Toxoplasma gondii infection in wildlife in China (1985-2024): A systematic review and meta-analysis with machine learning.

Abstract

The globally distributed zoonotic protozoan parasite Toxoplasma gondii presents escalating conservation and public health challenges through expanding wildlife reservoir infections. This systematic review with meta-analysis, integrated with machine learning approaches (Random Forest and SHAP analysis), aims to quantify aggregated prevalence, identify key epidemiological drivers, and map infection risks in Chinese wildlife. Across 55 studies published before 31 December 2024, the pooled T. gondii prevalence was (14.47 %, 95 % CI: 11.61 %-17.88 %), with striking variance across taxa (Felidae: 34.29 %, CI:30.86-37.72; Ochotonidae: 3.96 %, CI:0.16-7.76) and sample origins (zoo-originating specimens: 28.55 %, CI:26.50-30.69). Using restricted maximum likelihood meta-regression and RF modeling, we identified sampling period, sample origin, detection method, and species class as critical epidemiological drivers. Spatial projections further identify areas of highest current infection risk, with Jilin (22.77 %) and eastern/southern China emerging as high-risk regions based on current static variables. Our findings validate wildlife's sentinel role in toxoplasmosis transmission ecology, highlight the strong influence of anthropogenic captivity on infection dynamics, and demonstrate the value of machine learning for disentangling heterogeneous epidemiological patterns-urging standardized surveillance protocols to address methodological heterogeneity across existing studies.