Perturbations and heterogeneity: demographic rates of brant in western Alaska
Loading...
Authors
Riecke, Thomas Vance
Issue Date
2020
Type
Dissertation
Language
Keywords
Bayesian , Branta bernicla , Population Biology , Quantitative Ecology
Alternative Title
Abstract
As the earth’s climate rapidly changes, it remains unclear how populations will respond to novelsystems and anthropogenic inputs to said systems. The development of new applied conservationtools will be critical to the implementation of solutions to novel problems. My dissertation focuseson developing and implementing novel quantitative models to directly estimate demographic pa-rameters, and inform our understanding of ecological and evolutionary processes. I use a 29-year(1986�"2014) dataset on black brant geese (Branta bernicla nigricans, hereafter brant) collected onthe Yukon-Kuskokwim River Delta in western Alaska.In Chapter 2 (Riecke et al., 2018), I use simulated and real data to develop a novel fully con-ditional robust design likelihood which we implement in a Bayesian framework. I test my novellikelihood using simulated data, demonstrating its efficacy, and then use this model structure toexamine the effects of environmental conditions during growth on future lifetime fitness in brant.My results clearly indicate that more structurally developed goslings experienced increased breed-ing probability as an adult (β = 0.14; f = 0.94) with no effect on adult survival (β = 0.01;f = 0.62). I also provide evidence for long-term declines in apparent survival of breeding adultfemales (β = −0.01; f = 0.90). Thus, long-term declines in gosling growth rates (Lohman et al.,2019) may lead to reduced future reproductive potential at the population level. Further, temporalvariation in gosling growth rates may lead to substantial differences in reproductive potential amongcohorts.In my third chapter (Riecke et al., 2019a), I use simulated data to test the impacts of the inverseWishart prior for the covariance matrix of a multivariate normal distribution, and propose a novelalternative distribution. My simulations and analyses demonstrate the inverse Wishart prior distri-bution substantially affects estimates of covariances and variances of demographic parameters, andthat our proposed alternative distribution is substantially more effective than the conjugate inverseWishart prior. My simulations also reveal that substantial long-term data is required to estimate cor-relations regardless of the prior distribution chosen by demographers, where many previous studieshave failed to meet these minimal thresholds. I then use our novel variance-covariance parameteri-zation to estimate the correlation between survival of adult and juvenile female brant, where sharedenvironmental conditions lead to a strong positive correlation between these parameters (ρ = 0.563;95% CRI 0.181�"0.823). Given the potential for continued long-term habitat degradation in non-breeding areas and rapidly declining juvenile survival, our research provides strong evidence for theimportance of maintaining critical migratory stopover, staging, and wintering areas.In Chapter 4, I examine the impacts of long-term habitat degradation and climatic oscillationson adult phenotype in black brant. Specifically, we examine long-term changes in tarsus length ofadult female brant, where mean female tarsus of each cohort declined by 7.73 mm from 1986�"2013.gI then show that gosling growth rates have declined substantially (κ1 = −0.108 day) over the sametime period, explaining substantial variation in adult body size (R2 = 0.209; 85% CRI = 0.043�"0.418). I also used the methods developed in Chapter 2 to examine long-term changes in the effectsof body size on breeding probability and adult survival. At the beginning of the study, survival waspositively correlated with adult body size (β3 = 0.066, ηβ3 = 0.982), and body size had no effecton breeding probability. Over time, the relationships between body size, survival (β9 = −0.041,ηβ9 = 0.890), and breeding propensity (α9 = −0.096, ηβ9 = 0.847) reversed, where smallerindividuals now experience increased breeding and survival probability relative to large individuals.Thus, my results provide evidence that long-term changes in habitat can affect phenotype boththrough environmental conditions during growth, as well as changing selection pressures. Further,climatic oscillations might affect these processes.My fifth chapter combines nest monitoring data and capture-recapture data for gosling, juve-nile, and adult brant and their nests to estimate temporal variation in every brant demographicparameter across the entirety of the brant life cycle. I subsequently estimate the contributionsof each demographic parameter to population growth rates, as well as their covariance, to bet-ter understand the potential impacts of management actions on declining populations of adult fe-males (βηad = −151.96, 90% CRI -170.57, -133.62) and population growth rate (βλ = −0.0040,90% CRI -0.0055, -0.0025). There was weak evidence for long-term declines in adult survival(βφad = −0.013, ν = 0.863), nest survival (βφnest = −0.036, ν = 0.905), and clutch size(βξ = −0.007, ν = 0.836), and strong evidence for long-term declines in juvenile survival(βφjuv = −0.051, ν = 0.992). Conversely, there was strong evidence for increasing gosling survival(βφgos = 0.016, ν = 0.949) and breeding probability (βγ = 0.061, ν = 0.993). Our findingsindicate that brant are experiencing declining demographic rates in both breeding and non-breedingareas, where future conservation efforts will focus on positively impacting both areas.Collectively, these chapters highlight the importance of quantitative ecological tools, and theeffects of changing ecosystems and climate on life-history trade-offs, selection pressures, and pop-ulation trends on a long-lived specialist herbivore.