* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Regional and local impact on species diversity – from pattern to
Soundscape ecology wikipedia , lookup
Biodiversity wikipedia , lookup
Restoration ecology wikipedia , lookup
Introduced species wikipedia , lookup
Island restoration wikipedia , lookup
Unified neutral theory of biodiversity wikipedia , lookup
Habitat conservation wikipedia , lookup
Biogeography wikipedia , lookup
Biodiversity action plan wikipedia , lookup
Reconciliation ecology wikipedia , lookup
Occupancy–abundance relationship wikipedia , lookup
Molecular ecology wikipedia , lookup
Biological Dynamics of Forest Fragments Project wikipedia , lookup
Theoretical ecology wikipedia , lookup
Ecological fitting wikipedia , lookup
Latitudinal gradients in species diversity wikipedia , lookup
Oecologia (2002) 132:479–491 DOI 10.1007/s00442-002-0988-3 REVIEW Helmut Hillebrand · Thorsten Blenckner Regional and local impact on species diversity – from pattern to processes Received: 20 August 2001 / Accepted: 23 May 2002 / Published online: 16 July 2002 © Springer-Verlag 2002 Abstract The impact of regional factors (such as speciation or dispersal) on the species richness in local communities (SL) has received increasing attention. A prominent method to infer the impact of regional factors is the comparison of species richness in local assemblages (SL) with the total number of species in the region (SR). Linear relations between SR and SL have been interpreted as an indication of strong regional influence and weak influence of interactions within local communities. We propose that two aspects bias the outcome of such comparisons: (1) the spatial scale of local and regional sampling, and (2) the body size of the organisms. The impact of the local area reflects the scales of ecological interactions, whereas the ratio between local and regional area reflects the inherent moment of autocorrelation. A proposed impact of body size on the relation is based on the high dispersal and high abundance of small organisms. We predict strongest linearity between SR and SL for large organisms, for large local areas (less important ecological interactions) and for sampling designs where the local habitat area covers a high proportion of the regional area (more important autocorrelation). We conducted a meta-analysis on 63 relations obtained from the literature. As predicted, the linearity of the relationship between SL and SR increased with the proportion of local to regional sampling area. In contrast, neither the body size of the organisms nor the local area itself was significantly related to the relation between SL and SR. This indicated that ecological interactions played a minor role in the shape of local to regional richness plots, which instead was mainly influenced by the sampling design. We found that the studies published so far were highly biased towards larger organisms and towards high similarity between the local and regional area. The proposed prevalence of linear relationships may thus be an artefact and H. Hillebrand (✉) · T. Blenckner Erken Laboratory, Department of Limnology, Evolutionary Biology Centre, University of Uppsala, Norr Malma 4200, 76173 Norrtälje, Sweden e-mail: [email protected] Fax: +46-176-229315 plots of SL to SR are not a suitable tool with which to infer the strength of local interactions. Keywords Diversity · Dispersal · Body size · Area · Species pool Introduction The diversity in local assemblages can be regulated by local factors (such as competition, disturbance, abiotic conditions) and by regional factors (such as history of climate, evolution and migration). Much research has been devoted to local mechanisms which increase, maintain or decrease diversity (Tilman and Pacala 1993; Chesson 2000). In recent years, however, large-scale processes have been increasingly regarded as important determinants for the species richness of local communities (SL) (Ricklefs 1987; Cornell and Lawton 1992; Lawton 1999). These reports also comprised the question of how regional differences in diversity are generated and maintained during time (Rosenzweig 1995; Hillebrand and Azovsky 2001) and how these regional differences transfer into the diversity of local communities (Zobel 1997). Conceptually, the assembly of a local community can be visualised as species passing through a series of filters, which represent historical (e.g. dispersal, speciation) and ecological (e.g. competition, predation, disturbance, abiotic environmental factors) constraints on the arrival and survival of organisms at a certain locality (Zobel 1997; Lawton 1999). In this concept, the local diversity is related to the diversity of the regional pool if processes connected to the dispersal of organisms are mainly responsible for the assembly of local communities (Fig. 1). A dominant impact of the local environment (abiotic and biotic) was supposed to lead to independence between SL and regional species richness (SR) (Fig. 1). Several contributions tried to disentangle the regional and local constraints of SL (Cornell and Lawton 1992; Ricklefs and Schluter 1993; Cornell and Karlson 480 pact of both factors and connect these to a conceptual predictive model (see Background). We will then test these predictions in a quantitative synthesis of the literature available. For this meta-analysis, we will also revise the statistics used to distinguish between type I and type II communities. Background The impact of area Fig. 1 Conceptual model visualising the assembly of local communities from regional species pool (see text for more details) 1996, 1997; Cornell 1999; Srivastava 1999; Shurin 2001). Generally, these studies agree on an important influence of both regional and local factors, but the relative importance of these factors is still uncertain since they act on different temporal and spatial scales. Some of these scales are difficult to manipulate or are not at all tractable, reducing the possibility to experimentally test the predictions on regional and local influence. Therefore, the importance of regional and local processes has been derived from the analysis of patterns. A central method used for this discussion is the regression of SL on SR (Lawton 1999; Srivastava 1999). Significant linear regressions are interpreted as an indication of the high impact of regional factors on local diversity (type I communities), whereas saturating or nonlinear functions would indicate an upper limit of SL set by ecological interactions (type II communities) (Cornell and Lawton 1992). SL in type I communities thus tends to be higher in species-rich regions, which has also been called “regional enrichment” (Karlson and Cornell 1998, 2002). Analyses of SL to SR have been done in terrestrial, marine and freshwater systems, mainly for vertebrates and host-specific organisms like parasites. The studies published so far stressed the prevalence of type I communities (see reviews by Cornell and Karlson 1997; Cornell 1999; Lawton 1999; Srivastava 1999). In many cases, this strong linear relationship between SR and SL has been interpreted as an indication of “unsaturation” of local assemblages with species and of generally weak effects of local interactions on species richness (Oberdorff et al. 1998; Cornell 1999; Hugueny and Cornell 2000). Although this type of analysis seems to be straightforward, several pitfalls have been discussed (Srivastava 1999; Shurin et al. 2000). In this paper, we analyse two factors interfering with the analysis of local to regional richness plots, the spatial scale of the analysis and the body size of the organisms. We hypothesise that these two factors will affect the linearity between SL and SR more than the intensity of ecological interactions. In the following, we will thoroughly describe the proposed im- Previous studies showed that using different areas for censuses of SL or SR, respectively, resulted in highly different outcomes of the regression of SL to SR (Westoby 1993; Angermeier and Winston 1998; Karlson and Cornell 2002). Area plays an important role in these analyses: it affects the estimates of species richness (species area relationship), it reflects the scales on which species interact (ecological factors) and the size of the regional pool (SR), and it describes the degree of similarity if the local and regional area are close to each other (autocorrelation). Species-area relationships Increasing the area of sampling will generally result in an increase in species richness, as described by the wellknown species-area relationship (SAR). The different size of regions itself can be one explanation for different sizes of SR pools, e.g. on islands of different size (Ricklefs 1987). However, the use of naturally bound local habitats of different size for the determination of SL will result in a biased estimate of SL (Shurin et al. 2000). These differences can be avoided by removing the trend with area from the data (Karlson and Cornell 1998; Shurin et al. 2000) or by using standardised areas for localities and regions (Westoby 1993; Caley and Schluter 1997; Weiher 1999). Ecological factors The scale of local interactions should be used to determine local habitat size, since the local extinction of species due to competition or predation is implicitly assumed to cause the non-linearity between SL and SR. This scale may be a complete lake for relatively wellmixed pelagic communities, but only a few square metres for plants competing for soil nutrients (Huston and De Angelis 1994). In one of the most thorough analyses so far, Caley and Schluter (1997) presented strong evidence for a linear relationship between SR and SL for a variety of organisms. These results were independent of the size of the local community, which comprised either 1% or 10% of the 250,000 km2 regional squares. However, even the lower marginal of these two local assemblages covered an area on which animals or plants do not 481 interact (Huston 1999). Therefore, an upper limit of SL due to ecological interactions cannot be expected on these large scales. On the other hand, sampling areas for SL which are too low may result in increasing undersampling of rare species and thus the artificial creation of a non-linear pattern between SL and SR, which has been called pseudosaturation (Caley and Schluter 1997; Karlson and Cornell 2002). Regional species richness The delimitation of the regional species pool is an important factor in analyses of relationships between SL and SR. First, the suitable extent of the regional area is difficult to determine (Srivastava 1999; Shurin et al. 2000). Most studies used an arbitrarily chosen large area (Caley and Schluter 1997) or an area delimited by natural geography (Oberdorff et al. 1998). The regional pool should comprise all species able to invade the local community, which integrates a component of dispersal ability (see below) and a component of time. At extreme time scales, almost all species will at least possibly invade almost every community. To delimit a regional species pool, a theoretical ecological time period has to be assumed – but this has not been done explicitly so far. Secondly, all species able to reach a certain community may not be able to live there. Simply counting the number of species within a taxonomic group will overestimate the regional species pool and a correction for the suitability of the habitat would have to be done (Pärtel et al. 1996). Thirdly, different regional species pools are not independent, e.g. due to an overlap in species composition (Srivastava 1999; Ricklefs 2000). Autocorrelation The pattern between SL and SR will highly depend on the proportion of the region covered by the local communities. If the area of the local habitat compared to the region is extremely large, almost all species of a regional pool should be present in any “local” community and the linear relationship between SL and SR would be just a statistical consequence. This moment of autocorrelation is implicit in the SL to SR regressions and is more important if the proportion of local to regional area is large (Creswell et al. 1995). Loreau (2000) discussed this problem by dividing total regional diversity (γ) into a between-habitat (β–) and a within-habitat (α–) diversity component (see also Whittaker 1972; Godfray and Lawton 2001). Saturation of local α– with increasing γdiversity is only possible if the γ– and β– components are highly correlated, i.e. higher regional pools encompass more habitats, increasing the β-component of diversity. This is only possible if the area of the “local” assemblage is small, i.e. if higher regional areas easily include more habitats (Cornell and Lawton 1992; Loreau 2000). The impact of body size The probability that organisms will invade a local habitat is clearly related to dispersal ability and the abundance of propagules. If the regional species pool is supposed to reflect the species able to invade the target locality, both factors have to be taken into account. However, the studies presented so far did not adjust the definition of regions to dispersal ability, e.g. Caley and Schluter (1997) used the same regional area for each group. These differences become even more important by extending the approach to small organisms such as zooplankton, protozoans or unicellular algae (Shurin et al. 2000; Hillebrand et al. 2001). These organisms exhibit a very high abundance and high transportability, which will increase the statistical probability of successful dispersal (Finlay et al. 1996; Fenchel et al. 1997). Additionally, any individual unicellular organism can be seen as a propagule. Consequently, these organisms exhibit a comparably small global species richness, leading to a unimodal relation of global diversity with highest richness in intermediate body size classes (Godfray and Lawton 2001). The proposition of low global and high SL of unicellular organisms (Fenchel 1993) has repeatedly been criticised based on difficulties in applying morphospecies concepts to theses groups (Mann and Droop 1996; Foissner 1999; Sabbe et al. 2001). However, Godfray and Lawton (2001) summarise some evidence on reduced speciation rates of unicellular organisms by high dispersal ability and thus by high gene flow. Additionally, the high dispersal ability of small organisms is reflected by the similarity of species composition between distant sites (Hillebrand et al. 2001), the low proportion of unicellular endemism in remote lakes (Cocquyt 2000) and the rarity of large biogeographical patterns (Hillebrand and Azovsky 2001). These reports are contrasted by increasing evidence of dispersal limitation in higher plants (Tilman 1997; Cain et al. 1998; Hubbell et al. 1999; Turnbull et al. 2000) and benthic invertebrates (Roughgarden et al. 1988; Cowen et al. 2000). Body size is thus a rough proximate variable to indicate dispersal ability across different size classes (Fenchel 1993; Hillebrand and Azovsky 2001). The assumption that dispersal ability increases with decreasing body size is clearly valid only for a very large range of organism sizes. There are deviations from this assumption, both for some general groups (birds will be better dispersers than many smaller groups) and within groups (different dispersal ability of snails). Still, dispersal is often a chance-related event, combining transportability and abundance of propagules, both being higher for very small organisms. Decreasing body size will roughly increase the range from which organisms are able to reach the target locality and will thus influence the relation between SL and SR. 482 Fig. 2 Predicted effects of local area (a), the proportion of local to regional area (b) and body size (c) on the relationship between local species richness (SL) and regional species richness (SR). Lines represent extremes in a continuum of possible shapes of the regression of SL to SR Predictions In our analyses, we concentrate on two aspects of area: (1) the spatial scale of ecological interactions, and (2) the degree of autocorrelation. From the thoughts presented above, the following predictions can be derived. 1. The linearity of the relationship between SL and SR will increase with increasing area of the local habitat (Fig. 2a). The size of the local habitat determines the extent to which ecological interactions are covered. Increasing the area regarded as local habitat will thus decrease the strength of ecological interactions and increase the linearity. 2. The linearity of the relationship between SL and SR will increase with an increasing ratio between the area of the local habitat to the area of the region. The larger a locality is scaled in proportion to a region, the more it is influenced by factors regulating the regional pool (cf. Karlson and Cornell 2002) and the more Fig. 3a, b Conceptual model predicting the linearity between SL and SR. Linearity is measured as the slope of the log-transformed regression and depends on body size and the proportion of local to regional sampling (pL) (a) or on body size and the area of the local habitat (b). For abbreviations, see Fig. 2 SR is covered by SL. We introduce pL as the mean proportion of locality area to the area of the region and predict an increasing importance of the regional pool with increasing pL (Fig. 2b), reflecting the statistical impact of area. 3. The linearity of the relationship between SL and SR will increase with increasing body size of the organisms. The high dispersal ability and abundance of small organisms will result in high SL, which in turn should result in constant SL despite increasing SR. We thus expect that the regional species pool SR constrains SL mainly within groups of large organisms with a low dispersal ability and a low number of individuals (Fig. 2c). The impacts of area and body size are clearly interwoven, since diatoms or mammals differentially perceive the size of a certain local area. This is evident for both the ecological and the statistical aspect of area. We therefore created two conceptual models pairing the impact of body size with each of the area impacts (Fig. 3). For the statistical impact of area, the most linear relationships between SR and SL should be found at large body sizes and at high pL, i.e. if local areas cover a high proportion 483 of the regional area. If organisms are small, numerous and highly dispersed and if pL is small, the SL should be independent of the regional species pool. A three-dimensional graphical model emerges, giving the importance of SR for SL (i.e. the linearity of the relationship), depending on body size and pL (Fig. 3a). For the ecological impact of area, we predict lowest linearity for small organisms in small areas. Increasing the area will increase the linearity. For every given local area, the dependence on regional pools will be highest for intermediate body sizes, since the largest organisms may still be able to interact and the smallest may be highly dispersed. A second three-dimensional graphical model thus predicts a unimodal function of the slope b dependent on body size and local area (Fig. 3b) Materials and methods We use published data in a meta-analysis to test the predictions derived above. A central aspect of meta-analysis is the use of a suitable effect size (Osenberg et al. 1997), in this case a metric to measure the linearity between SL and SR. Therefore, we first revise the models used in the literature, before presenting the data basis of our analysis. Models to distinguish between type I and type II relationships There is no consistent statistical test used to distinguish between type I and type II relationships throughout the literature, but in most studies linear and non-linear regression results were compared (Cornell 1985; Hawkins and Compton 1992; Cornell and Karlson 1996). As non-linear regression models, second-order polynomial (quadratic) or power functions have been used (Creswell et al. 1995). Besides differences in the number of estimated parameters (Creswell et al. 1995), all curvilinear functions initially allow a nearly linear increase (Fig. 4a). Within the range of data used to compare SL and SR, the explained variance may thus be very similar and highly correlated between linear and saturation curves, giving little contrast to distinguish between models. However, a severe problem with these regression approaches is the reliance on model I regression and the corresponding estimates of explained variance (Cresswell et al. 1995; Pärtel et al. 1996). Model I regression assumes that the dependency of one variable on the other is known and that the independent variable is without error (Sokal and Rohlf 1995). Both assumptions are not met by the data sets presented in the literature. SR cannot be estimated without error and may also be influenced by local pools as vice versa. The lack of an error-free independent variable mostly affects the explained variance, and the error can be extremely large in non-linear regression models. The implicit autocorrelation due to the interdependence of SL and SR also biases the explained variance. Tests based on the variance explained by different regression models are thus statistically flawed. Also correlation analyses are not suitable to indicate the strength of a linear relationship due to autocorrelation (Cresswell et al. 1995; Pärtel et al. 1996). In addition to these statistical concerns, there is also a general lack of rigorous hypothesis testing in this approach: a slightly better fit in one type of relationship does not falsify the validity of the alternative regression approach. Several studies have tried to overcome these limitations. Pärtel et al. (1996) used correlation coefficients in connection with minimum strength null hypothesis and Monte-Carlo simulations. Griffiths (1997) used log-log regressions of SL on SR, arguing that slopes similar to 1 indicate proportional sampling (type I communities), while slopes <1 indicate levelling off in local richness (type II communities). Linear relationships going through the origin tend to have a slope b=1 in log-log space, whereas non-linear Fig. 4a, b Models used to distinguish between linear and non-linear regressions of SL on SR (see text for details). For abbreviations, see Fig. 2 saturating functions have a slope b<<1 (Fig. 4b). Using log-transformed variables, the deviation of b from 1 can thus be seen as a measure of linearity (see also Ricklefs 2000). This latter test is advantageous since estimates of the slope are not biased by the errors present in the independent variable (Sokal and Rohlf 1995) and the slope can be used as an effect metric in meta-analyses (Hillebrand et al. 2001). Therefore, we conducted a meta-analysis on the linearity of the relationship between SR and SL, calculating the slope b for every study (see Data basis) using the equation: (1) where SL is the mean SL and a and b the parameters to be estimated. We used b to test first each of the three single predictions concerning local area, pL, and body weight. Since a weighted metaanalysis was not possible due to biased estimates of SEs for the slope, we used Spearman rank-correlations between b and each of the three predicting variables. We then tested the predictions of the three-dimensional models by conducting a multiple regression analysis on b. We used a fixed non-linear model (second-order polynomials) to allow unimodal responses of the dependent variable. To avoid errors inclined with the errors in the independent variables, all variables were rank-transformed. Data Surveying the literature in addition to recent reviews by Cornell (1999) and Srivastava (1999), we found 54 studies on the relation- 484 ship between regional and SL (Appendix 1). It became obvious that to date most investigations were on vertebrate animals and host-specific organisms like galling insects, parasitoids or helminth parasites of fish. Studies on small organisms were rare, but we derived estimates of SR and SL for ciliates and diatoms, thus adding four more data sets (Appendix 1). The following criteria were used to select the data to be included in the analysis: 1. The minimum number of observations had to be n≥3. 2. The underlying model for SL to SR relationships analysed in the study should follow the argumentation in Srivastava (1999). Therefore, we included only studies investigating the richness of one community type in different regions, since Srivastava (1999) argued that using different habitat types in one region as regional pools (see Pärtel et al. 1996) confounds habitat differences and differences in SR. 3. Host-specific taxa like pathogens or phytophagous insects had to be excluded from our analysis, since their dispersal is not related to their body size but to the dispersal of their hosts. 4. When the original approach could not be used [e.g. due to pseudoreplication, cf. Srivastava (1999)], we obtained the original data from the study and recalculated the regression. If the presentation of the original study did not allow the recalculation, we deleted the study from the analysis. 5. If different local areas were used to estimate SL, we calculated the species area relationship and used the residuals as measures of SL. Our final data set comprised 63 analyses from 32 studies, which could be used to test our hypotheses. For each organism group, we estimated body mass from literature values and obtained estimates of local area from the original contribution. Furthermore, we calculated pL, for which alternative methods had to be used, since original approaches for defining regional and local pools differed. In most cases, the proportion could be calculated on the basis of the areas of local and regional sampling (Caley and Schluter 1997), whereas in some studies it was calculated on the basis of sample numbers (Van Valkenburgh and Janis 1993). In several studies, pL could not be obtained directly from the original analysis and had to be estimated (Appendix 1), but the impact of this error was minimised by log-transformation of the variables. Results We found positive correlations between the linearity of the relationship between SL and SR (i.e. slope b) and the proportion pL (rS=0.285, n=62, P=0.025). No significant relationship was found for the body weight of the organisms (rS=0.182, n=63, P=0.153) and the local area (rS=0.201, n=45, P=0.186), respectively. The combined impact of pL and body size on the linearity of the relationship between SL and SR was evident across the studies included in our analysis (Fig. 5, upper graph). We found highest linearity if organisms were large and if pL was high. This was especially evident if pL was ≥0.01 (1%). On the other hand, studies comprising small organisms and/or small pL revealed slopes <<1, indicating non-linear relationships (Fig. 5, upper graph). This pattern was reflected by a significant multiple regression model on slope b with pL and body size as independent variables (Table 1). Although the overall model was significant, the single factors were only significant for pL, but not for body weight (Table 1). Both factors comprised strong positive linear interactions and Fig. 5 Slope b of regressions of SL to SR dependent on pL and body size (upper graph) or on local area and body size (lower graph). For abbreviations, see Figs. 2 and 3 Table 1 Results of fixed non-linear (second-order polynomial) multiple regression on the slopes b of log-transformed regressions of local to regional species richness. The table gives the overall model statistics and the estimates for the independent variables for two different models. The first model comprised the proportion of local to regional sampling (pL) and body weight as linear and quadratic predictor variables, whereas the second model comprised local area and body weight. All variables were rank-transformed Model Overall Model 1 r2=0.2181 F4,57=3.98 P=0.006 Model 2 r2=0.056 F4,43=0.64 P=0.639 Variables Estimate of b P-level Intercept pL Linear pL Quadratic Weight linear Weight quadratic Intercept Local area linear Local area quadratic Weight linear Weight quadratic 0.206 0.042 –0.0006 0.005 –0.00007 0.899 0.002 –0.00004 –0.003 0.00009 0.261 <0.001 0.001 0.659 0.693 0.003 0.916 0.823 0.869 0.881 weaker negative quadratic terms, indicating a tendency towards non-linearity. Furthermore, it became obvious from Fig. 5, that there is strong bias in the investigations published so far 485 towards large organisms and high pL. Clearly, only few studies incorporated invertebrates and small relative local areas. Several studies revealed even slopes >>1, which indicated that low SR regions had too low estimates of SL which may be due to undersampling. The predicted impact of local area in connection to body weight was not shown in the data (Fig. 5, lower graph). While linearity still somewhat decreased for small organisms, there was no trend with the size of the local habitat. Also the multiple regression was clearly non-significant (Table 1). It became thus evident that the local impact of area was absent, whereas the statistical impact was the most important factor influencing the shape of the relation between SL and SR. Discussion Our results indicated that the linear relation between SL and SR is mainly driven by a strong impact of autocorrelation, that is increasing linearity was observed when the sizes of local and regional areas became more similar. Although there was some indication that SL becomes independent of SR for small organisms, this pattern was not significant. An ecological impact of area on the results could not be found. The proposed prevalence of linear type I relations (Cornell 1999; Lawton 1999; Srivastava 1999) was thus clearly based on a biased selection of test organisms and locality sizes. In the following, we first will discuss the use of regional to local richness plots to detect mechanisms of SL assembly, then proceed by focusing on processes instead of patterns and finally give some future implications of our analysis. The use of regressions between SL and SR Linear relationships between SR and SL have been interpreted as an indication that local assemblages are not saturated with species and SL is largely influenced by processes on regional scales (Ricklefs 1987; Cornell 1999). Additional evidence in favour of these regional imprints on SL has been taken from the equivocal results of experiments on the dependence of invasibility on species richness (Cornell 1999; Levine and D'Antonio 1999), and from the existence of large-scale gradients in regional as well as in local data sets (Gaston 2000; Hillebrand and Azovsky 2001). In this context, type I relationships have even been discussed as indicators of weak or absent ecological interactions (Lawton 1999; Hugueny and Cornell 2000) and a higher importance of dispersal limitation than ecological interactions (Cornell and Karlson 1997). Also our study revealed that most investigations found linear relationships between SL and SR (slopes ~1), thus corroborating previous statements that SL does not converge in similar habitats in regions of differing diversity (Lawton 1999; Srivastava 1999). However, the predominance of linear (type I) relationships indicated merely the scale of analysis, but not the unsaturation of local assemblages with species or the strength of ecological interactions (cf. Huston 1999; Loreau 2000). The clear increase in the linearity with pL and the generally high values of pL across the studies make it obvious that the linear relationship found in many investigations follows from the autocorrelation between SL and SR as defined in these studies. At high pL, the regression between SR and SL is linear without any underlying ecological mechanism, simply indicating that estimates of SL and SR are not independent (Loreau 2000). Only few studies tried to circumvent this problem by using very large regions and small local samples (Cornell and Karlson 1996; Karlson and Cornell 2002). In addition to the statistical artefact, our analysis strongly suggests that the prevalence of linear relationships between SL to SR is also based on a biased selection of test organisms (favouring organisms with comparably low dispersal ability). Thus, there seems to be a limited validity in the use of regressions of SL on SR to infer biological processes (Loreau 2000; Shurin et al. 2000). Shurin et al. (2000) found strong linear regressions of SL to SR for crustacean zooplankton, but simultaneously found evidence that dispersal was not an important factor influencing zooplankton diversity. Moreover, linear relations between SL and SR were not incompatible with strong local interactions (Shurin 2000; Shurin et al. 2000). But even if dispersal is high and local interactions are strong, SL may not be consistently reduced irrespective of SR. First, if regions contain only some localities where competitive exclusion is prevented, the mean SL will increase with the species pool SR (Huston 1999). Secondly, the reduction of SL requires time and disturbances may decrease the effects of biological interactions (Huston 1999; Karlson and Cornell 2002). In patch-occupancy models, even low rates of disturbance were shown to obscure the effect of strong competitive interactions on SL, leading to a linear relations between SL and SR (Caswell and Cohen 1993). Thirdly, in the context of saturation of local assemblages with species, the discussion of ecological interactions has been mainly restricted to one trophic level, i.e. competition (Cornell and Lawton 1992; Huston 1999). Other local interactions, such as grazing, can have a strong impact on species diversity (Proulx et al. 1996; Hillebrand et al. 2000) and should be included (Shurin and Allen 2001). Finally, even if local assemblages may saturate with species in time scales corresponding to ecological interactions, the saturation pattern may be invisible. In evolutionary time, new species may invade or evolve within the regional pool, which may lead to shifts in the species composition in local communities – and prevent SL from reaching hard limits (Cornell and Lawton 1992). In a thorough critical review on the use of local to regional richness regressions, Srivastava (1999) discussed additional pitfalls of the regional to local plots, such as: (1) in defining “regional” and “local” (cf. Huston 1999; Loreau 2000; Shurin et al. 2000), (2) in choosing the right type of data for this kind of comparison, and (3) in avoiding spatial and temporal pseudoreplication. The 486 sizes of the local and regional areas have been defined rather arbitrarily and differ widely. Not all studies have standardised the SL measurements to a certain area (Westoby 1993; Caley and Schluter 1997; Weiher 1999). Because of the increase in species richness with area, using different sizes of “local” and “regional” in one study will affect the results (Srivastava 1999). It would be preferable to standardise local size and to relate the size of the locality to the size of the organisms, their habitat use and especially to the scale of interactions. The size of the regional species pool is even more difficult to define. Counting the number of species in a certain taxa may combine species from different trophic guilds, which do not compete, (Huston 1999) or may comprise species which cannot survive under the present local conditions (Pärtel et al. 1996; Zobel 1997; Dupré 2000; Shurin 2000). The area used to calculate the regional pool should furthermore be adjusted to the dispersal ability of the organisms (since it should contain all species able to reach the local habitat), but most studies have used one arbitrarily defined large area for different groups (Caley and Schluter 1997). From these arguments it becomes clear that SL is connected to the regional species pool, irrespective of the importance of local interactions. Thus, the only conclusions to be drawn from the SR to SL plots is that local interactions are not strong enough to limit SL to a consistent level in all localities in different regions (Cornell and Lawton 1992; Loreau 2000). From pattern to process Our analysis revealed the limited applicability of SR to SL regressions to infer processes of diversity assembly. However, the statistical issues do not contradict the importance of regional processes on local communities. It is without question that total species richness differs between regions for many organism groups. Different regional diversity can be based on different processes, e.g. variation in net diversification rates due to higher physiogeographical heterogeneity and subsequent allopatric speciation (Ricklefs 1987; Qian and Ricklefs 2000). There is considerable discussion about the mechanisms creating and maintaining highly diverse regions (Gaston and Williams 1996; Rohde 1999; Hillebrand and Azovsky 2001). There is also good evidence beyond the SL to SR plots that high SR is connected to higher SL (Ricklefs 1987; Westoby 1993). Instead of inferring the saturation of local communities with species from patterns, it is more important to analyse how different levels of SR transfer into SL and which mechanisms affect this transfer. Recruitment limitation in fact seems to be an important process determining SL, influenced by organism traits (size of the dispersal stage, number of propagules produced, durability of the dispersal stages) and habitat characteristics (such as the degree of flow in aquatic communities). Propagule density does not only affect the colonisation of local communities (the supply side), but also the internal interactions within these communities (Menge and Sutherland 1987; Palmer et al. 1996). The continuous supply of propagules is proposed to intensify competitive and trophic interactions in local communities (Menge and Sutherland 1987). Palmer et al. (1996) made predictions about regional and local impacts under different scenarios of dispersal and disturbance. If both are high, they proposed a regional control of SL since random colonisation events will determine community composition. If disturbance and dispersal are low, a mainly local influence was proposed based on local interactions. At low disturbance but high dispersal, Palmer et al. (1996) assumed combined local and regional control. Menge and Sutherland (1987) made similar predictions for different trophic levels, finding important and distinct effects of recruitment on the relative importance of environmental stress, competition and predation for basic, intermediate and top trophic levels. The relative importance of these interacting forces and the mechanisms linking them can be best analysed by models and careful experimentation. Amarasekare and Nisbet (2001) explicitly considered how dispersal modified local competitive interactions and found that increased immigration rates can reduce coexistence. Rather than the patch occupancy approach with an emphasis on exclusion and colonisation, source-sink dynamics and spatial variation in competitive success were important features with regard to understanding coexistence [see Moore et al. (2001) and Mouquet and Loreau (2002) for similar approaches]. In a model linking predation, competition and dispersal, Shurin and Allen (2001) investigated the effects of predator-mediated coexistence on local and regional richness, proposing that processes other than dispersal limitation can promote linear relations between local and regional richness. If only competitive exclusion was included as a local interaction, the model predicted an asymptote of SL with increasing SR. However, if a keystone predator was introduced, regional coexistence was promoted due to an expanded array of conditions allowing competitors to coexist. Mean local diversity was affected positively or negatively, depending on the dispersal rates of the predator and its prey (Shurin and Allen 2001). Few experiments included the presence of a regional species pool (Palmer et al. 1996), but especially aquatic communities seem to be suitable for experimental manipulations. These experiments included recruitment and species pools on small laboratory scales (Long and Karel 2002) and larger field studies (Menge et al. 1997, 1999). The colonisation history and competitive dominance determined interactively the community structure in aquatic microcosms (Long and Karel 2002). In coastal benthic communities, differences in community structure can be tracked to differences in near-shore oceanography and thus food supply and density in recruits (Menge et al. 1997, 1999; Connolly and Roughgraden 1998). Regional and local factors also interacted when two different predators (fish, insect) were introduced into zooplankton as- 487 semblages that were or were not connected to a regional pool of zooplankton species (Shurin 2001). Both predators reduced zooplankton species richness if dispersal was not possible. With dispersal, however, the fish had positive effects on richness by facilitating the invasion by species from the regional species pool. Conclusions and future perspectives The linearity of relationships between SL and SR increased with the proportion of the region (pL) sampled to obtain SL. The strong relation between pL and b showed that the sampling design plays a major role in the outcome of these analyses. The body size as a proximate variable for abundance and dispersal ability of the organisms was not significantly related to the linearity and neither was the local area itself. Regressions of local to regional richness do not allow one to infer processes of community assembly and should be used cautiously. There is ample evidence for regional differences in diversity, but no evidence for the absence of strong ecological interactions that can be derived from linear SL to SR relations. Thus, the importance of regional impacts on SL has to be seen in conjunction with local processes. The models and experiments described above showed that the relative importance of regional and local influences and the mechanisms underlying this relation are analytically tractable. For such an analysis, the intensity of local interactions should be measured independently (e.g. as presence of competitive exclusion, resource limitation, density compensation, disturbance frequency or interaction strength) and compared to the regional species pool and the SR to SL relation (Srivastava 1999). Like other topics in macroecology, the question of regional impacts on SL leads to an inspiring discussion of community assembly with a strong impact on basic and applied issues. The saturation of local communities is linked to conservation issues in at least two aspects, first the success of invasive species in relation to SL (Levine and D'Antonio 1999), and second the usefulness of conserved areas if target species have a limited dispersal ability. Regional impacts may be important for the richness of local assemblages, but it is premature to infer a prevalence of regional factors from the studies presented so far. Studies beyond the plots of SL to SR have to be conducted to disentangle the scales on which the factors controlling community composition (inter-)act. Acknowledgements We are gratefully indebted to Robert E. Ricklefs and Ron Karlson for encouragement and critical discussion. Ron Karlson also provided us with original data. We thank all authors of the primary contributions on this topic for the possibility to conduct this meta-analysis. The manuscript profited from the comments of Stephanie Blenckner and Monika Feiling. We are moreover grateful to Russell Monson and two anonymous reviewers. This work has been funded by a grant to H. H. (Deutscher Akademischer Austauschdienst D/99/08944) and by the Erken Laboratory. Appendix 1 Studies on regional and local species richness analysed during this investigation thermore it is stated if the study was included (Incl.) in our analysis and how the proportion of local to regional sampling (pL) was calculated The table gives authors and references for each study, the regions and organisms studied, and the number of observations (n). FurAuthors Reference Abele (1984) In: Strong DR, et al. (eds) Worldwide Ecological communities. Princeton University Press, Princeton, N.J., pp 123–137 In: Esch GW, et al. (eds) Parasite communities: patterns and processes. Chapman and Hall, London, pp 157–195 In: Ricklefs RE, Schluter D (eds) North America Species diversity in ecological communities. University of Chicago Press, Chicago, Ill.,pp 185–193 Ecology 79:911–927 Virginia, USA Aho (1990) Aho and Bush (1993) Angermeier and Winston (1998) Caley and Schluter (1997) Clarke and Lidgaard (2000) Cornell and Karlson (1996) Cornell 1985) Region Organisms Decapods Parasites of amphibians Parasites of bass fish and sunfish Fish n Incl. Comments PL 4 Yes – Estimated area 13 No1 – – No1 – – 13/13 Yes10 – – 3/9 Ecology 78:70–80 Worldwide J Anim Ecol 69:799–814 North Atlantic Birds, mammals, 3–5 reptiles, fish, trees Bryozoans 14 J Anim Ecol 65:233–241 Worldwide Corals Ecology 66:1247–1260 California Galling insects Yes Yes 39/26 Yes2 9 No1 Two local sizes – Two estimates SL – Precise area Estimated area Estimated area – 488 Appendix 1 (Table continued) Authors Reference Region Organisms Cornell (1985) Dawah et al. (1994) Findley and Findley (2001) Foissner (1992–1996) Foissner (1996–1999) Frenzel and Brandl (2000) Gaston and Gauld (1993) Griffiths (1997) Hawkins and Compton (1992) Am Nat 126:565–569 J Anim Ecol 64:708–720 California Great Britain Ecol Monogr 71:69–91 Incl. Comments PL Galling insects 7 Parasitoids insects 15 No1 No1 – – – – Worldwide Fish 18 Yes Diversec Germany Aquatic ciliates 7 Yes Two region sizes – Estimated area Sample Diversed Worldwide Soil ciliates 16 Yes – Sample Global Ecol Biogeogr 9:293–309 Central Europe – No1 – – J Trop Ecol 9:491–499 Worldwide Phytophagous insects Insects 4 No1 – – J Anim Ecol 66:49–56 J Anim Ecol 61:361–372 North America South Africa 15 Yes 15/15 No1 – – – – –e Worldwide Fish, lacustrine Galling insects and their parasitoids Diatoms 10 Yes – –f Worldwide Diatoms 29 Yes – Estimated area Sample Am Nat 146:162–169 Cote d’Ivoire Fish 10 Yes2 – – Ecol Monogr 68:259–274 Worldwide Corals No7 – – J Biogeogr 26:825–841 Asia, N America Mammals Yes – Kennedy and Guegan (1994) Parasitology 109:175–185 Great Britain No1 – Estimated area – Lawes and Eeley (2000) J Biogeogr 27:1421–1435 Worldwide Lawes et al. (2000) Biodiv Conserv 9:683–705 South Africa Birds, mammals, Lepidoptera 5–15 Yes Lawton et al. (1993) In: Ricklefs RE, Schluter D (eds) Species diversity in ecological communities. University of Chicago Press, Chicago, Ill., pp 175–184 Divers Distrib 5:91–103 Great Britain Insects on fern 6 No1 Spain, South Africa Worldwide Coleoptera 4 Yes Ectoparasites of fish Lizards 35 No1 Two different Precise area regions – – 2 No8 – – Fish in rivers 9 Yes2 – 14 No9 27 No9 16, 203 Yes2 – – – Estimated area – – Estimated area H. Hillebrand, unpublished data H. Hillebrand, unpublished data Hugueny and Paugy (1995) Karlson and Cornell (1998) Kelt et al. (1999) Lobo and Davis (1999) Morand et al. (1999) Morton (1993) Oberdorff et al. (1998) Pärtel et al. (1996) Pärtel et al. (2000) Pearson and Juliano (1993) Pearson (1977) Int J Parasitol 29:663–672 In: Ricklefs RE, Schluter D (eds) Species diversity in ecological communities. University of Chicago Press, Chicago, Ill., pp 159–169 J Anim Ecol 67:472–484 Oikos 75:111–117 Oikos 90:191–193 In: Ricklefs RE, Schluter D (eds) Species diversity in ecological communities. University of Chicago Press, Chicago, Ill., pp 199–202 Condor 79:232–244 Australia and North America NW France Parasites of endemic and introduced fish Primates Estonia Plants Estonia Plants North America, Tiger beetles India, Australia Tropical forests Birds n 3 11–23 Yes 6 Yes2 Three different regions Three different regions – Estimated area Estimated area – Reanalysed Estimated by Srivastava area (1999) 489 Appendix 1 (Table continued) Authors Reference Region Organisms Pejler (1997) Worldwide Rotifers Richardson et al. (1985) Ricklefs (1987) Arch Hydrobiol [Suppl] 53:255–306 J Veg Sci 6:329–342 Australia Banksia (plants) Science 235:167–171 Caribbean Birds Ricklefs (2000) J Anim Ecol 69:1111–1116 Caribbean Birds Rørslett (1991) Aquat Bot 39:173–193 Scandinavia Roslin (2001) Shurin et al. (2000) Smith (2001) Ecography 24:511–524 Ecology 81:3062–73 Finland Worldwide Ecology 82:792–801 New Zealand Aquatic macrophytes Dung beetles Crustacean zooplankton Zoobenthos Soares et al. (2001) Stevens (1986) Stevens and Willig (2002) Terborgh and Faaborgh (1980) Tonn et al. (1990) Aust Ecol 26:187–192 Brazil Am Nat 128:35–46 Ecology 83:545–560 Incl. Comments PL 6 Yes – 40 Yes2 – Estimated area Precise area 5 No 7 Yes Reanalysed by Ricklefs (2000) – 18 No10 – Estimated area – 131 24 Yes Yes – – Precise area Precise area 24 Yes – Ants 10 Yes – Estimated area Precise area Eastern USA America Coleoptera Bats 20 32 No1 Yes – – Am Nat 116:178–195 Caribbean Birds 10 Yes – Am Nat 136:345–376 Finland and Wisconsin USA Fish in lakes 2 No8 – – Estimated area Estimated area – Mammals, fossils 24 Yes – Sample USA Bivalvia 14 Yes – Van Valkenburgh In: Ricklefs RE, Schluter D (eds) and Janis (1993) Species diversity in ecological communities. University of Chicago Press, Chicago, Ill., pp 330–340 Vaughn (1997) Ecography 20:107–115 Weiher (1999) Westoby (1993) Winkler and Kampichler (2000) Wisheu and Keddy (1996) Zobel and Liira (1997) n – J Ecol 87:1005–1011 In: Ricklefs RE, Schluter D (eds) Species diversity in ecological communities. University of Chicago Press, Chicago, Ill., pp 170–177 Ecography 23:385–392 Canada Australia and North America Plants, herbaceous Lizards, birds, plants, fish 10 2 Yes No9 – – Estimated area Precise area – Austria Collembola 10 Yes – Precise area Oikos 76:253–258 North America Plants 28 Yes – Precise area Oikos 80:325–332 Estonia Plants 27 No9 1. Organisms depend or presumably depend on their hosts for dispersal. 2. Original regression pseudreplicated, reanalysed. 3. Comprises data from Foissner (1997) (Limnologica 27:179– 238) and Foissner et al. (1992) (Limnologica 22:97–104). 4. Comprises data from Foissner (1996) (Biol Fertil Soils 23:282–291), Foissner (1997) (Biodiv Conserv 6:1627–1638), Foissner (1997) (Biol Fertil Soils 25:317–339), Foissner (1999) (Biodiv Conserv 8:319:389) and Foissner (1996) (Acta Protozool 35:95–123). 5. This data set contains the regional species richness of large scale floras (Central Europe, British Isles, Baltic Sea) and the mean number of species found in local assemblages (lakes, coastal sites) within these regions. – 6. This data set contains the species richness at one site as regional richness and the richness within one sample as local richness. 7. Data set identical to Cornell and Karlson (1996) (J Anim Ecol 65:233–241). 8. Too few data. 9. Data set used not suitable for the analysis (see Srivastava 1999) (J Anim Ecol 68:1–16). 10. Original regression pseudoreplicated, original data not obtainable. 490 References Amarasekare P, Nisbet RM (2001) Spatial heterogeneity, sourcesink dynamics, and the local coexistence of competing species. Am Nat 158:572–584 Angermeier PL, Winston MR (1998) Local vs. regional influences on local diversity in stream fish communities of Virginia. Ecology 79:911–927 Cain ML, Damman H, Muir A (1998) Seed dispersal and the holocene migration of woodland herbs. Ecol Monogr 68:325– 347 Caley MJ, Schluter D (1997) The relationship between local and regional diversity. Ecology 78:70–80 Caswell H, Cohen JE (1993) Local and regional regulation of species-area relations: a patch-occupancy model. In: Ricklefs RE, Schluter D (eds) Species diversity in ecological communities. University of Chicago Press, Chicago, Ill. pp 99–107 Chesson P (2000) Mechanisms of maintenance of species diversity. Ann Rev Ecol Syst 31:343–366 Cocquyt C (2000) Biogeography and species diversity of diatoms in the Northern Basin of Lake Tanganyika. Adv Ecol Res 31:125–150 Connolly SR, Roughgarden J (1998) A latitudinal gradient in Northeast Pacific intertidal community structure: evidence for an oceanographically based synthesis of marine community theory. Am Nat 151:311–326 Cornell HV (1985) Local and regional richness of cynipine gall wasps on Californian oaks. Ecology 66:1247–1260 Cornell HV (1999) Unsaturation and regional influences on species richness in ecological communities: a review of the evidence. Ecoscience 6:303–315 Cornell HV, Karlson RH (1996) Species richness of reef-building corals determined by local and regional processes. J Anim Ecol 65:233–241 Cornell HV, Karlson RH (1997) Local and regional processes as controls of species richness. In: Tilman D, Kareiva P (eds) Spatial ecology. Princeton University Press, Princeton, N. J. pp 250–268 Cornell HV, Lawton JH (1992) Species interactions, local and regional processes, and limits to the richness of ecological communities: a theoretical perspective. J Anim Ecol 61:1–12 Cowen RK, Lwiza KMM, Sponaugle S, Paris CB, Olson DB (2000) Connectivity of marine populations: open or closed. Science 287:857–859 Creswell JE, Vidal-Martinez VM, Crichton NJ (1995) The investigation of saturation in the species richness of communities: some comments on methodology. Oikos 72:301–304 Dupré C (2000) How to determine a regional species pool: a study in two Swedish regions. Oikos 89:128–136 Fenchel T (1993) There are more small than large species? Oikos 68:375–378 Fenchel T, Esteban GF, Finlay BJ (1997) Local versus global diversity of microorganisms: cryptic diversity of ciliated Protozoa. Oikos 80:220–225 Finlay BJ, Esteban GF, Fenchel T (1996) Global diversity and body size. Nature 383:132–133 Foissner W (1999) Protist diversity: estimates of the near-imponderable. Protist 150:363–368 Gaston KJ (2000) Global patterns in biodiversity. Nature 405:220–227 Gaston KJ, Williams PH (1996) Spatial patterns in taxonomic diversity. In: Gaston K (ed) Biodiversity – a biology of numbers and difference. Blackwell, Oxford, pp 202–229 Godfray HCJ, Lawton JH (2001) Scale and species numbers. Trends Ecol Evol 16:400–404 Griffiths D (1997) Local and regional species richness in North American lacustrine fish. J Anim Ecol 66:49–56 Hawkins BA, Compton SG (1992) African fig wasp communities: undersaturation and latitudinal gradients in species richness. J Anim Ecol 61:361–372 Hillebrand H, Azovsky AI (2001) Body size determines the strength of the latitudinal diversity gradient. Ecography 24:251–256 Hillebrand H, Worm B, Lotze HK (2000) Marine microbenthic community structure regulated by nitrogen loading and grazing pressure. Mar Ecol Prog Ser 204:27–38 Hillebrand H, Watermann F, Karez R, Berninger UG (2001) Differences in species richness patterns between unicellular and multicellular organisms. Oecologia 126:114–124 Hubbell SP, Foster RB, O'Brien ST, Harms KE, Condit R, Wechsler B, Wright SJ, Loo de Lao S (1999) Light-gap disturbances, recruitment limitation, and tree diversity in a neotropical forest. Science 283:554–557 Hugueny B, Cornell HV (2000) Predicting the relationship between local and regional species richness from a patch occupancy dynamics model. J Anim Ecol 69:194–200 Huston MA (1999) Local processes and regional patterns: appropriate scales for understanding variation in the diversity of plants and animals. Oikos 86:393–401 Huston MA, De Angelis DL (1994) Competition and coexistence: the effects of resource transport and supply rates. Am Nat 144:954–977 Karlson RH, Cornell HV (1998) Scale-dependent variation in local vs. regional effects on coral species richness. Ecol Monogr 68:259–274 Karlson RH, Cornell HV (2002) Species richness of coral assemblages: detecting regional influences at local spatial scales. Ecology 83:452–463 Lawton JH (1999) Are there general laws in ecology? Oikos 84:177–192 Levine JM, D'Antonio CM (1999) Elton revisited: a review of evidence linking diversity and invasibility. Oikos 87:15–26 Long ZT, Karel I (2002) Resource specialization determines whether history influences community structure. Oikos 96:62–69 Loreau M (2000) Are communities saturated? On the relationship between α-, β- and γ-diversity. Ecol Lett 3:73–76 Mann DG, Droop SJM (1996) Biodiversity, biogeography and conservation of diatoms. Hydrobiologia 336:19–32 Menge BA, Sutherland JP (1987) Community regulation: variation in disturbance, competition, and predation in relation to environmental stress and recruitment. Am Nat 130:730–757 Menge BA, Daley BA, Wheeler PA, Dahlhoff E, Sanford E, Strub PT (1997) Benthic-pelagic links and rocky intertidal communities: bottom-up effects and top-; down control? Proc Nat Acad Sci US 94:14530–14535 Menge BA, Daley BA, Lubchenco J, Sanford E, Dahlhoff E, Halpin PM, Hudson G, Burnaford JL (1999) Top-down and bottom-up regulation of New Zealand rocky intertidal communities. Ecol Monogr 69:297–330 Moore JL, Mouquet N, Lawton JH, Loreau M (2001) Coexistence, saturation and invasion resistance in simulated plant assemblages. Oikos 94:303–314 Mouquet N, Loreau M (2002) Coexistence in metacommunities: the regional similarity hypothesis. Am Nat 159:420–426 Oberdorff T, Hugueny B, Compin A, Belkessam D (1998) Non-interactive fish communities in the coastal streams of Northwestern France. J Anim Ecol 67:472–484 Osenberg CW, Sarnelle O, Cooper SD (1997) Effect size in ecological experiments: the application of biological models in meta-analysis. Am Nat 150:798–812 Palmer MA, Allan JD, Butman CA (1996) Dispersal as a regional process affecting the local dynamics of marine and stream benthic invertebrates. Trends Ecol Evol 11:322–326 Pärtel M, Zobel M, Zobel K, Van der Maarel E (1996) The species pool and its relation to species richness: evidence from Estonian plant communities. Oikos 75:111–117 Proulx M, Pick FR, Mazumder A, Hamilton PB, Lean DRS (1996) Experimental evidence for interactive impacts of human activities on lake algal species richness. Oikos 76:191–195 Qian H, Ricklefs RE (2000) Large-scale processes and the Asian bias in species diversity of temperate plants. Nature 407:180–182 Ricklefs RE (1987) Community diversity: relative roles of local and regional processes. Science 235:167–171 Ricklefs RE (2000) The relationship between local and regional species in birds of the Caribbena Basin. J Anim Ecol 69:1111– 1116 491 Ricklefs RE, Schluter D (1993) Species diversity: regional and historical influences. In: Ricklefs RE, Schluter D (eds) Species diversity in ecological communities. University of Chicago Press, Chicago, Ill. pp 350–363 Rohde K (1999) Latitudinal gradients in species diversity and Rapoport's rule revisited: a review of recent work and what can parasites teach us about the causes of the gradients? Ecography 22:593–613 Rosenzweig ML (1995) Species diversity in space and time. Cambridge University Press, Cambridge Roughgarden J, Gaines S, Possingham H (1988) Recruitment dynamics in complex life cycles. Science 241:1460–1466 Sabbe K, Vanhoutte K, Lowe RL, Bergey EA, Biggs BJF, Francoeur S, Hodgson D, Vyverman W (2001) Six new Actinella (Bacillariophyta) species from Papua New Guinea, Australia and New Zealand: further evidence for widespread diatom endemism, in the Australasian region. Eur J Phycol 36:321– 340 Shurin JB (2000) Dispersal limitation, invasion resistance, and the structure of pond zooplankton communities. Ecology 81:3074– 3086 Shurin JB (2001) Interactive effects of predation and dispersal on zooplankton communities. Ecology 82:3404–3416 Shurin JB, Allen EG (2001) Effects of competition, predation, and dispersal on species richness at local and regional scales. Am Nat 158:624–637 Shurin JB, Havel JE, Leibold MA, Pinel-Alloul B (2000) Local and regional zooplankton species richness: a scale-independent test for saturation. Ecology 81:3062–3073 Sokal RR, Rohlf FJ (1995) Biometry, 3rd edn. Freeman, New York Srivastava DS (1999) Using local-regional richness plots to test for species saturation: pitfalls and potentials. J Anim Ecol 68:1–16 Tilman D (1997) Community invasibility, recruitment limitation, and grassland biodiversity. Ecology 78:81–92 Tilman D, Pacala S (1993) The maintenance of species richness in plant communities. In: Ricklefs RE, Schluter D (eds) Species diversity in ecological communities. University of Chicago Press, Chicago, Ill. pp 13–25 Turnbull LA, Crawley MJ, Rees M (2000) Are plant populations seed-limited? A review of seed sowing experiments. Oikos 88:225–238 Van Valkenburgh B, Janis CM (1993) Historical diversity patterns in North American large herbivores and carnivores. In: Ricklefs RE, Schluter D (eds) Species diversity in ecological communities. University of Chicago Press, Chicago, Ill. pp 330– 340 Weiher E (1999) The combined effects of scale and productivity on species richness. J Ecol 87:1005–1011 Westoby M (1993) Biodiversity in Australia compared with other continents. In: Ricklefs RE, Schluter D (eds) Species diversity in ecological communities. University of Chicago Press, Chicago, Ill. pp 170–177 Whittaker RH (1972) Evolution and the measurement of species diversity. Taxon 21:213–251 Zobel M (1997) The relative role of species pools in determining plant species richness: an alternative explanation of species coexistence? Trends Ecol Evol 12:266–269