717q2003 The Society for the Study of Evolution. All rights reserved.Evolution, 57(4), 2003, pp. 717–745TESTING FOR PHYLOGENETIC SIGNAL IN COMPARATIVE DATA:BEHAVIORAL TRAITS ARE MORE LABILESIMONP. BLOMBERG,1THEODOREGARLAND,JR.,1,2ANDANTHONYR. IVES3,41Department of Biology, University of California, Riverside, California 925212E-mail: [email protected] of Zoology, University of Wisconsin, Madison, Wisconsin 537064E-mail: [email protected]. The primary rationale for the use of phylogenetically based statistical methods is that phylogenetic signal,the tendency for related species to resemble each other, is ubiquitous. Whether this assertion is true for a given traitin a given lineage is an empirical question, but general tools for detecting and quantifying phylogenetic signal areinadequately developed. We present new methods for continuous-valued characters that can be implemented witheither phylogenetically independent contrasts or generalized least-squares models. First, a simple randomization pro-cedure allows one to test the null hypothesis of no pattern of similarity among relatives. The test demonstrates correctType I error rate at a nominala50.05 and good power (0.8) for simulated datasets with 20 or more species. Second,we derive a descriptive statistic, K, which allows valid comparisons of the amount of phylogenetic signal across traitsand trees. Third, we provide two biologically motivated branch-length transformations, one based on the Ornstein-Uhlenbeck (OU) model of stabilizing selection, the other based on a new model in which character evolution canaccelerate or decelerate (ACDC) in rate (e.g., as may occur during or after an adaptive radiation). Maximum likelihoodestimation of the OU (d) and ACDC (g) parameters can serve as tests for phylogenetic signal because an estimate ofd or g near zero implies that a phylogeny with little hierarchical structure (a star) offers a good fit to the data.Transformations that improve the fit of a tree to comparative data will increase power to detect phylogenetic signaland may also be preferable for further comparative analyses, such as of correlated character evolution. Applicationof the methods to data from the literature revealed that, for trees with 20 or more species, 92% of traits exhibitedsignificant phylogenetic signal (randomization test), including behavioral and ecological ones that are thought to berelatively evolutionarily malleable (e.g., highly adaptive) and/or subject to relatively strong environmental (nongenetic)effects or high levels of measurement error. Irrespective of sample size, most traits (but not body size, on average)showed less signal than expected given the topology, branch lengths, and a Brownian motion model of evolution (i.e.,K was less than one), which may be attributed to adaptation and/or measurement error in the broad sense (includingerrors in estimates of phenotypes, branch lengths, and topology). Analysis of variance of log K for all 121 traits (from35 trees) indicated that behavioraltraits exhibit lower signal than body size, morphological, life-history, or physiologicaltraits. In addition, physiological traits (corrected for body size) showed less signal than did body size itself. For treeswith 20 or more species, the estimated OU (25% of traits) and/or ACDC (40%) transformation parameter differedsignificantly from both zero and unity, indicating that a hierarchical tree with less (or occasionally more) structurethan the original better fit the data and so could be preferred for comparative analyses.Key words.Adaptation, behavior, body size, branch lengths, comparative method, constraint, physiology.Received March 29, 2002. Accepted November 27, 2002.A great triumph of biology has been the demonstration thatorganisms are descended from common ancestors and henceare related in a hierarchical fashion (Mayr 1982). An obser-vation of almost equal importance is that phylogeneticallyrelated organisms tend to resemble each other for most as-pects of the phenotype (e.g., hummingbirds look like hum-mingbirds, elephants look like elephants). We term this re-semblance phylogenetic ‘‘signal’’ to avoid such terms as phy-logenetic ‘‘inertia’’ or ‘‘constraint,’’ which areinconsistentlydefined, both conceptually and operationally (e.g., see An-tonovics and van Tienderen 1991; McKitrick 1993; Wagnerand Schwenk 2000; Orzack and Sober 2001; Reeve and Sher-man 2001) and whose original use was rather different fromcurrent use (Blomberg and Garland 2002). As argued else-where (Blomberg and Garland 2002), we believe that theevolutionary processes implied by phylogenetic inertia andconstraint are difficult, if not impossible, to estimate fromcomparative data alone (but see Hansen 1997; Orzack andSober 2001; Schwenk and Wagner 2001). Therefore, we usephylogenetic signal simply to describe a tendency (pattern)for evolutionarily related organisms to resemble each other,with no implication as to the mechanism that might causesuch resemblance (process). This use of phylogenetic signalis consistent with recent usage in systematic biology (e.g.,Hillis and Huelsenbeck 1992) and is also similar to the ‘‘phy-logenetic effect’’ of Derrickson and Ricklefs (1988) and the‘‘phylogenetic conservatism’’ of Ashton (2001). We do notuse ‘‘phylogenetic conservatism,’’ however, because it seemsto connote less change than might be expected from the phy-logenetic structure of the taxa in question; as argued below,the existence of phylogenetic signal does not require the ex-istence of such processes as that may imply. We did not usethe ‘‘phylogenetic correlation’’ of Gittleman et al. (1996a)because they used statistical approaches different from thosepresented here and because our methods do not involve cor-relation per se.A crucial point that has not been sufficiently appreciatedin the literature is that phylogenetic signal will occur to someextent under most simple, stochastic models of character evo-lution along a tree with any amount of hierarchical structure.Under such models as Brownian motion, evolutionary chang-es are simply added to values present in the previous gen-eration or at the previous node on a phylogenetic tree. Thus,members of lineages that have only recently diverged willnecessarily (on average) tend to be similar, as compared withmore distantly related lineages. This effect carries on to the718SIMON P. BLOMBERG ET AL.values for
View Full Document
Unlocking...