Advantages and limitations in the use of extant xenarthrans (Mammalia) as morphological analogues for paleobiological reconstruction.
Extant species of Xenarthra represent a severely restricted sample of the total diversity achieved by the group. Given their shared history, the extant representatives of the three major groups of xenarthrans (Cingulata, Folivora, and Vermilingua) provide a valuable basis for paleobiological inference. However, many extinct taxa are morphologically so dissimilar from their extant relatives that they suggest very different ways of life. In these cases, extinct forms do not have modern models within the group and the application of a simplistic and strict approach can produce nonsensical reconstructions. In this contribution, we evaluate the limitations of the use of extant xenarthrans as morphological models for paleobiological reconstructions. A database of linear dimensions of the appendicular skeleton of extant and extinct xenarthrans and other mammals (marsupials, carnivorans, rodents, primates, perissodactyls, artiodactyls, and proboscideans) was constructed. Exploratory analyzes were performed on general morphometric similarity between existing and extinct xenarthrans (PCA) and the accuracy of body mass estimates of extinct xenarthrans based on their close relatives and other mammals (simple and multiple linear regressions) were tested. Extinct xenarthrans occupy similar relative positions in the morphospaces as extant mammals other than their closest relatives. Most allometric equations, particularly those based only on xenarthrans, produced remarkable underestimates. This can be explained by dimensional differences (up to four orders of magnitude) and shape differences between most of the extinct and extant xenarthrans. This does not invalidate actualism and the use of analogues, but suggests the need to apply other approaches, such as mechanics, that address formfunction relationships but are not necessarily based on known biological comparators.