We investigated fitness in natural populations of a managed solitary bee Osmia ribifloris Cockerell (Hymenoptera: Megachilidae) from sites separated from 400 to 2,700 km. Parental wild bees originated in central Texas (TX), central-northern Utah (UT), and central California (CA). They were then intercrossed and raised inside a mesh enclosure in southern Mississippi (MS). Females from all possible mated pairs of O. ribifloris produced F₁ broods with 30–40% female cocoons and outcrossed progeny were 30% heavier. Mitochondrial (COI) genomes of the four populations revealed three distinct clades, a TX-CA clade, a UT clade, and an MS clade, the latter (MS) representing captive progeny of CA and UT bees. Although classified as separate subspecies, TX and CA populations from 30° N to 38° N latitude shared 98% similarity in COI genomes and the greatest brood biomass per nest straw (600- to 700-mg brood). Thus, TX and CA bees show greater adaptation for southern U.S. sites. In contrast, UT-sourced bees were more distantly related to TX and CA bees and also produced ∼50% fewer brood. These results, taken together, confirm that adult O. ribifloris from all trap-nest sites are genetically compatible, but some phenotypic variation exists that could affect this species performance as a commercial blueberry pollinator. Males, their sperm, or perhaps a substance in their sperm helped stabilize our captive bee population by promoting legitimate nesting over nest usurpation. Otherwise, without insemination, 50% fewer females nested (they nested 14 d late) and 20% usurped nests, killing 33–67% of brood in affected nests.