Accuracy of environmental reconstruction based on a blind test of micromammal evidence from East Africa.

Paleoenvironmental reconstructions based on mammalian faunas provide the contextual basis for understanding evolutionary events in many branches of paleobiology. This paper presents the first blind test of an environmental reconstruction using extant African small mammals. We collected 1216 specimens representing a minimum of 332 individuals from Ngofila 3, Tanzania, then conducted a blind analysis of the taxonomic and community composition. We found agreement between the observed habitat and the predicted habitat along eight aspects of the environment: (1) mean annual precipitation, (2) surface water availability, (3) soil type, (4) vegetation, (5) land use, (6) topography, (7) habitat heterogeneity, (8) temperature. Of these, the first six had high agreement between the observed and predicted descriptions, two (topography and habitat heterogeneity) had moderate agreement and the last, (temperature) was discussed in the observed habitat description but not addressed in the predicted description and could not be compared. The high level of agreement demonstrates that paleoenvironmental reconstruction can accurately portray a known habitat. In addition, we suggest factors that should be considered in a standard method of habitat description, which would improve comparability of results between different analyses. [ABSTRACT FROM AUTHOR]