Abstract Uncrewed aerial systems (UAS) are increasingly used to survey wildlife populations, yet, until recently, most efforts were focused on developing technology rather than statistically defensible estimation of wildlife abundance. Therefore, we developed and applied an integrated framework to estimate waterfowl density and abundance from airborne imagery collected using 2 UAS platforms: a multirotor (DJI Mavic 2 Pro) in 2018 and a fixed wing (WingtraOne Gen II) in 2023, across wetlands in the Middle Rio Grande Valley, New Mexico, USA. We trained deep learning models to detect waterfowl in high‐resolution aerial imagery and quantify the sampled ground area required for density estimation. We corrected detection counts using platform‐specific confusion matrices to account for false detections and missed individuals. We used image footprints to quantify sampled areas and aggregated bias‐corrected observations at the transect level for analysis using design‐based ratio estimators, with uncertainty quantified through transect‐level bootstrap resampling. The fixed‐wing surveys estimated >20,000 ducks (95% CI = 13,321–29,384) and the multirotor surveys estimated >5,000 ducks (95% CI = 3,348–8,471) within surveyed areas, with additional populations of geese and cranes detected across platforms. Precision‐effort analyses quantified relationships between sampling intensity and estimator uncertainty, revealing precision dependence on platform‐specific spatial coverage and transect replication. By explicitly integrating automated detection, classification calibration, spatial sampling, and design‐based estimation, this workflow converts UAS imagery into statistically defensible estimates of density for broad taxonomic groups of waterfowl (e.g., ducks, geese, and cranes) with quantified uncertainty. This framework provides a practical and adaptable approach for incorporating UAS‐based surveys into wildlife monitoring and supports improved decision‐making in ecological management and conservation.

Technological and methodological advances in remote sensing and machine learning have created new opportunities for advancing wildlife surveys. We assembled a Community of Practice (CoP) to capitalize on these developments to explore improvements to the efficiency and effectiveness of aerial wildlife monitoring from a management perspective. The core objective of the CoP is to organize the development and testing of remote sensing and machine learning methods to improve aerial wildlife population surveys that support management decisions. Beginning in 2020, the CoP collaboratively identified the natural resource management decisions that are informed by wildlife survey data with a focus on waterbirds and marine wildlife. We surveyed our membership to establish 1) what management decisions they were using wildlife count data to inform; 2) how these count data were collected prior to the advent of remote sensing/machine learning methods; 3) the impetus for transitioning to a remote sensing/machine learning methodological framework; and 4) the challenges practitioners face in transitioning to this framework. This paper documents these findings and identifies research priorities for moving toward operational remote sensing-based wildlife surveys in service of wildlife management.

Deep learning shows promise for automating detection and classification of wildlife from digital aerial imagery to support cost-efficient remote sensing solutions for wildlife population monitoring. To support in-flight orthorectification and machine learning processing to detect and classify wildlife from imagery in near real-time, we evaluated deep learning methods that address hardware limitations and the need for processing efficiencies to support the envisioned in-flight workflow. We developed an annotated dataset for a suite of marine birds from high-resolution digital aerial imagery collected over open water environments to train the models. The proposed 3-stage workflow for automated, in-flight data processing includes: 1) image filtering based on the probability of any bird occurrence, 2) bird instance detection, and 3) bird instance classification. For image filtering, we compared the performance of a binary classifier with Mask Region-based Convolutional Neural Network (Mask R-CNN) as a means of sub-setting large volumes of imagery based on the probability of at least one bird occurrence in an image. On both the validation and test datasets, the binary classifier achieved higher performance than Mask R-CNN for predicting bird occurrence at the image-level. We recommend the binary classifier over Mask R-CNN for workflow first-stage filtering. For bird instance detection, we leveraged Mask R-CNN as our detection framework and proposed an iterative refinement method to bootstrap our predicted detections from loose ground-truth annotations. We also discuss future work to address the taxonomic classification phase of the envisioned workflow.

The R package popharvest was designed to help assess the sustainability of offtake in birds when only limited demographic information is available. In this article, we describe some basics of harvest theory and then discuss several considerations when using the different approaches in popharvest to assess whether observed harvests are unsustainable. Throughout, we emphasize the importance of distinguishing between the scientific and policy aspects of managing offtake. The principal product of popharvest is a sustainable harvest index (SHI), which can indicate whether the harvest is unsustainable but not the converse. SHI is estimated based on a simple, scalar model of logistic population growth, whose parameters may be estimated using limited knowledge of demography. Uncertainty in demography leads to a distribution of SHI values and it is the purview of the decision-maker to determine what amounts to an acceptable risk when failing to reject the null hypothesis of sustainability. The attitude toward risk, in turn, will likely depend on the decision-maker's objective(s) in managing offtake. The management objective as specified in popharvest is a social construct, informed by biology, but ultimately it is an expression of social values that usually vary among stakeholders. We therefore suggest that any standardization of criteria for management objectives in popharvest will necessarily be subjective and, thus, hard to defend in diverse decision-making situations. Because of its ease of use, diverse functionalities, and a minimal requirement of demographic information, we expect the use of popharvest to become widespread. Nonetheless, we suggest that while popharvest provides a useful platform for rapid assessments of sustainability, it cannot substitute for sufficient expertise and experience in harvest theory and management.

Aerial count surveys of wildlife populations are a prominent monitoring method for many wildlife species. Traditionally, these surveys utilize human observers to detect, count, and classify observations to species. However, given recent technological advances, many research groups are exploring the combined use of remote sensing and deep learning methods to replace human observers in order to improve data quality and reproducibility, reduce disturbance to wildlife, and increase aircrew safety. Given that deep learning detection and classification are not perfect and that statistical inference from ecological models is generally very sensitive to misclassification, we require study designs and statistical models to accommodate these observation errors. As part of an ongoing effort by the U.S. Fish and Wildlife Service, Bureau of Ocean Energy Management, and U.S. Geological Survey to survey marine birds and other marine wildlife using digital aerial imagery and deep learning object detection and classification, we developed a general hierarchical model for estimating species-specific abundance that accommodates object-level errors in classification. We consider hierarchical deep learning classification at multiple taxonomic levels subject to misclassification, hierarchically-structured human validation data subject to partial and erroneous misclassification, and an image censoring process leading to preferential sampling. We demonstrate that this model can estimate species-specific abundance and habitat relationships without bias when the assumptions are met, and we discuss the plausibility of these assumptions in practice for this study and others like it. Finally, we use this model to demonstrate the relevance of the features of the ecological systems under study to the classification task itself. In models that couple the ecological and classification processes into a single, hierarchical model, the true classes are treated as latent variables to be estimated and are informed by both the classification probability parameters and the ecological parameters that determine the expected frequencies of each class at the level the data are being modeled (e.g., site or site by occasion). We show that ignoring the expected frequencies of each class (when they are imbalanced) can cause correction for misclassification to produce biased parameter estimates, but coupling the ecological and classification models allows for the variability in relative class frequency across space and time due to ecological and sampling conditions to be accommodated with spatial or temporal covariates. As a result, bias is removed, classification is more accurate, and uncertainty is propagated between the ecological and classification models. We therefore argue that ability of deep learning classifiers, and classifiers more generally, to produce reliable ecological inference depends, in part, on the ecological system under study.

Abstract 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.

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.

Wildlife managers routinely seek to establish sustainable limits of sport harvest or other regulated forms of take while confronted with considerable uncertainty. A growing body of ecological research focuses on methods to describe and account for uncertainty in management decision-making and to prioritize research and monitoring investments to reduce the most influential uncertainties. We used simulation methods incorporating measures of demographic uncertainty to evaluate risk of overharvest and prioritize information needs for North American sea ducks (Tribe Mergini). Sea ducks are popular game birds in North America, yet they are poorly monitored and their population dynamics are poorly understood relative to other North American waterfowl. There have been few attempts to assess the sustainability of harvest of North American sea ducks, and no formal harvest strategy exists in the U.S. or Canada to guide management. The popularity of sea duck hunting, extended hunting opportunity for some populations (i.e., special seasons and/or bag limits), and population declines have led to concern about potential overharvest. We used Monte Carlo simulation to contrast estimates of allowable harvest and observed harvest and assess risk of overharvest for 7 populations of North American sea ducks: the American subspecies of common eider (Somateria mollissima dresseri), eastern and western populations of black scoter (Melanitta americana) and surf scoter (M. perspicillata), and continental populations of white-winged scoter (M. fusca) and long-tailed duck (Clangula hyemalis). We combined information from empirical studies and the opinions of experts through formal elicitation to create probability distributions reflecting uncertainty in the individual demographic parameters used in this assessment. Estimates of maximum growth (rmax), and therefore of allowable harvest, were highly uncertain for all populations. Long-tailed duck and American common eider appeared to be at high risk of overharvest (i.e., observed harvest < allowable harvest in 5-7% and 19-26% of simulations, respectively depending on the functional form of density dependence), whereas the other populations appeared to be at moderate risk to low risk (observed harvest < allowable harvest in 22-68% of simulations, again conditional on the form of density dependence). We also evaluated the sensitivity of the difference between allowable and observed harvest estimates to uncertainty in individual demographic parameters to prioritize information needs. We found that uncertainty in overall fecundity had more influence on comparisons of allowable and observed harvest than adult survival or observed harvest for all species except long-tailed duck. Although adult survival was characterized by less uncertainty than individual components of fecundity, it was identified as a high priority information need given the sensitivity of growth rate and allowable harvest to this parameter. Uncertainty about population size was influential in the comparison of observed and allowable harvest for 5 of the 6 populations where it factored into the assessment. While this assessment highlights a high degree of uncertainty in allowable harvest, it provides a framework for integration of improved data from future research and monitoring. It could also serve as the basis for harvest strategy development as management objectives and regulatory alternatives are specified by the management community.

Although monitoring data for sea ducks (Tribe Mergini) are limited, current evidence suggests that four of the most common species wintering along the eastern coast of the United States-long-tailed duck Clangula hyemalis, white-winged scoter Melanitta fusca, surf scoter Melanitta perspicillata, and black scoter Melanitta americana-may be declining, while the status of American common eider Somateria mollissima dresseri is uncertain. The apparent negative trends, combined with the fact that sea duck life histories are among the most poorly documented of North American waterfowl, have led to concerns for these species and questions about the impacts of human activities, such as hunting, as well as catastrophic events and environmental change. During winter, thousands of sea ducks are found along the U. S. Atlantic coast, where they may be affected by proposed wind-power development, changes to marine traffic, aquaculture practices, sand mining, and other coastal development. Possible impacts are difficult to quantify because traditional winter waterfowl surveys do not cover many of the marine habitats used by sea ducks. Thus, the U. S. Fish and Wildlife Service conducted an experimental survey of sea ducks from 2008 to 2011 to characterize their winter distributions along the U. S. Atlantic coast. Each year, data were collected on 11 species of sea ducks on >200 transects, stretching from Maine to Florida. In this paper, we describe distribution of common eider, long-tailed duck, white-winged scoter, surf scoter, and black scoter. Densities of the two species with the most northerly distribution, white-winged scoter and common eider, were highest near Cape Cod and Nantucket. Long-tailed duck was most abundant around Cape Cod, Nantucket Shoals, and in Chesapeake Bay. Surf scoter also concentrated within Chesapeake Bay; however, they were additionally found in high densities in Delaware Bay, and along the Maryland-Delaware outer coast. Black scoter, the most widely distributed species, occurred at high densities along the South Carolina coast and the mouth of Chesapeake Bay. Spatial patterns of high-density transects were consistent among years for all species except black scoter, which exhibited the most interannual variation in distribution. The distance from land, depth, and bottom slope where flocks were observed varied among species and regions, with a median distance of 3.8 km from land along the coastal transects and 75% of flocks observed over depths of,16 m. Common eider and long-tailed duck were observed closer to shore and over steeper ocean bottoms than were the three scoter species. Our results represent the first large-scale quantitative description of winter sea duck distribution along the U. S. Atlantic coast, and should guide the development of sea duck monitoring programs and aid the assessment of potential impacts of ongoing and proposed offshore development.

Reliable estimates of annual harvest rates are required for the implementation of mallard (Anas platyrhynchos) adaptive harvest management decision frameworks. Because not all standard bands recovered during the hunting season are reported, band reporting probabilities are needed to estimate mallard harvest rates. Information from birds recovered with bands that notify finders of a reward (i.e., reward bands) can be used to estimate band reporting rates. We analyzed reward banding data for 3 stocks of mallards to estimate reporting probabilities that can be used to estimate harvest rates for birds recovered with toll-free or web-address bands. Specifically, we explored spatial variability in reporting probabilities, and assessed whether reporting probabilities varied among years. Our analysis indicated that reporting probabilities varied among the 4 Flyways, eastern Canada, and western Canada and Alaska. We had difficulty interpreting temporal fluctuations and found little evidence for any meaningful trends in reporting rates between 2002 and 2010. We recommend that reporting probabilities of 0.67 in the Atlantic Flyway, 0.81 in the Mississippi Flyway, 0.70 in the Central Flyway, 0.76 in the Pacific Flyway, 0.50 in eastern Canada, and 0.57 in western Canada and Alaska be used to estimate harvest probabilities for birds recovered in these regions. (c) 2013 The Wildlife Society.

In Memoriam This report is dedicated to our colleague and friend, Thom Lewis, who died in the line of duty, while training for aerial surveys. Thom was a passionate wildlife biologist whose love and knowledge of nature, easy-going humor, attention to detail, and commitment to waterfowl conservation is deeply missed. His example continues to inspire us.

Climate change and its associated uncertainties are of concern to natural resource managers. Although aspects of climate change may be novel (e.g., system change and nonstationarity), natural resource managers have long dealt with uncertainties and have developed corresponding approaches to decision-making. Adaptive resource management is an application of structured decision-making for recurrent decision problems with uncertainty, focusing on management objectives, and the reduction of uncertainty over time. We identified 4 types of uncertainty that characterize problems in natural resource management. We examined ways in which climate change is expected to exacerbate these uncertainties, as well as potential approaches to dealing with them. As a case study, we examined North American waterfowl harvest management and considered problems anticipated to result from climate change and potential solutions. Despite challenges expected to accompany the use of adaptive resource management to address problems associated with climate change, we conclude that adaptive resource management approaches will be the methods of choice for managers trying to deal with the uncertainties of climate change. (C) 2011 The Wildlife Society.

Legal removal of migratory birds from the wild occurs for several reasons, including subsistence, sport harvest, damage control, and the pet trade. We argue that harvest theory provides the basis for assessing the impact of authorized take, advance a simplified rendering of harvest theory known as potential biological removal as a useful starting point for assessing take, and demonstrate this approach with a case study of depredation control of black vultures (Coragyps atratus) in Virginia, USA. Based on data from the North American Breeding Bird Survey and other sources, we estimated that the black vulture population in Virginia was 91,190 (95% credible interval = 44,520-212,100) in 2006. Using a simple population model and available estimates of life-history parameters, we estimated the intrinsic rate of growth (r(max)) to be in the range 7-14%, with 10.6% a plausible point estimate. For a take program to seek an equilibrium population size on the conservative side of the yield curve, the rate of take needs to be less than that which achieves a maximum sustained yield (0.5 X r(max)). Based on the point estimate for r(max) and using the lower 60% credible interval for population size to account for uncertainty, these conditions would be met if the take of black vultures in Virginia in 2006 was,3,533 birds. Based on regular monitoring data, allowable harvest should be adjusted annually to reflect changes in population size. To initiate discussion about how this assessment framework could be related to the laws and regulations that govern authorization of such take, we suggest that the Migratory Bird Treaty Act requires only that take of native migratory birds be sustainable in the long-term, that is, sustained harvest rate should be < r(max). Further, the ratio of desired harvest rate to 0.5 X r(max) may be a useful metric for ascertaining the applicability of specific requirements of the National Environmental Protection Act. (JOURNAL OF WILDLIFE MANAGEMENT 73(4): 556-565; 2009)

We evaluated double-observer methods for aerial surveys as a means to adjust counts of waterfowl for incomplete detection. We conducted our stud), in eastern Canada and the northeast United States utilizing 3 aerial-survey crews flying 3 different types of fixed-wing aircraft. We reconciled counts of front- and rear-seat observers immediately following an observation by the rear-seat observer (i.e., on-the-fly reconciliation). We evaluated 6 a priori models containing a combination of several factors thought to influence detection probability including observer, seat position, aircraft type, and group size. We analyzed data for American black ducks (Anas rubripes) and mallards (A. platyrhynchos), which are among the most abundant duck species in this region. The best-supported model for both black ducks and mallards included observer effects. Sample sizes of black ducks were sufficient to estimate observer-specific detection rates for each crew. Estimated detection rates for black ducks were 0.62 (SE = 0.10), 0.63 (SE = 0.06), and 0.74 (SE = 0.07) for pilot-observers, 0,61 (SE = 0.08), 0.62 (SE = 0.06), and 0.81 (SE = 0.07) for other front-seat observers, and 0.43 (SE = 0.05), 0.58 (SE = 0.06), and 0.73 (SE = 0.04) for rear-seat observers. For mallards, sample sizes were adequate to generate stable maximum-likelihood estimates of observer-specific detection rates for only one aerial crew. Estimated observer-specific detection rates for that crew were 0.84 (SE = 0.04) for the pilot-observer, 0.74 (SE = 0.05) for the other front-seat observer, and 0.47 (SE = 0.03) for the rear-seat observer. Estimated observer detection rates were confounded by the position of the seat occupied by an observer, because observers did not switch seats, and by land-cover because vegetation and landform varied among crew areas. Double-observer methods with on-the-fly reconciliation, although not without challenges, offer one viable option to account for detection bias in aerial waterfowl surveys where birds are distributed at low density in remote areas that are inaccessible by ground crews. Double-observer methods, however, estimate only detection rate of animals that are potentially observable given the survey method applied. Auxiliary data and methods must be considered to estimate overall detection rate.

This report summarizes information about the status of duck populations and wetland habitats during spring 2008, focusing on areas encompassed by the U.S. Fish and Wildlife (USFWS) and Canadian Wildlife Services’ (CWS) Waterfowl Breeding Population and Habitat Survey. This report does not include information from surveys conducted by state or provincial agencies. In the traditional survey area, which includes strata 1-18, 20-50, and 75-77 (Figure 1), the total duck population estimate (excluding scoters [Melanitta spp.], eiders [Somateria and Polysticta spp.], long-tailed ducks [Clangula hyemalis], mergansers [Mergus and Lophodytes spp.], and wood ducks [Aix sponsa]) was 37.3 ± 0.6 [SE] million birds. This estimate represents a 9% decline over last year’s estimate of 41.2 ± 0.7 million birds, but remains 11% above the 1955-2007 long-term averagea (Table 1). Estimated mallard (Anas platyrhynchos) abundance was 7.7 ± 0.3 million birds, which was similar to last year’s estimate of 8.3 ± 0.3 million birds and the long-term average (Table 2). Blue-winged teal (A. discors) abundance was 6.6 ± 0.3 million birds. This value is similar to last year’s estimate of 6.7 ± 0.4 million birds and 45% above the long-term average. Estimated abundances of gadwall (A. strepera; 2.7 ± 0.2 million) and Northern shovelers (A. clypeata; 3.5 ± 0.2 million) were below 2007 estimates (-19% and -23%, respectively) but remain well above their long-term averages (+56% and +56%, respectively). Estimated abundances of green-winged teal (A. crecca; 3.0 ± 0.2 million) and redheads (Aythya americana; 1.1 ± 0.1 million) were similar to last year’s and were >50% above their long-term averages. Estimates of canvasbacks (A. valisineria; 0.5 ± 0.05 million) were 44% below the 2007 estimate (0.9 ± 0.09 million) and 14% below the long-term average. The estimate for Northern pintails (Anas acuta) was 2.6 ± 0.1 million, which was 22% below the 2007 estimate of 3.3 ± 0.2 million, and 36% below the long-term average. The scaup estimate (Aythya affinis and A. marila combined; 3.7 ± 0.2 million) was similar to 2007, and remained 27% below the long-term average of 5.1 ± 0.2 million. Habitat conditions during the 2008 Waterfowl Breeding Population and Habitat Survey were characterized in many areas by a delayed spring in comparison with several preceding years. Drought in

This report summarizes information about the status of duck populations and wetland habitats during spring 2007, focusing on areas encompassed by the U.S. Fish and Wildlife (USFWS) and Canadian Wildlife Services’ (CWS) Waterfowl Breeding Population and Habitat Survey. This report does not include information from surveys conducted by State or Provincial agencies. In the traditional survey area, which includes strata 1-18, 20-50, and 75-77 (Fig. 1), the total duck population estimate (excluding scoters [Melanitta spp.], eiders [Somateria and Polysticta spp.], long-tailed ducks [Clangula hyemalis], mergansers [Mergus and Lophodytes spp.], and wood ducks [Aix sponsa]) was 41.2 ± 0.8 [SE] million birds. This was 14% greater than last year’s estimate of 36.2 ± 0.6 million birds and 24% above the 1955-2006 long-term average (Tables 1-12). Mallard (Anas platyrhynchos) abundance was 8.0 ± 0.3 million birds, which was 10% above last year’s estimate of 7.3 ± 0.2 million birds and 7% above the long-term average (Appendix A). Blue-winged teal (A. discors) abundance was 6.7 ± 0.4 million birds. This value was the third highest estimate since 1955, 14% greater than last year’s estimate of 5.9 ± 0.3 million birds, and 48% above the long-term average. Estimated abundances of gadwall (A. strepera; 3.4 ± 0.2 million) and Northern shovelers (A. clypeata; 4.6 ± 0.2 million) were also above those of last year (+19% and +24%, respectively) and well above their long-term averages (+96% and +106%, respectively). Estimated abundance of American wigeon (A. americana; 2.8 ± 0.2 million) was 29% greater than last year but similar to the long-term average. Estimated abundances of green-winged teal (A. crecca; 2.9 ± 0.2 million), redheads (Aythya americana; 1.0 ± 0.08 million), and canvasbacks (A. valisineria; 0.9 ± 0.09 million) were similar to last year’s, but were each >50% above their long-term averages. Abundances of Northern shovelers, redheads, and canvasbacks were the highest ever estimated in this survey area, and the