PALEOHABITAT RECONSTRUCTION FOR SOUTH AMERICAN CENOZOIC PALEOFAUNAS: ARE THERE APPROPRIATE MODERN ANALOGS?
Three ecomorphological variables were used for the EDA (body mass, locomotor style, and dietary preference), each subdivided into a number of categories (twelve, six, and five, respectively). A discriminant model was constructed to distinguish among 34 modern Neotropical mammal faunas based on the percentages of taxa in seven ecomorphological categories (or combinations thereof); this model resulted in 82.4% correct classification of modern faunas. Ecomorphological data for Santa Cruz and Tinguiririca taxa were compiled and the Neotropical discriminant model was used to infer the habitats of these paleofaunas. In both cases, the faunas were unequivocally classified as open habitats (posterior probabilities = 1.00), though centroid distances far exceeded those of the modern faunas (conditional probabilities ≤ 0.001). These results suggest that both Santa Cruz and Tinguiririca were structured very differently from modern Neotropical faunas.
Two potential explanations for the great dissimilarity between modern and fossil South American faunas include poor preservation of micromammals (< 500g) in fossil assemblages and/or relative paucity of large mammals (due to megafaunal extinction) in modern Neotropical faunas. To circumvent these two biases, comparisons were made with African macromammal faunas (i.e., mammals ≥ 500 g). The discriminant model classified 80% of modern African faunas correctly, yet classified Tinguiririca as a mixed habitat (slightly outside the range of modern faunas) and Santa Cruz as a closed (forest) habitat (but far outside the range of modern forests). These conflicting interpretations reinforce the unique structure of South American paleofaunas and indicate that paleohabitat reconstructions can be highly dependent on the comparative dataset used.