Climate, environment, and competition influenced human land preference in late pleistocene Australasia

Ecological niche models (ENM) have long been used in the biological sciences to predict species distributions and dispersal patterns, but this method remains relatively new to paleoanthropological reconstructions of hominin niches. Hindcasted predictions of hominin niche space in Europe and Central Asia have shown that key variables for predicting site preference are temperature, rainfall, and access to fresh water. However, recent work has demonstrated that models utilizing only abiotic climate factors may not provide a comprehensive picture of how a species utilizes its habitat. These studies have called for the inclusion of biotic (species-interaction) data to construct more comprehensive models. Here, for the first time, we calculate fossil estimates of human-carnivore competition (competition index) from five Late Pleistocene sites in Laos and Vietnam to be included in ENM analyses of human land preference. We also take traditionally used environmental predictors and apply them to Australasian landscapes to unpack human land preference during the Last Glacial Maximum (LGM) in a long-standing temperate space and then create mixed models combining abiotic and biotic data at both broad and localized scales. We validate ENM models using known fossil human occupation sites (n = 20) to construct human land preference maps in an area that retained tropical and temperate refugia during even the harshest glacial periods. Sources of assumptions and error become apparent in a temperate context – primarily interpreting ENM output using a curated set of variables. Only “distance to freshwater” remains an important traditional predictor variable in this temperate context. Because of this region’s longstanding temperate ecosystem, we also included terrain (slope) into models as a “non-traditional” variable that may be important in this mountainous region. Though competition across test sites was high, biotic interactions produced vague localized models, suggesting the need for additional species interaction data. Despite well-fitting models (AUC = 0.891, AUC = 0.924), overall land preference estimates are too broad to be of use for in depth evidence of human dispersal routes in this region. Therefore, further analysis should focus on additional non-traditional environmental variables that may disproportionally affect human movement in temperate spaces (e.g., vegetation type and cover, sea-crossings, large-scale species interactions, etc.).U of I OnlyAuthor requested U of Illinois access only (OA after 2yrs) in Vireo ETD syste