) stands were young prior to fire. Many of these characteristics were correlated with position in the landscape and associated conditions. We infer from the SEM results that postfire richness in this system is strongly influenced by local conditions and that these conditions are, in turn, predictably related to landscape-level conditions. For example, we observed that older stands of shrubs were characterized by more severe fires, which were associated with a low recovery of plant cover and low richness. These results may have implications for the use of prescribed fire in this system if these findings extrapolate to prescribed burns as we would expect.