Mapping predicted dung beetle diversity in the lowland tropical forests of Borneo by utilizing indices derived from a satellite image

In tropical regions, assessing biodiversity across expansive areas through direct data collection presents significant challenges. The advancement of remote sensing techniques for mapping predicted biodiversity offers a viable solution. To map predicted dung beetle diversity and its associated ecological functions using satellite images, we selected optimal multivariate regression models to predict overall species richness and abundance, forest species richness and abundance, biomass (logarithmic dry weight) associating ecological functions of the beetle, Simpson’s diversity index (1/D), and forest community integrity (measured as the score on axis 1 of nonmetric multidimensional scaling ordination analysis), utilizing the Akaike information criterion (AIC). The explanatory variables were extracted from a SPOT-5 satellite image utilizing varying grid cell configurations up to 290 m². The objective variables consisted of dung beetle collection data from intact, burnt, or artificially degraded natural forests; plantation forests; and grasslands within a 20 × 20 km area near Balikpapan, Indonesia. When the predicted and observed values were compared, the coefficients of determination of the linear model exhibited marked significance, with all relations having an <i>R</i>_<i>2</i> exceeding 0.25. Mapping these predicted values revealed that the areas with high values generally coincided with the large intact natural forests and the remnants surrounding burnt areas, except for overall and forest species abundances, which displayed higher values along the burnt ridges surrounding the intact natural forests. These findings demonstrate that the mapping of predicted dung beetle diversity and its associated ecological functions within tropical regions can be effectively achieved through the use of satellite images.

在热带地区，通过直接数据采集评估广阔区域的生物多样性面临巨大挑战。用于绘制预测生物多样性分布图的遥感技术的进步，为这一问题提供了可行的解决方案。为利用卫星图像绘制蜣螂预测多样性及其相关生态功能分布图，我们采用赤池信息准则（Akaike Information Criterion, AIC）筛选最优多元回归模型，以预测总物种丰富度与多度、森林物种丰富度与多度、与蜣螂生态功能相关的生物量（对数干重）、辛普森多样性指数（1/D），以及森林群落完整性（通过非度量多维尺度排序分析（nonmetric multidimensional scaling ordination analysis）的第一轴得分衡量）。解释变量提取自SPOT-5卫星图像，采用了最大为290平方米的不同网格单元配置。目标变量包括印度尼西亚巴厘巴板附近20×20公里区域内，来自原始、火烧或人工退化天然林、人工林及草地的蜣螂采集数据。对比预测值与观测值时，线性模型的决定系数表现出显著意义，所有关系的R²均超过0.25。绘制这些预测值分布图发现，高值区域通常与大片原始天然林及火烧区周边的残余植被重合，但总物种多度与森林物种多度除外——这两类多度在原始天然林周边的火烧山脊区域表现出更高值。这些研究结果表明，利用卫星图像可有效实现热带地区蜣螂预测多样性及其相关生态功能的分布图绘制。