Journal of Vegetation Science 24 (2013) 877–889 SPECIAL FEATURE: FUNCTIONAL DIVERSITY Partitioning phylogenetic and functional diversity into alpha and beta components along an environmental gradient in a Mediterranean rangeland Maud Bernard-Verdier, Olivier Flores, Marie-Laure Navas & Eric Garnier Keywords Abstract Beta dissimilarity; Calcareous rangelands; Community assembly; Environmental gradient; Questions: To what extent is the functional structure of plant communities Functional traits; Phylogenetic community captured by phylogenetic structure? Are some functional dimensions better rep- structure; Trait phylogenetic signal resented by phylogenetic relationships? In an empirical study, we propose to test the congruence between phylogenetic and functional structure at the alpha and Nomenclature the beta scale along an environmental gradient. Bernard (1996) Location: Causse du Larzac, southern France. Received 1 April 2012 Accepted 12 December 2012 Methods: We measured species abundances and eight key functional traits in Co-ordinating Editor: Francesco de Bello 12 plant communities distributed along a gradient of soil depth and resource availability in a Mediterranean rangeland. A phylogenetic super-tree of the spe- cies was assembled, and after quantifying the degree of phylogenetic signal pres- Bernard-Verdier, M. (corresponding author, ent in each trait, we quantified taxonomic (TD), phylogenetic (PD) and [email protected]): Universite´ functional (FD) diversity both within (alpha scale) and among (beta scale) com- Montpellier 2, Centre d’Ecologie Fonctionnelle munities, taking species abundances into account. We tested for trends in diver- et Evolutive (UMR 5175), 1919 route de Mende, sity along the environmental gradient, and looked for congruence among 34293 Montpellier, France different facets of diversity, both at the alpha and the beta scale. Garnier, E. ([email protected]): CNRS, Centre d’Ecologie Fonctionnelle et Evolutive Results: We found a significant phylogenetic signal for seven out of eight traits. (UMR 5175), 1919 route de Mende, 34293 However, when accounting for trends in taxonomic diversity (i.e. richness and Montpellier, France evenness), PD did not capture the strong functional structure observed within Flores, O. (olivier.fl[email protected]): UMR Peuplements Veg etaux et Bioagresseurs en and among the communities. At the alpha scale, we found an overall pattern of Milieu Tropical, UniversitedelaR eunion, phylogenetic convergence of abundant species, which did not reflect the 97715, St Denis Messageries, France observed functional divergence. At the beta scale, despite some congruence Navas, M.-L. ([email protected]): between betaPD and betaFD for three individual traits, phylogenetic dissimilari- Montpellier SupAgro, Centre d’Ecologie ties did not capture the overall environmental and functional sorting of species Fonctionnelle et Evolutive (UMR 5175), 1919 according to habitats. route de Mende, 34293 Montpellier, France Conclusions: We show that even when traits display a significant phylogenetic signal, PD does not capture the complex functional structure of communities in response to environmental gradients. Nevertheless, results suggest that phyloge- netic relationships may partially capture differences in the beta niche of species and provide additional insights on assembly processes not captured by the set of measured functional traits. Only by accounting for patterns in taxonomic diver- sity were we able to disentangle the functional and evolutionary determinants of species assembly along the gradient. netic structure, defined as the pattern of phylogenetic Introduction relatedness among species in communities (Cavender- Recent years have seen a growing interest in the use of Bares et al. 2009), has been used as a tool to understand phylogenetic information in community ecology (Webb community assembly (Webb 2000; Mayfield & Levine et al. 2002; Mouquet et al. 2012). Community phyloge- 2010), as well as to predict ecosystem functioning (Cadotte Journal of Vegetation Science Doi: 10.1111/jvs.12048 © 2013 International Association for Vegetation Science 877 Partitioning phylogenetic and functional diversity M. Bernard-Verdier et al. et al. 2009; Flynn et al. 2011). These approaches originally studies have investigated PD within communities, where stem from the concept of ecological niche conservatism processes of species co-existence, such as limiting similar- (reviewed in Wiens et al. 2010), a hypothesis stating that ity, are expected to shape local alpha diversity (Webb closely related species should share more ecological and 2000; Cavender-Bares et al. 2009). However, recent stud- functional similarities than distantly related ones. Based ies suggest that phylogenetic diversity may better capture on this assumption, a number of studies have interpreted assembly processes at the beta scale than at the alpha scale phylogenetic diversity (PD) as a surrogate for functional (Emerson & Gillespie 2008; Graham & Fine 2008; De Vic- diversity (FD) in communities (Webb 2000; Cadotte et al. tor et al. 2010). According to one line of reasoning, traits 2009; Violle et al. 2011). However, some authors have related to habitat preferences (i.e. beta niche traits) may be questioned this assumption (Cavender-Bares et al. 2004; more phylogenetically conserved within lineages than Silvertown et al. 2006; Losos 2008), and a consensus now determinants of local co-existence within communities seems to emerge from the literature stating that PD cannot (i.e. alpha niche traits), due to processes of sympatric niche in fact be considered as a simple proxy for FD (Mouquet differentiation among close relatives (Silvertown et al. et al. 2012). Studies comparing FD and PD tend to show 2001; Prinzing et al. 2008). Thus, phylogenetic differences an overall lack of consistency between phylogenetic and among communities, i.e. phylogenetic beta diversity, may functional patterns (Losos 2008; Swenson & Enquist 2009; reveal niche-based processes such as environmental but see Baraloto et al. 2012). Even in cases where clear sorting of species across habitats (Graham & Fine 2008; patterns of phylogenetic clustering or over-dispersion are Cavender-Bares et al. 2009). Although this hypothesis has detected, interpretations of assembly mechanisms are not found unequal support in the literature (Emerson & straightforward because different assembly processes may Gillespie 2008), quantifying phylogenetic beta diversity, create the same phylogenetic patterns (Losos 2008; May- and comparing it to the turnover in species and functional field & Levine 2010). traits along gradients, is a promising way to understand If PD cannot be interpreted as a simple proxy for FD, how ecological and evolutionary processes shape biodiver- then what do the non-random patterns of phylogenetic sity across landscapes and environmental gradients structure, consistently reported in very different systems (Graham & Fine 2008; Vamosi et al. 2009; Pavoine & (reviewed in Pavoine & Bonsall 2011), capture exactly? Bonsall 2011). One hypothesis is that PD may be related only to a few In this study, we investigate the functional information phylogenetically conserved traits. In this case, PD should captured by phylogenetic relationships both within and capture only assembly processes acting upon these traits among communities in a Mediterranean rangeland of (e.g. Cavender-Bares et al. 2004). However, if different southern France. To do this, we quantified and compared phylogenetically conserved traits display opposite patterns taxonomic, phylogenetic and functional diversity along a of convergence or divergence within a community, as seen gradient of soil depth and resource availability, along in recent trait-based studies (e.g. Cornwell & Ackerly which plant communities have been found to exhibit 2009; Spasojevic & Suding 2012; Bernard-Verdier et al. strong taxonomic and functional structures in previous 2012), then PD cannot be expected to capture such a mul- studies (Bernard-Verdier et al. 2012; Perez-Ramos et al. tiplicity of information. Another hypothesis is that PD may 2012). Using data for eight key functional plant traits, phy- be related to a complex combination of multiple ecological logenetic information from the latest published phyloge- traits capturing whole plant strategies, rather than individ- nies, and local species abundances, we aimed to answer ual traits (Flynn et al. 2011). In this case, one would the following questions: (i) can we detect phylogenetic expect PD to be more closely correlated to a multivariate structure, both within and among plant communities, in diversity score based on relevant traits, than to individual response to the environmental gradient; (ii) do phyloge- traits. A third hypothesis is that PD may in some cases cap- netic and functional diversities show congruent distribu- ture some ‘hidden’ aspects of plant functioning that are tions at the alpha and/or the beta scale; and (iii) is the not captured by any of the functional traits commonly existence of phylogenetic signal in traits a good indicator of measured in community studies (e.g. pathogen sensitivity; the correlation between PD and FD both within and Webb et al. 2006), and thus may show no congruence among communities? To answer these questions, we first with the measured FD. quantified the phylogenetic signal in each trait, and then Moreover, relationships between PD and FD are used a unified framework based on Rao’s quadratic expected to shift with