Taphonomic effects of faunal impoverishment and faunal mixing

Abstract

Tropical habitats are complex, and in areas of topographic or climatic variability they may vary greatly over short distances. Accumulations of animal remains may and usually do represent more than one habitat type. Moreover, many animal faunas from archaeological and/or palaeontological sites differ to some degree from the source fauna(s) from which they were derived, and many factors of accumulation and preservation may act to alter faunal composition, in particular processes of selection or rejection of species. This may result in faunal bias, either by reduction in species richness, when certain elements of the fauna are selected against, or by increase in richness when animals from more than one source are mixed together. Both forms of bias may operate together. The effects of habitat mixing and/or alterations in taxonomic composition in animal faunas are investigated here on a series of mammal assemblages of known habitat in order to investigate the extent to which they may be modified without losing their ecological signature. Ecological data from 19 faunas from a range of tropical African habitats have been ordinated, and habitat-based ordination scores have been calculated by averaging the habitat scores for all species in the habitats. The cumulative scores have then been progressively reduced to 25% of their original values by eliminating either small (< 1 kg) or large (> 90 kg) mammals. In addition, the ordination scores of the same faunas have been combined, either on the basis of equal numbers from pairs of different habitats, or on the basis of one quarter to three quarters of species from different habitats. The ordination method used is weighted averages, which is tested to estimate the significance of the bias and how far the ecological composition of the biased faunas differs from the original, unbiased, faunas. In most cases, faunal assemblages reduced by half still give a clear habitat ordination score, and in many cases the same is true of assemblages reduced to one quarter of the original assemblage. Mixed assemblages are more problematic, but in many cases the ordination score is still clear. Two fossil faunas from Laetoli, Tanzania are then investigated using the Taxonomic Habitat Index, which further averages the ordination scores taxonomically. The mammalian fauna from the Ndolanya beds (∼ 2.6 Ma) is most similar to a faunal mixture of 75% derived from bushland habitats and 25% derived from grassland habitats, suggesting an open bushed grassland habitat at this level; but the reconstruction of the habitat for the fauna from the upper Laetolil beds is most similar to equal mixtures of species derived from woodland and deciduous forest habitats, suggesting a heavily wooded environment with local patches of forest and few open grassland areas.