PurposeSphaerolecanium prunastri Boyer de Fonscolombe, as a polyphagous pest in wild fruit forests in Xinjiang Uygur Autonomous Region, China, exhibits notable characteristics such as fast reproduction, high reproductive rates, and rapid dissemination. These features highlight its substantial ecological adaptability and competitive edge in the ecosystem. Notably, it poses a severe threat not only to Prunus armeniaca L., commonly known as wild apricot, but also to the ecological environment and biodiversity of the wild fruit forests and local orchards. To obtain a more precise habitat distribution for S. prunastri, species distribution models were established and compared under various types of influencing factors.

MethodsThis study, integrating the R programming language with the Biomod2 package, species distribution models (SDMs) were established to predict the distribution patterns of S. prunastri: one only including climatic variables and the other incorporating climatic variables and host factors.

ResultsThe ecological niche results, incorporating biological variables, were consistent with actual distribution patterns. This approach eliminated the overestimation issue observed when using only climatic variables, leading to higher predictive accuracy. Therefore, the predictions based on the inclusion of host factors were selected for analysis. With the addition of host factors, the current total suitable habitat area was 15.06 × 10⁵ km², mainly distributed near the border with Kazakhstan, the northern foothills of the Tian Shan Mountains, and the lli River Valley. Under the SSP5-85 scenario in 2090s, the total suitable habitat area reached a maximum of 45.31 × 10⁵ km². Under different future climate scenarios, the total suitable habitat area for S. prunastri showed varying degrees of expansion, with significantly improved suitability observed in areas such as the lli Kazakh Autonomous Prefecture, Bortala Mongolian Autonomous Prefecture, Changji Hui Autonomous Prefecture, and Altay Region.

Conclusion The ensemble model comprising seven individual models constructed in this study had an average AUC value of 0.97 and an average TSS value of 0.88, indicating high predictive accuracy. After introducing the host factors to the distribution model, the suitable habitat was more consistent with the actual situation, and the result was more reliable than only considering the climate factor. These predictions offer a scientific basis for the early warning, monitoring, and management of S. prunastri in Xinjiang.