How natural habitat patchiness affects the distribution of diversity in Californian serpentine chaparral

Ecology, Sept, 1997 by Susan Harrison

INTRODUCTION

Patterns of biological diversity on serpentine soils have fascinated ecologists and evolutionary biologists for decades (reviews in Proctor and Woodell 1975, Kruckeberg 1984, Brooks 1987, Baker et al. 1992, Roberts and Proctor 1992). Serpentine or ultramafic soils, which occur patchily in zones of faulting and mountain uplift, are high in iron and magnesium and low in calcium. These harsh soils exclude most plants found on surrounding nonserpentine soils, but support a high proportion of endemic and regional indicator species (respectively, those globally or regionally restricted to serpentine). California has one of the temperate zone's richest serpentine floras, with 215 taxa (10% of all Californian endemics) restricted to serpentine soils (Brooks 1987, Skinner and Pavlik 1994). Biotic communities on serpentine have contributed greatly to our knowledge of speciation (e.g., Stebbins 1942, Kruckeberg 1954, 1967, Wild and Bradshaw 1977), microevolution (e.g., Kruckeberg 1967, Proctor and Wooddell 1975), and biogeography (e.g., Whittaker 1954, 1960).

Patchiness and biological distinctiveness also make serpentine outcrops suitable for testing current ecological theories about patchy populations (Hastings and Harrison 1994, Hanski and Gilpin 1996) and communities (e.g., Caswell and Cohen 1991, 1993, Nee and May 1992, Tilman et al. 1994). The spatial and ecological aspects of serpentine have been much less explored than the evolutionary ones, but Harrison et al. 1988 (Harrison 1989) documented local extinction and recolonization in a butterfly restricted to serpentine, and Murphy and Ehrlich (1989) measured species-area relationships on serpentine patches. Resource partitioning, disturbance dynamics, and other causes of local (alpha) diversity have been studied intensively in a Californian serpentine grassland (Huenneke et al. 1990, Hobbs and Mooney 1991, McCarten 1992, Wu and Levin 1994, Moloney and Levin 1996).

The first analysis of the local vs. regional components of diversity was a study of serpentine plants by Whittaker (1954, 1960), who defined beta, or differentiation diversity, as gamma (regional) divided by average alpha (local) diversity. His pioneering study demonstrated that the flora of serpentine soils had lower alpha diversity, but higher beta and gamma diversity, than the flora of granitic soils in the Siskiyou Mountains (California and Oregon). Higher beta and gamma diversity arose because the community on serpentine changed more rapidly along an elevational gradient (Whittaker 1960). Later authors pointed out that beta diversity actually has two components: spatial turnover of species along environmental gradients, as studied by Whittaker (1960), and spatial turnover due to the existence of different species in similar but separated habitats (called ecological equivalency by Shmida and Wilson 1985, and gamma diversity by Cody 1993). Recent metacommunity theory has stressed the importance of the second mechanism, showing that extinction and colonization among patches of equivalent habitat can generate high beta diversity within a region (Caswell and Cohen 1991, 1993).

Several authors have since tried to distinguish the role of spatial isolation from that of environmental gradients in producing beta diversity. The usual approach is to regress measures of floral or faunal dissimilarity among sites on first, the distance between sites, and second, the dissimilarity of sites in climate and/or vegetation (e.g., Shmida and Wilson 1985, Cody 1986, 1993, Harrison et al. 1992, Westoby 1993). However, this approach has the obvious weakness that sites may differ in some unmeasured way that is correlated with distance. A stronger approach that has been suggested (Cody 1993), but never used, is to compare beta diversity among similarly spaced patchy vs. continuous sites within a single type of habitat. The spatial structure of serpentine, with its very small to very large patches [ILLUSTRATION FOR FIGURE 1 OMITTED], makes it ideal for such an analysis. Comparisons of beta diversity among patchy vs. continuous sites at several spatial scales could reveal the scale at which discontinuity per se makes an important contribution to regional diversity.

This question has considerable importance beyond the study of serpentine, since habitat fragmentation is a central issue in conservation biology (reviews in Saunders et al. 1991, Groom and Schumaker 1993). Although the effects of fragmentation on communities are usually regarded as negative, metacommunity theory suggests the possibility of positive effects in some cases, especially in communities structured by strong competitive interactions. Examining this question in a naturally patchy habitat is of particular value, since diversity in recently fragmented habitats may be well above its eventual equilibrium (Tilman et al. 1994), and thus give misleading answers.

Here I conducted a spatially structured analysis of community diversity, asking whether within a three-county region, the components of diversity in the woody plant community were different among sets of small (0.5-3 ha) serpentine outcrops than among sites within large ([greater than]5 [km.sup.2]), continuous areas of serpentine. I also asked whether the woody plant community of serpentine had higher beta diversity than that of the adjacent nonserpentine soils. Each of these comparisons is performed at two spatial scales, within clusters of adjacent sites (50-3200 m apart) and among these clusters (16-45 km apart). Analyses are performed using two metrics: Whittaker's (1960) original beta, and Colwell and Coddington's (1994) "complementarity," or proportion of unshared species; advantages and disadvantages of these and other dissimilarity metrics are discussed by Shmida and Wilson (1985) and Colwell and Coddington (1994).