Local and regional diversity in a patchy landscape: native, alien, and endemic herbs on serpentine
Susan Harrison
INTRODUCTION
As once-continuous habitats continue to be fragmented, and naturally patchy habitats continue to be lost, it is essential to examine the effects of habitat discontinuity on the diversity of entire communities. Metacommunity models are a branch of ecological theory concerned with communities assembled through local extinction and recolonization on patches of habitat (Case 1991, Caswell and Cohen 1991, 1993, Tilman et al. 1994, Holt 1997). Many of these models build on longstanding ecological theory showing that pairs of competitors, or predators and prey, may coexist more readily in patchy than continuous habitat (reviewed in Harrison and Taylor 1997, Nee et al. 1997). Extending this idea of patchy coexistence to the community level, Caswell and Cohen (1991) showed that local extinction, recolonization, and competition in a patchy habitat can lead to high total species richness (gamma diversity, Whittaker 1960; here termed regional diversity), by promoting high variation in species composition among patches (beta diversity, Whittaker 1960; here termed differentiation diversity). However, Tilman et al. (1994) demonstrated that excessive habitat fragmentation can lead to lower species richness at both the local (alpha diversity, Whittaker 1960) and regional levels; in particular, the best competitors are lost from fragments, since these species are assumed to be the poorest dispersers.
Recently, Holt (1997) added a new dimension to metacommunity models by considering patches in a habitable matrix, and by allowing species to be either specialists or generalists upon these two (or more) habitat types. His model also did not assume competition to be a major force in the metacommunity. Rather, spatial structure in the metacommunity arises through the individual species’ different habitat breadths, colonization rates, and persistence abilities. One prediction from this model is that rarer patch types will support fewer habitat specialist species than will commoner types. Another prediction is that the rarer the habitat, the more the local community structure will be influenced by the spillover of generalist species from other habitats. This model may be an important step in the direction of realism for many patchy communities. However, there have been few attempts to compare metacommunity models with data, especially at a landscape scale.
Landscapes composed of a mixture of serpentine and nonserpentine soils provide promising settings for analyzing relationships between large-scale habitat patchiness and the components of species richness. Outcrops of serpentine or ultramafic soil are found in areas of ancient tectonic activity, such as along California’s major fault zones; they may range in size from a few square meters to many square kilometers. Serpentines support a distinctive flora because of an extremely high [Mg.sup.++]: [Ca.sup.++] ratio, which excludes many or most species from the surrounding community, and which has led to the evolution of many soil endemics (Kruckeberg 1984, Brooks 1987). Serpentine endemics constitute 10% of Californian endemic plants, and are well represented among listed sensitive or rare taxa because of their typically narrow geographic distributions (Skinner and Pavlik 1994). The flora of serpentine soils is particularly rich in the Northern Coast Ranges of California, where at least 90-100 species are endemics (Kruckeberg 1984). Environmental gradients in the diversity of plants on serpentine in this region have been studied by Whittaker (1960) and Wilson (1988). Serpentines also appear to be important refugia for many nonendemic native species, since they are relatively little invaded by the mediterranean species that now dominate Californian grasslands (Huenneke et al. 1990).
In this study I compare the species richness (henceforth “diversity”) of herbaceous plants in two settings: 24 small (mostly [less than]1 ha) and isolated ([greater than]1 km from large outcrops) patches of serpentine, and 24 similarly spaced and identically sampled sites within very large ([greater than]5 [km.sup.2]) serpentine areas. By comparing patchy and continuous sites within a single type of habitat and community, I can analyze the effects of patchiness per se with respect to the predictions of metacommunity theory, while avoiding the confounding effects that would arise in comparing two different communities with different evolutionary histories. For herbaceous plants on patchy vs. continuous serpentine, I compare Whittaker’s (1960) components of diversity: local (alpha), regional (gamma), and among-site differentiation (beta diversity, here measured by a metric proposed by Colwell and Coddington 1994).
In an earlier study, I performed a similar comparison for woody species on the same sites (Harrison 1997). For the woody species typical of serpentine, i.e., endemics plus “tolerators” of serpentine, regional diversity was similar on sets of patchy and continuous sites, but patches had significantly lower local diversity and higher among-site differentiation than continuous sites. Herbaceous species may be different in several important respects, however. Herbs have much faster population dynamics, and possibly narrower habitat breadths, than woody species. A smaller proportion of herbs than woody plants is endemic to serpentine, and, unlike woody endemics, all herbaceous endemics are sparse (none is a major space-holder in the community). Herbs also include a substantial component of nonnative species.
In the present study, I compare the components of diversity in three nested sets of the herbaceous flora on serpentine: (1) all herb species, (2) herbs that are native vs. alien to the study region, and (3) serpentine-endemic herbs, i.e., those that are strictly confined to serpentine within the study region. Also, I measure soil chemistry and other environmental variables to ascertain their relationships to patterns of diversity. It might be expected that endemics and nonendemics, and native vs. alien species, will respond differently to patchiness and other forms of environmental variation.
METHODS
The study was conducted in Lake, Napa, and Sonoma Counties, California, USA [ILLUSTRATION FOR FIGURE 1 OMITTED]. Geology and soils of this area have been mapped by Fox et al. (1973), Lambert and Kashiwagi (1978), and Wagner and Bortugno (1982). Serpentines here are part of the Coast Range Ophiolite, of Jurassic age. Flora and vegetation of this area are described by Barbour and Major (1988), Kruckeberg (1984), and Callizo (1992). Most serpentine soils in this region are poorly developed, and seldom support grassland. Instead they support leather oak chaparral (Sawyer and Keeler-Wolf 1995), dominated by four endemics (leather oak [Quercus durata, Fagaceae], whiteleaf manzanita [Arctostaphylos viscida, Ericaceae], musk brush [Ceanothus jepsoni var. albi-florus, Rhamnaceae], and silktassel bush [Garrya congdonii, Garryaceae]), and four nonendemics (gray or digger pine [Pinus sabiniana, Pinaceae], chamise [Adenostoma fasciculatum, Rosaceae], toyon [Heteromeles arbutifolia, Rosaceae], and bay laurel [Umbellularia californica, Lauraceae]). Herbs are sparse, and occur mainly on rocky outcrops and scree slopes interspersed with the chaparral. Nonserpentine soils in this region are mainly derived from sandstone and other sedimentary rocks of the Franciscan formation (Wagner and Bortugno 1982, see Plate 1). To the eastern (inland) side of the study area, the vegetation is predominantly blue oak (Quercus douglassii) woodland; toward the west, coastal mixed evergreen forest with live oak (Quercus agrifolia), Douglas-fir (Pseudotsuga menziesii, Pinaceae), madrone (Arbutus menziesii, Ericaceae), bay laurel, and other species (Sawyer and Keeler-Wolf 1995).
Most large tracts of serpentine in this region are publicly owned (U.S. Bureau of Land Management), but often lack roads and trails, while small outcrops of serpentine are embedded in private land. Using geologic maps, I located 24 small serpentine outcrops (“patches”) to which I was able to gain access. All patches were 0.5-3 ha in area, and isolated by [greater than]1 km from any large area of serpentine ([greater than]1 [km.sup.2]). The patches chosen were found in four clusters of five to seven patches, within which the patches were separated by 10-3200 m of nonserpentine habitat; distances among the four clusters ranged from 16 to 45 km. Large areas of serpentine were also selected on the basis of accessibility, although this constraint was less severe than in the case of small patches. The four large outcrops chosen were [approximately]6, 16, 30, and 55 [km.sup.2]. Each large outcrop was matched with one of the four clusters of patches, and within the large outcrop, a set of sampling sites was chosen with the same spatial configuration as that cluster of patches [ILLUSTRATION FOR FIGURE 2 OMITTED]. (Note that, since the purpose of this matching was simply to equalize the spacing of sampling sites, the resulting data were not analyzed pairwise.) The placement of sampling sites was done using topographic maps, without prior information on the vegetation. In choosing where to place the sampling sites, the criteria were: (1) to match the configuration of the small patches, (2) to avoid edges by [greater than]100 m, and (3) to minimize the distance to the sites from roads and trails. (However, the actual sampling was performed [greater than]50 m from roads, trails, or other human-disturbed areas.)
These sampling sites are henceforth referred to as the patchy and continuous “treatments” (N = 24 each). “Sites” refers to the individual sampling locations, and “clusters” to the groupings of sites; for the patchy treatment one cluster is a set of patches, and for the continuous treatment one cluster is all the sampling sites within one large serpentine outcrop [ILLUSTRATION FOR FIGURE 2 OMITTED]. The herb community was sampled at all sites during 2-25 April, 22-28 May, and 25-29 June 1996, and 26-31 March and 1-4 May 1997. (The flowering season for spring annuals was earlier and shorter in 1997 than in 1996; personal observation.) At each site, a central location was identified using topographic maps and a Global Positioning System. From this central location, three random compass directions were chosen, using the second hand of a watch, with the constraint that the directions be [greater than or equal to]90 [degrees] apart. In each direction, I performed one 5 x 50 m belt transect within which I recorded the presence or absence of all flowering herbaceous species; this follows the sampling technique advocated by the California Native Plant Society (Sawyer and Keeler-Wolf 1995). Plants were identified with the help of an expert regional botanist (Joseph Callizo, Napa Land Trust and California Native Plant Society); nomenclature follows Hickman (1993).
On the first sampling round (2-25 April 1996), I collected one soil sample from each site. Samples were gathered by scraping away the top 1 cm, collecting [approximately]100 mg of rock-free soil from the central location and from the end of each transect, and mixing these four subsamples for each site. These soils were oven-dried, ground, and analyzed by the Division of Agriculture and Natural Resources Analytic Laboratory (University of California, Davis). Analyses included exchangeable [Mg.sup.++] and [Ca.sup.++], using ammonium acetate extraction and atomic absorption/emission spectrometry; pH, using saturated paste and a pH meter; nitrate, using KCl extraction and a diffusion-conductivity analyzer; soil texture (percentage sand/silt/clay), using soil suspension and a hydrometer; and water-holding capacity at 30 kPa and 151 kPa, using a pressure plate system.
From field measurements and topographic maps, I recorded the slope and aspect, the elevation, and the distance from the coast of each site. Distance from the coast represents a climatic gradient in this region; moving inland, mean temperatures increase, total rainfall decreases, and seasonal variability increases (Barbour and Major 1988). Each site was scored for the presence or absence of cool northerly slopes ([greater than]10% slope, 330 [degrees]-90 [degrees] aspect), warm southerly slopes ([greater than] 10% slope, 150 [degrees]-270 [degrees] aspect), and neutral slopes (all others).
For each treatment, I computed regional diversity (i.e., species richness) as the total number of species recorded at all 24 sites in all five rounds of sampling. For each site, I computed its local diversity and its differentiation diversity. To measure differentiation diversity, I used a metric advocated by Colwell and Coddington (1994): the total number of unshared species between two species lists divided by the number of species in the two lists combined (i.e., one minus the Jaccard coefficient of similarity). Such metrics of differentiation or similarity are often calculated for pairs of sites, but here I measured the differentiation of each site with respect to all other sites in its grouping. Thus, for each site, I calculated its (1) “total differentiation,” the proportion of species unshared between that site and all other sites in its treatment, and (2) “within-cluster differentiation,” the proportion of species unshared between that site and all other sites in its cluster. I also repeated the analyses using Whittaker’s beta diversity, i.e., regional diversity divided by local diversity (Whittaker 1960), to measure differentiation, and [TABULAR DATA FOR TABLE 1 OMITTED] the results were qualitatively the same. I will report the results using the Colwell and Coddington metric because when calculated for an individual site, it is influenced by the number of unique species at that site, whereas Whittaker’s beta is influenced only by the total number of species at that site. (See Shmida and Wilson 1985 and Colwell and Coddington 1994 for excellent reviews of diversity metrics.)
To test the hypothesis that the components of diversity differ between patchy and continuous serpentine sites, I performed a MANOVA, with treatment as the independent variable and the components of diversity (local, total differentiation, and within-cluster differentiation) as dependent variables. To examine the relationships between diversity patterns and environmental variation, I first compared the levels of each environmental covariate (elevation, distance inland, and soil variables) between patchy and continuous sites using MANOVA. I then entered all environmental variables as covariates in analyses of covariance (ANCOVA) of the effect of treatment (patchy vs. continuous) on local diversity. Finally, I repeated the ANCOVAs using treatment and just those environmental covariates that showed significant effects on diversity, first checking to ensure that the corresponding treatment-covariate interaction terms were not significant, which would violate the assumptions of ANCOVA. These analyses were performed for all herbs, native vs. alien herbs, and serpentine-endemic herbs. All analyses were done using the Systat MGLH module (Wilkinson 1990).
RESULTS
Total diversity
In all, 209 herbaceous species were found on serpentine in the five rounds of sampling, including 33 alien species (Table 1) and 29 serpentine endemics (Table 2). [TABULAR DATA FOR TABLE 2 OMITTED] The overlap in species composition between patchy and continuous sites (i.e., shared species/total species x 100%) was 44% for native species, 24% for alien species, and 62% for serpentine endemics. The species accumulation curve [ILLUSTRATION FOR FIGURE 3 OMITTED], which combines spatial and temporal effects (three transects per date x five dates), indicates that the completeness of sampling differed little between patchy and continuous sites; total diversity in both treatments appeared to reach asymptotes. Thus, the potential problem of overestimating complementarity through undersampling sites (Colwell and Coddington 1994) should be minimal, and should not create bias between treatments.
Environmental variables
Presence or absence of warm and cool slopes did not differ significantly between patchy and continuous serpentine sites, and their interactions with elevation and distance inland did not enter significantly into any analyses of diversity, so they are dropped from further discussion. Preliminary analyses showed that water-holding capacities at 30 kPa and 151 kPa were highly correlated (r = 0.95), and that percentage of sand explained the most variation of any soil texture variable (sand, silt, clay); thus “water capacity” henceforth refers to water-holding capacity at 30 kPa, and “soil texture” to percent sand. Also, the [Mg.sup.++]: [Ca.sup.++] ratio did not differ between treatments and did not explain any more variation than the two variables individually, so it is not examined further.
Patches were significantly less far inland than continuous sites, but did not differ significantly in elevation. Patches had significantly higher levels of nitrate and percentage of sand than continuous sites, but did not differ significantly in levels of [Ca.sup.++], [Mg.sup.++], or any other soil variable (Table 3).
All herbaceous species
When all herbaceous species were considered together, regional and local diversity was higher on patchy than continuous sites, while differentiation diversity was not significantly different (Table 4). In an ANCOVA of local diversity on treatment plus all environmental covariates, local diversity was influenced only by elevation (F = 8.1, df = 1, 38, P = 0.007). The interaction of treatment and elevation was not significant (P [greater than] 0.05). In an ANCOVA including only treatment and elevation, both effects were significant [TABULAR DATA FOR TABLE 3 OMITTED] (treatment: F = 11.1, P = 0.002; elevation: F = 10.3, P = 0.002; df = 1, 45). Closer examination of the elevation effect revealed that local herb diversity decreased significantly with increasing elevation on continuous sites (r = -0.66, P [less than] 0.001), but not on small patches (r = -0.20, P = 0.34).
Native vs. alien species
For native herbaceous species, patches had substantially higher regional diversity than continuous sites; however, although the trend was the same at the local and differentiation levels, the treatments were not significantly different (Table 5). This result indicates that the higher local diversity of all herbs on small patches seen above (Table 4) must have been due to alien species. In fact, small patches had an average of 6.04 [+ or -] 3.88 alien species (mean [+ or -] 1 SD), while continuous sites averaged only 1.33 [+ or -] 1.76 alien species, a significant difference (F = 29.3; df = 1, 46; P [less than] 0.001). The occurrences of alien species on patches and continuous sites are given in Table 1. Since the treatments differed in the local diversity of alien but not native species, the rest of this section focuses on analyzing patterns in alien diversity.
For alien species, an ANCOVA of local diversity on treatment plus all environmental covariates revealed significant effects of treatment (F = 11.6, P = 0.002), distance inland (F = 6.1, P = 0.02), and calcium (F = 7.1, P = 0.01; df = 1, 38 in all cases). None of the corresponding interaction terms was significant (P always [greater than] 0.10). When the ANCOVA was repeated using only the above three variables, their effects remained significant (treatment, F = 17.1, P [less than] 0.001; distance inland, F = 7.2, P = 0.01; calcium, F = 6.2, P = 0.02; df = 1, 44). Alien diversity decreased with increasing distance inland and increased with increasing levels of calcium. However, the positive correlation between calcium levels and diversity of alien species was evident only on small patches (r = 0.62, P = 0.04), not on continuous sites (r = 0.13, P = 0.36).
Serpentine endemics
Serpentine endemics showed similar regional diversity on patchy and continuous sites; local diversity was [TABULAR DATA FOR TABLE 4 OMITTED] substantially lower on small patches, and differentiation diversity was higher at both the total and within-cluster levels (Table 6). Occurrences of endemics on patches and continuous sites are given in Table 2. The ANCOVA of local diversity of endemics on treatment and environmental variables revealed significant effects of treatment (F = 9.7, P = 0.004), distance inland (F = 7.2, P = 0.01), and calcium (F = 8.3, P = 0.006; df = 1, 38 in all cases). One of these variables, calcium, showed a significant interaction with treatment. Since class by covariate interactions violate the assumptions of ANCOVA, the ANCOVA was repeated using only treatment and distance inland, both of which remained significant (treatment, F = 15.0, P = 0.001; distance inland, F = 13.4, P = 0.001; df = 1, 45). Endemic diversity increased with increasing distance inland.
To analyze further the effects of calcium, the local diversity of endemic species was regressed separately on calcium level for patches and continuous sites. Local diversity declined significantly with calcium levels on continuous sites (r = -0.74, T = -5.15, P [less than] 0.001) but not on small patches (r = -0.22, T = -1.07, P = 0.29) [ILLUSTRATION FOR FIGURE 4 OMITTED].
Across all sites, the local diversities of endemics and aliens were weakly negatively correlated (r = -0.33), which could either signify that invasion by aliens adversely affects endemics, or could be the indirect consequence of their opposite responses to treatment and calcium. In an analysis of variance on local diversity of endemics, including alien diversity as an independent variable, treatment and calcium were highly significant (P [less than] 0.001), while the diversity of aliens was not (P = 0.31).
Following the completion of sampling, I discovered [TABULAR DATA FOR TABLE 5 OMITTED] that two small patches (shown as separate on a geologic map), on a steep and densely chaparral-covered slope, were actually connected at their upper ends. I eliminated one of these from the data set and repeated the analyses reported above; this did not change the significance levels of any of the above results.
DISCUSSION
These results demonstrate the importance of the large-scale spatial structure of the landscape, as well as climate and soil variation, in shaping the components of diversity in a rich regional flora. These effects only became visible when the various components of the flora were differentiated. In particular, the relatively small subset of soil “specialists” (serpentine endemics) responded differently and often in opposite directions than soil “generalists” (nonendemics, particularly aliens). Since serpentine endemics are of great ecological, evolutionary and conservation interest, I discuss them first.
For serpentine-endemic herbs, small serpentine patches were virtually equal to large continuous sites in regional diversity, but were poorer in diversity at the local level, and higher in among-site differentiation. Surprisingly, no endemics were found on all continuous sites but no small patches, nor on all patches but no continuous sites (Table 2). The higher differentiation diversity on patchy sites occurred at the within-cluster level as well as the total level, and so was probably not the result of the climatic gradient associated with distance inland.
Serpentine endemics are generally considered to be inferior competitors able to tolerate the stress of unusually low levels of calcium (Brooks 1987). Thus, [TABULAR DATA FOR TABLE 6 OMITTED] even though previous studies have focused on individual plant performance rather than on diversity, it is not surprising that increasing calcium levels were associated with fewer serpentine-endemic species. However, the interactive effect of patchiness and calcium levels on the diversity of endemics is less straightforward to interpret. Calcium levels do not simply explain the effect of patchiness on endemic diversity, because, to do so, they would have to be higher on patchy than continuous sites, which was not the case. Instead, calcium had a strong negative correlation with endemic diversity on continuous but not on patchy sites; thus, calcium levels would appear to determine the effect of patchiness on diversity, or vice versa. One possible way to explain this is that small patches are typically poor in endemic species because of low rates of colonization and/or high rates of extinction. For the same reason, large continuous sites can be richer in endemic species, but are only so if they are also low in calcium. A large site that is unusually calcium-rich may be dominated by certain highly competitive nonendemic species such as chamise. However, experiments will be required to disentangle this interactive effect.
Serpentine endemics formed only 14% of total herbaceous diversity, and the above patterns were obscured when all species were considered. For all herbs combined, the small patches had higher regional and local diversity than continuous sites. In addition, local herb diversity declined with increasing elevation, but more so on continuous sites than on small patches. Considering native herbs only, regional diversity was 40% higher on patches than on continuous sites, and this increase occurred at both the local and differentiation levels, although it was not statistically significant at either. These results suggest that “generalist” herb species are more diverse on small patches. However, this tendency was not strong, perhaps because of the wide variation among generalist species in the sign and magnitude of their affiliation with serpentine. Patterns were stronger for alien species, which appear to be more uniform in having a low tolerance for serpentine. Aliens were considerably more diverse on small patches than in the interiors of large continuous ones. One conceivable explanation involves soil nitrate differences, since small patches had higher levels of nitrate, and since N fertilization has been shown to make some serpentine soils more suitable for alien grasses (Huenneke et al. 1990). However, the majority of studies indicate that low calcium is more important than low nitrogen or phosphorus in causing serpentine intolerance in many plant species, both native and alien (Kruckeberg 1984, Brooks 1987). In any case, the local diversity of alien species was not correlated with nitrate levels in this study. Conversely, alien diversity was positively correlated with soil calcium, but calcium levels did not differ between patchy and continuous sites in the direction that would explain their different diversities of aliens. Instead, the diversity of alien species increased strongly with calcium levels on small patches but not on continuous sites. Here, one possible biological explanation is that high calcium makes serpentine soils more invasible by alien species, but that invasion occurs only if a nearby source of propagules is also available. In other words, both patchiness and high calcium may be required for aliens to invade serpentine.
These results suggest that herb diversity on serpentine outcrops is shaped by multiple interacting forces, both spatial and nonspatial. For the specialist (endemic) species, small patches may have low diversity because of low rates of colonization and high rates of local extinction, and, for the same reason, endemic diversity can be higher on large outcrops. However, endemic diversity is depressed on large outcrops that are not “good serpentine,” i.e., that are unusually high in calcium. On the other hand, the lower local diversity of endemics on small patches is offset almost precisely by the patches’ higher among-site differentiation in species composition, which could be the result of colonization and extinction dynamics, habitat variation, or both. In contrast, generalist species, particularly aliens, are more prevalent on small patches; this may be because of the proximity of propagules from the non-serpentine matrix. However, small patches that are especially low in calcium remain resistant to aliens. To summarize: large patch size plus low calcium levels lead to local communities rich in serpentine endemics; edge effects plus high calcium levels lead to the invasion of serpentine by aliens.
These results differ in several ways from the patterns seen in woody plants (Harrison 1997). First, for herbs, the generalists formed a higher proportion of the total flora (86% vs. 72%). Alien species in particular had important effects on the relative herbaceous diversity of patchy and continuous sites, while among woody species there were no aliens. Hence, for herbs but not for woody plants, the presumed edge effects involving alien species were a dominant force in shaping local communities on small patches. This is especially true to the extent that competition from aliens may have adverse effects on the diversity of native herbs. Second, herb diversity was influenced by variation in soil calcium, in opposite directions for native and endemic species. In contrast, the diversity of woody species was not correlated with soil chemistry (S. P. Harrison, unpublished data).
For the habitat specialist species in this study, patchiness was not associated with major differences in regional diversity, only with differences in its partitioning into local vs. differentiation components. From a theoretical perspective, this result does not support models that predict, implicitly for habitat specialists, that patchy environments support more species regionally by permitting strong competitors to coexist (e.g., Caswell and Cohen 1991, 1993). Neither is there evidence for lower regional diversity in a patchy setting through the loss of superior competitors (Tilman et al. 1994). In retrospect, the indifferent fit of these results to competition-based metacommunity models is perhaps not surprising, since serpentine-endemic herbs usually seem too sparse to compete strongly. Moreover, rarely does competition appear to influence large-scale patterns of local and regional diversity (Cornell 1993). The above metacommunity models probably apply better to strongly competing species at smaller spatial scales (e.g., Bengtsson 1991, Wu and Levin 1994).
The results of this study are in better agreement with a model that does not assume competition affects the presence or absence of habitat specialists, and that does assume the inter-patch matrix is habitable and influential. Holt’s (1997) metacommunity model predicts that sparse habitat types will support a lower diversity of habitat specialists and a higher diversity of generalists, relative to widespread habitats. The present study, comparing the same community where it is sparse vs. widespread, provides evidence consistent with this idea. Holt’s (1997) model also makes another prediction that may fit this system, although data to test it are presently lacking. A generalist species may persist in a landscape because of its ability to use several habitat types, where one habitat is widespread but too low in quality to support persisting populations, while the other is suitable but too sparse to permit adequate dispersal among patches. Such a caricature may fit the serpentine “refugee” species, the nonendemics that grow abundantly on serpentine patches but are sparse on nonserpentine because of competition with mediterranean grasses.
For conservation purposes, these results suggest that large serpentine patches are the most effective places to conserve serpentine endemics, since compared to small patches they have an equal regional diversity and often a higher local diversity of endemic herbs, and are less invaded by exotic species. However, small serpentine patches are more diverse in the sense of being more different from one another, and also support slightly more nonendemic native species. This illustrates that the successful conservation of species in heterogeneous landscapes requires going beyond such oversimplifications as “single large or several small,” and understanding in detail how patchiness and habitat quality interact to shape the multiple components of regional diversity. This demonstrates that many more empirical studies are needed to assess the fit between popular spatial theories and real patchy landscapes.
ACKNOWLEDGMENTS
This study would not have been possible without the generous help of Joe Callizo in identifying plants, the access to sites kindly granted by numerous landowners, field assistance by Nicole Jurjavcic and Amy Wolf, and the programming skills of Rose Cook. Charles Canham, Martha Hoopes, John Maron, Amy Wolf, and two anonymous reviewers provided many helpful comments on the manuscript. This study was supported by NSF grant DEB 94-24137.
LITERATURE CITED
Barbour, M. G., and J. Major, editors. 1988. Terrestrial vegetation of California. California Native Plant Society, Sacramento, California, USA. Reprinted from 1977 edition. Wiley-Interscience, New York, New York, USA.
Bengtsson, J. 1991. Interspecific competition in metapopulations. Pages 219-237 in M. E. Gilpin and I. Hanski, editors. Metapopulation dynamics: empirical and theoretical investigations. Academic Press, London, UK.
Brooks, R. R. 1987. Serpentine and its vegetation: a multidisciplinary approach. Dioscorides, Portland, Oregon, USA.
Callizo, J. 1992. Serpentine habitats for the rare plants of Lake, Napa and Yolo Counties, California. Pages 35-51 in A. J. M. Baker, J. Proctor, and R. D. Reeves, editors. The vegetation of ultramafic (serpentine) soils. Intercept, Andover, Hampshire, UK.
Case, T. J. 1991. Invasion, resistance, species buildup and community collapse in metapopulation models with interspecies competition. Pages 239-266 in M. E. Gilpin and I. Hanski, editors. Metapopulation dynamics: empirical and theoretical investigations. Academic Press, London, UK.
Caswell, H., and J. E. Cohen. 1991. Disturbance, interspecific interaction and diversity in metapopulations. Pages 193-218 in M. E. Gilpin and I. Hanski, editors. Metapopulation dynamics: empirical and theoretical investigations. Academic Press, London, UK.
Caswell, H., and J. E. Cohen. 1993. Local and regional regulation of species-area relations: a patch-occupancy model. Pages 99-107 in R. E. Ricklefs and D. Schluter, editors. Species diversity in ecological communities: historical and geographical perspectives. University of Chicago Press, Chicago, Illinois, USA.
Colwell, R. K., and J. A. Coddington. 1994. Estimating terrestrial biodiversity through extrapolation. Pages 101-118 in Philosophical Transactions of the Royal Society of London, B 345.
Cornell, H. V. 1993. Unsaturated patterns in species assemblages: the role of regional processes in setting local species richness. Pages 243-252 in R. E. Ricklefs and D. Schluter, editors. Species diversity in ecological communities. University of Chicago Press, Chicago, Illinois, USA.
Fox, K. F., J. D. Sims, J. A. Barlow, and E. J. Helley. 1973. Preliminary geologic map of eastern Sonoma and western Napa Counties, California. United States Geological Survey, Denver, Colorado, USA.
Harrison, S. 1997. How natural habitat patchiness affects the distribution of diversity in Californian serpentine chaparral. Ecology 78:1898-1906.
Harrison, S., and A.D. Taylor. 1997. Empirical evidence for metapopulation dynamics. Pages 27-42 in I. Hanski and M. E. Gilpin, editors. Metapopulation dynamics: ecology, genetics and evolution. Academic Press, New York, New York, USA.
Hickman, J. C. 1993. The Jepson manual: higher plants of California. University of California Press, Berkeley, California, USA.
Holt, R. D. 1997. From metapopulation dynamics to community structure: some consequences of environmental heterogeneity. Pages 149-165 in I. Hanski and M. E. Gilpin, editors. Metapopulation dynamics: ecology, genetics and evolution. Academic Press, New York, New York, USA.
Huenneke, L., S. Hamburg, R. Koide, H. Mooney, and P. Vitousek. 1990. Effects of soil resources on plant invasion and community structure in Californian serpentine grassland. Ecology 71:478-491.
Kruckeberg, A. R. 1984. California serpentines: flora, vegetation, geology, soils and management problems. University of California Press, Berkeley, California, USA.
Lambert, G., and J. Kashiwagi. 1978. Soil survey of Napa County, California. Soil Conservation Service, Washington D.C., USA.
Nee, S., R. M. May, and M.P. Hassell. 1997. Two-species metapopulation models. Pages 123-148 in I. Hanski and M. E. Gilpin, editors. Metapopulation dynamics: ecology, genetics and evolution. Academic Press, New York, New York, USA.
Sawyer, J. O., and T. Keeler-Wolf. 1995. A manual of California vegetation. California Native Plant Society, Sacramento, California, USA.
Shmida, A., and M. V. Wilson. 1985. Biological determinants of species diversity. Journal of Biogeography 12:1-20.
Skinner, M. W., and B. M. Pavlik. 1994. California Native Plant Society’s inventory of rare and endangered plants of California. California Native Plant Society Special Publication Number 1, Sacramento, California, USA.
Tilman, D., R. M. May, C. L. Lehman, and M. A. Nowak. 1994. Habitat destruction and the extinction debt. Nature 371:65-66.
Wagner, D. L., and E. J. Bortugno. 1982. Geologic map of the Santa Rosa Quadrangle, California, 1:250,000. State of California, Resources Agency, Sacramento, California, USA.
Whittaker, R. H. 1960. Vegetation of the Siskiyou Mountains, Oregon and California. Ecological Monographs 30:279-338.
Wilkinson, L. 1990. Systat: the system for statistics. Systat Incorporated, Evanston, Illinois, USA.
Wilson, M. V. 1988. Within-community vegetation structure in the conifer woodlands of the Siskiyou Mountains, Oregon. Vegetatio 78:61-72.
Wu, J., and S. A. Levin. 1994. A spatial patch dynamic modeling approach to pattern and process in an annual grassland. Ecological Monographs 64:447-464.
COPYRIGHT 1999 Ecological Society of America
COPYRIGHT 2000 Gale Group
