Genome-wide architecture of prolonged starvation adaptation in experimentally evolved Drosophila and comparative enrichment in human orthologs

Abstract Long-term starvation stress represents a strong evolutionary constraint across taxa, yet the genetic architecture underlying adaptation to sustained nutrient deprivation remains poorly resolved. We experimentally evolved Drosophila melanogaster under starvation stress for 60 generations, maintaining four starvation-selected populations and four matched controls, and then performed whole-genome resequencing. Starvation-selected populations exhibited extended survival and pronounced genome-wide restructuring, including the expansion of low-heterozygosity regions, reduced nucleotide diversity, and recurrent sweep signatures across replicates. Drift-aware allele-frequency modeling identified 3,578 single-nucleotide polymorphisms (SNPs) whose shifts exceeded neutral expectations, indicating widespread, parallel responses to sustained starvation. Mitochondrial pathways emerged as prominent targets: nuclear-encoded mitochondrial genes were significantly enriched among low-diversity, sweep-associated, and drift-exceeding loci, and the mitochondrial origin of replication harbored a sharply differentiated variant, suggesting a mito–nuclear component of starvation adaptation. Comparative analyses further showed that human orthologs of starvation-responsive fly genes were enriched for highly differentiated variants in selected 1000 Genomes populations, with regulators of TOR/S6K signaling recurring among loci in the extreme tails of population differentiation. Together, these results define a replicate-convergent and functionally coherent genomic response to prolonged starvation in Drosophila, comprising regional signatures of linked selection and distributed allele-frequency shifts exceeding neutral expectations, and connect laboratory selection in flies to population differentiation in human orthologs. Competing Interest Statement The authors have declared no competing interest.