Reproductive success and female survival affect local population density of canvasbacks

Long-term data (1949-55 and 1959-80) from a canvasback (Aythya valisineria) population near Minnedosa, Manitoba suggested that breeding density was limited to ≤4.5 pairs/km² by annual variation in reproductive success (the local-recruitment hypothesis). We attempted to test this hypothesis by experimentally increasing reproductive success on a 15.8-km² study block during 1983-90. Manipulations included barrier fences to protect canvasback nests from mammalian predators and egg additions to replace canvasback eggs that had been displaced due to parasitic egg-laying by redheads (Aythya americana). We monitored breeding pair density and duckling production on the experimental study block and on an adjacent 15.8-km² control block using complete pair and brood counts (1983-90), nest and brood monitoring (1983-90), and roadside 'pair and brood surveys (1983-94). We nasal-marked pre-fledgling and adult female canvasbacks and used mark-resighting analysis to estimate juvenile survival, adult survival, and resighting probabilities. Reproductive success varied substantially among years, primarily in response to variable wetland conditions. The experimental study block had higher per capita fledging success during 6 out of 8 years (P = 0.08), but survival and resighting rates did not differ between study blocks (P = 0.28-0.80). Indices of canvasback density were similar during 1983-84, but pair populations became higher on the experimental block during 1985-90 (P < 0.01). By 1988, canvasback density on the experimental block reached 8.6 pairs/km², which was 1.8 times the highest density recorded on the control block. Differences in pair density persisted during the posttreatment years of 199194 (P < 0.003; except 1993: P = 0.72), although posttreatment duckling production did not differ between the 2 study blocks (P > 0.16). Annual changes in pair density were correlated positively with adult survival (r = 0.69, P = 0.007) and fledged ducklings/hen (r = 0.62, P = 0.02), although neither correlation remained significant when the other factor was controlled with partial correlation (P = 0.24 and 0.63). Population simulations suggested that both factors were important in affecting population changes. Management actions that enhance local breeding success or annual survival should lead to larger local breeding populations.