Abstract Context Populations of native mammals are declining at an alarming rate in many parts of tropical northern Australia. Fire regimes are considered a contributing factor, but this hypothesis is difficult to test because of the ubiquity of fire. Aims This preliminary study investigated relative abundance and richness of small mammals on a gradient of fire regimes in the Uunguu Indigenous Protected Area (north Kimberley, Australia). Methods Species were sampled using 40 unbaited camera traps, positioned for a year on 20 transects crossing the rainforest–savanna boundary at locations with comparable environment and geology but varying fire history. The relative importance of the factors ‘fire frequency’, ‘late dry season fire frequency’, ‘time since burnt’ and ‘vegetation type’ as predictors of the number of small mammal species and detections was tested using Spatial Generalised Linear Mixed Models to account for spatial autocorrelation. Key results Nine species of small mammals were observed. Mammals were more abundant and diverse in locations with low overall fire frequency, which was a better predictor than late dry season fire frequency or time since burnt. The model including fire frequency and vegetation explained the highest proportion of total variation in mammal diversity (R2 = 42.0%), with most of this variation explained by fire frequency alone (R2 = 40.5%). The best model for number of detections (R2 = 20.9%) included both factors. Conclusions In the north Kimberley, small mammals are likely to be more abundant and diverse in areas with low fire frequency. Implications This natural experiment supports the theory that frequent fires are contributing to the decline of small mammals observed across northern Australia.

中文翻译：

---

与澳大利亚金伯利北部经常被烧毁的雨林 - 稀树草原马赛克相比，小型哺乳动物的多样性很少见

摘要背景在澳大利亚北部热带地区的许多地区，本地哺乳动物的数量正在以惊人的速度下降。火势被认为是一个促成因素，但由于火无处不在，这一假设难以检验。目标 这项初步研究调查了 Uunguu 土著保护区（澳大利亚金伯利北部）在不同火势条件下小型哺乳动物的相对丰度和丰富度。方法 使用 40 个无诱饵的相机陷阱对物种进行采样，这些陷阱在跨越雨林-稀树草原边界的 20 个横断面上放置了一年，这些横断面位于环境和地质相似但火灾历史不同的位置。“火灾频率”、“旱季晚期火灾频率”等因素的相对重要性 使用空间广义线性混合模型测试“自燃烧后的时间”和“植被类型”作为小型哺乳动物物种和检测数量的预测因子以解释空间自相关。主要结果 观察到九种小型哺乳动物。哺乳动物在总体火灾频率较低的地区更加丰富多样，这比旱季晚期火灾频率或自燃烧时间更能预测。包括火灾频率和植被的模型解释了哺乳动物多样性总变异的最高比例 (R2 = 42.0%)，其中大部分变化仅由火灾频率解释 (R2 = 40.5%)。检测次数的最佳模型 (R2 = 20.9%) 包括这两个因素。结论 在金伯利北部，火灾频率低的地区小型哺乳动物可能更加丰富多样。