A major challenge in ecology is to explain species diversity in ecological communities. At least some variation in community diversity can be explained by evolutionary processes (Taylor et al. 1990; Zobel 1992; Pärtel 2002; Pärtel et al. 2007c) and dispersal limitation (Foster 2001; Foster et al. 2004; Pärtel & Zobel 2007). The species pool concept claims that large-scale processes like speciation, extinction and dispersal shape the regional flora or fauna, and species are then further filtered to the species pool of local communities based on the species’ ecological requirements (Zobel 1992, 1997). The species pool is thus a set of species available in the region that are ecologically suitable for growing in the particular type of environment occupied by a particular local community (Taylor et al. 1990; Eriksson 1993).
- 1The regional species pool is a set of species available in a region and ecologically suitable for growing in the particular environment occupied by a local community. As species pools are largely influenced by evolutionary processes such as the conservation of ecological niches within lineages we hypothesize that the size of the regional species pool increases with the variety of distinct phylogenetic lineages represented in a local community. We contrast this with hypotheses invoking diversification of individual lineages within environments or stochastic present-day assembly of local communities.
- 2We calculated phylogenetic distinctness for a local community as the number of nodes separating two species averaged over all pairwise comparisons across a phylogenetic topology of a regional flora. We calculated the size of the regional species pool for a local community as the number of species in the regional flora that share the ecological niche position of the species constituting the local community.
- 3Analysing field-layer communities across a wide range of environments, we indeed found that local communities composed of phylogenetically highly distinct species recruit from larger species pools than communities of low phylogenetic distinctness. Accounting for the presence of two particularly diversifying lineages (Poaceae and Cyperaceae) confirmed these results.
- 4These results help us to understand how the species pool was assembled throughout evolution in different types of environments (immigration vs. in situ radiation of individual lineages).
- 5The phylogenetic approach is of large practical value to infer the size of the regional species pool because phylogenies have become available for many groups of species worldwide, while knowledge of the species’ ecological requirements or habitat affiliation (needed for the classical definition of species pools) is often still lacking.
- 6Synthesis. We show that the size of the regional species pool can be predicted by the average phylogenetic distinctness between the species present in a local community. This approach contributes to the understanding of the causes of species richness in regional species pools and local communities. The approach is also an important tool for determining the size of the regional species pool when parameters other than species phylogeny are not known.
Revisiting methods to determine species pools
Knowledge of the size of species pools is of major importance for both community and conservation ecology (Zobel et al. 1998). For instance, species-pool size has been used to standardize local richness for protection purposes (Ingerpuu et al. 2001) or to detect the imprint of recent human impact (Pärtel et al. 2007a). In order to quantify the size of the species pool for plant communities, the ecological method is the most direct method used to date (Pärtel et al. 1996). This method first involves assessment of the mean ecological requirements of species in a local community for light, soil moisture, fertility and soil reaction (e.g. using Ellenberg indicator values; Ellenberg et al. 1991), and then defining the size of the species pool as the number of species in the regional flora that share the same ecological requirements. Other authors have characterized local communities by their phytosociological position, by their habitat affiliation, or the life-history traits of their constituent species (van der Valk 1981; Keddy 1992; Weiher & Keddy 1995; Dupré 2000). Then, the number of species from the entire regional flora that matched the same phytosociological or habitat affiliation or shared the same traits was quantified and was defined as the size of the species pool. Zobel et al. (1998) have pointed out the shortcomings of these approaches, such as the arbitrary choice of life-history traits. Moreover, in most regions of the world we lack sufficient knowledge of the typology and geographical distribution of habitats, and of the ecological requirements and life history of species. Finally, all these methods are essentially ‘post hoc’ in that they identify how many species from a given region match the species present in a given community. They do not explain why some species pools are richer than others, that is, they do not formulate a priori hypotheses on the factors driving the assembly of species pools.
The species pool as outcome of evolution
By definition, species pools are influenced by large-scale regional and evolutionary processes (Ricklefs & Schluter 1993). Focusing on the evolutionary history of different environments, Pärtel (2002) has shown that the size of the species pool of plant communities on acidic soils indeed increases with the evolutionary age of acidic soils in different regions of the world. An alternative approach would be to focus on the evolutionary history of the species constituting a local community.
As species traits (both functional and ecological) are an outcome of evolution, there should be a link between species’ phylogenetic and ecological similarity. Phylogenetic conservatism of ecological niches has been shown for higher plants (e.g. Prinzing et al. 2001): closely related taxa have relatively similar ecological niches as a legacy of descent from a common ancestor during evolutionary history. This view offers a new dimension to predict the size of a community's species pool by a phylogenetic approach. A local community may comprise a range from phylogenetically highly distinct species to phylogenetically closely related species (Warwick & Clarke 1998; Webb 2000; Cavender-Bares et al. 2004). Given niche conservatism within lineages, a local community assembled from distinct lineages recruits from a wide phylogenetic species pool, that is, a wide range of lineages, each contributing many ecologically similar species. As a consequence, we hypothesize that a high phylogenetic distinctness (PD) of species in a local community predicts a large regional species pool. We call this the Niche Conservatism Hypothesis of species pools.
Alternatively, evolution may affect species pools by massive diversification of individual lineages in particular environments, thereby increasing the species pools in these environments. Classical examples are the radiations of plant lineages in oceanic island environments (e.g. Emerson 2002), but also the radiation of Poaceae and Cyperaceae in grasslands (Behrensmeyer et al. 1992). Local communities sampled from such species pools enriched by diversification will be characterized by a low PD of their constituent species. We call this the Diversification Hypothesis of species pools. Finally, the phylogenetic structure of a local community may be the idiosyncratic outcome of stochastic movement and local extinction of species across present-day communities, blurring any link to the size of the evolved species pool. This is our Null Hypothesis.
Testing the hypotheses
To test these hypotheses, we studied the relationship between PD of species in 26 herb-layer plant communities and the size of the respective regional species pools in Estonia (original data from Zobel & Liira 1997). The sampled communities covered a wide range of regional community types from grasslands to forests. The herb-layer species richness of the sampled communities ranged from 16 to 136. We estimated the size of the regional species pool for each local community on the basis of the ecological requirements of the species in the local community, creating a four-dimensional ‘habitat space’ with the Ellenberg factors for light, soil moisture, pH and nitrogen content as axes (Pärtel et al. 1996, 2000). The regional species pool for each community was compiled by including from the regional flora all species for which the position in the ‘habitat space’ did not differ more than 1.5 relative units from the community mean, because an average ecological amplitude of a species is regarded to be 1.5 relative units (ter Braak & Gremmen 1987; van der Maarel 1993). Quantified by this method, the size of the regional species pool indeed successfully predicts the richness of local communities (Pärtel et al. 1996, 2000).
The phylogeny of the species in local communities was based on the phylogenetic topology for higher plants of Central Europe (Durka 2002, checked against Bremer et al. 2003 and Davies et al. 2004). This topology is about 70% dichotomously resolved, including the sub-family and sub-genus level. Such a high degree of resolution is essential when characterizing the phylogenetic structure of vegetation composed of a limited number of families as is the case in most temperate regions, including our present study (Hardy & Senterre 2007). To achieve this high degree of resolution dozens of original studies had to be compiled, many of which use different molecular markers, and thus only the topological information and not the branch lengths can be retained (Durka 2002).
The PD of species in a local community corresponds to ‘average taxonomic distinctness’ developed by Warwick & Clarke (1998), that is, average distance per all pairwise species combinations, applied to a less arbitrary phylogenetic topology (Hardy & Senterre 2007). We quantified distinctness in terms of number of nodes separating two species averaged over all pairwise comparisons across a phylogenetic topology (Webb 2000) of the regional flora. Such nodal distances across a topology in fact correspond well to ecological and niche distances between lineages (Prinzing et al. 2001). We chose this parameter as, contrary to others (e.g. Vane-Wright et al. 1991, Faith 1992, Izsak & Papp 2000), it was not significantly correlated to the species richness of the local communities (n = 26, r = 0.362, P = 0.0695), and thus represents a truly independent source of information. We used random sampling of species from the regional flora into each local community as a null model in order to estimate random PD of a local community. We found the random number of phylogenetic nodes per community to be 25.1 ± 0.04 and used it to standardize the observed distinctness by taking the difference from the null expectation.
PD in a local community indeed correlated to the size of the regional species pool, that is, number of species potentially capable of growing in the local community (Fig. 1). The larger the PD of species within a local community the larger the regional species pool that the community recruits from. PD of a local community described 45% (adjusted R2) of the variation of regional species-pool size by univariate linear regression analysis (t = 4.6, d.f. = 24, P < 0.0001). This confirms our Niche Conservatism Hypothesis and rejects the Diversification Hypothesis of species pools as well as the Null Hypothesis.
A graphical exploration of the results showed that the relationship between PD and species-pool size was not perfectly linear (Fig. 1). In fact, a saturation regression (Species-pool size = b0 – b0/(1 + (PD/b2)b1) explained a larger proportion of the variance (R2 = 0.57; b0 = 355, t = 15.0, d.f. = 23, P < 0.0001; b2 = 21, t = 50.9, d.f. = 23, P < 0.0001; b1 = 16, t = 2.1, d.f. = 23, P = 0.0483). The nonlinear relationship may be the result of the constrained size of the regional flora, for example, there are only 418 species characteristic to wooded meadows, the most species rich community in Estonia (Pärtel et al. 2007b).
The above relationship between PD of local communities and the size of the regional species pool may depend on the presence of a few particularly diversifying lineages, such as the Poaceae and Cyperaceae (accounting for an average of 28% of the species in our local communities). We thus repeated the analyses above including the percentage of Poaceae + Cyperaceae as an additional independent variable. The linear model obtained explained 53% of the variation of species-pool size. The inclusion of Poaceae + Cyperaceae even strengthened the observed relationship between PD and regional species-pool size (linear regression analysis: t = 4.9, d.f. = 23, P < 0.0001; linear component in saturation regression analysis: t = 47.4, d.f. = 22, P < 0.0001). Interestingly, the percentage of Poaceae + Cyperaceae had a negative effect on species-pool size (linear regression analysis: t = –2.2, d.f. = 23, P = 0.0375; linear component in saturation regression analysis: t = –1.9, d.f. = 22, P = 0.0646). This contradicts clearly the Diversification Hypothesis.
Conclusions and applications
Our results indicate that, across a wide range of environments, the evolutionary history of the species constituting a local community predicts the size of the regional species pool from which the community is recruiting. We were able to show that a large PD of species in a local community predicts a large regional species pool. This means that local communities open to colonization from multiple lineages profit from a larger species pool. Using the phylogenetic structure of local communities to predict the regional species pool would be a major advance. While knowledge of the species’ ecological requirements (needed for the classical definition of species pools) is often still lacking, phylogenies have become available for many groups of species worldwide (Judd et al. 2002; Sitte et al. 2002; Kühn et al. 2004).
Additionally, prediction of the size of the regional species pool from the PD of a local community can be an important tool for conservation ecologists. A local community rich in evolutionary lineages has a newly recognized conservation value because it comprises genetic resources needed to ensure continued evolution and thus future biodiversity (Mace et al. 2003; Forest et al. 2007). Phylogenetic richness of local communities has been used alongside species richness to prioritize areas for conservation in the last decade (Posadas et al. 2001; Sechrest et al. 2002). We propose that our method can be used for planning conservation areas in regions with limited information about the geographical dispersion of habitats, species affiliation to habitats, and species functional and ecological traits.
This work was supported by the Estonian Science Foundation (grants #0534J, #6614 and #7610) and the Estonian-French co-operational program PARROT. Comments of two anonymous referees and Bryan L. Foster improved the manuscript.