Heavily male-biased adult sex ratio is associated with lower female survival in the federally endangered Anas wyvilliana (Koloa Maoli or Hawaiian Duck) on north Kaua'i, Hawai'i, USA

Abstract: Anas wyvilliana (Koloa Maoli or Hawaiian Duck) is a federally endangered, non-migratory dabbling duck that experienced substantial population declines and range contraction during the 20th century. A critical obstacle in assessing threats and developing effective conservation actions involves the lack of information on population abundance, structure, and key demographic rates. We used a 5-year capture-mark-recapture (CMR) dataset and robust design CMR models to estimate abundance, sex ratio, and survival of adult and first-year A. wyvilliana at Hanalei National Wildlife Refuge (NWR), a core wetland site, on north Kaua‘i, Hawai‘i, USA. During 2010–2015, 792 A. wyvilliana (nfemale = 150, nmale = 642) were captured 1,236 times. Model-averaged, sex-specific abundance estimates revealed a heavily distorted, male-biased sex ratio (3:1) that was relatively consistent across seasons and years. Annual apparent survival was lower for females (0.43) compared to males (0.69). We found strong evidence for temporary emigration, though the particular form (i.e., random vs. Markovian) was not clearly distinguishable. The top-ranked model indicated temporary emigration followed a random process but varied seasonally, with individuals having a higher probability of being unavailable for capture during summer (0.76) compared to winter (0.64). Seasonal abundance of adult and first-year A. wyvilliana available for capture at Hanalei NWR averaged 407 individuals (range: 252–625). Assuming random temporary emigration, the total number of A. wyvilliana available and temporarily unavailable for capture on the refuge averaged 1,283 individuals (range: 929–1,822). Our study provides robust estimates for several demographic parameters that establish a quantitative foundation for future population modeling and targeted, evidence-based conservation decisions for the recovery of A. wyvilliana. General description of files: This repository contains the encounter history data and occasion-specific covariates required to construct robust design capture-mark-recapture (RDCMR) models in Program MARK, along with the R script necessary to process model outputs and replicate all secondary analyses. These data represent a 5-year capture-mark-recapture study of Anas wyvilliana (Koloa Maoli) conducted at Hanalei National Wildlife Refuge on north Kaua‘i, Hawai‘i, USA, during 2010–2015, as reported in Malachowski et al. (2026). Software required: Program MARK (available at: http://www.phidot.org/software/mark/) The R Project for Statistical Computing (available at: https://www.r-project.org/) File descriptions Malachowski_2026_RDCMR_Koloa encounter histories.inp Encounter history data for analysis of koloa maoli apparent survival, temporary emigration, and capture/recapture probabilities using robust design capture-mark-recapture (RDCMR) analysis in Program MARK. Malachowski_2026_RDCMR_Covariates.csv Occasion-specific environmental and sampling covariate values used to model capture and recapture probabilities within the robust design capture-mark-recapture (RDCMR) framework. Fields include: prim_occ: Sequential index of the primary sampling occasion. sec_occ: Sequential index identifying the unique secondary sampling occasion across the entire study period. This serves as an identifier to align the occasion-specific covariates. season: Season during which the primary occasion occurred (0 = winter; 1 = summer). effort: Total sampling effort (i.e., number of trap-days) during the secondary occasion. effort_log: Natural logarithm of sampling effort. rainPO: Mean daily rainfall (mm/day) during the primary occasion. rainPO_log: Natural logarithm of rainPO. rain30: Total cumulative rainfall (mm) during the 30 days preceding the primary occasion. rain30_log: Natural logarithm of rain30. rainSO: Mean daily rainfall (mm/day) within each secondary occasion. rainSO_log: Natural logarithm of rainSO. Malachowski_2026_RDCMR_Time intervals.csv Temporal sampling framework required to structure the robust design capture-mark-recapture (RDCMR) models. It defines the elapsed (scaled) time between each sequential sampling occasion. Fields include: interval: The sequential index number representing the specific gap between sampling occasions. duration_6mo: The elapsed time of the interval, standardized to a 6-month unit (where 1 unit = 6 months). A value of 0 represents an interval between secondary sampling occasions within the same primary occasion, and values > 0 represent the interval between primary sampling occasions. Malachowski_2026_Koloa abundance, sex ratio, and survival.R R script for processing parameter estimates from robust design capture-mark-recapture models and performing secondary analyses, including: Calculate annual survival probabilities Calculate bounded log-normal confidence intervals for model-averaged local abundance estimates Perform a mixed-effects meta-regression to evaluate patterns in local abundance estimates Calculate superpopulation sizes Calculate effective population sizes Evaluate correlation between local abundance estimates and concurrent monthly count surveys