The Quantitative Assessment of Phylogenetic Constraints in Comparative Analyses: Sexual Dimorphism in Body Weight Among Primates

Abstract

We have presented a formal model for the quantitative analysis of phylogenetic and specific effects on the distribution of trait values among species. Total trait values are divided into phylogenetic values, inherited from an ancestral species, and specific values, the result of independent evolution. This allows a quantitative assessment of the strength of the phylogenetic inertia, or burden, displayed by a character in a lineage, so that questions concerning the relative importance of phylogenetic constraints in evolution can be answered. The separation of phylogenetic from specific effects proposed here also allows phylogenetic factors to be explicitly included in cross-species comparative analyses of adaptation. This solves a long-standing problem in evolutionary comparative studies. Only species' specific values can provide information concerning the independent evolution of characters in a set of related species. Therefore, only correlations among specific values for traits may be used as evidence for adaptation in cross-species comparative analyses. The phylogenetic autocorrelation model was applied to a comparative analysis of the determinants of sexual dimorphism in weight among 44 primate species. In addition to sexual dimorphism in weight, mating system, habitat, diet, and size (weight itself) were included in the analysis. All of the traits, except diet, were substantially influenced by phylogenetic inertia. The comparative analysis of the determinants of sexual dimorphism in weight indicates that 50% of the variation among primate species is due to phylogeny. Size, or scaling, could account for a total of 36% of the variance, making it almost as important as phylogeny in determining the level of dimorphism displayed by a species. Habitat, mating system, and diet follow, accounting for minor amounts of variation. Thus, in attempting to explain why a particular modern primate species is very dimorphic compared to other primates, we would say first because its ancestor was more dimorphic than average, second because it is a relatively large species, and third because it is terrestrial, polygynous, and folivorous.

Comparative methods for studying primate adaptation and allometry

“…10 Phylogenetic autocorrelation methods can be used to assess whether a trait is correlated with phylogeny and can generate "phylogeny-free" values for comparative studies. 92,108 This approach can be implemented in the computer package COMPARE or by using the program Phylogenetic Autocorrelation (ftp://ftp.math.utk.edu/pub/luh/PA.hqx). Grafen's 54 Phylogenetic Regression program may be useful for some comparative studies (http:// users.ox.ac.uk/ϳgrafen/phylo/).…”

Padrões e processos na evolução do tamanho do corpo em carnívoros (Mammalia) da América do Sul

“…Esse padrão foi também detectado para outros grupos de organismos (Cheverud et al, 1985;Gittleman et al, 1996). Gittleman et al (1996) encontraram que cerca de 76% na variação do tamanho do corpo pode ser explicada pela filogenia, para 39 espécies de carnívoros de todo o mundo (incluindo algumas da América do Sul).…”

“…A principal implicação desses valores elevados de inércia filogenética no caráter é que as correlações entre caracteres ecológicos e de história de vida com o tamanho do corpo, ao longo dessas espécies, podem estar claramente viesadas pela estrutura filogenética nos dados. Nesses casos, os valores de S, em substituição aos valores totais (T), devem ser utilizados na estimativa das correlações a fim de corrigir esse viés (Cheverud et al, 1985;Martins, 1996).…”

“…Foi realizada então uma análise de autoregressão filogené-tica (Cheverud et al, 1985; 1990; Martins 1996), que objetiva particionar o valor total de um caráter (T) em um componente filogenético (P), que expressa a parte da variação compartilhada entre as espécies, e em um componente específico (S), que contém a parte da variação única (equivalente a uma autapomorfia) de cada espécie, de modo que:…”

“…A principal implicação desses padrões é que o teste das correlações entre as variáveis torna-se mais complexo, pois é necessário corrigir os efeitos da inércia filogenética. A probabilidade de erro Tipo I (aceitar a correlação quando ela de fato não existe) passa a ser consideravelmente maior nos casos de forte inér-cia filogenética, pois a correlação entre os caracteres pode indicar apenas covariação de ambos com a filogenia (Felsenstein, 1985;Cheverud et al, 1985;Gittleman & Luh, 1992;Martins, 1996).…”