Alternatives to partial Mantel test in the study the enviromental factors shaping morphological variation
PEREZ S.I.; FELIZOLA DINIZ-FILHO JA; BERNAL V.; GONZALEZ P.
JOURNAL OF HUMAN EVOLUTION
ACADEMIC PRESS LTD-ELSEVIER SCIENCE LTD
Año: 2010 vol. 59 p. 698 - 703
The results obtained by partial Mantel tests should be treatedwith caution because these tests can sometimes remove the effectone is trying to detect, including the influence of environmentalfactors. Spatial regression techniques, on the other hand, could beused to provide more accurate statistical estimates of the associationbetween morphological and ecological variables. The spatialand phylogenetic regression techniques are not new to interspecificresearch (Ives and Zhu, 2006; Bini et al., 2009), but they have beenunderused in biological anthropology. In our view, spatial regressiontechniques are more flexible and they can incorporate similarityamong populations (i.e., autocorrelation) as well as broadscalegeographic trends (i.e., spatial dependence) into the models,depending on how the data are modeled. We do not claim thatprevious studies aiming to test ecological factors are misguided, butwe wonder if the use of more explicit spatial regression modelscould also be used to study the importance of ecological factorsdriving morphological variation among human populations