Classifying Goliath Grouper (Epinephelus itajara) Behaviors from a Novel, Multi-Sensor Tag.

Inertial measurement unit sensors (IMU; i.e., accelerometer, gyroscope and magnetometer combinations) are frequently fitted to animals to better understand their activity patterns and energy expenditure. Capable of recording hundreds of data points a second, these sensors can quickly produce large datasets that require methods to automate behavioral classification. Here, we describe behaviors derived from a custom-built multi-sensor bio-logging tag attached to Atlantic Goliath grouper (Epinephelus itajara) within a simulated ecosystem. We then compared the performance of two commonly applied machine learning approaches (random forest and support vector machine) to a deep learning approach (convolutional neural network, or CNN) for classifying IMU data from this tag. CNNs are frequently used to recognize activities from IMU data obtained from humans but are less commonly considered for other animals. Thirteen behavioral classes were identified during ethogram development, nine of which were classified. For the conventional machine learning approaches, 187 summary statistics were extracted from the data, including time and frequency domain features. The CNN was fed absolute values obtained from fast Fourier transformations of the raw tri-axial accelerometer, gyroscope and magnetometer channels, with a frequency resolution of 512 data points. Five metrics were used to assess classifier performance; the deep learning approach performed better across all metrics (Sensitivity = 0.962; Specificity = 0.996; F1-score = 0.962; Matthew's Correlation Coefficient = 0.959; Cohen's Kappa = 0.833) than both conventional machine learning approaches. Generally, the random forest performed better than the support vector machine. In some instances, a conventional learning approach yielded a higher performance metric for particular classes (e.g., the random forest had a F1-score of 0.971 for backward swimming compared to 0.955 for the CNN). Deep learning approaches could potentially improve behavioral classification from IMU data, beyond that obtained from conventional machine learning methods.

Visitation patterns of two ray mesopredators at shellfish aquaculture leases in the Indian River Lagoon, Florida.

The Indian River Lagoon is a primary location of field-based "grow-out" for bivalve shellfish aquaculture along Florida's Atlantic coast. Grow-out locations have substantially higher clam densities than surrounding ambient sediment, potentially attracting mollusk predators to the area. Inspired by clammer reports of damaged grow-out gear, we used passive acoustic telemetry to examine the potential interactions between two highly mobile invertivores-whitespotted eagle rays (Aetobatus narinari) and cownose rays (Rhinoptera spp.)-and two clam lease sites in Sebastian, FL and compared these to nearby reference sites (Saint Sebastian River mouth, Sebastian Inlet) from 01 June 2017 to 31 May 2019. Clam lease detections accounted for 11.3% and 5.6% of total detections within the study period, for cownose and whitespotted eagle rays, respectively. Overall, the inlet sites logged the highest proportion of detections for whitespotted eagle rays (85.6%), while cownose rays (11.1%) did not use the inlet region extensively. However, both species had significantly more detections at the inlet receivers during the day, and on the lagoon receivers during the night. Both species exhibited long duration visits (> 17.1 min) to clam lease sites, with the longest visit being 387.5 min. These visit durations did not vary substantially between species, although there was individual variability. Based on generalized additive mixed models, longer visits were observed around 1000 and 1800 h for cownose and whitespotted eagle rays, respectively. Since 84% of all visits were from whitespotted eagle rays and these longer visits were significantly longer at night, this information suggests that observed interactions with the clam leases are potentially underestimated, given most clamming operations occur during daytime (i.e., morning). These results justify the need for continued monitoring of mobile invertivores in the region, including additional experimentation to assess behaviors (e.g., foraging) exhibited at the clam lease sites.

Lesion recovery of two scleractinian corals under low pH conditions: Implications for restoration efforts.

Some coral restoration efforts are involving cultivation of coral microfragments in land-based pools under controlled conditions until they reach viable size for outplanting. However, gaps in knowledge with these efforts include effects of changing pH on regeneration rates of tissue lesions and other physiological responses on different size fragments. To address this, two fragment sizes of Porites porites and Porites astreoides were artificially inflicted with lesions and incubated in two pH treatments to follow effects on recovery and physiological performance. Recovery was significantly reduced at reduced pH for P. porites in both fragment sizes; while recovery of P. astreoides was reduced only in the larger fragments. Different responses were also seen for Symbiodinium density and total protein with pH and fragment size. Effects on lesion recovery rate from pH and fragment size were species specific and may be related to morphology and/or energetic constrains.

The role of P450 metabolism in the estrogenic activity of bifenthrin in fish.

Bifenthrin, a pyrethroid pesticide, is estrogenic in vivo in fishes. However, bifenthrin is documented to be anti-estrogenic in vitro, in the ER-CALUX (estrogen receptor) cell line. We investigated whether metabolite formation is the reason for this incongruity. We exposed Menidia beryllina (inland silversides) to 10ng/l bifenthrin, 10ng/l 4-hydroxy bifenthrin, and 10ng/l bifenthrin with 25µg/l piperonyl butoxide (PBO) - a P450 inhibitor. Metabolite-exposed juveniles had significantly higher estrogen-mediated protein levels (choriogenin) than bifenthrin/PBO-exposed, while bifenthrin alone was intermediate (not significantly different from either). This suggests that metabolites are the main contributors to bifenthrin's in vivo estrogenicity.