UMD Theses and Dissertations
Permanent URI for this collectionhttp://hdl.handle.net/1903/3
New submissions to the thesis/dissertation collections are added automatically as they are received from the Graduate School. Currently, the Graduate School deposits all theses and dissertations from a given semester after the official graduation date. This means that there may be up to a 4 month delay in the appearance of a given thesis/dissertation in DRUM.
More information is available at Theses and Dissertations at University of Maryland Libraries.
Browse
27 results
Search Results
Item ANTHROPOGENIC INFLUENCES ON BOTTOM-UP AND TOP-DOWN REGULATION OF ANIMAL DISTRIBUTIONS, POPULATIONS, AND BEHAVIORS IN URBAN ENVIRONMENTS(2024) Herrera, Daniel Joseph; Gallo, Travis; Environmental Science and Technology; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Animal populations are simultaneously governed by both bottom-up (e.g., habitat availability) and top-down (e.g., predation) regulation. While ecologists historically sought to differentiate the roles of bottom-up and top-down regulation on ecosystems, the two are not so easily defined in urban ecosystems due to the immense influence humans have on ecological processes in cities. In Chapter One, I present this argument from a philosophical perspective and comment on how this philosophy has shaped my worldview. In Chapter Two, I examine the legacy of historical park planning on urban bird assemblages using archived municipal maps and historical bird data. My analysis found a positive correlation between percent park area and both species richness and functional richness of birds. Additionally, I found the effect size of park area was larger than the effect of certain life history traits thought to facilitate urban exploitation. These results indicate that landscape features and life history traits are equally responsible for the success of synurbic species. Chapter Three explores the effect of urbanization on animal behavior by analyzing anti-predator behavior of white-tailed deer (Odocoileus virginianus) in relation to ambient light, noise, and human activity. Despite negligible predation risk in my study area, deer expressed higher vigilance behavior in dark and noisy conditions, and increased their foraging group size during noisy conditions. These results suggest that anti-predator behaviors are a response to the perception of predation risk rather than a response to the actual presence of predators. Although predation of deer is rare in urban ecosystems, predation of smaller wildlife species by mesopredators, such as non-native domestic cats (Felis catus) is common. Chapter Four examines the potential for predation and zoonotic disease transmission between cats and eight native mammals by estimating the spatial and temporal overlap between species. I found that cat distribution was largely driven by anthropogenic features, whereas native wildlife was generally deterred by anthropogenic features and instead occupied forested areas. I also found that cats, as a species, were active on the landscape during the full 24-hour cycle. As a result, while spatial overlap between cats and wildlife varied across the study area, temporal overlap was possible anywhere cats and wildlife co-occurred. Chapter Five expands on Chapter Four and investigates predation directly by using observations of cats carrying prey documented by motion-activated cameras. I found that predation by cats was higher in areas where supplemental cat food was prevalent, but declined near forested areas. Additionally, my results indicate that cats within 250 meters of a forest edge predominantly preyed on native wildlife, whereas cats generally preyed on non-native rats (Rattus norvegicus) when greater than 250 meters from a forest edge. Each chapter provides applied recommendations to the management and conservation of urban wildlife, but together, my work demonstrates the entanglement of bottom-up, top-down, and anthropogenic forces in urban ecosystems. In light of these findings, I advocate for a more nuanced understanding of ecosystem regulation through a socio-ecological lens.Item OVARIAN STROMAL CELLS IMPROVE SURVIVAL, BUT NOT GROWTH, IN PRE- AND EARLY ANTRAL FELINE FOLLICLES(2024) Marks, Batsheva Naomi; Keefer, Carol; Animal Sciences; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Ovarian stromal cells act as crucial support and regulators for in vivo folliculogenesis; however, less is known about their effect on in vitro grown follicles. The objective of this study was to investigate the impact of ovarian stromal cell co-culture or conditioned medium (CM) on survival and development of cat pre-, early, and antral follicles in vitro. Ovaries were obtained from cats older than six months (n = 3), then enzymatically digested to release stromal cells. The ovarian stromal cells were allowed to grow to confluency in a T75 flask, before being cryopreserved for long term storage in liquid nitrogen. Cells were thawed one week prior to follicular culture onset, and passaged once before CM collection. CM was subsequently removed 24 - 48 hours after feeding, and stored at -80C until used. Ovarian follicles were mechanically isolated from cats older than six months (n = 23 cats, 155 follicles), encapsulated in 0.5% alginate hydrogel. The isolated follicles were then divided into five treatment groups (control, ovarian stromal cell co-culture, 20% CM, 50% CM, and 100% CM in Endothelial Cell Growth Medium), and classified based on initial diameter as preantral (224.4 + 4.7 m), early antral (394.8 + 7.4 m), or antral (592.2 + 18.8 m). Culture subsequently lasted for 13 days, and survival and growth of the follicles were evaluated on Days 0, 4, 6, 8, 11 and 13. At the end of culture, follicles were assessed via qRT-PCR for expression of CYP19A, FSHR, and GDP9 to further quantify development. Statistical analysis was done in R software. Follicles in 100% CM had higher survival up to Day 11 of culture as compared to other treatment groups (Cox proportional hazards model, p < 0.01). Initial stage also influenced survival, with antral follicle survival significantly lower than that of pre- and early antral follicles (p < 0.0001). However, no differences in growth were detected across the treatment groups, nor across initial size classifications (Kruskal-Wallis test, p > 0.05). Post culture qRT-PCR analysis of the three selected genes showed upregulation of CYP19A in 50% CM follicles compared to the control (ANOVA, p < 0.05). However, there were no differences in CYP19A expression between the control and other treatment groups, or in GDF9 and FSHR expression among culture groups (p > 0.05). In summary, the findings demonstrated that conditioned medium collected from primary culture of ovarian stromal cells improves in vitro survival and modulates CYP19A expression of isolated cat follicles. Further research to identify paracrine factors present in conditioned medium will elucidate the roles of ovarian stromal cells pertaining to follicle survival during in vitro folliculogenesis.Item Monitoring Aboveground Biomass in Forest Conservation and Restoration Areas Using GEDI and Optical Data Fusion(2024) Liang, Mengyu; Duncanson, Laura I; Geography; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Forests play a critical role in the global carbon cycle by sequestering carbon in the form of aboveground biomass. Area-based conservation measures, such as protected areas (PAs), are a cornerstone conservation strategy for preserving some of the world's most at-risk forest ecosystems. Beyond PAs, tree planting and forest restoration have been lauded as solutions to combat climate change and criticized as ways for polluters to offset carbon emissions. Consistent monitoring and quantification of forest restoration can impact decisions on future restoration activities. In this dissertation, I utilized a fusion of remote sensing assets and a combination of remote sensing with impact assessment techniques, to obtain objective baseline information for reconstructing past forest biomass conditions, and for monitoring and quantifying the patterns and success of forest regrowth in areas that underwent different forest management interventions. This overarching research goal is approached in three studies corresponding to chapters 2-4. In chapter 2, PAs’ effectiveness in storing biomass carbon and preserving forest structure is assessed on a regional scale using Global Ecosystem Dynamics Investigation (GEDI) lidar data in combination with a counterfactual analysis using statistical matching. This chapter provides an assessment of the reference condition of the biomass carbon storage capacity by one of the most stringent forest management means. The study finds that analyzed PAs in Tanzania possess 24.4% higher biomass densities than their unprotected counterparts and highlights that community-governed PAs are the most effective category of PAs at preserving forest structure and aboveground biomass density (AGBD). In chapter 3, empirical models are developed to link current (2019-2020) AGBD estimates from the GEDI with Landsat (2007-2019) at a regional scale. This will allow both current wall-to-wall biomass mapping and estimation of biomass dynamics across time. We demonstrate the utility of the method by applying it to quantify the AGBD dynamics associated with forest degradation for charcoal production. In chapter 4, the same modeling framework laid out in chapter 3 will be used to derive AGBD trajectories for 27 forest restoration sites across three biomes in East Africa. To assess the effectiveness of and compare Assisted Natural Regeneration (ANR) and Active Restoration (AR) in enhancing forest AGBD growth compared to natural regeneration (NR), we used staggered difference-in-difference (staggered DiD) to analyze the average annual AGBD change. We controlled for pre-intervention AGBD change rate between AR/ANR and NR and estimated the effectiveness with explicit consideration of intervention duration. This study finds that AR and ANR outperform NR during long-term restoration. Using the most suitable restoration interventions in each biome and timeframe, 4% suitable areas could enhance 2.40 ± 0.78 Gt (billion metric tons) forest carbon uptake over 30 years, equivalent to 3.6 years of African-wide emissions. Overall, this dissertation develops remote sensing methodological frameworks for using GEDI data and its fusion with Landsat time series to quantify and monitor forest AGBD. Moreover, by combining remote sensing-derived AGBD dynamics with impact assessment techniques, such as statistical matching and staggered DiD, the dissertation further assesses and compares different conservation and restoration means’ effectiveness in increasing AGBD and carbon uptake in forests. The dissertation therefore advances the applications of state-of-the-art remote sensing data and techniques for sustainably managing forests towards climate mitigation targets.Item EVALUATING ROADSIDE INTEGRATED VEGETATION MANAGEMENT (IVM) TECHNIQUES TO IMPROVE POLLINATOR HABITAT(2023) Kuder, Lisa Jennifer; vanEngelsdorp, Dennis; Entomology; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Pollinator declines have spurred interest in managing road verges as early successional habitats that promote wildflowers. Reduced mowing is a common cost-effective pollinator habitat improvement strategy. Studies on verge management demonstrate the benefits of this method, but also highlight downsides including dispersal of weed seeds by mowing machinery that facilitates establishment and persistence of noxious weeds. Herbicides are commonly used on electric powerline rights-of-way (ROW) to manage invasive plants, as well as establish and maintain low-growing pollinator habitat. This method, however, has not been tested for its suitability for verge management.To explore this knowledge gap, we compare pollinator habitat quality and bee abundance in six Maryland, USA roadsides under three different vegetation management types: frequent mowing, reduced mowing, and spot-spray. For two growing seasons, we took standardized photographs of road verge transects to measure foliar cover using image analysis software (percent cover sampling), counted floral units in fixed quadrats (quadrat sampling), and recorded all insect-pollinated blooms throughout treatment areas and assigned perceived abundance categories to each species (scanning sampling). Management types were compared with respect to the density, native status and diversity of roadside flowers, the proportion of bare ground and leaf litter, and bee abundance. We found that road verges managed under spot spray and fall mow resulted in improved pollinator forage with higher vascular plant cover (4x), floral density (4 and 7x), proportion of native plant species (1.9 and 2.5x), and Hill number’s diversities (3 and 4x) compared to the control. The two treatment groups did not differ in measured metrics, with three exceptions: Spot spray sites had higher proportions of native floral units and potential nesting sites (bare and leaf covered ground); whereas fall mow increased plant species diversity. Surprisingly, we did not detect a significant difference in bee abundance among the three management regimes. However, season, site and year were major predictors of bee abundance. It is not unusual for bee populations to differ substantially throughout and between years Our study demonstrates that during the early transition stage from turf to meadow, spot spraying was similar to fall mowing at promoting roadside pollinator forage. Also, spot spraying resulted in higher proportions of bare ground and leaf litter that could potentially provide nesting and overwintering habitat. Thus there are alternatives to reduced mowing that may have an important role in future conservation efforts. Proper and selective use of herbicides may be especially relevant for areas where noxious weeds are pervasive, or species of concern are at risk from destructive biomass removal (i.e., nesting birds, turtles, and butterfly larvae).Item MULTISCALE, MULTITEMPORAL ASSESSMENT OF CHIMPANZEE (Pan troglodytes) HABITAT USING REMOTELY SENSED DATASETS(2023) Jantz, Samuel M; Hansen, Matthew C; Geography/Library & Information Systems; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)All four sub-species of our closest living relative, the chimpanzee, are listed as endangered by the International Union for the Conservation of Nature (IUCN), and their populations continue to decline due to human activities. Effective conservation efforts require information on their population status and distribution. Traditional field surveys are expensive and impractical for covering large areas at regular time intervals, making it difficult to track population trends. Given that chimpanzees occupy a large range (2.3 x 106 km2), new cost-effective methods and data are needed to provide relevant information on population status and trends across large geographic and time scales. The objective of this dissertation is to help fill this gap by leveraging freely available and regularly updated remotely sensed datasets to map and monitor chimpanzee habitat across their range. This research begins by first producing annual forest cover and change maps for the Greater Gombe (GGE) and Greater Mahale ecosystems (GME) in western Tanzania using Landsat phenological metrics and machine learning methods. Canopy cover was predicted at 30-meter resolution and the Cumulative Sums (CuSum) algorithm was applied to the canopy cover time series to detect forest loss and gain events between 2000-2020. An accuracy assessment showed the CuSum algorithm was able to detect forest loss well but had more difficulty detecting gradual forest gain events. A total of 276,000 ha (+/- 27,000 ha) of gross forest loss was detected between 2000 and 2020 in the GGE and GME; however, loss was not spread equally among the two ecosystems. The results show widespread forest loss in the GME, contrasted with net forest cover gain in the GGE. Next, the annual forest cover maps, and additional derived variables, were used to train an ensemble model to predict the relative encounter rate of chimpanzee nest sightings in the GGE and GME. Model output exhibited a strong linear relationship to chimpanzee abundances and population density estimated from a recent ground survey, enabling model output to be linearly transformed into population estimates. The model predicted the two ecosystems harbor just over 3,000 individuals, which agrees with the upper limit of population estimates from ground surveys. Most importantly, the model can be applied to annually updated variables enabling the detection of potential population shifts caused by changes in landscape condition. Model output indicates a possible population reduction in portions of the GME, while the GGE is predicted to have increased its ability to sustain a larger population. Finally, Random Forests regression was used to relate predictor variables, primarily derived from Landsat data to a coarse resolution, range-wide habitat suitability map enabling the prediction of habitat suitability at 30 meter resolution. The model showed good agreement with the calibration data; however, there was considerable variation in predictive capability among the four chimpanzee sub-species. Elevation, Landsat ETM+ band 5 and Landsat derived canopy cover were the strongest predictors; highly suitable areas were associated with dense tree canopy cover for all but the Nigeria-Cameroon and Central Chimpanzee sub-species. The model can detect changes in suitability to support monitoring and conservation planning across the chimpanzee range. Results from this dissertation highlight the promise of integrating continuously updated satellite data into habitat suitability models to detect changes through time and inform conservation efforts for chimpanzees at multiple scales.Item TO WHAT EXTENT DO MODE OF REPRODUCTION, LEVELS OF GENOTYPIC DIVERSITY, AND CONNECTIVITY IN Vallisneria americana MICHX. CONFER RESILIENCE TO A CHANGING CLIMATE?(2023) Perkins, Carrie; Neel, Maile C.; Plant Science and Landscape Architecture (PSLA); Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)The macrophyte Vallisneria americana Michx. (Hydrocharitaceae) is a foundational submersed aquatic vegetation (SAV) species that provides valuable ecosystem services, such as nutrition for waterfowl and shelter for fish. When healthy, V. americana can absorb excess nutrients from the water and stabilize sediments, but many of its meadows, which span freshwater to oligohaline environments in eastern North America, have been declining since European settlers cleared the land. Declines only intensified in the 1950s due to chronic environmental stressors and major storm events. To determine the extent to which remaining populations can adapt through natural selection or acclimate to novel environmental conditions, I combined observational field data, greenhouse experiments, and spatial modeling to quantify V. americana reproduction at local to regional scales, evaluate evidence of local adaptation and acclimation to environmental stress, and assess the extent to which high levels of connectivity in a V. americana-dominated landscape can absorb environmental stress.I quantified reproduction at 15 sites in the Chesapeake Bay and 14 sites in the Hudson River, with sites in each geographic region spanning the portion of the salinity gradient in which V. americana grows (0-12 ppt). Numbers of inflorescences, sex ratios, and distances among male and female inflorescences varied greatly across latitude and along salinity gradients. Hudson V. americana had fewer inflorescences across two sampling seasons than Chesapeake Bay V. americana but delayed phenology, skewed sex ratios, and large distances among males and females relative to the Chesapeake Bay were more pronounced in 2018. In 2018, warmer spring and summer water temperatures in the Chesapeake coincided with our findings of higher flowering, fruiting, and potential for pollination at the three Chesapeake sites that served as means of comparison to the Hudson. By contrast, in 2020 Hudson plants were larger and produced more inflorescences in July than Chesapeake plants produced in June, indicating that the regional difference in phenology may be smaller than our hypothesis of approximately 23 days, although it is difficult to estimate how much smaller. We attribute this result to sites in the Hudson – mainly those in the tidal-fresh zone of the river – being highly responsive to unusually warm 2020 spring water temperatures. But not all sites experienced this warmth. The tidal-saline zone of the Hudson and the non-tidal zone of the Chesapeake had the fewest flowers and fruits of either region, likely due to the synergistic effects of cold temperatures and high salinity and turbidity in the former and fast currents in the latter inhibiting growth and reproduction. Through greenhouse experiments evaluating growth and reproduction of Chesapeake and Hudson V. americana grown in different salinity conditions, we found evidence of one-way local adaptation in plants sourced from brackish waters of both the Chesapeake and Hudson. In the first experiment (parental-generation), brackish-source plants demonstrated phenotypic buffering, a stress-induced version of phenotypic plasticity. When exposed to three salinity treatments (0 ppt, 6 ppt, and 12 ppt) applied after plants had sprouted, brackish-source plants buffered the effects of salt stress via increased vegetative growth in the form of many ramets and turions at the cost of small stature. By contrast, plants sourced from fresh waters of both regions grew tall in fresh water, but photosynthetic leaf material declined from the time of salt application (June) to the end of the experiment (September). The most severe salinity treatment, 18 ppt, was lethal to most individuals regardless of source habitat. Unfortunately, neither phenotypic buffering nor phenotypic plasticity sensu stricto was carried over via transgenerational plasticity (TGP), when turions were exposed to 12 ppt immediately upon planting (offspring generation). This early-development salt exposure proved lethal for some individuals and sublethal (had a negative effect on growth but did not result in mortality) for others, with turions either failing to sprout or growing a single shoot that was minuscule in stature. Parental-generation salt exposure only exacerbated these offspring effects, producing a non-adaptive TGP effect, resulting in even lower chance of sprouting, higher chance of mortality, and smaller stature. Evidence of local adaptation and acclimation to salinity only when exposure begins later in development suggests that populations have potential for resilience to saltwater intrusion (movement of saline water into fresh water) only if salinities do not remain elevated during the time of early plant development (spring/early summer) and across multiple seasons. In the event of prolonged salinity stress, much habitat (~10,000 hectares) that is currently mesohaline (5-12 ppt) but within the range of tolerance for V. americana will become unsuitable. In our spatial model of SAV persistence in the V. americana-dominated Upper Chesapeake Bay, high connectivity and high probability of SAV presence were found not only in the freshwater head of the Bay, but also in mesohaline (5-12 ppt) and oligohaline (0.5-5 ppt) waters near Middle River. Persistence of predominantly freshwater aquatic macrophytes in Middle River suggests that either 1) plants are locally adapted to brackish waters or 2) existing connectivity buffers the stress of low-quality habitat. Excess nitrogen, an anthropogenic environmental stressor that remains at high levels in Baltimore Harbor and other tributaries, was correlated with a decreased probability of SAV presence in the southern portion of our study area. As expected, low nitrogen, low salinity, and high landscape connectivity at the head of the Bay coincided with the highest predicted probabilities of SAV presence, particularly in the core of the one of the largest SAV beds in the entire Chesapeake Bay, the Susquehanna Flats.Item INFORMING CONSERVATION OF THREATENED BAT SPECIES USING GENOMICS AND ACOUSTICS(2022) Nagel, Juliet Joy; Nelson, David; Marine-Estuarine-Environmental Sciences; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Bats are vital to healthy ecosystems, providing billions of dollars of services in the form of forest and crop pest control. Unfortunately, North American bat populations have faced novel pressures during the past decade that may threaten their persistence. First, several species of tree-roosting bats (primarily hoary [Lasiurus cinereus], eastern red [L. borealis], and silver-haired [Lasionycteris noctivagans] bats) are experiencing large numbers of fatalities at industrial wind-energy facilities. Second, several species of cave-dependent bats have experienced large-scale mortality as the result of infection by a fungal pathogen that causes white-nose syndrome (WNS). As bats are generally long-lived and have low reproductive rates, such increases in mortality can cause significant population declines from which they may be unable to recover. Basic questions about population trends, size and structure remain largely unanswered for these species because of challenges in applying traditional wildlife monitoring approaches to bats. This lack of understanding impedes conservation and management efforts. In my dissertation, I use genomic and acoustic survey techniques to investigate questions related to the threats that wind-energy development and WNS are posing to bat species in North America. In my first chapter, I evaluate range-wide population structure and effective population size (Ne) for hoary, eastern red, and silver-haired bats. Using genotyping-by-sequencing (GBS), I genotyped single-nucleotide polymorphism (SNP) data from 173 hoary, 113 eastern red, and 89 silver-haired bats from multiple locations spread across their geographic distributions. Hoary bats and eastern red bats showed no geographic structure in genetic diversity, whereas silver-haired bats displayed longitudinal population variation. Coalescent modeling suggested that eastern red bats have the largest evolutionary Ne, followed by hoary bats, then silver-haired bats. In my second chapter, I used GBS to assess the population structure of two federally endangered cave bat species: Indiana bats (Myotis sodalis) and gray bats (M. grisescens). Using tissue samples from 45 Indiana bats and 47 gray bats spread across their ranges, I showed that Indiana bats display no geographic genetic structure, whereas gray bats exhibit east–west population variation across the Mississippi River Valley. In my final chapter, I used acoustic surveys across the State of Maryland to investigate bat community changes in the decade following the arrival of WNS. From 2010 through 2019, I conducted annual mobile acoustic routes each summer, for a total of 344 completed routes resulting in 426 hours of recordings and 24,375 identified bat passes. I detected massive (> 92%) declines of little brown bats (M. lucifugus), northern long-eared bats (M. septentrionalis), eastern small-footed bats (M. leibii), and tricolored bats (Perimyotis subflavus), with no evidence of recovery in recent years. Trends in hoary bats and eastern red bats were non-significant during this period. Bat community composition varied among Maryland’s physiographic regions, with eastern red bats comprising a larger percentage in the east. Species composition across the state likely reflects the impact of several factors, including mortality from WNS and wind-energy development, and perhaps reduced interspecific competition. Overall, my results illustrate the unique insights, but also distinct limitations, that genomic and acoustic data can provide regarding the conservation of bats in North America.Item Algorithms for scalable and efficient population genomics and metagenomics(2022) Javkar, Kiran Gajanan; Pop, Mihai; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Microbes strongly impact human health and the ecosystem of which they are a part. Rapid improvements and decreasing costs in sequencing technologies have revolutionized the field of genomics and enabled important insights into microbial genome biology and microbiomes. However, new tools and approaches are needed to facilitate the efficient analysis of large sets of genomes and to associate genomic features with phenotypic characteristics better. Here, we built and utilized several tools for large-scale whole-genome analysis for different microbial characteristics, such as antimicrobial resistance and pathogenicity, that are important for human health. Chapters 2 and 3 demonstrate the needs and challenges of population genomics in associating antimicrobial resistance with genomic features. Our results highlight important limitations of reference database-driven analysis for genotype-phenotype association studies and demonstrate the utility of whole-genome population genomics in uncovering novel genomic factors associated with antimicrobial resistance. Chapter 4 describes PRAWNS, a fast and scalable bioinformatics tool that generates compact pan-genomic features. Existing approaches are unable to meet the needs of large-scale whole-genome analyses, either due to scalability limitations or the inability of the genomic features generated to support a thorough whole-genome assessment. We demonstrate that PRAWNS scales to thousands of genomes and provides a concise collection of genomic features which support the downstream analyses. In Chapter 5, we assess whether the combination of long and short-read sequencing can expedite the accurate reconstruction of a pathogen genome from a microbial community. We describe the challenges for pathogen detection in current foodborne illness outbreak monitoring. Our results show that the recovery of a pathogen genome can be accelerated using a combination of long and short-read sequencing after limited culturing of the microbial community. We evaluated several popular genome assembly approaches and identified areas for improvement. In Chapter 6, we describe SIMILE, a fast and scalable bioinformatics tool that enables the detection of genomic regions shared between several assembled metagenomes. In metagenomics, microbial communities are sequenced directly without culturing. Although metagenomics has furthered our understanding of the microbiome, comparing metagenomic samples is extremely difficult. We describe the need and challenges in comparing several metagenomic samples and present an approach that facilitates large-scale metagenomic comparisons.Item ECOLOGICAL APPLICATIONS OF MACHINE LEARNING TO DIGITIZED NATURAL HISTORY DATA(2022) Robillard, Alexander John; Rowe, Christopher; Bailey, Helen; Marine-Estuarine-Environmental Sciences; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Natural history collections are a valuable resource for assessment of biodiversity and species decline. Over the past few decades, digitization of specimens has increased the accessibility and value of these collections. As such the number and size of these digitized data sets have outpaced the tools needed to evaluate them. To address this, researchers have turned to machine learning to automate data-driven decisions. Specifically, applications of deep learning to complex ecological problems is becoming more common. As such, this dissertation aims to contribute to this trend by addressing, in three distinct chapters, conservation, evolutionary and ecological questions using deep learning models. For example, in the first chapter we focus on current regulations prohibiting the sale and distribution of hawksbill sea turtle derived products, which continues internationally in physical and online marketplaces. To curb the sale of illegal tortoiseshell, application of new technologies like convolutional neural networks (CNNs) is needed. Therein we describe a curated data set (n = 4,428) which was used to develop a CNN application we are calling “SEE Shell”, which can identify real and faux hawksbill derived products from image data. Developed on a MobileNetV2 using TensorFlow, SEE Shell was tested against a validation (n = 665) and test (n = 649) set where it achieved an accuracy between 82.6-92.2% correctness depending on the certainty threshold used. We expect SEE Shell will give potential buyers more agency in their purchasing decision, in addition to enabling retailers to rapidly filter their online marketplaces. In the second chapter we focus on recent research which utilized geometric morphometrics, associated genetic data, and Principal Component Analysis to successfully delineate Chelonia mydas (green sea turtle) morphotypes from carapace measurements. Therein we demonstrate a similar, yet more rapid approach to this analysis using computer vision models. We applied a U-Net to isolate carapace pixels of (n = 204) of juvenile C. mydas from multiple foraging grounds across the Eastern Pacific, Western Pacific, and Western Atlantic. These images were then sorted based on general alignment (shape) and coloration of the pixels within the image using a pre-trained computer vision model (MobileNetV2). The dimensions of these data were then reduced and projected using Universal Manifold Approximation and Projection. Associated vectors were then compared to simple genetic distance using a Mantel test. Data points were then labeled post-hoc for exploratory analysis. We found clear congruence between carapace morphology and genetic distance between haplotypes, suggesting that our image data have biological relevance. Our findings also suggest that carapace morphotype is associated with specific haplotypes within C. mydas. Our cluster analysis (k = 3) corroborates past research which suggests there are at least three morphotypes from across the Eastern Pacific, Western Pacific, and Western Atlantic. Finally, within the third chapter we discuss the sharp increase in agricultural and infrastructure development and the paucity of widespread data available to support conservation management decisions around the Amazon. To address these issues, we outline a more rapid and accurate tool for identifying fish fauna in the world's largest freshwater ecosystem, the Amazon. Current strategies for identification of freshwater fishes require high levels of training and taxonomic expertise for morphological identification or genetic testing for species recognition at a molecular level. To overcome these challenges, we built an image masking model (U-Net) and a CNN to mask and classify Amazonian fish in photographs. Fish used to generate training data were collected and photographed in tributaries in seasonally flooded forests of the upper Morona River valley in Loreto, Peru in 2018 and 2019. Species identifications in the training images (n = 3,068) were verified by expert ichthyologists. These images were supplemented with photographs taken of additional Amazonian fish specimens housed in the ichthyological collection of the Smithsonian’s National Museum of Natural History. We generated a CNN model that identified 33 genera of fishes with a mean accuracy of 97.9%. Wider availability of accurate freshwater fish image recognition tools, such as the one described here, will enable fishermen, local communities, and citizen scientists to more effectively participate in collecting and sharing data from their territories to inform policy and management decisions that impact them directly.Item Ecological Restoration Drives Functional Composition and Diversity in Urban Forest Patches(2020) Do, Sara Miya; Johnson, Lea R; Plant Science and Landscape Architecture (PSLA); Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Urbanization greatly alters environmental conditions, affecting biodiversity in cities and ecological processes. To restore processes and native biodiversity, land managers have turned to ecological restoration of urban forest patches. Urban forest patches, nested within urban ecosystems, are subject to urban influences during ecological succession. Building on a long-term study evaluating outcomes of ecological restoration in New York City, I examined the effects of urban conditions, restoration, and forest succession on functional composition and diversity of restored and unrestored urban forest patches after 15-20 years. Functional traits play an essential role in community assemblages and influence the resilience and ecosystem functioning of urban ecosystems. I found that restored plots had greater functional evenness. Differences in functional composition indicated direct influence from restoration, succession, urban conditions, and success in meeting restoration goals. These results demonstrate that ecological restoration drives changes in functional composition and diversity of urban forest patches.
- «
- 1 (current)
- 2
- 3
- »