Abstract:

【Background】The Luopanshan pig, a rare indigenous breed native to Chongqing, is recognized as a key genetic resource at both national and regional levels. However, due to factors such as changes in breeding structures and outbreaks of African swine fever, its population has declined sharply, leading to a significant loss of genetic diversity. Urgent conservation measures are required to prevent its extinction. While in situ conservation remains the primary strategy, establishing ex situ conservation farms serves as a crucial supplementary approach to mitigate risks associated with habitat degradation and disease outbreaks. 【Objective】In accordance with the Measures for the Administration of Conservation Farms, Protected Areas, and Gene Banks for Livestock and Poultry Genetic Resources issued by the Ministry of Agriculture and Rural Affairs, ex-situ conservation farms should be established in locations where the natural ecological conditions closely match those of the breed’s original habitat. To facilitate the formulation of evidence-based strategies for the ex-situ conservation of the Luopanshan pig, this study aimed to identify potential areas within Chongqing Municipality that were environmentally suitable for its survival and reproduction. The findings were expected to provide a scientific basis for future conservation farm site selection and to support the sustainable development of large-scale breeding programs. 【Method】Utilizing 105 georeferenced occurrence records of Luopanshan pigs collected through field surveys in Chongqing, a Random Forest machine learning model via the scikit-learn library was employed to assess the relative importance of 16 environmental and anthropogenic variables, including soil properties (Available Phosphorus, Available Potassium, Available Nitrogen, Cation Exchange Capacity, Soil Organic Matter and Soil pH), land surface temperature, precipitation, digital elevation model, and STREAM. SHapley Additive exPlanations (SHAP) were used to quantify the marginal contribution of each variable to the model's predictions. Pearson correlation coefficients (PCC) were calculated to evaluate multicollinearity among variables, and variance inflation factors (VIF) were applied for further collinearity diagnostics. The refined set of variables and species occurrence data were then input into the Maximum Entropy Model (MaxEnt) model to predict habitat suitability and identify key environmental factors influencing the distribution of Luopanshan pigs. 【Result】The MaxEnt model achieved a high predictive performance, with an average Area Under the Curve (AUC) value of 0.995, indicating excellent model accuracy. The primary environmental factors influencing the habitat suitability of wild boars in Luopanshan, Chongqing, were precipitation (39.7%), pH (24.1%), available phosphorus(18.9%), land surface temperature (6.2%), and distance to buildings(5.7%), with a total contribution of 94.6%. Secondary environmental factors include available potassium(1.9%), digital elevation model(1.4%), STREAM(1.2%), cation exchange capacity(0.8%), and available nitrogen(0.1%), contributing 5.4% in total. Suitable habitats for Luopanshan pigs were predominantly scattered across Tongnan, Hechuan, Dazu, and Rongchang districts, encompassing a total area of 4 217.59 km², which represented 5.1% of Chongqing's land area. Among these, highly suitable habitats covered 189.44 km², notably in Xingsheng, Wofu, and Tangba towns of Tongnan District; Gaoping, Sanqu, Zhuxi, and Youting towns of Dazu District; Fenggao Subdistrict and Wanling Town of Rongchang District; Weixin Town of Tongliang District; and Chashan Zhuhai Subdistrict of Yongchuan District. Moderately suitable habitats spanned 554.97 km², while low suitability areas covered 1 034.78 km², and marginally suitable areas accounted for 2 438.39 km². 【Conclusion】This study provided a scientific basis for the strategic planning of ex situ conservation farms for Luopanshan pigs in Chongqing. The methodology and findings also offered a reference framework for habitat suitability assessments and conservation management of other indigenous and rare livestock breeds.

Key words: Luopanshan pig, MaxEnt model, prediction of suitable habitat distribution, Chongqing City