INFLUENCE OF GROOMING ON PERMANENT ARTHROPOD ASSOCIATES OF BIRDS: CATTLE EGRETS, LICE, AND MITES.

Birds have a diverse community of "permanent" arthropods that complete their entire life cycle on the body of the host. Because some of these arthropods are parasites that reduce host fitness, birds control them by grooming, which consists of preening with the beak and scratching with the feet. Although preening is the primary component of grooming, scratching is essential for controlling arthropods on the head and neck, which cannot be preened. Several unrelated groups of birds have evolved comb-like pectinate claws on the middle toenail of each foot. We tested the role of these claws in the control of arthropods by experimentally removing teeth from the claws of captive western cattle egrets (Bubulcus ibis) infested with chewing lice (Insecta: Phthiraptera), feather mites (Acari: Sarcoptiformes), and nasal mites (Acari: Mesostigmata). After a period of 4 mo, we compared the abundance of arthropods on experimental birds to that of control birds with intact teeth. We used video to quantify the grooming rates of the captive birds, which groomed twice as much as wild birds. Experimental and control birds did not differ significantly in grooming time. Both groups virtually eradicated the chewing lice, but not feather mites or nasal mites. We found no support for the hypothesis that pectinate claws increase the efficiency of arthropod control by grooming. Experiments with wild birds are needed to test the hypothesis further under conditions in which birds devote less time to grooming.

An approach to rapid processing of camera trap images with minimal human input.

Camera traps have become an extensively utilized tool in ecological research, but the manual processing of images created by a network of camera traps rapidly becomes an overwhelming task, even for small camera trap studies.We used transfer learning to create convolutional neural network (CNN) models for identification and classification. By utilizing a small dataset with an average of 275 labeled images per species class, the model was able to distinguish between species and remove false triggers.We trained the model to detect 17 object classes with individual species identification, reaching an accuracy up to 92% and an average F1 score of 85%. Previous studies have suggested the need for thousands of images of each object class to reach results comparable to those achieved by human observers; however, we show that such accuracy can be achieved with fewer images.With transfer learning and an ongoing camera trap study, a deep learning model can be successfully created by a small camera trap study. A generalizable model produced from an unbalanced class set can be utilized to extract trap events that can later be confirmed by human processors.