Geology Reports⌕ Search

Geology topics

Therese M. Donovan

Publications and source records attributed to Therese M. Donovan.

At least 19 recordsLinked to original sources
USGS2026

Evaluating seedling traits, maintenance and effective costs to support tropical forest restoration practices

While large-scale restoration is essential, tropical forest restoration often depends on small-scale practitioners and landowners who produce, plant and maintain trees. Their collective impact on restoring forest diversity and function is substantial, making it critical to address the constraints they face through efficient, effective techniques. We used generalized linear mixed models to evaluate how seedling traits, weeding maintenance and a fertilizer application influenced the 5-year survival of 1796 seedlings from 11 tree species planted in Monteverde, Costa Rica. Estimates of annual survival increased with seedling age (time after planting) and leaf count at planting. Weeding effects varied by species, improving survival in 7 of 10 species and reducing survival in one. All species reached >80% annual survival after 2 years regardless of maintenance, and seedlings planted with ≥25 leaves showed increased survival that decreased the need for weeding. Weeding is costly: all effective costs per 4-year surviving seedlings were higher with weeding than without and varied by species. Costs were calculated by species, treatment and seedling leaf counts when planted (5, 10, 15, 25, 35, 45). Beyond the 2-year threshold, where mortality is highest, weeding effort and costs no longer appear justified. A single fertilizer application did not improve survival in either of the species tested. Practical implications . These findings support landowner restoration by giving restoration technicians accessible tools to improve seedling survival—increasing leaf counts before planting—and reducing post-planting weeding costs by targeting only species for which it benefits and limiting weeding to 2 years.

Los Llanos, Monteverde↗
USGS2026

Advancing ecosystem recovery with diverse species plantings in tropical forest restoration

Tropical forest restoration has increased in the past decades, with possible advancements given the UN declaration of the “Decade of Ecosystem Restoration”. However, robust assessments to compare ecosystem functions among restored forest stages are essential. We evaluated 13 actively restored forest stands ranging from 3 to 21 years of age and compared measures of forest biodiversity, structure, and ecosystem function to four 70+ year old “reference” stands that serve as restoration “targets” in the study region of the Premontane wet forest of Costa Rica. The restored stands were planted with an average of 13 tree species on abandoned pastures that were fallow for at least two years. Sixteen tree-stand attributes and six ecosystem function estimates were assessed, including: annual biomass (C) accumulation, N-fixation potential, threatened species conservation, and the provision of avian frugivore forage, insect habitat, and insect pollination. Using Principal Component Analysis, linear modeling, and Mahalanobis distance analyses, we learned that planting a diversity of tree species sets the stage for forest recovery at early restoration ages, with an inflection point at 15 years towards older reference forest characteristics and functions. Given that all restoration ages provided tree diversity and some level of ecosystem functions, the value of all restored stands in the landscape is notable. The assessment methods are easily employed, thereby providing an accessible tool to restoration practitioners.

Monteverde↗
USGS2026

Two-stage approach to automatic detection with machine learning for improved surveillance of the invasive Cuban treefrog

The Cuban treefrog ( Osteopilus septentrionalis ), as an invasive species in the southern United States, presents a need for effective surveillance. Automated detection expedites processing of audio data for large-scale surveillance and monitoring programs. However, current available methods commonly used for anuran species have not been sufficient to detect Cuban treefrogs. Here, we present results from a two-stage method for automated detection that employs both cross-correlation template matching and secondary supervised learning classifiers. In the first stage, audio data are screened for initial detections using template matching, in which the detections contain both true and false positives. In the second stage, the false positives are screened out using classifier algorithms. We used this method to process 139,985 audio recordings, consisting of 596,046 total minutes, collected at 13 locations in Louisiana and Florida from 2014 to 2022. From the stage 1 template matching, we detected 83,191 Cuban treefrog signals across recordings. The stage 2 machine learning model was able to identify stage 1 false positive detections with a testing accuracy of 98.46% and a testing false positive rate of 1.116%. After pruning false positive detections, a total of 20,271 individual Cuban treefrog detections remained, distributed mainly across 3 sites in an area with known presence. Locations with presumed absence had an easily verifiable number of false positive detections ( n = 109 across all other sites). The two-stage methodology utilizing both template matching and machine learning algorithms can be integrated into wildlife surveillance or monitoring programs for species with distinctive, conserved calls as an effective way to achieve sensitive species detection with a low incidence of false positives.

Florida, Louisiana↗
USGS2025

Observations of tear-drinking by lepidopterans on moose ( Alces alces americana ) in northeastern North America

Lepidoptera have long been known to feed on the tears of vertebrates as a presumed source of minerals or nutrients. While this unusual behavior has been observed in a variety of species, only a single previous record has been documented outside of the tropics. Here, we present the first documentation of moths visiting the eyes of a bull moose ( Alces americanus americanus ), captured via trail camera in Green Mountain National Forest, Vermont, United States. We discuss the biogeography of this behavior, how it may differ between tropical and temperate climates, and its potential impact on moose health.

Vermont↗
USGS2025

DeepFaune New England: A species classification model for trail camera images in northeastern North America

The DeepFaune New England model classifies wildlife species in trail camera images, identifying 24 taxa from northeastern North America with high (97%) accuracy. The model was adapted from the DeepFaune model for identifying European wildlife, demonstrating the practicality of transfer learning across continents. The majority of training data is openly licensed, and the model itself is open source, enabling easy integration into camera trapping workflows. The open source software is available at ( https://code.usgs.gov/vtcfwru/deepfaune-new-england ), and has been further integrated into the PyTorch-Wildlife framework.

New England↗
USGS2025

Two-stage models improve machine learning classifiers in wildlife research: A case study in identifying false positive detections of Ruffed Grouse

Autonomous recording units are increasingly being used to monitor wildlife on large geographic and temporal scales, paired with machine learning (ML) to automate detection of wildlife. However, false positive detections from ML classifiers can result in erroneous ecological models that can lead to misguided management and conservation actions. We used a two-stage general approach to understand and reduce false positive detections, a technique in which outputs of the primary classification model are passed to a secondary classification model to yield the probability that a detection from the primary model is a true positive detection. This approach is demonstrated on two open-source models that detect Ruffed Grouse ( Bonasa umbellus ). We analyzed over 9500 h of acoustic data collected in 2022–2023 from the Green Mountain National Forest in Vermont, USA, and found the two models detected different types of acoustic signals associated with differing life history traits. The first model yielded 4106 detections (71.5 % true positives) while the second model yielded 524 detections (17.0 % true positives). Secondary logistic regression models separated true positives and false positives with high accuracy (84.5 % and 89.8 % respectively). Our findings go beyond improving Ruffed Grouse monitoring and conservation efforts to, more broadly, illustrate how two-stage ML approaches can improve the use of model-derived detections in wildlife research.

Vermont↗
USGS2024

Net carbon sequestration implications of intensified timber harvest in Northeastern U.S. forests

U.S. forests, particularly in the eastern states, provide an important offset to greenhouse gas (GHG) emissions. Some have proposed that forest-based natural climate solutions can be strengthened via a number of strategies, including increases in the production of forest biomass energy. We used output from a forest dynamics model (SORTIE-ND) in combination with a GHG accounting tool (ForGATE) to estimate the carbon consequences of current and intensified timber harvest regimes in the Northeastern United States. We considered a range of carbon pools including forest ecosystem pools, forest product pools, and waste pools, along with different scenarios of feedstock production for biomass energy. The business-as-usual (BAU) scenario, which represents current harvest practices derived from the analysis of U.S. Forest Service Forest Inventory and Analysis data, sequestered more net CO 2 equivalents than any of the intensified harvest and feedstock utilization scenarios over the next decade, the most important time period for combatting climate change. Increasing the intensity of timber harvest increased total emissions and reduced landscape average forest carbon stocks, resulting in reduced net carbon sequestration relative to current harvest regimes. Net carbon sequestration “parity points,” where the regional cumulative net carbon sequestration from alternate intensified harvest scenarios converge with and then exceed the BAU baseline, ranged from 12 to 40 years. A “no harvest” scenario provides an estimate of an upper bound on forest carbon sequestration in the region given the expected successional dynamics of the region's forests but ignores leakage. Regional net carbon sequestration is primarily influenced by (1) the harvest regime and amount of forest biomass removal, (2) the degree to which bioenergy displaces fossil fuel use, and (3) the proportion of biomass diverted to energy feedstocks versus wood products.

Connecticut, Maine, Massachusetts, New Hampshire, ↗
USGS2024

occupancyTuts: Occupancy modelling tutorials with RPresence

1. The occupancy modelling framework offers tremendous flexibility in estimating species abundance and distribution patterns while accounting for imperfect detection, and has seen rapid growth and adoption since its introduction at the beginning of the century. 2. At the same time, in an era of big data, there are increasing demands on developing quantitative skills and proficiency in young ecologists, many of whom lack the quantitative training needed to conduct research professionally. 3. We introduce occupancyTuts , an R package that features 28 learnr tutorials that teach the statistical underpinnings of several occupancy models. The tutorials include written content, instructional videos, R exercises, and quiz elements, covering a range of topics including statistical underpinnings, single- and dynamic-occupancy models, study design and several of the ‘spin-off’ models that extend the basic framework. 4. We plan for development of new tutorials that use RPresence as the analysis engine, and welcome new tutorial contributions that use other R packages as the analysis engine as well.

Methods in Ecology and Evolution↗
USGS2023

Advances in wildlife abundance estimation using pedigree reconstruction

The conservation and management of wildlife populations, particularly for threatened and endangered species are greatly aided with abundance, growth rate, and density measures. Traditional methods of estimating abundance and related metrics represent trade-offs in effort and precision of estimates. Pedigree reconstruction is an emerging, attractive alternate approach because its use of one-time, noninvasive sampling of individuals to infer the existence of unsampled individuals. However, advances in pedigree reconstruction could improve its utility, including forming a measure of precision for the method, establishing required spatial sampling effort for accurate estimates, ascertaining the spatial extent of abundance estimates derived from pedigree reconstruction, and assessing how population density affects the estimator's performance. Using established relationships for a stochastic, spatially explicit simulated moose ( Alces americanus ) population, pedigree reconstruction provided accurate estimates of the adult moose population size and trend. Novel bootstrapped confidence intervals performed as expected with intensive sampling but underperformed with moderate sampling efforts that could produce abundance estimates with low bias. Adult population estimates more closely reflected the total number of adults in the extant population, rather than number of adults inhabiting the area where sampling occurred. Increasing sampling effort, measured as the proportion of individuals sampled and as the proportion of a hypothetical study area, yielded similar asymptotic patterns over time. Simulations indicated a positive relationship between animal density and sampling effort required for unbiased estimates. These results indicate that pedigree reconstruction can produce accurate abundance estimates and may be particularly valuable for surveying smaller areas and low-density populations.

Ecology and Evolution↗
USGS2023

Genetic diversity and connectivity of moose (Alces americanus americanus) in eastern North America

Genetic diversity is critical to a population’s ability to overcome gradual environment change. Large-bodied wildlife existing in regions with relatively high human population density are vulnerable to isolation-induced genetic drift, population bottlenecks, and loss of genetic diversity. Moose ( Alces americanus americanus ) in eastern North America have a complex history of drastic population changes. Current and potential threats to moose populations in this region could be exacerbated by loss of genetic diversity and connectivity among subpopulations. Existing genetic diversity, gene flow, and population clustering and fragmentation of eastern North American moose are not well quantified, while physical and anthropogenic barriers to population connectivity already exist. Here, single nucleotide polymorphism (SNP) genotyping of 507 moose spanning five northeastern U.S. states and one southeastern Canadian province indicated low diversity, with a high proportion of the genomes sharing identity-by-state, with no consistent evidence of non-random mating. Gene flow estimates indicated bidirectionality between all pairs of sampled areas, with magnitudes reflecting clustering and differentiation patterns. A Discriminant Analysis of Principal Components analysis indicated that these genotypic data were best described with four clusters and indicated connectivity across the Saint Lawrence River and Seaway, a potential physical barrier to gene flow. Tests for genetic differentiation indicated restricted gene flow between populations across the Saint Lawrence River and Seaway, and between many sampled areas facing expanding human activity. These results document current genetic variation and connectivity of moose populations in eastern North America, highlight potential challenges to current population connectivity, and identify areas for future research and conservation.

New York, Vermont, New Hampshire, Maine, and Massa↗
USGS2023

Evaluating a tandem human-machine approach to labelling of wildlife in remote camera monitoring

Remote cameras (“trail cameras”) are a popular tool for non-invasive, continuous wildlife monitoring, and as they become more prevalent in wildlife research, machine learning (ML) is increasingly used to automate or accelerate the labor-intensive process of labelling (i.e., tagging) photos. Human-machine hybrid tagging approaches have been shown to greatly increase tagging efficiency (i.e., time to tag a single image). However, those potential increases hinge on the extent to which an ML model makes correct vs. incorrect predictions. We performed an experiment using a ML model that produces bounding boxes around animals, people, and vehicles in remote camera imagery (MegaDetector) to consider the impact of a ML model’s performance on its ability to accelerate human labeling. Six participants tagged trail camera images collected from 12 sites in Vermont and Maine, USA (January–September 2022) using three tagging methods (one with ML bounding box assistance and two without assistance). We used a generalized linear mixed model to examine the influence of ML model performance and tagging method on tagging efficiency. We found that ML bounding boxes offer significant improvement in tagging efficiency when labelling data compared to unassisted tagging. Additionally, the time taken to label with bounding boxes was not statistically different from an unassisted tagging approach. However, we found that gains in efficiency are contingent on the ML algorithm’s performance and that incorrect ML predictions, particularly the 4.2% false positive and 3.6% false negative predictions, can slow the tagging process compared to a non-hybrid approach. These findings indicate that although practitioners usually forgo the production of bounding boxes when selecting a data labelling process due to the increased effort, ML bounding box-assisted tagging can offer an efficient method for labeling. More broadly, ML-assisted data labelling offers an opportunity to accelerate the analysis of trail camera imagery, but an assessment of the ML model’s performance can illuminate whether the hybrid-tagging approach is ultimately a help or hinderance.

Maine, Vermont↗
USGS2023

Birth rates and neonate survival in a parasite rich moose population in Vermont, USA

Moose ( Alces alces ) populations are declining across much of their southern geographic range in North America. In Vermont and other northeastern states, measurable declines are attributed to low calf survival and reduced productivity associated with persistent winter tick ( Dermacentor albipictus ) parasitism. In 2017–2020, we studied 75 radio-collared female moose (38 calves and 37 adults) in Vermont to examine physiological, spatial, and temporal parameters relative to calf survival and adult productivity. Physiological measures included concentration of fecal glucocorticoid metabolites (fGCM) which reflects stress, and urea nitrogen:creatinine ratios in urine (UN:C) which proxy nutritional state. The pregnancy rate at capture across years was 0.67 (95% CI = 0.50 – 0.80), and was negatively related to presence of lungworm ( Dictyocaulus spp.). The birth rate calculated as the average number of offspring delivered per adult female was <1.0 overall (2017–2020, LCI = 0.22, UCI = 0.86), similar across years, but increased with age. Logistic exposure models indicated that daily calf survival to 60 d increased as Julian birth date and days since birth increased (log odds = 0.0819, SE = 0.0215). The per capita independence rate, or rate that adult females add independent calves to the population, was negatively related to UN:C ratios and positively with fGCM. Further, this rate was related to autumnal habitat use of adult females; it was greater in home ranges characterized by large amounts of mature (canopy) evergreen forests and wetland habitats, and small amounts of mixed forests and elevation than in ranges with abundant levels of mixed forest at high elevation. We conclude that winter ticks can negatively affect moose fecundity, and efforts to reduce host (moose) density through harvest or parasite (host) abundance through habitat manipulation may improve productivity and recruitment in local moose populations.

Vermont↗
USGS2021

Effects of winter ticks and internal parasites on moose survival in Vermont, USA

Moose ( Alces alces ) have experienced considerable declines along the periphery of their range in the northeastern United States. In Vermont, the population declined 45% from 2010 to 2017 despite minimal hunter harvest and adequate habitat. Similarly, nearby populations recently experienced epizootics characterized by >50% mortality. Declines have largely been associated with the effects of winter ticks ( Dermacentor albipictus ), but uncertainty exists about the effects of environmental and other parasite-related conditions on moose survival. We examined patterns of moose survival among a radio-collared population ( n = 127) in Vermont from 2017 to 2019. Our objectives were to estimate causes of mortality and model survival probability as a function of individual and landscape variables for calves (<1 yr) and adults (≥1 yr). Observed adult survival was 90% in 2017, 84% in 2018, and 86% in 2019, and winter calf survival was 60% in 2017, 50% in 2018, and 37% in 2019. Winter tick infestation was the primary cause of mortality (91% of calves, 25% of adults), and 32% of all mortalities had evidence of meningeal worm ( Parelaphostrongylus tenuis ). Other sources of mortality such as vehicles, harvest, predation, deep snow, and other parasitic infections were negligible. The best supported calf model included sex differences and negative effects of tick engorgement (%/week) and parasite level (roundworm and lungworm). The best supported adult model included the effect of cumulative tick engorgement (cumulative %/week), which negatively affected survival. Our results indicate that winter tick engorgement strongly affects survival, and is probably compounded by the presence of meningeal worm and other parasites. Reduced tick effects may be achieved by decreasing moose density through harvest and managing late winter habitat to minimize tick density. Management of white-tailed deer ( Odocoileus virginianus ) density may also affect the transmission of meningeal worm.

Vermont↗
USGS2021

Juvenile moose stress and nutrition dynamics related to winter ticks, landscape characteristics, climate-mediated factors and survival

Moose populations in the northeastern United States have declined over the past 15 years, primarily due to the impacts of winter ticks. Research efforts have focused on the effects of winter tick infestation on moose survival and reproduction, but stress and nutritional responses to ticks and other stressors remain understudied. We examined the influence of several environmental factors on moose calf stress hormone metabolite concentrations and nutritional restriction in Vermont, USA. We collected 407 fecal and 461 snow urine samples from 84 radio-collared moose calves in the winters of 2017–2019 (January–April) to measure fecal glucocorticoid metabolites (fGCM) concentrations and urea nitrogen:creatinine (UN:C) ratios. We used generalized mixed-effects models to evaluate the influence of individual condition, winter ticks, habitat, climate and human development on stress and nutrition in calf moose. We then used these physiological data to build generalized linear models to predict calf winter survival. Calf fGCM concentrations increased with nutritional restriction and snow depth during adult winter tick engorgement. Calf UN:C ratios increased in calves with lighter weights and higher tick loads in early winter. Calf UN:C ratios also increased in individuals with home ranges composed of little deciduous forests during adult winter tick engorgement. Our predictive models estimated that winter survival was negatively related to UN:C ratios and positively related to fGCM concentrations, particularly in early winter. By late March, as winter ticks are having their greatest toll and endogenous resources become depleted, we estimated a curvilinear relationship between fGCM concentrations and survival. Our results provide novel evidence linking moose calf stress and nutrition, a problematic parasite and challenging environment and winter survival. Our findings provide a baseline to support the development of non-invasive physiological monitoring for assessing environmental impacts on moose populations.

Vermont↗
USGS2021

Moose habitat selection and fitness consequences during two critical winter tick life stages in Vermont, United States

The moose ( Alces alces ) is a charismatic species in decline across much of their southern distribution in North America. In the northeastern United States, much of the reduction has been attributed to winter tick ( Dermacentor albipictus ) infestations. Winter ticks are fairly immobile throughout all life stages, and therefore their distribution patterns at any given time are shaped largely by the occurrence of moose across the landscape during the peak of two critical time periods: fall questing (when ticks latch onto moose) and spring drop-off (when engorged female ticks detach from moose). We used recent land cover and lidar data within a dynamic occupancy modeling framework to estimate first-order habitat selection (use vs. non-use) of female moose ( n = 74) during the tick questing and drop-off periods. Patch extinction and colonization rates between the fall questing and spring drop-off periods were strongly influenced by habitat and elevation, but these effects were diminished during the fall questing period when moose were more active across the landscape. From the fall questing period to the spring drop-off period, patches where colonization was high and extinction was low had higher proportions of young (shrub/forage) mixed forest at higher elevations. Further, we evaluated the fitness consequences of habitat selection by adult females during the fall questing period, when females and their calves acquire ticks. We compared Resource Selection Functions (RSF) for five females that successfully reared a calf to age 1 with five females whose calves perished due to ticks. Adult female moose whose offspring perished selected habitats in the fall that spatially coincided with areas of high occupancy probability during the spring tick drop-off period. In contrast, adult female moose whose offspring survived selected areas where the probability of occupancy during the spring drop-off was low; at present, natural selection may favor female adults who do not select the same habitats in fall as in spring. Our model coefficients and mapped results define “hotspots” that are likely encouraging the deleterious effects of the tick-moose cycle. These findings fill knowledge gaps about moose habitat selection that may improve the effectiveness of management aimed at reversing declining population trends.

Vermont↗
USGS2021

Remote ecological monitoring with smartphones and tasker

Researchers have increasingly used autonomous monitoring units to record animal sounds, track phenology with timed photographs, and snap images when triggered by motion. We piloted the use of smartphones to monitor wildlife in the Riverside East Solar Energy Zone (California) and at Indiana Dunes National Park (Indiana). For both efforts, we established remote autonomous monitoring stations in which we housed an Android smartphone in a weather-proof box mounted to a pole and powered by solar panels. We connected each smartphone to a Google account, and the smartphone received its recording/photo schedule daily via a Google Calendar connection when in data transmission mode. Phones were automated by Tasker, an Android application for automating cell phone tasks. We describe a simple approach that could be adopted by others who wish to use nonproprietary methods of data collection and analysis.

California, Indiana↗
USGS2021

Modeling moose habitat use by age, sex, and season in Vermont, USA using high-resolution lidar and national land cover data

Moose ( Alces alces ) populations have experienced unprecedented declines along the southern periphery of their range, including Vermont, USA. Habitat management may be used to improve the status of the population and health of individuals. To date, however, Vermont wildlife managers have been challenged to effectively use this important tool due to the lack of fine-scale information on moose space use and habitat characteristics. To assess habitat use, we combined more than 40,000 moose locations collected from radio-collared individuals (n = 74), recent land cover data, and high resolution, 3-dimensional lidar ( light detection and ranging ) data to develop Resource Utilization Functions (RUF) by age (mature and young adult), season (dormant and growth), and sex. Each RUF linked home range use to average habitat conditions within 400 m or 1 km of each 30 m 2 pixel within the home range. Across analyses, the top RUF models included both composition (as measured through the National Land Cover Database) and structure (as measured through lidar) variables, and significantly outperformed models that excluded lidar variables. These findings support the notion that lidar is an effective tool for improving the ability of models to estimate patterns of habitat use, especially for larger bodied mammals. Generally speaking, female moose actively used areas with proportionally more regenerating forest (i.e., forage < 3.0 m) and more mature forest (i.e., canopy structure > 6.0 m), while males actively used more high elevation, mixed forest types. Further, moose exhibited important seasonal differences in habitat use that likely reflect temporal changes in energetic and nutritional requirements and behavior across the year. Moose used areas with proportionally more regenerating forest (i.e., forage < 3.0 m) during the growth period and female moose had strong positive associations with lidar-derived canopy structure during the growth (but not the dormant) period. Ultimately, the resultant maps of habitat use provide a means of informing management activities (e.g., the restoration or alteration of habitats to benefit moose) and policies around land use that may contribute to population recovery.

Vermont↗