Table 1 Ecological and methodological variables tested to understand the magnitude of trait effects on historical dispersal rates across clades

From: Body size and life history shape the historical biogeography of tetrapods

Explanatory variables

Hypotheses relating variables to the magnitude of trait effects in historical dispersal rates

Influence on the magnitude of body size effects

Influence on the magnitude of life history effects

Ecological variables

Variability of body size within clade

Traits with low variability rarely generate large differences in dispersal potential between species110,111

  

Variability of life history within clade

 

coef −0.09 (P = 0.01)

Number of dispersal events per lineage

Proxy for strength of biogeographic barrier.

Too few dispersal events decrease statistical power (a problem of small sample size112), too many dispersal events indicate weak barriers113 and thus reduced potential for traits to influence dispersal. This could either lead to a quadratic effect (if both hypotheses are supported) or a linear effect (if only one of them is confirmed)

coef −0.03 (P = 0.17)

coef −0.14 (P < 0.001)

(Number of dispersal events per lineage)²

coef −0.04 (P = 0.08)

coef −0.05 (P = 0.07)

Proportion of past oceanic dispersal

Continental dispersal might be less selective than oceanic dispersal

 

coef −0.11 (P = 0.05)

(Proportion of past oceanic dispersal)²

Oceanic and continental barriers might filter species differently6,114, which could confound the results when both types of dispersal are combined

 

coef 0.07 (P = 0.08)

Methodological variables

Number of species within clade

Too few species may limit statistical power (a problem of small sample size112)

  

Trait data coverage

Little available trait data makes it difficult to correctly identify species’ life histories, and may obscure trait–dispersal relationships

  

Average node resolution

Phylogenetic uncertainty might affect estimations of biogeographic histories

  

Crown age

Phylogenetic scale can affect biogeographic estimations; particularly estimations of deep nodes may be biased which might lead to decreased trait effects with increasing clade age115,116

  

Biogeographic base model

Differences in types of vicariance allowed at a node may introduce variation in dispersal events inferred

  

Full model-adjusted R²

  

5%

27%

  1. The hypothesis underlying each variable is described. Effect size (standardized coefficient (coef)) and P values are reported if selected in the final models (stepwise variable selection procedure on multivariate regressions, no adjustments for multiple comparisons were made). Trait data coverage was not tested for body size–dispersal relationships since body size was available for all species. Variability of body size within clade was not tested for the effect magnitude of life history, and inversely, variability of life history within clade was not tested for the effect magnitude of body size since we do not assume a causal link between these variables and effects. Variables with a superscript 2 were included to test for quadratic effects. Variables with a P value equal or lower than 0.05 are highlighted in bold.
Back to article page