Geolocators as Tools for Inferring Waterfowl Movements and Breeding Phenology in the Pacific Flyway
AdvisorShoemaker, Kevin T
Natural Resources and Environmental Science
StatisticsView Usage Statistics
Understanding the geographical extent and timing of wildlife movements enables resource managers to meet the habitat needs of target species efficiently and effectively. Historically, information about waterfowl movements has been derived from mark-recovery data in which birds are trapped and marked with metal leg bands that are subsequently reported via hunter harvest. Such banding efforts typically yield two data points per individual (banding location and harvest location), and are poorly suited for estimating migration paths, stopover sites and timing of movements. Here, we use light-level geolocators – which enable researchers to track individual locations on the basis of estimated daily twilight times – to build a more complete understanding of the geography and timing of migratory movements for canvasbacks (Aythya valisineria) in the Pacific Flyway. During 2015-2017, 151 geolocators were placed on canvasbacks using two alternative attachment methods (leg-band vs. nasal-saddle mounts) during spring migration (February - March) near Reno, NV. Five of these geolocators (three males and two female) were successfully recovered from hunters. Four of the five tagged canvasbacks (2 males and 2 female) migrated to breeding sites in southern Canada (Alberta and Saskatchewan, via stopover sites in Utah and Montana), while one male migrated to a breeding site in Alaska. After initial capture in early spring, two canvasbacks (one Canada-breeding male and the Alaska-breeding male) remained near the capture site in Western Nevada or California for over a month (40 and 77 days, respectively), until resuming migration in late March or April. During spring migration, canvasbacks made an average of 3 stops, with an average stopover duration of 18±4 days. Three canvasbacks made a distinct molt migration after breeding. For fall migration, the canvasbacks made an average of 2.3±0.7 stops, lasting an average of 7.1±1.3 days, on their way to wintering sites in California’s Central Valley and San Francisco Bay area. Recovery of nasal saddle-mounted geolocators was significantly lower than leg band-mounted devices. This study demonstrates the value of geolocators for assessing year-round habitat use for waterfowl populations. This information complements standard band-recovery approaches, and enables waterfowl managers to ensure that the spatial and temporal distributions of individuals are identified so that habitat conservation efforts reflect the full annual habitat use cycle.Avian ecologists typically undertake costly and time intensive nest surveys to estimate key parameters such as nesting propensity, clutch size, nest success and hatching success. However, nesting surveys can be prohibitively difficult and/or costly in many cases. Less time consuming and invasive tools to supplement reproductive parameter estimates are needed. However, use of geolocators to infer reproductive parameters has not been rigorously validated. We attached light level geolocators to wood ducks (Aix sponsa) in northern Nevada, in a population that also was intensively monitored via artificial nest boxes. To designate each day as either nesting or non-nesting from geolocator data, we used a Bayesian mixture model, enabling estimation of life-history traits such as nesting propensity, annual nesting attempts, incubation duration and nest success. We also attempted to distinguish among nest prospecting, laying, incubation, and brood-rearing behaviors using an unsupervised time-series clustering algorithm, and to further estimate clutch size and nest fate. Using artificial nest box monitoring as validation data, we confirmed the accuracy of geolocator-derived estimates of the number and timing of nesting attempts and nest success. We were unable to confirm the validity of geolocator-derived estimates of clutch size (number of laying days). Using geolocator data, we estimated a nesting propensity of 0.92, with 87% apparent nest success for surviving females (100% success in natural cavities vs. 73% success in artificial nest boxes). Our clustering assignments showed some ability to discriminate among nesting behaviors, but were unable to effectively predict clutch size or nest fate. Our study is the first to rigorously validate the use of incubation length for predicting success of individual nests. We contend that widespread deployment of geolocators on waterfowl and other birds would enable estimates of reproductive parameters for species and populations not amenable to standard nesting surveys, after controlling for nesting season adult female survival.