Accurate Identification of Native Asian Honey Bee Populations in Jilong (Xizang, China) by Population Genomics and Deep Learning.

The Jilong Valley, situated in Rikaze, Xizang, China, is characterized by its complex topography and variable climatic conditions, providing a suitable habitat for Fabricius, 1793. To facilitate the conservation of germplasm resources and maintain genetic diversity, it is imperative to elucidate the population structure and lineage differentiation of within this ecologically distinct region. In this study, we collected specimens from 12 geographically disparate locations across various altitudinal gradients within the Jilong Valley, and also integrated publicly available sequencing data of from various regions across mainland Asia. In total, our analysis encompassed sequencing data from 296 individuals. Population structure analyses based on SNP data revealed that in Jilong represents a genetically distinct population that differs markedly from other regional populations in terms of genetic lineage, although its subspecies identity remains to be confirmed. Through screening based on F values, we identified SNP loci that contribute significantly to distinguishing between Jilong and non-Jilong . Using these loci, the convolutional neural network model TraceNet was trained, which demonstrated specific recognition capabilities for Jilong versus non-Jilong . This further confirmed the universality and efficiency of TraceNet in identifying honey bee lineages. These findings contribute valuable insights for the identification and conservation of germplasm resources in specific geographical regions.