ABSTRACT

Detection of key factors driving fire frequency, especially in protected areas, is important for effective management of ecosystems. In this study, we used Generalized Linear Models to test the contribution of natural (NDVI, wind speed, dry matter productivity, soil moisture, percentage tree cover, elevation and temperature) and anthropogenic (distance from settlements) factors as predictors of fire frequency in Hwange National Park and adjacent areas. We used the Akaike Information Criterion (AIC) to evaluate the variable contribution to fire frequency. The model results indicated that all the variables that were used contributed significantly to fire recurrence (p < 0.05). Distance from settlements contributed the most to the model whilst dry matter productivity and annual average temperature were second and third respectively. Removal of distance from settlements from the model increased the AIC value to 1411.2 while removal of dry matter productivity and temperature resulted in AICs of 1269.9 and 1265.8 respectively. Results showed that settlements which are found in the vicinity of the protected area influence the recurrence of fires. Findings from this study can be used for strategic fire management and for the development of effective measures to minimize fire recurrence in a protected area.

摘要

检测导致火灾频率的关键因素，特别是在保护区，对于有效管理生态系统非常重要。在这项研究中，我们使用广义线性模型来测试自然（NDVI、风速、干物质生产力、土壤湿度、树木覆盖率、海拔和温度）和人为（与定居点的距离）因素作为火灾频率预测因子的贡献。万基国家公园及邻近地区。我们使用 Akaike 信息准则 (AIC) 来评估变量对火灾频率的贡献。模型结果表明，所有使用的变量都对火灾复发有显着贡献（p < 0.05）。与聚落的距离对模型的贡献最大，而干物质生产力和年平均温度分别位居第二和第三。从模型中去除与定居点的距离将 AIC 值增加到 1411.2，而去除干物质生产力和温度导致 AIC 分别为 1269.9 和 1265.8。结果表明，在保护区附近发现的定居点会影响火灾的再次发生。这项研究的结果可用于战略性火灾管理和制定有效措施以尽量减少保护区内的火灾复发。