The demography of Atlantic brant (Branta bernicla hrota)

Animal population dynamics are driven by variation in survival and productivity. Long-lived species such as Arctic-nesting geese often are characterized by high adult survival and low but highly variable annual reproductive success. Productivity is commonly the mechanism for population change in long-lived species, but minor perturbations in survival can strongly influence trajectories. Events and processes during one season of the annual cycle that influence population demography during another period are termed cross-seasonal effects (CSEs) and can be a result of environmental conditions such as temperature or precipitation. Thus, robust conservation planning for animal populations depends on a complete understanding of both survival and productivity across the full annual cycle. My thesis is split into two chapters that describe survival and productivity in Atlantic brant (Branta bernicla hrota), which are an Arctic-nesting goose species that breed in the Foxe Basin of Nunavut, Canada, stage during fall and spring migration in James Bay, and winter on the Atlantic coast primarily in heavily urbanized landscapes of New Jersey and Long Island, New York. In chapter 1, I tested the extent to which environmental conditions at different scales throughout the annual cycle influenced Atlantic brant productivity over the past 44 years using generalized linear mixed models. I modeled the effects of the North Atlantic Oscillation Index, temperature and precipitation, and regional snow and ice cover during winter on the Atlantic coast, spring staging areas at James Bay, and breeding areas in the Foxe Basin on the fall Atlantic brant age-ratio. I predicted that harsh conditions would negatively influence productivity throughout the annual cycle, and that the strongest effects would occur during the breeding season. My results suggested CSEs during winter and spring, as well as the conditions during the breeding season explained variation in Atlantic brant productivity over the past 44 years, and conditions during spring had the strongest effect. Favorable spring conditions at all scales (local weather, regional snow and ice cover, and climatic indices) and only higher local temperatures during the breeding season positively influenced Atlantic brant productivity. Notably, I documented contrasting effects of winter regional snow and ice cover and local temperature conditions on productivity, such that lower temperatures positively influenced productivity while increased snow and ice cover negatively influenced productivity. I attributed this to greater levels of anthropogenic disturbance when temperatures were warmer during winter. These results emphasize the importance of evaluating conditions at multiple scales and throughout the annual cycle for greatest understanding of population level processes and to inform prioritization of conservation efforts for Atlantic brant. Future brant research should focus on nutrient dynamics of James Bay staging areas, identifying core breeding areas and quantifying reproductive metrics, and determining wintering habitat and space use. Our approach of evaluating seasonal environmental conditions at various scales can similarly be applied to other species with productivity datasets for holistic perspective of drivers of demography across space and time. In chapter 2, I sought to quantify improvements in survival estimates for Atlantic brant given implementation of a color-marking and resighting program, and addition of a winter operational banding program to supplement the existing summer metal banding program in the Arctic. I used a two-season joint encounter (JE) survival modeling framework which incorporated all existing metal banding and recovery data with color-marking and resighting data to estimate Atlantic brant survival from 2000 to 2021. Then, I used demographic estimates from empirical models to develop a suite of simulations with varied capture and resighting efforts under both a two-season and single-season framework to draw inference on the utility of a winter color-marking and resighting program. I quantified improvements in precision of survival estimates from JE simulations compared to traditional dead-recovery (DR) models under the existing summer metal banding effort. From 2000 to 2021, adult survival during the hunting season and non-hunting season was 0.91 (95 percent Credible Interval [CRI] 0.87, 0.94) and 0.96 (95 percent CRI 0.90, 0.99) respectively and juvenile survival during the hunting season and non-hunting season was 0.89 (95 percent CRI 0.83, 0.94) and 0.70 (95 percent CRI 0.43, 0.92) respectively. Reported mortality probability for metal banded brant was 0.43 (95 percent CRI 0.31, 0.63) and for double color-marked brant was 0.55 (95 percent CRI 0.37, 0.87). The reported mortality probability estimates for brant and two-season simulations were biased high because the data collection framework for Atlantic brant did not provide adequate information for the second season of the model (i.e., marked individuals could not be resighted or recovered in one of two seasons). In the single-season modeling approach, small sample size limited utility of additional resighting data in JE models. Under all simulations, precision in survival estimates was not increased in JE models compared to DR models. I recommend further development of a single-season model that leverages resighting information but is simpler than the two-season framework. As the color-marking and resighting program is still relatively new, I recommend continued color-marking to establish a larger dataset which can be used to quantify band targeting in hunter harvest and explore additional uses of resighting data such as estimation of lifetime reproductive success. Overall, I suggest that practitioners interested in estimating Atlantic brant survival should use single season DR or JE models for continued conservation planning and management of this species.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.