Explore the vast range of titles available.

Operating wind-power projects often includes protecting volant wildlife. One method for doing this uses an automated system to detect, identify (through use of artificial intelligence; AI), track animals (targets) and curtail turbines when risk of a collision is high. However, assessments of the effectiveness, in terms of identification accuracy and subsequent turbine curtailment of such systems are lacking. Over 1 year, we assessed such an automated system installed at a wind project in California, USA to determine its identification accuracy and rates at which \"virtual” curtailments were ordered (without slowing turbines), for eagles (intended targets) and non-eagle targets. The system correctly identified 77% of eagles and 85% of non-eagles. Curtailment orders occurred 6 times more frequently for non-eagle targets (5,439) than for eagle targets (850). Greater abundance of common ravens that were misidentified as eagles influenced the effectiveness of the system by greatly increasing unintended curtailment orders. The balance between costs (price of the IdentiFlight system, reduced energy generation, turbine wear and maintenance) and benefits (reduced collisions between intended target species and turbines) may depend upon the biological setting, speed at which operators can curtail turbines, and the objectives of the operator when considering the IdentiFlight system.

To maximize fitness, flying animals should maximize flight speed while minimizing energetic expenditure. Soaring speeds of large-bodied birds are determined by flight routes and tradeoffs between minimizing time and energetic costs. Large raptors migrating in eastern North America predominantly glide between thermals that provide lift or soar along slopes or ridgelines using orographic lift (slope soaring). It is usually assumed that slope soaring is faster than thermal gliding because forward progress is constant compared to interrupted progress when birds pause to regain altitude in thermals. We tested this slope-soaring hypothesis using high-frequency GPS-GSM telemetry devices to track golden eagles during northbound migration. In contrast to expectations, flight speed was slower when slope soaring and eagles also were diverted from their migratory path, incurring possible energetic costs and reducing speed of progress towards a migratory endpoint. When gliding between thermals, eagles stayed on track and fast gliding speeds compensated for lack of progress during thermal soaring. When thermals were not available, eagles minimized migration time, not energy, by choosing energetically expensive slope soaring instead of waiting for thermals to develop. Sites suited to slope soaring include ridges preferred for wind-energy generation, thus avian risk of collision with wind turbines is associated with evolutionary trade-offs required to maximize fitness of time-minimizing migratory raptors.

Accurately predicting behavioural modes of animals in response to environmental features is important for ecology and conservation. Supervised learning (SL) methods are increasingly common in animal movement ecology for classifying behavioural modes. However, few examples exist of applying SL to classify polytomous animal behaviour from environmental features especially in the context of millions of animal observations. We review SL methods (weighted k‐nearest neighbours; neural nets; random forests; and boosted classification trees with XGBoost) for classifying polytomous animal behaviour from environmental predictors. We also describe tuning parameter selection and assessment strategies, approaches for visualizing relationships between predictors and class outputs, and computational considerations. We demonstrate these methods by predicting three categories of risk to bald eagles from colliding with wind turbines using, as predictors, 12 environmental state features associated with 1.7 million GPS telemetry data points from 57 eagles. Of the SL methods we considered, XGBoost yielded the most accurate model with 86.2% classification accuracy and pairwise‐averaged area under the ROC curve of 90.6. Computational time of XGBoost scaled better to large data than any other SL method. We also show how SHAP values integrated in the R package (xgboost) facilitate investigation of variable relationships and importance. For big data applications, XGBoost appears to provide superior classification accuracy and computational efficiency. Our results suggest XGBoost should be considered as an early modelling option in situations where the intent is to classify millions of animal behaviour observations from environmental predictors and to understand relationships between those predictors and movement behaviours. We also offer a tutorial to assist researchers in implementing this method.

Summary Animals respond to a variety of environmental cues, including weather conditions, when migrating. Understanding the relationship between weather and migration behaviour is vital to assessing time‐ and energy limitations of soaring birds. Different soaring modes have different efficiencies, are dependent upon different types of subsidized lift and are weather dependent. We collected GPS locations from 47 known‐age golden eagles that moved along 83 migration tracks. We paired each location with weather to determine meteorological correlates of migration during spring and fall as birds crossed three distinct ecoregions in north‐east North America. Golden eagle migration was associated with weather conditions that promoted thermal development, regardless of season, ecoregion or age. Eagle migration showed age‐ and season‐specific responses to weather conditions that promoted orographic lift. In spring, adult eagles migrated earlier, over fewer days, and under more variable weather conditions than did pre‐adults, suggesting that adults were time limited and pre‐adults made choices to conserve energy. In fall, we found no difference in the time span of migration or when each age class migrates; however, we saw evidence that pre‐adults were less efficient migrants than adults. The decision by soaring birds to migrate when thermals developed allowed individuals to manage trade‐offs between migratory speed and migratory efficiency. When time was limited (i.e. spring movement of adults speeding towards nesting territories), use of whatever lift was available decreased the time span of migration. When migration was not time limited (e.g. spring movements by pre‐adults, all movements in fall), eagles avoided suboptimal flight conditions by pausing migration, thus increasing the time span of migration while reducing energetic costs. Lay Summary

Recent advances in digital data collection have spurred accumulation of immense quantities of data that have potential to lead to remarkable ecological insight, but that also present analytic challenges. In the case of biologging data from birds, common analytical approaches to classifying movement behaviors are largely inappropriate for these massive data sets. We apply a framework for using K‐means clustering to classify bird behavior using points from short time interval GPS tracks. K‐means clustering is a well‐known and computationally efficient statistical tool that has been used in animal movement studies primarily for clustering segments of consecutive points. To illustrate the utility of our approach, we apply K‐means clustering to six focal variables derived from GPS data collected at 1–11 s intervals from free‐flying bald eagles (Haliaeetus leucocephalus) throughout the state of Iowa, USA. We illustrate how these data can be used to identify behaviors and life‐stage‐ and age‐related variation in behavior. After filtering for data quality, the K‐means algorithm identified four clusters in >2 million GPS telemetry data points. These four clusters corresponded to three movement states: ascending, flapping, and gliding flight; and one non‐moving state: perching. Mapping these states illustrated how they corresponded tightly to expectations derived from natural history observations; for example, long periods of ascending flight were often followed by long gliding descents, birds alternated between flapping and gliding flight. The K‐means clustering approach we applied is both an efficient and effective mechanism to classify and interpret short‐interval biologging data to understand movement behaviors. Furthermore, because it can apply to an abundance of very short, irregular, and high‐dimensional movement data, it provides insight into small‐scale variation in behavior that would not be possible with many other analytical approaches. We demonstrate the use of K‐means clustering to classify bird behavior using points from short time interval GPS tracks. We illustrate how these data can be used to identify behaviors and life stage‐ and age‐related variation in behavior. Our work has implications for understanding movement behaviors from GPS data.

Despite the ecological importance of avian scavengers such as vultures, demographic information that is essential to their conservation and management remains limited. The goal of this study was to evaluate survival and mortality risk in black vultures (Coragyps atratus), a protected native species of conflict management concern in the United States. Here, we combined monitoring data from a 28‐year period to estimate annual survival rates among age classes and test for seasonal and age‐related patterns in mortality risk. Using dead recovery information, we also summarized the causes and timing of annual mortalities. Additionally, we tested whether mortality risk was affected by aspects of landscape composition and configuration, as well as human development. Average annual survival was high overall (0.95, 95% CI: 0.92–0.98), with estimate precision markedly improved by combining datasets (72.1%–84.2% increase). Mortality risk differed by season and age class such that vultures experienced 68.7% more hazard during the breeding season, and adults experienced 66.2% less hazard than juveniles. Among the mortality causes, 67% were anthropogenic, 4% were natural, and the remaining 29% were unknown. Additionally, greater land cover diversity (Shannon diversity index) reduced mortality risk, whereas measures of landscape configuration and human development had no effect. High survival rates help explain this species' population growth and range expansion and further inform allowable take for sustainable management practices. Moreover, the identified seasonal and age‐related vulnerabilities may help guide lethal control of human–vulture conflicts in an ecologically relevant manner. Maintaining diverse landscapes may also enhance survival overall, facilitating conservation of this species and other avian scavengers. The goal of this study was to evaluate survival and mortality risk in black vultures (Coragyps atratus), a protected native species of conflict management concern in the United States. High survival rates from this study help explain this species' population growth and range expansion and further inform allowable take for sustainable management practices.

There is increasing pressure on wind energy facilities to manage or mitigate for wildlife collisions. However, little information exists regarding spatial and temporal variation in collision rates, meaning that mitigation is most often a blanket prescription. To address this knowledge gap, we evaluated variation among turbines and months in an aspect of collision risk—probability of entry by an eagle into a rotor‐swept zone (hereafter, “probability of entry”). We examined 10,222 eagle flight paths identified and recorded by an automated bird monitoring system at a wind energy facility in Wyoming, USA. Probabilities of entry per turbine–month combination were 4.03 times greater in some months than others, ranging 0.15 to 0.62. The overall probability of entry for the riskiest turbine (i.e., the one with the greatest probability of entry) was 2.39 times greater than the least‐risky turbine. Our methodology describes large variation across turbines and months in the probability of entry. If subsequently combined with information on other sources of variation (i.e., weather, topography), this approach can identify risky versus safe situations for eagles under which cost of management, curtailment prescriptions, and collision risk can be simultaneously minimized. Rates that eagles entered rotor‐swept zones of wind turbines varied by turbine across a wind farm in Wyoming, USA. Such variation demonstrates that curtailment criteria should vary per turbine and season.

Human activity influences wildlife. However, the ecological and conservation significances of these influences are difficult to predict and depend on their population‐level consequences. This difficulty arises partly because of information gaps, and partly because the data on stressors are usually collected in a count‐based manner (e.g., number of dead animals) that is difficult to translate into rate‐based estimates important to infer population‐level consequences (e.g., changes in mortality or population growth rates). However, ongoing methodological developments can provide information to make this transition. Here, we synthesize tools from multiple fields of study to propose an overarching, spatially explicit framework to assess population‐level consequences of anthropogenic stressors on terrestrial wildlife. A key component of this process is using ecological information from affected animals to upscale from count‐based field data on individuals to rate‐based demographic inference. The five steps to this framework are (1) framing the problem to identify species, populations, and assessment parameters; (2) field‐based measurement of the effect of the stressor on individuals; (3) characterizing the location and size of the populations of interest; (4) demographic modeling for those populations; and (5) assessing the significance of stressor‐induced changes in demographic rates. The tools required for each of these steps are well developed, and some have been used in conjunction with each other, but the entire group has not previously been unified together as we do in this framework. We detail these steps and then illustrate their application for two species affected by different anthropogenic stressors. In our examples, we use stable hydrogen isotope data to infer a catchment area describing the geographic origins of affected individuals, as the basis to estimate population size for that area. These examples reveal unexpectedly greater potential risks from stressors for the more common and widely distributed species. This work illustrates key strengths of the framework but also important areas for subsequent theoretical and technical development to make it still more broadly applicable.

Birds select critical resources to meet needs that vary in response to spatial, temporal, and individual variation. As an example, perch or roost sites may be at locations that provide protection from predators, mobbing, or inclement weather. Applied to large, soaring predators, this theory suggests that they may select perch and roost sites near food resources or at sites where environmental updrafts develop. To test these theories, we characterized selection of nonflight locations throughout the annual cycle for Golden Eagles (Aquila chrysaetos) in eastern North America. We determined factors associated with selection of perching (daytime) and roosting (nighttime) sites by eagles by comparing land cover and topographic characteristics of GPS telemetry locations for eagles (used) with random (available) locations. We separately assessed selection for perch and roost sites during each of 4 seasons (winter, summer, and spring and fall migration). Golden Eagles showed different selection patterns for perching by season and age. Throughout the year, eagles selected perch sites on steep slopes. The direction these slopes faced differed among seasons, with eagles selecting south-facing slopes in summer and east-facing slopes during migration. Adults showed greater preferences for broadleaf forests in summer and for ridges in fall. Patterns of perch-site use were consistent with selection for sites that provide thermal protection and access to thermal updrafts. We found few patterns of selection for roosting sites. Our analysis provides insight into decision-making by a longdistance migrant across its annual cycle and throughout its geographic range, and thus into how resource selection changes seasonally.

Knowledge of the distribution and movements of populations of migratory birds is useful for the effective conservation and management of biodiversity. However, such information is often unavailable because of the difficulty of tracking sufficient numbers of individuals. We used more easily obtained feather stable hydrogen isotope ratios (δ2H) to predict the summer grounds of the small, threatened, and migratory population of Golden Eagles (Aquila chrysaetos) in eastern North America. We then identified summer locations and the extent of migratory connectivity for this population. We collected δ2H (δ2Hf), stable carbon isotope (δ13C), and stable nitrogen isotope (δ15N) data from the body feathers of 47 juvenile, subadult, and adult Golden Eagles. Values of δ13C and δ15N suggested that all but 2 birds obtained food from terrestrial-based food webs and therefore that δ2H data were appropriate for inferring the geographic region of molt for the majority of birds. There was relatively large interfeather variation in the δ2H values of subadults vs. adults, suggesting that these groups molted at different times and places. The most negative δ2Hf values from birds with known summering grounds exhibited (1) a negative correlation with their summering latitude, and (2) a positive correlation with amount-weighted δ2H values of May–August precipitation at the summer location. These data validate the use of δ2Hf values for inferring the summer locations of Golden Eagles of unknown origin. Likelihood-of-origin maps derived from δ2Hf values revealed that (1) the majority of birds spent the breeding season in central Québec and Labrador, and (2) birds that wintered at southern latitudes, from approximately northern Alabama to southwestern Virginia, migrated about twice the distance of birds that wintered at northern latitudes, from Pennsylvania to New York. We observed a positive relationship between δ2Hf values and the latitude of the wintering location, which, along with the likelihood-of-origin maps, revealed moderate patterns of leapfrog migration and migratory connectivity.