The Vertebrate Integrative Physiology (VIP) lab monitors the population of northern elephant seals at the largest mainland breeding colony, located at Piedras Blancas (San Simeon, CA). As the population expands, more human-seal interactions and conflicts over land use occur. The VIP lab's work informs California State Parks and helps with the management of the rookery. Currently, members of the VIP lab fly a drone over the beaches, capture multiple images, and manually count the seals, which takes around 14 to 21 hours of analysis per survey. Machine learning methods such as Convolutional Neural Networks (CNN) and Region-based Convolutional Neural Networks (RCNN) have been shown to quickly and accurately determine the count but require lots of data, which is not feasible for this task due to the 79 available beach images. By dividing larger beach images into smaller sub-images, it is possible to generate more data, facilitating the use of deep learning techniques. This thesis outlines a pipeline to use these sub-images and determine the seal count of a beach image.

Bayou St. John (BSJ) and City Park Lakes and Lagoons (CPLL) are urban waterways in New Orleans, Louisiana. I studied habitat selection of red drum in BSJ, and fish assemblage change in BSJ and CPLL over 40 years. Temperature was found to be the best predictor of red drum habitat selection in Bayou St. John, while salinity and change in depth also were found to be good predictors for certain sites. Potential prey item abundance did not appear to influence habitat selection. Using data from 1971 – 2010, nearshore habitats in CPLL were affected by Hurricane Katrina, but have sense recovered and nearshore habitats in BSJ were found to have decreased diversity. Pelagic habitats in both areas were found stable across 40 years. Since 2006, nearshore assemblages were similar for CPLL and BSJ with a decrease in fishes from Order: Cyprinodontiformes and an increase in other fishes seen across years.

Web mashup is a lightweight technology used to integrate data from remote sources without direct access to their databases. As a data consumer, a Web mashup application creates new contents by retrieving data through the Web application programming interface (API) provided by the external sources. As a data provider, the service program publishes its Web API and implements the specified functions. In the project reported by this thesis, we have implemented two Web mashup applications to enhance the Web site oystersentinel.org: the Perkinsus marinus model and the Oil Spill model. Each model overlay geospatial data from a local database on top of a coastal map from Google Maps. In addition, we have designed a Web-based data publishing service. In this experimental system, we illustrated a successful Web mashup interface that allows outside developers to access the data about the local oyster stock assessment.

Quantifying the impacts of restoration on coastal waterways is crucial to understanding their effectiveness. Here, I look at the impacts of multiple restoration projects on urban waterways within the city limits of New Orleans, LA, with an emphasis on the response of fishes. First I report the effects of two projects designed to improve exchange down estuary on the hydrologic characteristics of Bayou St. John (BSJ). Within BSJ, flow is dominated by subtidal wind driven processes. Removal of an outdated flood control structure did not appear to alter exchange in BSJ, but removal combined with sector gate openings did. I also refined a three dimensional hydrodynamic model of this system to have accurate predictions of velocity and elevation. Temperature and salinity were difficult to constrain with this model. Solutions of this model were used to compare flow metrics, along with linearly interpolated temperature, and other variables to Red Drum (Sciaenops ocellatus) activity and movement patterns. Relationships between Red Drum activity and velocity suggested a response to subtidal, wind driven flow. Overall, high Sedentariness, a measure of inactivity, was found suggesting high levels of site fidelity. Higher mean Sedentariness during the night was also found. I also used a pseudo-BACI design to analyze the fish assemblage response to removal of an outdated flood control structure and the impacts of sector gate openings on fish guild species richness in BSJ. Limited differences were found when comparing fish assemblages before and after removal, but these differences were likely due to a decrease in salinity not restoration efforts. No significant differences in Freshwater or Estuarine fish guild species richness was observed for any of the control or impact sites. Marine fish species richness was found to be higher immediately following sector gate openings at the site closest to the structure, suggesting an initial pulse of young marine organisms is provided via these events. The findings here can be used to optimize management of exchange flow in coastal impounded waterways.

This thesis examines the environmental damages done by humans, such as plastic pollution, harmful energy production, and habitat destruction, and the things we are doing to correct them. It analyzes the costs and benefits government policies such as the ban of plastic products, along with creative innovations such as a sea vacuum and a cannon that shoots fish. Are we doing enough to heal the damage we have caused? Furthermore, can we change our behaviors to prevent more damage in the future?

In the southeastern United States, skimmer trawls, pusher-head trawls, wing nets, and small try nets (headrope length less than 12-ft (3.66-m)) are exempt from using a turtle excluder device (TED) and instead must adhere to tow time restrictions as a mode to mitigate sea turtle bycatch. However, observer and stranding data indicate that these tow times may often be exceeded and result in mortality of sea turtles. The National Marine Fisheries Service (NMFS) published a notice of proposed rulemaking in December 2016 to extend TED requirements to other trawl types, however there has been limited development of specialized TEDs for these smaller trawls. In anticipation of a regulatory change, we developed and identified multiple versions of a top-opening TED with a minimum width of 28-in (71-cm) and height of 24-in (61-cm) as the best option for small trawl gears. Prototypes were initially equipped within small try nets and tested for gear performance and sea turtle exclusion in Panama City, Florida. The final designs were then tested for target shrimp retention, bycatch reduction, and general usability of TEDs in the commercial fisheries. A paired comparison test was conducted in 8-ft (2.44-m) and 10-ft (3.05-m) try nets aboard the NMFS R/V Caretta, three commercial skimmer trawls in vesselsFarfantepenaeus duorarum) fishery. There was a general reduction of shrimp and bycatch averaging from a minimum loss of 3.31% in the Miami wing net fishery (FV FL-01) to 22.07% in the skimmer trawl fishery (FV LA-01). This dissertation research indicates that TEDs can function properly in small trawl types, however additional studies are recommended to minimize the shrimp loss and improve the overall TED effectiveness specific to each fishery and trawl type evaluated.

This study used an in situ heart preparation to analyze the power and work of spontaneously beating hearts of four anurans (R. marina, L. catesbeianus, X. laevis, P. edulis) and three urodeles (N. maculosus, A. tigrinum, A. tridactylum) in order to elucidate the meaning of relative ventricle mass (RVM) in terms of specific cardiac performance variables. This study also tests two hypotheses: 1) the ventricles of terrestrial species (R. marina, P. edulis, A. tigrinum) of amphibians are capable of greater maximum power outputs (Pmax) compared to aquatic species (X. laevis, A. tridactylum, N. maculosus, L. catesbeianus) and, 2) the ventricles of Anuran species (R. marina, P. edulis, L. catesbeianus, X. laevis) are capable of greater maximum power output compared to aquatic species (A. tigrinum, A. tridactylum, N. maculosus). The data supported both hypotheses. RVM was significantly correlated with Pmax, stroke volume, cardiac output, afterload at Pmax, and preload at Pmax. Preload at Pmax and afterload at Pmax also correlated very closely with each other, suggesting that an increase blood volume and/or increased modulation of sympathetic tone may influence interspecific variation RVM and may have played a role in supporting higher rates of metabolism, as well as dealing with hypovolemic stresses of life on land.

Effects of exercise, social facilitation, and delayed conditioning after vi exercise on the learning behavior of 5 to 6-inch rainbow trout, Salmo gairdneri, was measured in a conditioned avoidance response apparatus. The conditioning schedule contained an intertrial interval between alternate presentations of conditioned (light) and unconditioned (shock) stimuli. This randomization eliminated learning losses found in a previous study. Mean percentage avoidance, the measurement of learning, did not decrease significantly during conditioning trials. Exercised fish learned avoidance better than did non-exercised fish. Mean percentage avoidance for fish exercised at 0.5 ft/sec was 66.3; at 1.0 ft/ sec, 1.5 ft/sec, and no-exercise, mean percentage avoidance was 68.2, 68.9 and 65.0, respectively. Social facilitation affected learning in the one, two, and three fish per cell tests, where mean percentage avoidance was 55.5, 68.9, and 81.0 percent, respectively. A delay of 24 hours between exercise and conditioning resulted in decreased learning levels. Mean percentage avoidance was 60.3, 63.5, 67.7, and 53.7 for the 1-, 2-, 2-, and 24-hour delay tests, respectively; however, mean percentage avoidance for the last 60 trials of each test indicated the 1-, 2-, and 4-hour delay tests were all similar, over 70 percent, while mean percentage avoidance for the 24-hour delay test was only 57.8 percent.

Rainbow trout lactate dehydrogenase (LDH) B2' B2' , B2' B2" and B2" B2" phenotypes were tested under a variety of conditions for swimming endurance, oxygen consumption rates, and blood pH, p02, and lactate. The B2" B2" phenotype exhibited a lower swimming endurance, under a limited oxygen supply, than the B2' B2' and B2' B2" phenotypes. Blood pH at fatigue did not differ among the LOH phenotypes that swam under low oxygen conditions (2 mg/1). Blood p02 at fatigue was significantly higher for females than for males under low oxygen conditions. Oxygen consumption rates were significantly higher for males than for females at dissolved oxygen concentrations of 3 mg/1 to 8 mg/1. The depression in oxygen consumption rates at low oxygen levels was greater for males than for females. The increase in blood lactate with swimming under low oxygen concentrations and static exposure to low oxygen depended on LOH phenotype. The adaptive significance of the LDH B2' and B2" alleles was discussed. Thirteen populations of Intermountain cutthroat trout were sampled for biochemical genetic variation. Variation was observed in muscle aspartate aminotransferase for all populations which would fit a genetic model of two loci and three alleles. Unique muscle A group lactate dehydrogenase variation was observed in Snake Valley cutthroat trout which suggested the presence of several variant alleles. Allele frequencies for tetrazolium oxidase changed during three consecutive years in Bear Lake cutthroat trout that were trapped and spawned artificially. The possible causes of those changes were discussed. No variation was observed in alphaglycerophosphate dehydrogenase, isocitrate dehydrogenase, eye lactate dehydrogenase or muscle malate dehydrogenase in cutthroat trout.

The lack of timely reporting of commercial fisheries landings interferes with effective management of fisheries in United States Virgin Islands (USVI). Federal law requires that landings be limited to prevent annual catch limits (ACLs) from being exceeded. Previous attempts to predict total landings have used historic data from prior fishing seasons to predict future landings rather than leveraging available in-season data to provide a more real-time prediction of landings. This study presents an in-season model that predicts total landings using partial reports from the current fishing year. This estimate of total landings, including error bounds around that estimate, can then be compared to the ACL established for the species to estimate potential deviations from the allowable landings and adjust effort accordingly. The performance of the model was tested in a retrospective analysis on historical commercial landings data. Differences between predicted and observed fishing year landings by defined cut-off dates were used to identify reasonable deadlines for fishery managers to begin making reliable predictions on total annual landings. On average, predictions can be made with less than 9% error with at least four months of partial data, and with less than 5% error with at least seven months of partial data. This model's in-season predictions should be useful to managers to prevent ACL overages, and to guide fishers in their application of effort within and among components of the fishery, for example, to shift effort from one fishery management unit to another in response to excessive landings.