Aerial Wildlife Image Repository for animal monitoring with drones in the age of artificial intelligence.

Drones (unoccupied aircraft systems) have become effective tools for wildlife monitoring and conservation. Automated animal detection and classification using artificial intelligence (AI) can substantially reduce logistical and financial costs and improve drone surveys. However, the lack of annotated animal imagery for training AI is a critical bottleneck in achieving accurate performance of AI algorithms compared to other fields. To bridge this gap for drone imagery and help advance and standardize automated animal classification, we have created the Aerial Wildlife Image Repository (AWIR), which is a dynamic, interactive database with annotated images captured from drone platforms using visible and thermal cameras. The AWIR provides the first open-access repository for users to upload, annotate, and curate images of animals acquired from drones. The AWIR also provides annotated imagery and benchmark datasets that users can download to train AI algorithms to automatically detect and classify animals, and compare algorithm performance. The AWIR contains 6587 animal objects in 1325 visible and thermal drone images of predominantly large birds and mammals of 13 species in open areas of North America. As contributors increase the taxonomic and geographic diversity of available images, the AWIR will open future avenues for AI research to improve animal surveys using drones for conservation applications. Database URL: https://projectportal.gri.msstate.edu/awir/.

Fire season matters for midstory hardwood control: Impacts of fire season and firing technique on plant communities.

In the southeastern USA, lack of historical fire regimes often leads to hardwood encroachment into early successional plant communities and managed pine stands, reducing wildlife value and timber yields. Land managers lack information on how firing technique interacts with fire season to influence plant communities. We designed an experiment to quantify these interactions in east-central Mississippi with pairs of 4 m × 8 m plots randomly assigned a backing and heading fire in each of three seasons: February (Feb), May-June (May/Jun), and September-October (Sep/Oct). We used thermocouples to monitor fire temperature and tagged midstory trees to monitor response. We lit heading fires with an 18-25 kph wind generated by a backpack blower and backing fires into the ambient wind. Despite backing fires producing longer residence times than heading fires and raising temperature above the lethal threshold of 60 °C an average of 54 s longer, firing technique did not influence midstory response one growing season post-fire. Backing and heading fires produced similar maximum temperatures. For both firing techniques, May/Jun resulted in the highest midstory mortality rates which were 3-fold greater than Sep/Oct and 4-fold greater than Feb. Among all three fire seasons, trees with a 2.5 cm diameter at breast height (DBH) had approximately a 75% chance of top-kill which decreased to <20% as trees approached 6.5 cm DBH. We found no effects of fire season on fire temperature, rate of spread, flame height, or percent crown scorch. We found no significant interactions between fire season and firing technique. Understory analysis revealed Sep/Oct produced the greatest increase in forb coverage, May/Jun resulted in the most grass coverage, and Feb produced the most brambles (Rubus spp.). On sites with similar species, weather, and fuel conditions to ours, land managers should emphasize fire season over firing technique for midstory control and understory manipulation. Where midstory hardwood control with fire is a priority, fire return intervals should be frequent enough to prevent trees from exceeding 2.5 cm DBH to avoid trees escaping fire's reach. These data can help managers reduce midstory competition with crop trees and promote understory development for wildlife.

What drives wild pig (Sus scrofa) movement in bottomland and upland forests?

BACKGROUND: The wild pig (Sus scrofa) is an exotic species that has been present in the southeastern United States for centuries yet continues to expand into new areas dominated by bottomland and upland forests, the latter of which are less commonly associated with wild pigs. Here, we aimed to investigate wild pig movement and space use attributes typically used to guide wild pig management among multiple spatiotemporal scales. Our investigation focused on a newly invaded landscape dominated by bottomland and upland forests. METHODS: We examined (1) core and total space use using an autocorrelated kernel density estimator; (2) resource selection patterns and hot spots of space use in relation to various landscape features using step-selection analysis; and (3) daily and hourly differences in movement patterns between non-hunting and hunting seasons using generalized additive mixed models. RESULTS: Estimates of total space use among wild pigs (n = 9) were smaller at calculated core (1.2 ± 0.3 km2) and 90% (5.2 ± 1.5 km2) isopleths than estimates reported in other landscapes in the southeastern United States, suggesting that wild pigs were able to meet foraging, cover, and thermoregulatory needs within smaller areas. Generally, wild pigs selected areas closer to herbaceous, woody wetlands, fields, and perennial streams, creating corridors of use along these features. However, selection strength varied among individuals, reinforcing the generalist, adaptive nature of wild pigs. Wild pigs also showed a tendency to increase movement from fall to winter, possibly paralleling increases in hard mast availability. During this time, there were also increases in anthropogenic pressures (e.g. hunting), causing movements to become less diurnal as pressure increased. CONCLUSIONS: Our work demonstrates that movement patterns by exotic generalists must be understood across individuals, the breadth of landscapes they can invade, and multiple spatiotemporal scales. This improved understanding will better inform management strategies focused on curbing emerging invasions in novel landscapes, while also protecting native natural resources.

Fusion of visible and thermal images improves automated detection and classification of animals for drone surveys.

Visible and thermal images acquired from drones (unoccupied aircraft systems) have substantially improved animal monitoring. Combining complementary information from both image types provides a powerful approach for automating detection and classification of multiple animal species to augment drone surveys. We compared eight image fusion methods using thermal and visible drone images combined with two supervised deep learning models, to evaluate the detection and classification of white-tailed deer (Odocoileus virginianus), domestic cow (Bos taurus), and domestic horse (Equus caballus). We classified visible and thermal images separately and compared them with the results of image fusion. Fused images provided minimal improvement for cows and horses compared to visible images alone, likely because the size, shape, and color of these species made them conspicuous against the background. For white-tailed deer, which were typically cryptic against their backgrounds and often in shadows in visible images, the added information from thermal images improved detection and classification in fusion methods from 15 to 85%. Our results suggest that image fusion is ideal for surveying animals inconspicuous from their backgrounds, and our approach uses few image pairs to train compared to typical machine-learning methods. We discuss computational and field considerations to improve drone surveys using our fusion approach.

Improving Animal Monitoring Using Small Unmanned Aircraft Systems (sUAS) and Deep Learning Networks.

In recent years, small unmanned aircraft systems (sUAS) have been used widely to monitor animals because of their customizability, ease of operating, ability to access difficult to navigate places, and potential to minimize disturbance to animals. Automatic identification and classification of animals through images acquired using a sUAS may solve critical problems such as monitoring large areas with high vehicle traffic for animals to prevent collisions, such as animal-aircraft collisions on airports. In this research we demonstrate automated identification of four animal species using deep learning animal classification models trained on sUAS collected images. We used a sUAS mounted with visible spectrum cameras to capture 1288 images of four different animal species: cattle (Bos taurus), horses (Equus caballus), Canada Geese (Branta canadensis), and white-tailed deer (Odocoileus virginianus). We chose these animals because they were readily accessible and white-tailed deer and Canada Geese are considered aviation hazards, as well as being easily identifiable within aerial imagery. A four-class classification problem involving these species was developed from the acquired data using deep learning neural networks. We studied the performance of two deep neural network models, convolutional neural networks (CNN) and deep residual networks (ResNet). Results indicate that the ResNet model with 18 layers, ResNet 18, may be an effective algorithm at classifying between animals while using a relatively small number of training samples. The best ResNet architecture produced a 99.18% overall accuracy (OA) in animal identification and a Kappa statistic of 0.98. The highest OA and Kappa produced by CNN were 84.55% and 0.79 respectively. These findings suggest that ResNet is effective at distinguishing among the four species tested and shows promise for classifying larger datasets of more diverse animals.

Availability and assessment of microplastic ingestion by marsh birds in Mississippi Gulf Coast tidal marshes.

Millions of tons of plastic enter the environment every year, where much of it concentrates in environmental sinks such as tidal marshes. With prior studies documenting harm to marine fauna caused by this plastic pollution, the need to understand how this novel type of pollution affects estuarine fauna is great. Yet, research on the fate and uptake of plastic pollutants in estuarine ecosystems is sparse. Therefore, we quantified plastic prevalence and ingestion by two species of resident marsh bird, Clapper Rails (Rallus crepitans) and Seaside Sparrows (Ammospiza maritima), in coastal marsh ecosystems within Mississippi. We detected microplastics (plastics smaller than 5 mm) in 64% of marsh sediment samples, 83% of Clapper Rail and 69% of Seaside Sparrow proventriculus samples. Dominant types of microplastics detected in sediment and bird samples were fibers. This study provides the first evidence of microplastic ingestion by marsh birds and its distribution in coastal marshes within Mississippi.

Multiple paternities increase genetic diversity of offspring in Brandt's voles.

Mating system and philopatry influence the genetic structure of a social group in mammals. Brandt's vole (Lasiopodomys brandtii) lives in social groups year-round and has male biased dispersal, which makes the vole a model system for studies of genetic consequences of mating system and philopatry. This study aimed to test the hypotheses that: (1) multiple paternity (MP) would exist in Brandt's voles, enhance offspring genetic diversity and reduce genetic relatedness between littermates; (2) promiscuity would occur in this species in that males and females mate with multiple partners; and (3) plural breeders of a social group would be genetically related because of philopatry of female juveniles in Brandt's voles. Paternity analysis indicated that MP occurred in 11 (46%) of 24 social groups examined and that promiscuity existed in this species. Multiple paternity litters had twice the offspring genetic diversity and half the average within-litter genetic relatedness of single paternity litters. We also found plural breeding females in six social groups. Average pairwise genetic relatedness of plural breeders ranged from 0.41 to 0.72 in four social groups, suggesting first-order kinship. Future studies need to investigate effects of reproductive skew and MP on population genetic structure of Brandt's voles.