Accurate estimates of population size are often crucial to determining status and planning recovery of endangered species. The ability to detect trends in survival and population size over time enables conservation managers to make effective decisions for species and refuge management. During 2004–2007, the translocated population of endangered Laysan Teal (Anas laysanensis; also Laysan Duck) was fitted with radio transmitters providing known (―gold standard‖) measures of survival and reproduction. However, as the population grew, statistically rigorous monitoring protocols were needed that were less labor intensive than radio telemetry. A population die-off and alarmingly high number of carcasses (181) were recorded during a botulism epizootic in August–October 2008, which further reinforced the need for effective monitoring protocols since this endangered species is vulnerable to catastrophic population declines. In fall 2008, we initiated a pilot study using standardized surveys with uniquely marked birds to monitor abundance and estimate the population growth rate of the reintroduced Laysan Teal. Since few birds carried marks (leg bands) after the 2008 botulism die-off (only about 15% of the population), and standardized surveys were not yet implemented, the magnitude of the die-off on the population size was unknown.

To learn more about this endangered species' status and develop monitoring protocols useful to refuge managers and recovery planners in the U.S. Fish and Wildlife Service (USFWS), we marked (banded) 252 new Laysan Teal for this pilot project. With skilled refuge staff and trained volunteers, we conducted counts of marked, unmarked, and unknown birds during bimonthly surveys from Oct 2008 to Jan 2010. We recorded the identities of marked birds observed, recovered carcasses, and then used the last date a bird was detected alive and the median resight frequency to conclude if a bird was likely to be alive on a given survey date. Using mark-resight data and individual resight frequencies, we produced a series of abundance estimates from surveys that met accuracy criteria and approached ―closed population‖ assumptions. Since only one year of standardized, atoll-wide surveys were conducted, we analyzed data selected from multiple surveys using Lincoln-Petersen (LP) estimates instead of multi-year likelihood estimators. We adjusted surveys to account for unknown birds (e.g., swimming birds), temporary band loss, and described the frequency of double counting. Double counting is an important consideration in the population estimate because we found a maximum of 13% of marked birds were counted multiple times during a survey.

These survey protocols allowed us to estimate the species' post-fledging population (combined adults and juveniles), and the methods are comparable to those used on Laysan Island. The Laysan Teal population increased 91% from 247 (95% CI, 233–260) in 2007 to 439–508 in early 2010. There was no change from 2009 to 2010 indicating that there was no population growth, however, our 2010 estimate should be considered preliminary since only one month of 2010 resight data was used. We compared a series of direct counts to their corresponding population estimates during 2008–2009 to evaluate if counts could serve as an unbiased ―index‖ of population abundance. There was a moderate correlation between abundance estimates and total birds counted (r² = 0.51) during resight surveys but a low correlation with all-wetland counts (r² = 0.02). This indicated that using direct all-wetland counts to predict abundance would result in confidence intervals on the order of ± 200 birds, which is equal to 50% of the estimate. With such large confidence intervals, it would be unlikely to detect annual changes in abundance or determine the magnitude of a catastrophic decline.

To improve the Laysan Teal population estimates, we recommend changes to the monitoring protocol. Additional years of data are needed to quantify inter-annual seasonal detection probabilities, which may allow the use of standardized direct counts as an unbiased index of population size. Survey protocols should be enhanced through frequent resights, regular survey intervals, and determining reliable standards to detect catastrophic declines and annual changes in adult abundance. In late 2009 to early 2010, 68% of the population was marked with unique color band combinations. This allowed for potentially accurate adult population estimates and survival estimates without the need to mark new birds in 2010, 2011, and possibly 2012. However, efforts should be made to replace worn or illegible bands so birds can be identified in future surveys. It would be valuable to develop more sophisticated population size and survival models using Program MARK, a state-of-the-art software package which uses likelihood models to analyze mark-recapture data. This would allow for more reliable adult population and survival estimates to compare with the ―source‖ Laysan Teal population on Laysan Island. These models will require additional years of resight data (> 1 year) and, in some cases, an intensive annual effort of marking and recapture. Because data indicate standardized all-wetland counts are a poor index of abundance, monitoring efforts could be improved by expanding resight surveys to include all wetlands, discontinuing the all-wetland counts, and reallocating some of the wetland count effort to collect additional opportunistic resights. Approximately two years of additional bimonthly surveys are needed to validate the direct count as an appropriate index of population abundance. Additional years of individual resight data will allow estimates of adult population size, as specified in recovery criteria, and to track species population dynamics at Midway Atoll.