and Webb et al. (2007); www.phylodiversity.net/phylocom]. As Webb et al. (2002) and Cavender-Bares and Wilczek (2003) reasoned, where abiotic habitat filtering is the dominant force shaping coexistence, PNC should result in phylogenetic clustering in the phylogeny of the regional species pool. On the other hand, where biotic competitive exclusion is the dominant ecological force, PNC should result in a more even (overdispersed) distribution of species on the regional tree than expected by chance.
These relations may hold in the abstract, and have oriented the interpretation of a number of studies [e.g., Cavender-Bares et al. (2004), Ackerly et al. (2006), Slingsby and Verboom (2006)], but there are a variety of complications or necessary extensions. For example, as Webb et al. (2002) and Cavender-Bares and Wilczek (2003) appreciated, an overdispersed phylogenetic pattern can also result from abiotic filtering from an underlying phylogeny showing convergent niche evolution. This observation simply highlights the need to couple such studies with independent phylogenetic tests of the extent of PNC in the clades under consideration [cf. Cavender-Bares et al. (2006) and Slingsby and Verboom (2006)]. Likewise, “ecological facilitation,” rather than competition, might underlie an over-dispersed phylogenetic pattern (Valiente-Banuet and Verdu, 2007), again emphasizing that there is not a simply one-to-one relationships between a phylogenetic pattern and an underlying cause.
It is also clear that possible causal processes will vary in intensity, and even in kind, as a function of scale (Swenson et al., 2007). For example, Webb et al. (2006) hypothesized that seedling phylodiversity patterns within small rainforest plots reflect the sharing of fungal pathogens among close relatives, whereas at a larger scale in the same forest they found evidence of habitat filtering. Cavender-Bares et al. (2004, 2006), in studies centered on oaks (Quercus) in northern Florida, showed evidence for phylogenetic evenness at smaller spatial and taxonomic scales (interpreted as the outcome of competition), but phylogenetic clustering at larger scales (interpreted as habitat filtering of phylogenetically conserved ecological traits). Clearly, sorting out among such possibilities requires the development of appropriate null models, and simulations to evaluate the power to distinguish alternative explanations [e.g., Kembel and Hubbell (2006) and Kraft et al. (2007)].
For present purposes it is especially important to note that entirely different causal factors become relevant as such studies scale up to much broader regions, or focus on clades that have moved around the globe. For example, Forest et al. (2007) reported lower phylogenetic diversity (despite higher species diversity) in the western Cape flora of South Africa, in part as a function of multiple rapid radiations (Linder, 2005). In contrast, the eastern Cape showed higher phylodiversity, in part because it interdigitates with another biodiversity hotspot. The key point is that at such larger