Explore the vast range of titles available.

Ecologists routinely fit complex models with multiple parameters of interest, where hundreds or more competing models are plausible. To limit the number of fitted models, ecologists often define a model selection strategy composed of a series of stages in which certain features of a model are compared while other features are held constant. Defining these multi‐stage strategies requires making a series of decisions, which may potentially impact inferences, but have not been critically evaluated. We begin by identifying key features of strategies, introducing descriptive terms when they did not already exist in the literature. Strategies differ in how they define and order model building stages. Sequential‐by‐sub‐model strategies focus on one sub‐model (parameter) at a time with modeling of subsequent sub‐models dependent on the selected sub‐model structures from the previous stages. Secondary candidate set strategies model sub‐models independently and combine the top set of models from each sub‐model for selection in a final stage. Build‐up approaches define stages across sub‐models and increase in complexity at each stage. Strategies also differ in how the top set of models is selected in each stage and whether they use null or more complex sub‐model structures for non‐target sub‐models. We tested the performance of different model selection strategies using four data sets and three model types. For each data set, we determined the \"true\" distribution of AIC weights by fitting all plausible models. Then, we calculated the number of models that would have been fitted and the portion of \"true\" AIC weight we recovered under different model selection strategies. Sequential‐by‐sub‐model strategies often performed poorly. Based on our results, we recommend using a build‐up or secondary candidate sets, which were more reliable and carrying all models within 5–10 AIC of the top model forward to subsequent stages. The structure of non‐target sub‐models was less important. Multi‐stage approaches cannot compensate for a lack of critical thought in selecting covariates and building models to represent competing a priori hypotheses. However, even when competing hypotheses for different sub‐models are limited, thousands or more models may be possible so strategies to explore candidate model space reliably and efficiently will be necessary.

Differential habitat use and intraguild competition are both thought to be important drivers of animal population sizes and distributions. Habitat associations for individual species are well-established, and interactions between particular pairs of species have been highlighted in many focal studies. However, community-wide assessments of the relative strengths of these two factors have not been conducted. We built multi-scale habitat occupancy models for five carnivore taxa of New York's Adirondack landscape and assessed the relative performance of these models against ones in which co-occurrences of potentially competing carnivore species were also incorporated. Distribution models based on habitat performed well for all species. Black bear (Ursus americanus) and fisher (Martes pennanti) distribution was similar in that occupancy of both species was negatively associated with paved roads. However, black bears were also associated with larger forest fragments and fishers with smaller forest fragments. No models with habitat features were more supported than the null habitat model for raccoons (Procyon lotor). Martens (Martes americana) were most associated with increased terrain ruggedness and elevation. Weasel (Mustela spp.) occupancy increased with the cover of deciduous forest. For most species dyads habitat-only models were more supported than those models with potential competitors incorporated. The exception to this finding was for the smallest carnivore taxa (marten and weasel) where habitat plus coyote abundance models typically performed better than habitat-only models. Assessing this carnivore community as whole, we conclude that differential habitat use is more important than species interactions in maintaining the distribution and structure of this carnivore guild.

Mesocarnivores fill a vital role in ecosystems through effects on community health and structure. Anthropogenic-altered landscapes can benefit some species and adversely affect others. For some carnivores, prey availability increases with urbanization, but landscape use can be complicated by interactions among carnivores as well as differing human tolerance of some species. We used camera traps to survey along a gradient of urban, rural, and forest cover to quantify how carnivore landscape use varies among guild members and determine if a species was a human exploiter, adapter, or avoider. Our study was conducted in and around Corvallis, Oregon from April 2018 to February 2019 (11,914 trap nights) using 47 camera trap locations on a gradient from urban to rural. Our focal species were bobcat ( Lynx rufus ), coyote ( Canis latrans ), gray fox ( Urocyon cinereoargenteus ), opossum ( Didelphis virginiana ), raccoon ( Procyon lotor ), and striped skunk ( Mephitis mephitis ). Raccoon and opossum were human exploiters with low use of forest cover and positive association with urban and rural developed areas likely due to human-derived resources as well as some refugia from larger predators. Coyote and gray fox were human adapters with high use of natural habitats while the effects of urbanization ranged from weak to indiscernible. Bobcat and striped skunk appeared to be human avoiders with negative relationship with urban cover and higher landscape use of forest cover. We conducted a diel temporal activity analysis and found mostly nocturnal activity within the guild, but more diurnal activity by larger-bodied predators compared to the smaller species. Although these species coexist as a community in human-dominated landscapes throughout much of North America, the effects of urbanization were not equal across species. Our results, especially for gray fox and striped skunk, are counter to research in other regions, suggesting that mesopredator use of urbanized landscapes can vary depending on the environmental conditions of the study area and management actions are likely to be most effective when decisions are based on locally derived data.

Ambient acoustic conditions shape how animals perceive and interact with their environments, yet their role in structuring space use remains underexplored. Noise can mask biologically informative sounds which can impact foraging success, physiological fitness, and displace animals from viable habitat. Here, we test how landscape variables and soundscape characteristics across a mixed-use forest landscape affect spatial distribution of a nocturnal predator with specialized hearing. We used passive acoustic recordings to assess the effects of noise levels within biologically relevant frequency ranges on landscape use of the northern saw-whet owl ( Aegolius acadicus ) across 276 sites in Oregon, USA. Owl landscape use declined with increasing noise levels in the 1.60–7.10 kHz band, corresponding with species peak auditory sensitivity. In contrast, general low-frequency sound (0.25–1.00 kHz) was a poor predictor of landscape use but negatively affected acoustic detection probability. These results provide evidence that sensory masking from ambient soundscapes can constrain the realized acoustic niche and drive avoidance of otherwise suitable habitat. Our findings highlight the importance of considering full-spectrum acoustic environments in spatial ecology and suggest that species distributions are shaped not only by physical habitat but also by the perceptual accessibility of ecological information.

The biodiversity crisis necessitates spatially extensive methods to monitor multiple taxonomic groups for evidence of change in response to evolving environmental conditions. Programs that combine passive acoustic monitoring and machine learning are increasingly used to meet this need. These methods require large, annotated datasets, which are time‐consuming and expensive to produce, creating potential barriers to adoption in data‐ and funding‐poor regions. Recently released pre‐trained avian acoustic classification models provide opportunities to reduce the need for manual labelling and accelerate the development of new acoustic classification algorithms through transfer learning. Transfer learning is a strategy for developing algorithms under data scarcity that uses pre‐trained models from related tasks to adapt to new tasks. Our primary objective was to develop a transfer learning strategy using the feature embeddings of a pre‐trained avian classification model to train custom acoustic classification models in data‐scarce contexts. We used three annotated avian acoustic datasets to test whether transfer learning and soundscape simulation‐based data augmentation could substantially reduce the annotated training data necessary to develop performant custom acoustic classifiers. We also conducted a sensitivity analysis for hyperparameter choice and model architecture. We then assessed the generalizability of our strategy to increasingly novel non‐avian classification tasks. With as few as two training examples per class, our soundscape simulation data augmentation approach consistently yielded new classifiers with improved performance relative to the pre‐trained classification model and transfer learning classifiers trained with other augmentation approaches. Performance increases were evident for three avian test datasets, including single‐class and multi‐label contexts. We observed that the relative performance among our data augmentation approaches varied for the avian datasets and nearly converged for one dataset when we included more training examples. We demonstrate an efficient approach to developing new acoustic classifiers leveraging open‐source sound repositories and pre‐trained networks to reduce manual labelling. With very few examples, our soundscape simulation approach to data augmentation yielded classifiers with performance equivalent to those trained with many more examples, showing it is possible to reduce manual labelling while still achieving high‐performance classifiers and, in turn, expanding the potential for passive acoustic monitoring to address rising biodiversity monitoring needs.

Wildlife monitoring programs designed to inform forest management and conservation decisions in the face of climate change benefit from long-term datasets with consistent methodology. Nevertheless, many monitoring programs may seek to transition to alternative methods because emerging technologies can improve trend tracking and expand the number of target populations, increase spatial scale, and reduce long-term costs. Integrated models strengthen the capacity to adapt long-term monitoring programs to next generation methods. Here we present a case study of northern spotted owl ( Strix occidentalis caurina ) population monitoring that is under transition. The first monitoring phase focused on territory occupancy and mark-resighting individual owls. Owing to rapidly declining populations and increasing costs, traditional methods are less viable for long-term monitoring. A non-invasive approach, passive acoustic monitoring, is effective for detecting spotted owl presence, estimating occupancy rates, distinguishing sex, detecting trends in populations, and monitoring many additional species. A key component to support transition to passive acoustic monitoring was the development of machine learning models to automate species detections that enable rapid and effective data processing and analysis workflows. Coupling passive acoustic monitoring networks with Forest Inventory and Analysis (FIA) and gradient nearest neighbor (GNN) datasets provide powerful tools for predicting forest change impacts on wildlife populations and identify winners and losers in dynamic landscapes. The second monitoring phase will leverage new technologies, expand the scope of inference, link forest inventory and remote sensing datasets, and transition the program to broad biodiversity monitoring that assists managers as they face myriad challenges in dynamic landscapes.

A major threat to small mammalian carnivore populations is human‐induced land use change, but conservation and management are inhibited by limited knowledge about their ecology and natural history. To fill a key knowledge gap of the western spotted skunk (Spilogale gracilis), we investigated their spatial ecology at the landscape and home range scale in the temperate rainforests of the Oregon Cascades during 2017–2019. For the landscape scale analysis, we used detections of western spotted skunks at 112 baited camera traps and fitted a dynamic occupancy model to investigate spatial distribution and drivers of inter‐seasonal and inter‐annual changes in occupancy. Concurrently, we radio‐collared 25 spotted skunks (9 female, 16 male) and collected 1583 relocations. Using continuous‐time movement models, we estimated large home range sizes for both male and female spotted skunks, relative to their body mass, and highly overlapping home ranges that indicated a lack of territoriality. Using these home ranges, we fitted a resource selection function using environmental covariates that we assigned to various hypotheses such as resources, predator avoidance, thermal tolerance, and disturbance. Overall, western spotted skunks were widely distributed across our study area (seasonal occupancy up to 63.7 ± 5.3%) and highly detectable (weekly detection probability = 41.2%). At both spatial scales, spotted skunks selected wetter areas and local valleys, which we attributed to areas with more food resources. At the home range scale, spotted skunks selected locations with lower predation risk and areas surrounded by more previously logged forests. In this montane environment, inter‐seasonal contractions in the spatial distributions of spotted skunks were strongly driven by their response to cold temperature and accumulated snow. This was especially evident when seasonal occupancy declined significantly following a severe heavy snow event in February 2019. Given that there is little information available on the natural history of the western spotted skunk, these results provide essential information about their ecology to focus future monitoring efforts and may help identify potential threats (e.g., forest management, severe snow events, or wildfires) to this species.

Passive acoustic monitoring is an emerging approach to wildlife monitoring that leverages recent improvements in automated recording units and other technologies. A central challenge of this approach is the task of locating and identifying target species vocalizations in large volumes of audio data. To address this issue, we developed an efficient data processing pipeline using a deep convolutional neural network (CNN) to automate the detection of owl vocalizations in spectrograms generated from unprocessed field recordings. While the project was initially focused on spotted and barred owls, we also trained the network to recognize northern saw‐whet owl, great horned owl, northern pygmy‐owl, and western screech‐owl. Although classification performance varies across species, initial results are promising. Recall, or the proportion of calls in the dataset that are detected and correctly identified, ranged from 63.1% for barred owl to 91.5% for spotted owl based on raw network output. Precision, the rate of true positives among apparent detections, ranged from 0.4% for spotted owl to 77.1% for northern saw‐whet owl based on raw output. In limited tests, the CNN performed as well as or better than human technicians at detecting owl calls. Our model output is suitable for developing species encounter histories for occupancy models and other analyses. We believe our approach is sufficiently general to support long‐term, large‐scale monitoring of a broad range of species beyond our target species list, including birds, mammals, and others. Passive acoustic monitoring is a powerful tool for wildlife conservation and research but depends on researchers’ ability to isolate signals of interest in the resulting data. We developed a data‐processing pipeline to automate this task using a deep convolutional neural network to identify the calls of six owl species in field recordings, with precision ranging from <0.4 to 77.1% and recall ranging from 63.1 to 91.5% by species. Our approach is applicable to a variety of taxa and is suitable for developing species encounter histories for occupancy models and other analyses.

Passive acoustic monitoring is widely used in conservation but often limited by delayed results due to manual retrieval and post‐processing. To explore the feasibility of real‐time monitoring in remote landscapes, we conducted field evaluations of a prototype automated surveying unit (ASU). The ASU integrated onboard convolutional neural network‐based sound classification with satellite data transmission for near real‐time species detection. We deployed ASUs alongside conventional autonomous recording units at eight sites to survey for northern spotted owls (Strix occidentalis caurina) and marbled murrelets (Brachyramphus marmoratus) in old‐growth forests of the Pacific Northwest, USA. We compared detection rates, background noise levels, and operational performance. ASUs successfully transmitted detection summaries and system status, though the elevated background noise in ASU recordings reduced detection rates relative to conventional units. These preliminary findings suggest that real‐time autonomous monitoring systems demonstrate potential to support time‐sensitive conservation efforts, while highlighting current limitations and areas for improvement. We evaluated a prototype acoustic monitoring device that performs onboard species classification and transmits detection summaries via satellite, enabling near real‐time data access from remote field sites. Deployed alongside conventional autonomous recording units at eight Northern Spotted Owl territories, the system demonstrated successful transmission and detection capability in dense forest environments, though with lower sensitivity due to elevated background noise. Our findings highlight the promise and current limitations of real‐time acoustic monitoring for conservation in remote landscapes.

Species coexistence is governed by availability of resources and intraguild interactions including strategies to reduce ecological overlap. Gray foxes are dietary generalist mesopredators expected to benefit from anthropogenic disturbance, but populations have declined across the midwestern USA, including severe local extirpation rates coinciding with high coyote and domestic dog occurrence and low red fox occurrence. We used data from a large‐scale camera trap survey in southern Illinois, USA to quantify intraguild spatial and temporal interactions among the canid guild including domestic dogs. We used a two‐species co‐occurrence model to make pairwise assessments of conditional occupancy and detection rates. We also estimated temporal activity overlap among species and fit a fixed‐effects hierarchical community occupancy model with the four canid species. We partitioned the posterior distributions to compare gray fox occupancy probabilities conditional on estimated state of combinations of other species to assess support for hypothesized interactions. We found no evidence of broadscale avoidance among native canids and conclude that spatial and temporal segregation were limited by ubiquitous human disturbance. Mean guild richness was two canid species at a site and gray fox occupancy was greater when any combination of sympatric canids was also present, setting the stage for competitive exclusion over time. Domestic dogs may amplify competitive interactions by increasing canid guild size to the detriment of gray foxes. Our results suggest that while human activities can benefit some mesopredators, other species such as gray foxes may serve as bellwethers for habitat degradation with trophic downgrading and continued anthropogenic homogenization. We directly assesses the effects of intraguild interactions and possible competitive exclusion among native canids and domestic dogs in a human‐dominated but rural landscape. The results demonstrate that human influences severely limit niche segregation among native canids potentially contributing to the decline of gray foxes, contrary to common assumptions that mesopredators benefit from human‐dominated landscapes, and highlighting the need to consider co‐occurrence and guild interactions over time as we assess the outcomes of trophic downgrading.