- Functional traits and phylogeny - 381 Journal of Vegetation Science 19: 381-392, 2008 doi: 10.3170/2008-8-18378, published online 10 March 2008 © IAVS; Opulus Press Uppsala. Functional traits and phylogeny: What is the main ecological process determining species assemblage in roadside plant communities? Valladares, F. 1,2* ; Tena, D. 1,5 ; Matesanz, S. 1,6 ; Bochet, E. 3,7 ; Balaguer, L. 4,8 ; Costa-Tenorio, M. 4,9 ; Tormo, J. 3,10 ; García-Fayos, P. 3,11 1 Instituto de Recursos Naturales. Centro de Ciencias Medioambientales, CSIC. Serrano 115 Bis, ES-28006 Madrid, Spain 2 Departamento de Biología y Geologia, Escuela Superior de Ciencias Experimentales y Tecnológicas, Universidad Rey Juan Carlos, c/ Tulipán s/n, ES-28933 Móstoles, Spain; 3 Centro de Investigaciones sobre desertiicación (CSIC, Universidad Valencia, Generalitat Valenciana), Camí de la Marjal s/n, ES-46470 Albal, Valencia, Spain; 4 Departamento de Biología Vegetal I, Facultad de Biología, Universidad Complutense de Madrid, C/ Jose Antonio Novais s/n, ES-280040, Madrid, Spain; 5 E-mail tiotena@hotmail.com; 6 E-mail silvia@ccma.csic.es; 7 E-mail esther.bochet@uv.es; 8 E-mail balaguer@bio.ucm.es; 9 E-mail mct@bio.ucm.es; 10 E-mail jaume.tormo@uv.es; 11 E-mail patricio.garcia-fayos@uv.es; * Corresponding author; Fax +34 913940800; E-mail valladares@ccma.csic.es Abstract Question: What is the main ecological process determining spe- cies assemblage in roadside herbaceous plant communities? Location: Roadside slopes (roadcuts and embankments) in the south (Málaga, mesic Mediterranean) and east (Valencia, dry and continental) of the Iberian Peninsula. Methods: We identiied 417 plant species, from which we selected the 331 most abundant (within the 70th abundance percentile) at each site. We compiled information on 28 func- tional traits and on the biogeographic range of each of these 331 species. We quantiied the phylogenetic signal of each trait for the species of each community and determined the number of functional convergences or divergences over the phylogenetic tree for each of the four situations (roadcuts and embankments in the two sites). Results: There was a signiicant phylogenetic signal in many traits, being positive in Valencia embankments and negative in Valencia roadcuts with almost no signal in any type of slope in Málaga. Each trait was signiicantly correlated with 20% - 35% of all other traits but correlation coeficients were low. No sig- niicant phylogenetic signal was found for the species’ distribu- tion range in any of the four communities studied, which might be the consequence of the complex mixture of biogeographic origins of the species found in these communities. Conclusion: The lack of a phylogenetic signal in most traits in Málaga, a climatically favourable locality, suggests that competitive exclusion was the main process involved in the assemblage of these communities. The signiicant and either positive or negative phylogenetic signal (in embankments and roadcuts respectively), the latter coupled with a signiicant number of functionally convergent nodes in the phylogenetic tree, suggests that environmental iltering is the most likely process involved in the harsh locality of Valencia. Keywords: Environmental gradient; Mediterranean; Phyloge- netic signal; Plant community; Road slope. Abbreviation: AOT = Analysis of traits. Introduction We have little knowledge of the mechanisms deter- mining the assemblage of the species that make up a community and, although it has been known for some time now that neither plants nor their functional traits are distributed randomly (Raunkiær 1934; Diamond 1975), numerous studies are still attempting to determine the processes that give rise to community structure (e.g. Keddy & Weiher 2001; Temperton & Hobbs 2004). From an ecological point of view, two main hypotheses might account for the non-random distribution of a group of species and their functional traits (Tofts & Silvertown 2000). On one hand, the classical theory of competitive exclusion states that competition is directly proportional to the similarity among species, which brings co-existing species to maximise the difference for a given functional trait (Johansson & Keddy 1991). This trait is, in turn, distributed among the species of the community in an overdispersed manner (i.e. exhibiting phenotypic repul- sion) regarding a null hypothesis of random distribution of species and functional traits (Armbruster et al. 1994; Wilson & Watkins 1994; Wilson & Gitay 1995). On the other hand, community structure might result from the inluence of environmental ilters and only those species possessing speciic attributes would persist in the community (Montalvo et al. 1991; Fernández-Alés et al. 1993; Díaz & Cabido 1997; Díaz et al. 2001). In this case, the species would be more similar than what would be expected by chance and a given trait would be clustered (i.e. exhibiting phenotypic attraction) regarding a null hypothesis of randomness. Neither hypothesis is exclusive, so that competitive exclusion and environmen- tal ilters concur, although their relative importance can differ depending on environmental conditions (Weiher &