In protected areas (PAs), the lack of tourism impact prediction models of vegetation is a shortcoming in PA management. Now, the main question are how recovery can be accelerated, or which ecological factors are associated with the rehabilitation of vegetation density? We aimed to compare the multilayer perceptron (MLP), radial basis function neural network (RBFNN), and support vector machine (SVM) models to predict tourism impact on land vegetation density changes. Three old national parks in Iran with diversity in tourist pressure and ecological condition were selected for analysis. We recorded 12 ecological and tourist variables in 400 sample plots, which are classified by topography, plot soil, and tourist pressure factors. We developed the tourism impact assessment model (TIAM) by MLP, RBFNN, and SVM techniques. Comparing with RBFNN and SVM, the MLP model (TIAM_MLP) is introduced as the most accurate model for vegetation density changes for tourism impact assessment in PAs. The MLP model represents the highest value of R² in training (.969), test (.806), and all datasets (.876). Sensitivity analysis proved that the values of the tourist pressure, soil organic matters, soil moisture, soil porosity, and soil electrical conductivity are respectively as the most significant inputs, which influence TIAM_MLP in PAs. We concluded that habitats with higher organic matter and moisture in the soil would likely tolerate more tourists' pressure. The MLP model, as a tool for PAs managers, is able to predict vegetation density changes under tourism pressure precisely.

中文翻译：

---

利用数据挖掘技术对保护区植被密度的旅游影响评估模型

在保护区（PA）中，缺乏植被的旅游影响预测模型是PA管理中的一个缺点。现在，主要问题是如何加快恢复速度，或者哪些生态因素与植被密度的恢复有关？我们旨在比较多层感知器（MLP），径向基函数神经网络（RBFNN）和支持向量机（SVM）模型，以预测旅游业对土地植被密度变化的影响。选择了伊朗三个游客压力和生态条件各异的老国家公园进行分析。我们在400个样地中记录了12个生态和游客变量，这些样地按地形，样地土壤和游客压力因素分类。我们通过MLP，RBFNN和SVM技术开发了旅游业影响评估模型（TIAM）。与RBFNN和SVM相比，引入_MLP作为最准确的植被密度变化模型，以评估PA中的旅游影响。MLP模型代表训练（.969），测试（.806）和所有数据集（.876）中R ^2的最大值。敏感性分析表明，旅游压力，土壤有机质，土壤湿度，土壤孔隙度和土壤电导率的值分别是最重要的输入，这会影响PA中的TIAM _MLP。我们得出的结论是，土壤中有机质和水分较高的栖息地可能会承受更多游客的压力。作为保护区管理者的工具，MLP模型能够准确预测旅游压力下的植被密度变化。