Mapping the Mangrove Forest Restoration Potential and Conservation Gaps in China Based on Random Forest Model

• DOI: 10.11628/ksppe.2022.25.4.425
• Publication Date: 2022-09-06T23:55:56Z
• Authors: Zhonghua Yu; Wei Li; Shaowei Zhang; Buqing Zhong; Jun Wang; Shi-young Lee; Jaehyuck Choi; Shulin Deng
• Keywords: Restoration Ecology

Background and objective: The area of mangroves is gradually decreasing globally, are already one the most threatened ecosystems. Despite net growth in mangrove areas China, restoration potential still insufficient. This study proposed Random forest model as an excellent data mining method to map based on predicted probability habitat suitability.Methods: We demonstrated vital environmental variables influencing suitability. de-cisive advantages RFM were parsimonious (variables selection), cost-effective (us-ing existing open-source data), accurate (training AUC was 0.89, testing 0.91), highly efficient (fast-training speed); its results had high explanatory power. Here, we first mapped conservation gaps using RFM.Results: showed that temperature important factor suit-ability mangroves. northern limit suitable around 24°44' N. theoretical for 196,566.6 ha (the 32,551.4 ha, medium 164,015.2 ha). 176,264 (Guangdong with 104215.4 Guangxi 65957.5 ha).Conclusion: 24 sites forests nine examples further plan. took example site explanation: how key factors influence suitability use infor-mation guide strategies. can be used a algo-rithm utmost presence-only ecological data, objectively evaluating species distribution, providing scientifically technical planning.