Domestication simplifies genomes. Cultivated crops have less genetic diversity than their wild ancestors. Domesticated animals lose alleles with each selective bottleneck. The pattern is so consistent it has a name — the domestication syndrome — and a mechanism: strong selection on a few visible traits erodes the variation in everything else.
O'Donnell, Rezende, Vernadet, Snirc, and Ropars at Université Paris-Saclay (Genome Biology and Evolution, 2025) found the opposite in fungi. Domesticated strains of Aspergillus and Penicillium — the fungi used to make cheese, soy sauce, sake — harbor significantly more Starship elements than their wild relatives. Starships are giant transposable elements, up to 700 kilobases, that carry dozens to hundreds of cargo genes and move between species via horizontal transfer. The domesticated strains are not simplified. They are genomically richer in the specific machinery that enables rapid adaptation.
The Starships in domesticated strains are enriched for genes involved in adaptation to food environments — genes for secondary metabolite production, stress tolerance, nutrient utilization. Human cheesemakers selected for flavor. The genome accumulated architecture for evolvability. The cheesemakers could not see the Starships, did not intend to select for them, and would not have understood what they were selecting for if told. But every generation that preferentially cultured the strain with better flavor was, at the genomic level, preferentially propagating the strain with more mobile genetic cargo.
The structural insight: selection for a product inadvertently selects for the genomic architecture that generates products. Flavor is a chemical output. Chemical novelty comes from new gene combinations. New gene combinations come from horizontal transfer. Horizontal transfer comes from Starships. So selection for flavor is, at several removes, selection for Starship content. The visible trait and the invisible architecture are coupled through a causal chain the selector cannot see.
This inverts the standard domestication narrative. Wild bananas have genes for disease resistance that cultivated bananas lost. Wild fungi have fewer tools for rapid adaptation than the fungi we domesticated. The difference is what we selected on. When humans select on physical traits (size, yield), they narrow the genome. When they select on chemical traits (flavor, fermentation), they inadvertently widen the genome's capacity to generate novelty — because the chemical output and the genomic mobility are downstream of the same architecture.