Using Ecosystem-Based Modeling to Describe an Oil Spill and Assess the Long-Term Effects

The goal of the research conducted in this dissertation was to define and test methods to incorporate oil spill effects into an ecosystem-based assessment model. It was instigated by the Deepwater Horizon oil spill, an unprecedented oil spill in the United States for both depth and volume, with unknown implications for the health of the region. Using an ecosystem-based assessment model like Atlantis, with integrated oil spill dynamics, was the ideal candidate to predict long-term impacts such as decreased abundance or population recovery time. However no previous methodology existed for doing so in any ecosystem-based assessment model. Therefore, first I conducted a literature review to gather data across fish species on lesion frequency and fish body growth impacts from oil exposure. The two data sets were then fitted to four different dose-response models, and an effect threshold log-linear “hockey-stick” model was selected as the best fit and most parsimonious for both lesions and growth. Next, I conducted a similar analysis comparing macrofaunal and meiofaunal abundances to oil exposure concentrations in the Gulf of Mexico collected after Deepwater Horizon. I confirmed that these data had the domed relationship between invertebrate abundances and oil concentration observed in previous invertebrate oil studies. This domed relationship indicates that abundance increases at low to moderate oil levels, and declines at high oil levels. To drive this relationship in an Atlantis ecosystem model, three scenarios were tested in combination with oil toxicity: 1) Mississippi nutrient loading, 2) increased detritus from marine oil snow sedimentation and flocculent accumulation, and 3) predators altering their behavior to avoid oil exposure. At the Atlantis polygon resolution, only scenario 2, increased detritus from marine oil snow sedimentation and flocculent accumulation, generated the domed relationship for invertebrate abundances. Lastly, the “hockey-stick” model for fish mortality and growth was applied to both fishes and invertebrates in combination with scenario 2 for an integrated long-term assessment of the Gulf of Mexico. Newly available fish exposure data were used to generate an uptake-depuration model for this assessment. The combined effect forcings on vertebrates and invertebrates proved to have more severe long-term implications on population size and recovery than simulations with only fish forcings. Large demersal fishes, including elasmobranchs, were the most severely impacted by large biomass declines in the model spill region. Sensitivity analyses indicated that there was the potential for no recovery during 50 years of simulation in the spill region for many functional groups. Analysis of the synergy between fishing mortality F and toxicity from an oil spill identified that some guilds are more sensitive in an oil spill simulation to varied F than others. Snappers are the most sensitive to increased fishing mortality, while groupers respond the most to a reduction in fishing mortality. The invertebrate guild and small pelagic fishes responded the least to different values of F. Changing F also had implications for guild recovery – some guilds only fully recovered to control scenario biomass when F was reduced. A few functional groups were unable to survive with the combined effects of oil toxicity and increased F, and went extinct before the end of the 50-year simulation. Overall, this work provided the first framework for initial integrated modeling of oil spill impacts in an ecosystem-based assessment model, a potentially important component to future ecosystem-based fisheries management. The “hockey-stick” dose response model is applicable beyond Atlantis modeling, and can be tuned to fit specific events based on available data. I have also identified the importance of including marine oil snow sedimentation and flocculent accumulation to accurately drive the response of benthic invertebrates. Findings from the combined vertebrate and invertebrate simulations should help inform research efforts in the Gulf of Mexico and future oil spill response efforts