Aim: Local environmental selection and lineage admixture have long been accepted as important adaptive mechanisms in adjusting widespread taxa to new environments. We studied a pond skater, Gerris latiabdominis, to explore the relative roles played by these two mechanisms in its process of adaptation to heterogeneous landscapes. Location: East Asia. Taxon: Gerris latiabdominis Miyamoto, 1958. Methods: Mitochondrial (COI, COII) and nuclear (ddRAD-seq) markers were sequenced from 202 individuals. Phylogenetic and phylogeographical analyses were conducted to reveal the population genetic structure. The demographical history was simulated by approximate Bayesian computation (ABC). The single-nucleotide polymorphisms (SNPs) that may have been subjected to natural divergent selection among the populations were assessed, and the recent migration rates were estimated. Finally, shifts in suitable habitat from the last interglacial (LIG) to the present were predicted through ecological niche modelling (ENM). Results: We found population structures inferred from both mitochondrial and nuclear genomes to be almost coincident and to correspond to temporal or spatial heterogeneity. The early Pliocene splitting event between the northern (NO) and southeastern (SE) groups is consistent with the ‘arid belt hypothesis’, followed by local environmental selection. Approximate Bayesian computation and gene flow estimation strongly suggested that the southwestern (SW) groups originated from a lineage admixture event between the SE and western (WE) groups and subsequently underwent ongoing gene flow from the SE group during the Pleistocene. The mito-nuclear discordance pattern in the SW group was probably due to the ancient mitochondrial differentiation. Main conclusion: Our results suggest that both local environmental selection and lineage admixture acted as significant mechanisms of G. latiabdominis adaptation to heterogeneous landscapes in East Asia. Furthermore, this study showed that conclusions based exclusively on mitochondrial data might mislead taxonomic, phylogenetic and phylogeographical studies.