Deep Learning for Automated Detection and Identification of Migrating American Eel Anguilla rostrata from Imaging Sonar Data

Adult American eels (Anguilla rostrata) are vulnerable to hydropower turbine mortality during outmigration from growth habitat in inland waters to the ocean where they spawn. Imaging sonar is a reliable and proven technology for monitoring of fish passage and migration; however, there is no efficient automated method for eel detection. We designed a deep learning model for automated detection of adult American eels from sonar data. The method employs convolution neural network (CNN) to distinguish between 14 images of eels and non-eel objects. Prior to image classification with CNN, background subtraction and wavelet denoising were applied to enhance sonar images. The CNN model was first trained and tested on data obtained from a laboratory experiment, which yielded overall accuracies of >98% for image-based classification. Then, the model was trained and tested on field data that were obtained near the Iroquois Dam located on the St. Lawrence River; the accuracy achieved was commensurate with that of human experts.

Terrestrial Locomotion in American Eels (Anguilla rostrata): How Substrate and Incline Affect Movement Patterns

Synopsis Fishes overcome a variety of challenges in order to invade the terrestrial environment. Terrestrial invasions by fish occur over a variety of environmental contexts. In order to advance their bodies on land, fishes capable of terrestrial excursions tend to use one of three different types of locomotor modes: axial-based, appendage-based, or axial-appendage-based. Elongate species with reduced appendages, such as the American eel, Anguilla rostrata, rely on axial based locomotion in water and on land. When eels move from water to land as part of their complex life cycle, they inevitably encounter a variety of substrates and must traverse variable degrees of incline. The aim of this study was to determine the effect of substrate and incline on the terrestrial locomotion of the American eel. In order to do this, eels were filmed from a dorsal view on three substrates and four inclines: sand, loose pebbles, and fixed (glued) pebbles at 0°, 5°, 10°, and 15°. We digitized 20 evenly spaced points along the body to examine the following characteristics of locomotion: velocity, distance ratio (DR), and wave parameters such as wave amplitude, frequency, and length and assessed whether substrate, incline, or body position affected these parameters. DR, our metric of movement efficiency, was highest on the flat sand condition and lowest on 15° pebble conditions. Efficiency also varied across the body. Velocity followed a similar pattern being highest on sand at 0° and lowest at the steepest inclines. Wave amplitude generally increased toward the tail but was similar across substrates and inclines. Wave frequency was relatively consistent across the body on both pebble substrates, but on sand, frequency was higher toward the head but decreased toward the tail. Wavelengths on sand were the longest at 0° near the head and shorter wavelengths were observed on steeper inclines. Both pebble substrates elicited lower wavelengths that were more similar across the body. Overall, A. rostrata were more effective in navigating compliant substrates but struggled at steeper inclines. Our findings provide insight into locomotor challenges that American eels may encounter as they move from and between bodies of water.