Abstract Ecological niche models (ENM) of species distributions and dispersal patterns are well established in the biological sciences. Their use in paleoanthropological reconstructions of hominin niches is relatively recent, successfully focusing on out of Africa dispersals and human land preference in Europe and Central Asia. These studies have suggested that some of the most important variables for predicting human site use in these regions are moderate annual temperature and rainfall. Here, we used ENM to combine these long-used abiotic predictors of human land preference with landform data (slope) and one potentially important biotic variable (human-carnivore competition quantified using a competition index) in an Australasian test case. We constructed ENMs in the program Maxent to investigate the impact of these abiotic and biotic variables on human land preference patterns in Late Pleistocene Australasia. Though calculated competition across test sites was high, models including this biotic data produced ill-fitting localized models (AUC = 0.695) that relied on mean annual temperature. Large-scale models including solely temperature and rainfall fit well (AUC = 0.84) but are poor predictors of land preference compared to models including slope in this mountainous region (AUC = 0.924) showcasing a discrepancy between accuracy and precision in abiotic models. While the biotic data included in these models was considered unimportant to predictions of human land preference, the inclusion of additional landform data in temperate ENMs should be pursued given the importance of slope as a predictor in large-scale models.

中文翻译：

---

在澳大利亚测试案例中使用生态位模型了解晚更新世人类土地偏好

摘要 物种分布和扩散模式的生态位模型 (ENM) 在生物科学中得到了很好的建立。它们在古人类学重建古人类生态位中的使用是相对较新的，成功地专注于欧洲和中亚的非洲散布和人类土地偏好。这些研究表明，预测这些地区人类场地使用的一些最重要的变量是年温和的气温和降雨量。在这里，我们使用 ENM 将这些长期使用的人类土地偏好的非生物预测因子与地形数据（坡度）和一个潜在的重要生物变量（使用竞争指数量化的人类-食肉动物竞争）在澳大利亚测试案例中结合起来。我们在 Maxent 程序中构建了 ENM，以研究这些非生物和生物变量对晚更新世大洋洲人类土地偏好模式的影响。尽管跨测试站点的计算竞争很高，但包含此生物数据的模型产生了依赖于年平均温度的不合适的局部模型 (AUC = 0.695)。仅包括温度和降雨量的大型模型拟合良好 (AUC = 0.84)，但与包括该山区坡度 (AUC = 0.924) 的模型相比，土地偏好的预测指标较差，这表明非生物模型的准确性和精确度之间存在差异。虽然这些模型中包含的生物数据被认为对人类土地偏好的预测不重要，