The hemlock woolly adelgid (HWA, Adelges tsugae Annand) is an invasive insect that threatens to eradicate native eastern hemlock (Tsuga canadensis (L.) Carr.) across the eastern United States. In southern New England and southern Appalachian forests, HWA-induced hemlock mortality has impacted carbon (C) flux by altering stand age, litter composition, species composition, and coarse woody debris levels. However, no one has examined how total C storage and sequestration may be impacted by these changes. Further, while projections are that HWA will ultimately infest hemlock across its entire geographic range, the majority of studies have been limited to southern New England and Appalachian forests where HWA infestation has been ongoing. To address these gaps, we examined how HWA might alter C dynamics in northern New England forests using the Forest Vegetation Simulator (FVS) and Forest Inventory Analysis (FIA) data to model C storage and successional pathways under three different scenarios: preemptive harvesting of hemlock, HWA-induced hemlock mortality, and a control mimicking natural stand development absent of disturbance. Our 150 year simulation showed that, while all treatments differed significantly in C storage in the short term, there was no significant difference in total C stocks between HWA infestation and presalvage treatments by the 75th year. Compared to the control, both simulated treatments resulted in a significant decrease in total C storage, with greater impacts on stands with higher hemlock densities. However, net C losses over the 150 year simulation were significantly higher for the presalvage scenario, indicating that allowing HWA infestation to progress naturally through a stand may result in the least impact to long-term C sequestration for the region's forests.

This research spans a variety of research topics with a common theme, providing decision support through the development and analysis of methods that assist decision making for natural resource and wildlife management. I used components of structured decision making and decision analysis to address natural resources management problems, specifically monitoring and estimating the status of harvested populations, as well as data collection decisions for landscape conservation. My results have implications for the way populations are monitored and their status is estimated. I find that the inclusion of error in data collection can have a substantial impact of the performance of abundance and growth rate estimates of harvested species and that the selection of estimation methods depends on what management objectives are most important. For example, the Sex-Age-Kill population estimation method best estimates the size of populations, while the Downing population reconstruction method better estimates trends in population growth rates. I provide a framework to support selection of the best estimation method while considering a monitoring program as a whole. Based on this framework the Vermont Fish and Wildlife Department will obtain the most benefits from a monitoring program including necropsy analysis that uses the Downing method to track population status. Finally, I demonstrated the use of value of information analysis as a tool to determine the relative expected benefits of addition spatial data collection for use in landscape mapping and conservation. This type of analysis can provide conservation agencies with a planning tool to direct budgets and mapping efforts.

Rising sea levels present an ongoing threat to communities and resources around the Chesapeake Bay, east coast, USA, where tide gauges indicate that the relative rise of sea level is approximately twice the rate of average, eustatic sea-level rise. This has significantly compromised the health and viability of salt marsh habitat on the Eastern Shore during the 20th century, and the biologists who are charged with managing coastal resources in the coming decades need to understand the nature and causes of high rates of regional sea-level rise to develop suitable adaptation plans. Dated geologic deposits and geophysical models suggest that sea-level rise is relatively high on mid-Atlantic coastlines because the land surface is subsiding due to a collapsing glacial forebulge following the Last Glacial Maximum (LGM). To fully understand this process, past sea-level indicators such as dated shoreline deposits are needed to reconstruct regional sea-level behavior in the past, but rigorous age control on geologic deposits is largely restricted to the Holocene and to marine isotope stage (MIS) 5, so the rates and timescales over which these processes operate remain unknown. This research provides long-term paleoenvironmental records from ancient environments under east-central Chesapeake Bay to place the current sea-level threats into the context of a long geologic history of sea-level fluctuations. First, the Pleistocene geologic framework of the region is reconstructed through borehole drilling. Sediments from boreholes provided material for interpreting depositional environments, and for establishing age control for deposits, so that the entire stratigraphy was constrained both in space and time. The geologic framework and ages indicate that Chesapeake Bay alternated between a deeply incised fluvial system and a filled estuary repeatedly in response to major climate fluctuations since at least the early Pleistocene, ~2 Ma. The ages and sedimentology indicate that the field area was submerged intermittently in a shallow estuary until nearly the end of marine isotope stage 3. Because global sea-level proxies suggest that sea level was ~40-80 meters lower than present at that time, these ages suggest that the penultimate glacial forebulge must have remained significantly lowered for nearly 100 ky following the retreat of ice. The implication of this time lag is that mid-Atlantic coastlines are still in a relatively early state of forebulge collapse, and subsidence following retreat of ice from the Last Glacial Maximum will likely continue for the foreseeable future. Ongoing subsidence will continue to exacerbate projected eustatic sea-level rise due to changing global climate, and coastal adaptation plans must remain focused on encouraging the migration of vital habitat toward higher elevations in the landscape.

Light detection and ranging (LiDAR) data can provide detailed information about three-dimensional forest horizontal and vertical structure that is important to forest productivity and wildlife habitat. Indeed, LiDAR data have been shown to provide accurate estimates to forest structural parameters and measures of higher trophic levels (e.g., avian abundance and diversity). However, links between forest structure and tree function have not been evaluated using LiDAR. This study was designed and scaled to assess the relationship of LiDAR to multiple aspects of forest structure and higher trophic levels (arthropod and bird populations), which included the ground-based collection of percent crown and understory closure, as well as arthropod and avian abundance and diversity data. Additional plot-based measures were added to assess the relationship of LiDAR to forest health and productivity. High-resolution discrete-return LiDAR data (flown summer of 2009) were acquired for the Hubbard Brook Experimental Forest (HBEF) in New Hampshire, USA. LiDAR data were classified into four canopy structural categories: 1) high crown and high understory closure, 2) high crown and low understory closure, 3) low crown and high understory closure, and 4) low crown and low understory closure. Nearby plots from each of the four LiDAR categories were grouped into "blocks" to assess the spatial consistency of data. Ground-based measures of forest canopy structure, site, stand and individual tree measures were collected on nine 50 m-plots from each LiDAR category (36 plots total), during summer of 2012. Analysis of variance was used to assess the relationships between LiDAR and a suite of tree function measures. Our results show the novel ability of LiDAR to assess forest health and productivity at the stand and individual tree level. We found significant correspondence between LiDAR categories and our ground-based measures of tree function, including xylem increment growth, foliar nutrition, crown health, and stand mortality. Furthermore, we found consistent reductions in xylem increment growth, decreases in foliar nutrition and crown health, and increases in stand mortality related to high understory closure. This suggests that LiDAR measures can reflect competitive interactions, not just among overstory trees for light, but also interactions between overstory trees and understory vegetation for resources other than light (e.g., nutrients). High-resolution LiDAR data show promise in the assessment of forest health and productivity related to tree function.

The black bear (Ursus americanus) is an iconic species with cultural, economic and ecological importance in Vermont, USA. Bears exhibit a highly variable diet, and few studies have described bear diet in the state. Information on diet may provide insight into foraging behavior, thus allowing managers to better assess patterns of human-bear conflict. My objectives were to estimate the relative contribution of food items to bear diet and how factors including sex, habitat, food availability, and nuisance status describe patterns of consumption. I collected samples from bears and major food groups including C3 plants, white-tailed deer (Odocoileus virginianus), corn (Zea mays), and human foods, then quantified diet using stable isotope analysis. Samples were collected from 71 bears, 547 plants, and 38 deer throughout Vermont. I also collected 12 corn samples, and 20 human hair samples to represent anthropogenic foods. I determined δ13C and δ15N isotope values for all samples, then used Bayesian mixing models to estimate the contribution of foods and effect of each factor on proportional contribution estimates. Nuisance status best described patterns of diet over other factors. Median percent contributions for non-nuisance bears were 73.2% C3 plants, 23.8% corn, 1.9% human foods, and 0.5% deer. Median percent contributions for nuisance bears were 64.6% C3 plants, 28.9% corn, 3.2% human foods, and 0.7% deer. Factors such as sex, habitat, and food availability exerted less effect on diet than expected. Proportional contribution of meat was lower than in some other parts of North America, suggesting bears forage differently in Vermont. Results provide the first statewide estimate of bear diet and indicate corn may represent a much larger component of diet than previously thought. In particular, bears labeled as nuisance animals may forage on greater proportions of corn throughout the year.

The role of collaboration within conservation is of increasing interest to scholars, managers and forest communities. Collaboration can take many forms, but one under-studied topic is the form and content of public discourses across conservation project timelines. To understand the discursive processes that influence conservation decision-making, this research evaluates the use of collaborative rhetoric and claims about place within discourses of conservation in the Adirondacks. Local newspaper articles and editorials published from January 1996 to December 2013 and concerning six major conservation projects were studied using content analysis. Results show that collaborative rhetoric increased during the study period, and conflict discourses declined, in concert with the rise of collaborative planning efforts. Data also show an increasing convergence between conservation sponsors and local communities regarding the economic benefits of conservation and the importance of public participation. The study has value in examining representations of place and media claims-making strategies within conservation discourses, an important topic as natural resource managers increasingly embrace community-based natural resource management.

Wildlife species that rely on early successional habitat are showing long-term region-wide declines, including songbirds such as Eastern Towhee (Pipilo erythrophthalmus), Field Sparrow (Spizella pusilla), Blue-winged Warbler (Vermivora cyanoptera), Golden-winged Warbler (Vermivora chrysoptera), Prairie Warbler (Setophaga discolor), and Brown Thrasher (Toxostoma rufum). All six species are listed as Species of Greatest Conservation Need on a majority of the New England state's Wildlife Action Plans and in 2011, the Golden-winged Warbler was placed under review for federal listing. In areas where vegetation structure is actively maintained in early successional stages, such as powerline rights-of-way (ROWs), there is an opportunity to provide habitat over long time frames. This study focused on habitat use by six early successional bird species in the Champlain Valley in Vermont along powerline ROWs. Thirty sites that included potential habitat were established. During the breeding season, trained citizen scientist volunteers spot-mapped species distribution patterns along transects. Focal habitat use by individuals was mapped during timed field observations (n=83). Vegetation species composition and structure was evaluated within 1m2 vegetation plots across each site (n=965). Logistic regression models of study species' abundance patterns and focal habitat use were compared using Akaike's information criterion (AIC). Surrounding landscape composition, vegetation species composition and vegetation structure had the greatest influence on species abundance and focal habitat use based on best-fit models. All species occurrence, except Blue-winged Warbler and Brown Thrasher, decreased with greater development and fragmentation in the surrounding landscape. While the covariates that influenced habitat use patterns by the species were similar, the effect size and direction of influence varied. These results suggest that a universal management approach for shrubland songbird habitat would not support a broad range of species. The findings from this study determined species-specific habitat preferences that can improve management practices to benefit these declining species.

This dissertation examines the language-based, discursive processes through which meanings and experiences are socially constituted in outdoor recreation and natural resource environments. Language use and discourse are seen as interactive, constructive processes, approached through the theoretical perspectives of argumentation, social constructionism, and performance. Three qualitative studies, based in data collected at Acadia National Park and forest-related sites throughout Vermont, comprise this dissertation. The first study uses rhetorical analysis to examine the ways National Park Service managers and community leaders argue for the meanings and management of dark night skies in and around Acadia. The second study examines how national park visitors socially construct meanings of night sky experiences, focusing on the structure, functions and styles of language. The third study evaluates forest-oriented environmental interpretation materials produced by Vermont-based agencies through an analysis of performance. Each study analyzes a different type of discourse: semi-formal "expert" language solicited in interviews with managers and leaders (study 1), semi-formal "naive" language solicited in interviews with park visitors (study 2), and formal, written texts produced by agencies (study 3). Results show how language is used to forge agreement across competing ideals; construct meanings despite undeveloped vocabularies and intangible values; and direct visitors to perform forests in ways that develop the meanings of place. These studies contribute to the understanding of how individuals and organizations use language within discourse practices to create the reality in which socially- and culturally-important natural resource environments are managed and experienced, forming a body of work that informs theory and practice.

The need to protect diverse biological resources from ongoing development pressures is one of today's most pressing environmental challenges. In response, "ecosystem services" has emerged as a conservation framework that links human economies and natural systems through the benefits that people receive from nature. In this dissertation, I investigate the science-policy interface of ecosystem services in order to understand the use of ecosystem service decision support tools and evaluate the pathways through which ecosystem services knowledge impacts decisions. In the first paper, I track an ecosystem service valuation project in California to evaluate how the project changes the social capacity to make conservation-oriented decisions and how decision-makers intend to use ecosystem services knowledge. In a second project, I analyze a global sample of cases and identify factors that can explain the impact of ecosystem services knowledge on decisions. I find that the perceived legitimacy of knowledge (whether it is unbiased and representative of many diverse viewpoints) is an important determinant of whether the knowledge impacts policy processes and decisions. For the third project, I focus on the global use of spatial ecosystem service models. I analyze country-level factors that are associated with use and the effect of practitioner trainings on the uptake of these decision support tools. Taken together, this research critically evaluates how ecosystem service interventions perform. The results can inform the design of boundary organizations that effectively link conservation science with policy action, and guide strategic efforts to protect, restore, and enhance ecosystem services.

The long-term success of wildlife conservation depends on maximizing the benefits of limited funds and data in pursuit of population and habitat objectives. The ultimate currency for wildlife management is progress toward long-term preservation of ample, wild, free wildlife populations and to this end, funds must be wisely spent and maximal use made from limited data. Through simulation-based analyses, I evaluated the efficacy of various models for estimating population abundance from harvest data. Because managers have different estimators to choose from and can also elect to collect additional data, I compared the statistical performance of different estimation strategies (estimator + dataset) relative to the financial cost of data collection. I also performed a value of information analysis to measure the impact that different strategies have on a representative harvest management decision. The latter analysis is not based on the cost of data, but rather on the management benefit derived from basing decisions on different datasets. Finally, I developed a hybrid modeling framework for mapping habitat quality or suitability. This framework makes efficient use of expert opinion and empirical validation data in a single, updatable statistical structure. I illustrate this method by applying it across an entire state.