COS 30-6: Traits of invasive and noninvasive Commelinaceae in a phylogenetic context
Jean H. Burns, Washington University in St. Louis, Robert B. Faden, Smithsonian Institution, and Scott J. Steppan, Florida State University.
A major theme in ecology is the incorporation of phylogenetic information in studies of ecological behavior. While it is well appreciated that an understanding of phylogeny may yield a more accurate type I error rate for traits that have evolved few times, it is less often appreciated that incorporating phylogeny may also improve the ability to detect trait associations. We examined the relationship between biological traits, historical factors, and invasive status in 68 species of Commelinaceae across five genera with invasive and noninvasive taxa. We generated a novel phylogenetic hypothesis within and among these five genera using data from chloroplast and nuclear ribosomal DNA regions. While most trait-associations are consistent between non-phylogenetic and phylogenetic analyses, we find that some trait-associations are dependent on phylogenetic context. Traits consistently associated with invasiveness in these species include weediness, annual life history, and the presence of vegetative reproduction. Capsule dimorphism is only associated with invasiveness in phylogenetically naive analyses and self-compatibility is only associated with invasiveness in tests incorporating phylogeny. We also present a novel phylogenetic comparative analysis, using logistic regression to analyze contrasts where continuous traits (like specific leaf area) might predict a discrete response variable (like invasiveness). This method has lower power than non-phylogenetically corrected logistic regression, and the need for more sophisticated logistic regression comparative methods is suggested. Incorporating phylogeny in tests of trait associations can help us make stronger inferences about causal relationships, and can help us to detect trait associations that might otherwise be obscured by phylogenetic signal.