| Abstract: | Nest success, the proportion of clutches resulting in one or more fledglings, is a key indicator for assessing the effect of management on bird populations. However, the figures reported for New Zealand populations are usually "apparent nest success", the number of successful nests divided by the total number found. Apparent nest success invariably overestimates the true success rate, and the degree of bias depends on the population and monitoring regime. Consequently, apparent nest success rates cannot be reliably compared. We used Stanley?s (2000) method for estimating stage-specific daily survival probabilities for New Zealand robin (Petroica australis) nests at Tiritiri Matangi, Paengaroa, Boundary Stream and Pureora. We show how Stanley?s method can be used to eliminate biases, to calculate point estimates and confidence intervals for nest success, and to model the factors affecting nest success. At Pureora, where monitoring was extremely intensive, the apparent overall nest success (39%) was close to that estimated from daily survival probabilities (37%). Apparent nest success rates were extremely biased for the other populations due to less intensive monitoring, with the bias exacerbated by changes in survival probabilities with season and/or stage of the nesting cycle. Modelling the data showed that failure rates were: (1) higher early in the breeding season for at least some mainland populations, (2) different for incubation and nestling stages, with the pattern depending on the season (early or late) and type of predator, and (3) substantially lowered by predator control at Pureora, with the impact varying between sites and stages. Taking these factors into account, the estimated nest success (and 95% confidence limits) was 60% (44- 74) at Pureora after predator control, 47% (23-73) at Boundary Stream, 37% (26-49) at Tiritiri Matangi, 25% (11- 48) at Paengaroa, and 25% (17-35) at Pureora without predator control. |
