USGS Science⌕ Search

SEARCH · USGS Science

Results for “Conservation Ecology”

Search indexed USGS publications on groundwater, aquifers, geologic maps, mineral resources and earthquakes. Explore source records by subject and place.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 343 records · Page 19Linked to original sources
USGS2022

Movement of Canada geese in urban and rural areas of Iowa, USA

Temperate-breeding Canada Goose ( Branta canadensis maxima ) abundance has increased to previously unrecorded levels, providing social, ecological, and economic value. However, there are also costs associated with abundant Canada Geese. Although hunter harvest is a valued, sustainable use of Canada Geese, the adaptability of geese to urban areas may result in lower susceptibility of geese to hunters, potentially reducing the contribution of hunter harvest to conflict reduction. Our goal was to compare movement of geese marked in urban and rural areas to assess efficacy of hunter harvest in managing urban goose populations. We marked 71 adult female Canada Geese during brood-rearing with GPS-GSM transmitters in urban (n = 45) and rural (n = 26) locations in Iowa, USA to monitor movement during the 2018-19 and 2019-20 Mississippi Flyway goose hunting season frameworks. We estimated the mean proportion of locations each group was available for hunter harvest and examined factors affecting home range areas using generalized linear mixed models. Additionally, we estimated habitat selection of urban- and rural-marked geese using a step-selection analysis. Urban geese had a lower proportion of locations in areas available to hunters (0.07 [95% CI: 0.04-0.13]) than rural geese (0.56 [95% CI: 0.34-0.76]), but median home range area was similar for each group and decreased in size from autumn to late winter. Canada Geese marked in urban areas were more likely to select developed areas and less likely to select wetlands than rural geese, and they had high selection of agricultural fields within city limits during goose hunting seasons. Although Canada Geese breeding in urban areas may be less available for hunter harvest, movement data show when and where opportunity exists to increase harvest susceptibility. Canada Goose management could require actions in addition to hunter harvest to achieve goals in urban areas.

Iowa↗
USGS2020

Predicting wildlife distribution patterns in New England USA with expert elicitation techniques

Understanding the impacts of landscape change on species distributions can help inform decision-making and conservation planning. Unfortunately, empirical data that span large spatial extents across multiple taxa are limited. In this study, we used expert elicitation techniques to develop species distribution models (SDMs) for harvested wildlife species (n = 10) in the New England region of the northeastern United States. We administered an online survey that elicited opinions from wildlife experts on the probability of species occurrence throughout the study region. We collected 3396 probability of occurrence estimates from 46 experts, and used linear mixed-effects methods and landcover variables at multiple spatial extents to develop SDMs. The models were in general agreement with the literature and provided effect sizes for variables that shape species occurrence. With the exception of gray fox, models performed well when validated against crowdsourced empirical data. We applied models to rasters (30 × 30 m cells) of the New England region to map each species’ distribution. Average regional occurrence probability was highest for coyote (0.92) and white-tailed deer (0.89) and lowest for gray fox (0.42) and moose (0.52). We then stacked distribution maps of each species to estimate and map focal species richness. Species richness ( s ) varied across New England, with highest average richness in the least developed states of Vermont ( s = 7.47) and Maine (s = 7.32), and lowest average richness in the most developed states of Rhode Island (s = 6.13) and Massachusetts (s = 6.61). Our expert-based approach provided relatively inexpensive, comprehensive information that would have otherwise been difficult to obtain given the spatial extent and range of species being assessed. The results provide valuable information about the current distribution of wildlife species and offer a means of exploring how climate and land-use change may impact wildlife in the future.

Connecticut, Maine, Massachusetts, New Hampshire, ↗
USGS2022

Automating sandhill crane counts from nocturnal thermal aerial imagery using deep learning

Population monitoring is essential to management and conservation efforts for migratory birds, but traditional low-altitude aerial surveys with human observers are plagued by individual observer bias and risk to flight crews. Aerial surveys that use remote sensing can reduce bias and risk, but manual counting of wildlife in imagery is laborious and may be cost-prohibitive. Therefore, automated methods for counting are critical to cost-efficient application of remote sensing for wildlife surveys covering large areas. We conducted nocturnal surveys of sandhill cranes ( Antigone canadensis ) during spring migration in the Central Platte River Valley of Nebraska, USA, using midwave thermal infrared sensors. We developed a framework for automated counting of sandhill cranes from thermal imagery using deep learning, assessed and compared the performance of two automated counting models, and quantified the effect of spatial resolution on counting accuracy. Aerial thermal imagery data were collected in March 2018 and 2021; 40 images were analyzed. We applied two deep learning models: an object detection approach, Faster R-CNN and a recently developed pixel-density estimation approach, ASPDNet. Model performance was determined using data independent of the training imagery. The effect of spatial resolution was quantified with a beta regression on relative error. Our results showed model accuracy of 9% mean percent error for ASPDNet and 18% for Faster R-CNN. Most error was related to the undercounting of sandhill cranes. ASPDNet had < 50% of the error of Faster R-CNN as measured by mean percent error, root-mean-squared error and mean absolute error. Spatial resolution affected accuracy of both models, with error rate increasing with coarser resolution, particularly with Faster R-CNN. Deep learning models, particularly pixel-density estimators, can accurately automate counting of migratory birds in a dense, aggregate setting such as nocturnal roosting sites.

Nebraska↗
USGS2025

Breaking down Palila decline: Assessing the role of drought and vegetation health in the population loss of an endangered Hawaiian honeycreeper

The Palila ( Loxioides bailleui ), the last member of the once speciose finch-billed Hawaiian honeycreeper clade (Drepanidinae) in the main Hawaiian Islands, faces critical conservation challenges as an endangered species. Understanding the drivers of its decline is essential for effective management. We used additive decomposition models to examine temporal trends in climatic variables (temperature, precipitation, drought) and Normalized Difference Vegetation Index (NDVI), a vegetation health metric hypothesized to be associated with long-term trends in Palila abundance at landscape (250 m) scales on the Island of Hawai'i. A breakpoint analysis identified 2005–2009 as critical years of Palila decline. Vegetation health metrics at the 250 m scale lined up well both spatially and temporally with trends in Palila declines, with a significant browning from January 2004 to January 2014. Given the strong correlation between vegetation health and drought metrics at the landscape scale (r = 0.75, p < 0.001), NDVI changes appeared driven by drought. To enable the future projection of habitat quality in this area, we explored a stepwise linear regression to explain the variation in MODIS NDVI in recent years. We found that 87 % of the variability in NDVI can be explained by wet season precipitation and vapor pressure deficit from the previous dry season. The model is largely driven by a strong positive correlation between wet season precipitation and NDVI (r = 0.72, adjusted p < 0.001). Areas that maintained a low likelihood of NDVI decline throughout the time series and experienced increases in predicted Palila count represent potential drought microrefugia for the species. This higher elevation microrefugia is likely resilient against decreases in wet season precipitation through supplemental water retention from fog drip. While NDVI rebounded after 2014, Palila have not recovered. Our analysis highlights the importance of trend decomposition for monitoring endangered species with limited rebound potential due to small population dynamics and indicate continued warm, dry conditions may prevent Palila recovery without intervention.

Hawaii↗
USGS2020

Social attraction used to establish Caspian tern nesting colonies in San Francisco Bay

Conservation of colonial waterbird breeding populations often includes restoring historic nesting habitat or establishing new nesting habitat in protected areas. However, colonization of new or restored nesting habitat may be hindered by the lack of social cues from nesting conspecifics to attract prospecting birds. Social attraction, whereby decoys and colony sound recordings are used to mimic active nesting colonies, has been used successfully to establish waterbird nesting colonies throughout the world. We constructed islands, modified the substrate so that it was attractive to nesting Caspian terns (Hydroprogne caspia), and then used social attraction to establish nesting colonies within two managed ponds in San Francisco Bay, California where Caspian terns had not previously nested. During the 2015–2017 breeding seasons, we deployed decoys of adult Caspian terns, broadcasted colony sound recordings, and monitored Caspian tern response. Caspian terns formed nesting colonies within weeks of social attraction deployment at each of the two ponds in 2015, and the size of these colonies increased in each subsequent year of the study. In 2017, the final year of the study, we estimated a minimum of 501 breeding pairs between the two colonies, making them two of the three largest Caspian tern colonies in the San Francisco Bay estuary. In total, these two colonies produced 1343 nests and 531 fledglings over the three-year study period. Nest densities were low (mean: 0.29 nests/m2 of active colony area) compared to other studies, and greater than 80% of the modified island habitat remained unused by nesting Caspian terns in 2017, suggesting that there is additional space for future colony growth. The successful establishment of two of the largest Caspian tern nesting colonies in the San Francisco Bay estuary in just three years demonstrates the potential of using island construction and habitat modifications, combined with social attraction measures to establish waterbird nesting colonies.

California↗
USGS2021

Refining the coarse filter approach: Using habitat-based species models to identify rarity and vulnerabilities in the protection of U.S. biodiversity

Preserving biodiversity and its many components is a priority of conservation science and how to efficiently allocate resources to preserve healthy populations of as many species, habitats, and ecosystems as possible. We used the U.S. Geological Survey (USGS) Gap Analysis Project (GAP) species models released in 2018, which identify predicted habitats for terrestrial vertebrates in the conterminous United States, to illustrate hotspots of biodiversity for the major taxonomic groups. This collection represents the first complete compilation of terrestrial vertebrate species models for the conterminous United States (U.S. Geological Survey (USGS), 2018a). We used the species models but not the available subspecies models; this resulted in the inclusion of 282 amphibian models, 621 bird models, 365 mammal models, and 322 reptiles in our analysis. We also used population trend information and made spatial queries to characterize species in three dimensions: geographic range (small or large), habitat breadth (narrow or wide), and population trend (decreasing vs stable or increasing). This characterization allowed us to divide the species into eight groups (A-H) with similar characteristics. Group A species (large geographic range, wide habitat breadth, and stable or increasing population trend) are species that are common now with no indication of becoming rare. Species B-H have theoretical or known characteristics that could lead them to become rare with the H species exhibiting small geographic range, narrow habitat breadth, and decreasing population trend. Finally, we evaluated the prevalence of mapped habitat on protected lands for each species, exploring the patterns of representation in the rare species groups by ecoregion. The species we identified with population and habitat use characteristics that potentially predispose them to being or becoming rare represented a large percentage of each taxon. Potentially rare species were widely distributed among ecoregions. Of the 20 ecoregions in the country, 14 have a greater number of rare species than the national average for at least one taxon. Protection of the habitat for the majority of these rare species is below that recommended (17% of available habitat) by the Convention on Biological Diversity (CBD). The Everglades ecoregion was the only ecoregion that protected more than half of its rare or potentially rare species.

Global Ecology and Conservation↗
USGS2022

Comparison of methods for estimating density and population trends for low-density Asian bears

Populations of bears in Asia are vulnerable to extinction and effective monitoring is critical to measure and direct conservation efforts. Population abundance (local density) or growth (λ ) are the most sensitive metrics to change. We discuss the value in implementing spatially explicit capture-recapture (SCR), the current gold standard for density estimation, and open population SCR (OPSCR) to monitor changes in density over time. We provide guidance for designing studies to provide estimates with sufficient power to detect changes. Because of the wide availability of camera traps and interest in their use, we consider six density estimation methods and their extensions developed for use with camera traps, with specific consideration of assumptions and applications for monitoring Asian bears. We conducted a power analysis to calculate the precision in estimates needed to detect changes in populations with reference to IUCN Red List criteria. We performed a systematic review of empirical studies implementing camera trap abundance estimation methods and considered sample sizes, effort, and model assumptions required to achieve adequate precision for population monitoring. We found SCR and OPSCR, reliant on “marked” individuals, are currently the only methods with enough power to reliably detect even moderate to major (20–80%) declines. Camera trap methods with unmarked individuals rarely achieved precision sufficient to detect even large declines (80–90%), although with some exceptions (e.g., situations with moderate population densities, large number of sampling sites, or inclusion of ancillary local telemetry data. We describe additional estimation options including line transects, direct observations, monitoring age-specific survival and reproductive rates, and hybrid/integrated methodologies that may have potential to work for some Asian bear populations. We conclude monitoring changes in abundance or density is possible for most Asian bear populations but will require collaboration among researchers over broad spatial extents and extensive financial investment to overcome biological and logistical constraints. We strongly encourage practitioners to consider study design and sampling effort required to meet objectives by conducting simulations, power analyses, and assumption checks prior to implementing monitoring efforts, and reporting standardized dispersion measures such as coefficients of variation to allow for assessment of precision. Our guidance is relevant to other low-density and wide-ranging species.

Global Ecology and Conservation↗
USGS2022

Review of field methods for monitoring Asian bears

Efficient and effective monitoring methods are required to assess population status and gauge efficacy of conservation actions for threatened species. Here we review the spectrum of field methods useful for monitoring distribution, occupancy, abundance, and population trend for the five species of Asian terrestrial bears. Methods reviewed include expert opinion, local knowledge, bear sign, visual observations, camera traps, DNA-based methods (hair and scat derived), and radio telemetry. We examine the application of each method in terms of realizing specific monitoring objectives, their assumptions, challenges, and advantages. Our goal is to assist researchers in matching appropriate field methods with sought-after project objectives and to highlight shortfalls and trade-offs. Methods vary greatly in terms of cost, logistics, required number and expertise of staff, and the reliability of the data they provide. Many Asian bear population assessments have relied on expert opinion, local interviews, and sign surveys to provide estimates of distribution, abundance, and trend, in part because these are inexpensive and relatively easy to employ. However, increasing use of camera traps and DNA-based methods now allow for better monitoring via occupancy or rigorous capture–recapture population estimation, with the caveat that these methods may be restricted by inadequate budgets or logistical constraints. For distribution monitoring, camera traps and DNA yield the most definitive records of presence, but in low density bear populations, sign and local knowledge may be more effective. For occupancy, camera traps and DNA are advantageous in providing definitive detections in known time periods. For abundance/density or population trend monitoring in relatively small areas (<10,000 km 2 ), bears must be individually identified and used in a mark-recapture design. This requires DNA from collections of hair or scat, or a camera-based survey in which natural chest marks are clearly visible and individually distinguishable. DNA-methods or camera traps within individual identification is best for population trend when sufficient funding is available. Alternatively, careful use of local knowledge or expert opinion may be viable options, but come with greater uncertainty. For the foreseeable future, we believe that expert opinion will likely continue to play a large part in monitoring Asian bears, but these opinions should be informed by more rigorous data from the other methods we discuss.

Global Ecology and Conservation↗
USGS2023

A regionally varying habitat model to inform management for greater sage-grouse persistence across their range

Identifying habitat needs for species with large distributions is challenging because species-habitat associations may vary across scales and regions (spatial nonstationarity). Furthermore, management efforts often cross jurisdictional boundaries, complicating the development of cohesive conservation strategies among management entities. The greater sage-grouse ( Centrocercus urophasianus ) is a rapidly declining species that spans 11 U.S. states and responds to habitat conditions across a wide range of spatial scales and regions. Allowing for regional variance in species-habitat associations and suitability predictions could systematically identify important habitats at levels relevant to management. We collaboratively developed a model with Bureau of Land Management (BLM) biologists that: (1) evaluated the scale of effect for different environmental covariates; (2) accounted for regional differences in population-level responses; and (3) predicted probabilities of persistence across the U.S. occupied range. We modeled range-wide lek persistence data (6615 communal breeding sites classified as active or inactive) as a function of environmental covariates. Environmental covariates included sagebrush cover, pinyon-juniper cover, topography, precipitation, point and line disturbance densities, and landscape configuration metrics. Our model treated habitat assessment areas – regionally delineated by BLM biologists – as random intercepts and slopes that allowed for geographic variation in species-habitat associations and predicted probabilities of lek persistence. Our final model indicated support for 12 environmental covariates predicting lek persistence at scales extending between 1- to 15-km radii from lek centers, and a covariate measuring distance to the occupied range boundary. Five of these covariates showed significant regionally varying responses: sagebrush clumpiness (a measure of habitat aggregation), pinyon-juniper cover, point disturbance of anthropogenic features such as energy infrastructure and communication towers, elevation, and a topographic index associated with mesic habitats. This spatial nonstationarity indicates unitary range-wide recommendations, or rules-of-thumb with respect to their effects on lek persistence, may be problematic for these environmental conditions. For covariates that did not include random slopes, and which were potentially amenable to management actions, we found that leks were predicted to become extirpated when sagebrush cover fell below 9.6 % (summarized at the 3.2-km radius extent), and the proportion of classified sagebrush habitat fell below 0.7 (1-km). We produced a continuous predictive probability surface of lek persistence which we binned based on model sensitivity thresholds to produce habitat quality categories. The highest quality habitat (capturing 50 % of active leks) covered 25.5 % of the occupied range, while the combined lowest through highest quality habitats (capturing 95 % of active leks) covered 65.0 %. Accommodating regional environmental differences in models that are relevant to habitat management planning will help ensure their applicability to targeted goals. Continuous collaboration between modelers and land managers early in the modeling process increases the likelihood of this outcome.

Global Ecology and Conservation↗
USGS2019

Invasive buffelgrass detection using high-resolution satellite and UAV imagery on Google Earth Engine

Methods to detect and monitor the spread of invasive grasses are critical to avoid ecosystem transformations and large economic costs. The rapid spread of non‐native buffelgrass (Pennisetum ciliare) has intensified fire risk and is replacing fire intolerant native vegetation in the Sonoran Desert of the southwestern US. Coarse‐resolution satellite imagery has had limited success in detecting small patches of buffelgrass, whereas ground‐based and aerial survey methods are often cost prohibitive. To improve detection, we trained 2 m resolution DigitalGlobe WorldView‐2 satellite imagery with 12 cm resolution unmanned aerial vehicle (UAV) imagery and classified buffelgrass on Google Earth Engine, a cloud computing platform, using Random Forest (RF) models in Saguaro National Park, Arizona, USA. Our classification models had an average overall accuracy of 93% and producer's accuracies of 94–96% for buffelgrass, although user's accuracies were low. We detected a 2.92 km 2 area of buffelgrass in the eastern Rincon Mountain District (1.07% of the total area) and a 0.46 km 2 area (0.46% of the total area) in the western Tucson Mountain District of Saguaro National Park. Buffelgrass cover was significantly greater in the Sonoran Paloverde‐Mixed Cacti Desert Scrub vegetation type, on poorly developed Entisols and Inceptisol soils and on south‐facing topographic aspects compared to other areas. Our results demonstrate that high‐resolution imagery improve on previous attempts to detect and classify buffelgrass and indicate potential areas where the invasive grass might spread. The methods demonstrated in this study could be employed by land managers as a low‐cost strategy to identify priority areas for control efforts and continued monitoring.

Arizona↗
USGS2019

UAV-derived estimates of forest structure to inform ponderosa pine forest restoration

Restoring forest ecosystems has become an increasingly high priority for land managers across the American West. Millions of hectares of forest are in need of drastic yet strategic reductions in density (e.g., basal area). Meeting the restoration and management goals requires quantifying metrics of vertical and horizontal forest structure, which has relied upon field‐based measurements, manned airborne or satellite remote sensing datasets. We used unmanned aerial vehicle (UAV ) image‐derived Structure‐from‐Motion (SfM) models and high‐resolution multispectral orthoimagery in this study to quantify vertical and horizontal forest structure at both the fine‐ (<4 ha) and mid‐scales (4–400 ha) across a forest density gradient. We then used these forest structure estimates to assess specific objectives of a forest restoration treatment. At the fine‐scale, we found that estimates of individual tree height and canopy diameter were most accurate in low‐density conditions, with accuracies degrading significantly in high‐density conditions. Mid‐scale estimates of canopy cover and forest density followed a similar pattern across the density gradient, demonstrating the effectiveness of UAV image‐derived estimates in low‐ to medium‐density conditions as well as the challenges associated with high‐density conditions. We found that post‐treatment conditions met a majority of the prescription objectives and demonstrate the UAV image application in quantifying changes from a mechanical thinning treatment. We provide a novel approach to forest restoration monitoring using UAV ‐derived data, one that considers varying density conditions and spatial scales. Future research should consider a more spatially extensive sampling design, including different restoration treatments, as well as experimenting with different combinations of equipment, flight parameters, and data processing workflows.

Western United States↗
USGS2020

Estimating inundation of small waterbodies with sub-pixel analysis of Landsat imagery: Long-term trends in surface water area and evaluation of common drought indices

Small waterbodies are numerically dominant in many landscapes and provide several important ecosystem services, but automated measurement of waterbodies smaller than a standard Landsat pixel (0.09 ha) remains challenging. To further evaluate sub‐Landsat pixel techniques for estimating inundation extent of small waterbodies (basin area: 0.06–1.79 ha), we used a partial spectral unmixing method with matched filtering applied to September 1985–2018 Landsat 5 and eight imagery from southern Arizona, USA. We estimated trends in modeled surface water area each September and evaluated the ability of several common drought indices to explain variation in mean water area. Our methods accurately classified waterbodies as dry or inundated (Landsat 5: 91.3%; Landsat 8: 98.9%) and modeled and digitized surface water areas were strongly correlated ( R 2 = 0.70–0.92; bias = −0.024 to −0.015 ha). Estimated surface water area was best explained by the 3‐month seasonal standardized precipitation index (SPI03; July‒September). We found a wide range of estimated relationships between drought indices (e.g. SPI vs. Palmer Drought Severity Index) and estimated water area, even for different durations of the same drought index (e.g. SPI01 vs. SPI12). Mean waterbody surface area decreased by ~14% from September 1985 to September 2018, which matches declines in local annual precipitation and regional trends of reduced inundation extent of larger waterbodies. These results emphasize the importance of understanding local systems when relying on drought indices to infer variation in past or future surface water dynamics. Several challenges remain before widespread application of sub‐pixel methods is feasible, but our results provide further evidence that partial spectral unmixing with matched filtering provides reliable measures of inundation extent of small waterbodies.

Arizona↗
USGS2020

Ultra‐high‐resolution mapping of biocrusts with Unmanned Aerial Systems

Biological soil crusts (biocrusts) occur in drylands globally where they support ecosystem functioning by increasing soil stability, reducing dust emissions and modifying soil resource availability (e.g. water, nutrients). Determining biocrust condition and extent across landscapes continues to present considerable challenges to scientists and land managers. Biocrusts grow in patches, cover vast expanses of rugged terrain and are vulnerable to physical disturbance associated with ground‐based mapping techniques. As such, remote sensing offers promising opportunities to map and monitor biocrusts. While satellite‐based remote sensing has been used to detect biocrusts at relatively large spatial scales, few studies have used high‐resolution imagery from Unmanned Aerial Systems (UAS) to map fine‐scale patterns of biocrusts. We collected sub‐centimeter, true color 3‐band imagery at 10 plots in sagebrush and pinyon‐juniper woodland communities in a semiarid ecosystem in the southwestern US and used object‐based image analysis (OBIA) to segment and classify the imagery into maps of light and dark biocrusts, bare soil, rock and various vegetation covers. We used field data to validate the classifications and assessed the spatial distribution and configuration of different classes using fragmentation metrics. Map accuracies ranged from 46 to 77% (average 65%) and were higher in pinyon‐juniper (average 70%) versus sagebrush (average 60%) plots. Biocrust classes showed generally high accuracies at both pinyon‐juniper plots (average dark crust = 70%; light crust = 80%) and sagebrush plots (average dark crust = 69%; light crust = 77%). Point cloud density, sun elevation and spectral confusion between vegetation cover explained some differences in accuracy across plots. Spatial analyses of classified maps showed that biocrust patches in pinyon‐juniper plots were generally larger, more aggregated and contiguous than in sagebrush plots. Pinyon‐juniper plots also had greater patch richness and a lower Shannon evenness index than sagebrush plots, suggesting greater soil cover heterogeneity in this plant community type.

Utah↗
USGS2021

Global application of an unoccupied aerial vehicle photogrammetry protocol for predicting aboveground biomass in non-forest ecosystems

Non-forest ecosystems, dominated by shrubs, grasses and herbaceous plants, provide ecosystem services including carbon sequestration and forage for grazing, and are highly sensitive to climatic changes. Yet these ecosystems are poorly represented in remotely sensed biomass products and are undersampled by in situ monitoring. Current global change threats emphasize the need for new tools to capture biomass change in non-forest ecosystems at appropriate scales. Here we developed and deployed a new protocol for photogrammetric height using unoccupied aerial vehicle (UAV) images to test its capability for delivering standardized measurements of biomass across a globally distributed field experiment. We assessed whether canopy height inferred from UAV photogrammetry allows the prediction of aboveground biomass (AGB) across low-stature plant species by conducting 38 photogrammetric surveys over 741 harvested plots to sample 50 species. We found mean canopy height was strongly predictive of AGB across species, with a median adjusted R 2 of 0.87 (ranging from 0.46 to 0.99) and median prediction error from leave-one-out cross-validation of 3.9%. Biomass per-unit-of-height was similar within but different among, plant functional types. We found that photogrammetric reconstructions of canopy height were sensitive to wind speed but not sun elevation during surveys. We demonstrated that our photogrammetric approach produced generalizable measurements across growth forms and environmental settings and yielded accuracies as good as those obtained from in situ approaches. We demonstrate that using a standardized approach for UAV photogrammetry can deliver accurate AGB estimates across a wide range of dynamic and heterogeneous ecosystems. Many academic and land management institutions have the technical capacity to deploy these approaches over extents of 1–10 ha −1 . Photogrammetric approaches could provide much-needed information required to calibrate and validate the vegetation models and satellite-derived biomass products that are essential to understand vulnerable and understudied non-forested ecosystems around the globe.

Remote Sensing in Ecology and Conservation↗
USGS2021

Improved fire severity mapping in the North American boreal forest using a hybrid composite method

Fire severity is a key driver shaping the ecological structure and function of North American boreal ecosystems, a biome dominated by large, high-intensity wildfires. Satellite-derived burn severity maps have been an important tool in these remote landscapes for both fire and resource management. The conventional methodology to produce satellite-inferred fire severity maps generally involves comparing imagery from 1 year before and 1 year after a fire, yet environmental conditions unique to the boreal have limited the accuracy of resulting products. We introduce an alternative method – the ‘hybrid composite’ – based on deriving mean severity over time on a per-pixel basis within the cloud-computing environment of Google Earth Engine. It constructs the post-fire image from satellite data composited from all valid images (i.e., clear-sky and snow-free) acquired in the time period immediately after fire through the early growing season of the following year. We compare this approach to paired-scene and composite approaches where the post-fire time period is from the growing season 1 year after fire. Validation statistics based on field-derived data for 52 fires across Alaska and Canada indicate that the hybrid composite method outperforms the other approaches. This approach presents an efficient and cost-effective means to monitor and explore trends and patterns across broad spatial domains, and could be applied to fires in other regions, especially those with frequent cloud cover or rapid vegetation recovery.

Alaska↗
USGS2023

Challenges and solutions for automated avian recognition in aerial imagery

Remote aerial sensing provides a non-invasive, large geographical-scale technology for avian monitoring, but the manual processing of images limits its development and applications. Artificial Intelligence (AI) methods can be used to mitigate this manual image processing requirement. The implementation of AI methods, however, has several challenges: (1) imbalanced (i.e., long-tailed) data distribution, (2) annotation uncertainty in categorization, and (3) dataset discrepancies across different study sites. Here we use aerial imagery data of waterbirds around Cape Cod and Lake Michigan in the United States to examine how these challenges limit avian recognition performance. We review existing solutions and demonstrate as use cases how methods like Label Distribution Aware Marginal Loss with Deferred Re-Weighting, hierarchical classification, and FixMatch address the three challenges. We also present a new approach to tackle the annotation uncertainty challenge using a Soft-fine Pseudo-Label methodology. Finally, we aim with this paper to increase awareness in the ecological remote sensing community of these challenges and bridge the gap between ecological applications and state-of-the-art computer science, thereby opening new doors to future research.

Remote Sensing in Ecology and Conservation↗
USGS2023

Beyond presence mapping: Predicting fractional cover of non-native vegetation in Sentinel-2 imagery using an ensemble of MaxEnt models

Non-native species maps are important tools for understanding and managing biological invasions. We demonstrate a novel approach to extend presence modeling to map fractional cover (FC) of non-native yellow sweet clover Melilotus officinalis in the Northern Great Plains, USA. We used ensembles of MaxEnt models to map FC across landscapes from satellite imagery trained from regional aerial imagery that was trained by local unmanned aerial vehicle (UAV) imagery. Clover cover from field surveys and classified UAV imagery were nearly identical ( n = 22, R 2 = 0.99). Two classified UAV images provided training data to map clover presence with MaxEnt and National Agricultural Imagery Program (NAIP) aerial imagery. We binned cover predictions from NAIP imagery within each Sentinel-2 pixel into eight cover classes to create pure (100%) and FC (20%–95%) training data and modeled each class separately using MaxEnt and Sentinel-2 imagery. We mapped pure clover with one classification threshold and compared its performance to 15 candidate maps that included FC predictions outside pure predictions. Each FC map represented alternative combinations of five MaxEnt thresholds and three approaches to assign cover to pixels with multiple predictions from the FC ensemble. Evaluations of performance with independent datasets revealed maps including FC corresponded to field ( n = 32, R 2 range: 0.39–0.68) and UAV ( n = 20, R 2 range: 0.61–0.84) data better than pure clover maps ( R 2 = 0.15 and 0.31, respectively). Overall, the pure clover map predicted 3.2% cover, whereas the three best performing FC maps predicted 6.6%–8.0% cover. Including FC predictions increased accuracy and cover predictions which can improve ecological understanding of invasions. Our method allows efficient FC mapping for vegetative species discernible in UAV imagery and may be especially useful for mapping rare, irruptive or patchily distributed species with poor representation in field data, which challenges landscape-level mapping.

Montana, South Dakota↗
USGS2023

Both Landsat- and LiDAR-derived measures predict forest bee response to large-scale wildfire

Large-scale disturbances such as wildfire can have profound impacts on the composition, structure, and functioning of ecosystems. Bees are critical pollinators in natural settings and often respond positively to wildfires, particularly in forests where wildfire leads to more open conditions and increased floral resources. The use of Light Detection and Ranging (LiDAR) provides opportunities for quantifying habitat features across large spatial scales and is increasingly available to scientists and land managers for post-fire habitat assessment. We evaluated the extent to which LiDAR-derived forest structure measurements can predict forest bee communities after a large, mixed-severity fire. We hypothesized that LiDAR measurements linked to post-fire forest structure would improve our ability to predict bee abundance and species richness when compared to satellite-based maps of burn severity. To test this hypothesis, we sampled wild bee communities within the Douglas Fire Complex in southwestern Oregon, USA. We then used LiDAR and Landsat data to quantify forest structure and burn severity, respectively, across bee sampling locations. We found that the LiDAR forest structure model was the best predictor of abundance, whereas the Landsat burn severity model had better predictive ability for species richness. Furthermore, the Landsat burn severity model was better at predicting the presence and species richness of bumble bees (Bombus spp.), an ecologically distinct and economically important group within the Pacific Northwest. We posit that the divergent responses of the two modeling approaches are due to distinct responses by bee taxa to variation in forest structure as mediated by wildfire, with bumble bees in particular depending on closed-canopy forest for some portions of their life cycle. Our study demonstrates that LiDAR data can provide information regarding the drivers of bee abundance in post-wildfire conifer forest, and that both remote sensing approaches are useful for predicting components of wild bee diversity after large-scale wildfire.

Oregon↗