Bayesian approaches are increasingly being used to embed mechanistic process models used into statistical state space frameworks for environmental prediction and forecasting applications. In this study, I focus on Bayesian State Space Models (SSMs) for modeling the temporal dynamics of carbon in terrestrial ecosystems. In Chapter 1, I provide an introduction to Ecological Forecasting, State Space Models, and the challenges of using State Space Models for Ecosystems. In Chapter 2, we provide a brief background on State Space Models and common methods of parameter estimation. In Chapter 3, we simulate data from an example model (DALECev) using driver data from the Talladega National Ecosystem Observatory Network (NEON) site and perform a simulation study to investigate its performance under varying frequencies of observation data. We show that as observation frequency decreases, the effective sample size of our precision estimates becomes too small to reliably make inference. We introduce a method of tuning the time resolution of the latent process, so that we can still use high-frequency flux data, and show that this helps to increase sampling efficiency of the precision parameters. Finally, we show that data cloning is a suitable method for assessing the identifiability of parameters in ecosystem models. In Chapter 4, we introduce a method for embedding positive process models into lognormal SSMs. Our approach, based off of moment matching, allows practitioners to embed process models with arbitrary variance structures into lognormally distributed stochastic process and observation components of a state space model. We compare and contrast the interpretations of our lognormal models to two existing approaches, the Gompertz and Moran-Ricker SSMs. We use our method to create four state space models based off the Gompertz and Moran-Ricker process models, two with a density dependent variance structure for the process and observations and two with a constant variance structure for the process and observations. We design and conduct a simulation study to compare the forecast performance of our four models to their counterparts under model mis-specification. We find that when the observation precision is estimated, the Gompertz model and its density dependent moment matching counterpart have the best forecasting performance under model mis-specification when measured by the average Ignorance score (IGN) and Continuous Ranked Probability Score (CRPS), even performing better than the true generating model across thirty different synthetic datasets. When observation precisions were fixed, all models except for the Gompertz displayed a significant improvement in forecasting performance for IGN, CRPS, or both. Our method was then tested on data from the NOAA Dengue Forecasting Challenge, where we found that our novel constant variance lognormal models had the best performance measured by CRPS, and also had the best performance for both CRPS and IGN for one and two week forecast horizons. This shows the importance of having a flexible method to embed sensible dynamics, as constant variance lognormal SSMs are not frequently used but perform better than the density dependent models here. In Chapter 5, we apply our lognormal moment matching method to embed the DALEC2 ecosystem model into the process component of a state space model using NEON data from University of Notre Dame Environmental Research Center (UNDE). Two different fitting methods are considered for our difficult problem: the updated Iterated Filtering algorithm (IF2) and the Particle Marginal Metropolis Hastings (PMMH) algorithm. We find that the IF2 algorithm is a more efficient algorithm than PMMH for our problem. Our IF2 global search finds candidate parameter values in thirty hours, while PMMH takes 82 hours and accepts only .12% of proposed samples. The parameter values obtained from our IF2 global search show good potential for out of sample prediction for Leaf Area Index and Net Ecosystem Exchange, although both have room for improvement in future work. Overall, the work done here helps to inform the application of state space models to ecological forecasting applications where data are not available for all stocks and transfers at the operational timestep for the ecosystem model, where large numbers of process parameters and long time series provide computational challenges, and where process uncertainty estimation is desired. / Doctor of Philosophy / With ecosystem carbon uptake expected to play a large role in climate change projections, it is important that we make our forecasts as informed as possible and account for as many sources of variation as we can. In this dissertation, we examine a statistical modeling framework called the State Space Model (SSM), and apply it to models of terrestrial ecosystem carbon. The SSM helps to capture numerous sources of variability that can contribute to the overall predictability of a physical process. We discuss challenges of using this framework for ecosystem models, and provide solutions to a number of problems that may arise when using SSMs. We develop methodology for ensuring that these models mimic the well defined upper and lower bounds of the physical processes that we are interested in. We use both real and synthetic data to test that our methods perform as desired, and provide key insights about their performance.

Conceptual and theoretical advancements have been developed in recent years to break down the assumptions and traditional boundaries that establish seemingly independent disciplines, and the research outlined in this dissertation aspires to build on these advancements to provide innovative solutions to a broad array of modern problems in social-ecological. I used a variety of techniques to address challenges ranging from disconnections between theory and application, perceived versus realized roles of prescribed fire in resprouting shrublands, and the need for broader participation in research as part of undergraduate education. The chapters in this dissertation serve as a case-study approach across multiple scientific disciplines that overcome the traditions and assumptions that conflict with our ability to develop innovative solutions to modern social-ecological problems. First, I bridge theoretical and applied concepts by showing how recent theoretical advancements in resilience can be integrated into a predictive framework for environmental managers. Second, experimental data from multiple experiments were collected in two ecological regions of Texas to assess the potential for using extreme fire, in isolation and in combination with herbicide, as a novel intervention approach in resprouting shrublands of the southern Great Plains. The findings from these experiments demonstrate the importance of moving past traditional assumptions of when prescribed fire should be applied to demonstrate new patterns of woody plant responses to the applications of “more extreme” prescribed fires while not causing undesirable invasions by exotic grasses and exotic insects. Finally, I initiated a PhD instructed course on undergraduate research that sought to increase undergraduate participation while lowering the costs of conducting research. This chapter shows how traditional approaches of supporting undergraduate research are incapable of meeting the broader goals established by society and reveal a novel approach that can provide an additional pathway for supporting undergraduate student participation at large, research-based universities. Ultimately, this research suggests that our capacity to enhance services in social-ecological systems ultimately hinges upon the integration of theoretical and applied concepts that drive policy and governance and overcoming the assumptions and traditions that limit their integration.

A fundamental focus in ecology is understanding interactions between environmental heterogeneity and ecological community structure, both of which are currently undergoing unprecedented alterations due to global change. In particular, many freshwater phytoplankton communities are experiencing multiple global change stressors, altering phytoplankton community composition, biomass, and spatial distribution. I used multiple approaches to characterize the interactions between spatial distribution and community structure of phytoplankton and quantify uncertainty in predictions of phytoplankton temporal dynamics. First, I analyzed data from 51 lakes to determine the environmental drivers of phytoplankton vertical distributions across the water column for different phytoplankton groups. I show that the relative importance of environmental drivers varies according to the functional traits of each phytoplankton group. Second, I conducted whole-ecosystem experiments in a reservoir to assess phytoplankton responses to surface water mixing events, which may become more prevalent as storms increase under global change. My results demonstrate that aggregated phytoplankton biomass has inconsistent responses to mixing over the short term, but responses of morphology-based functional groups of phytoplankton to mixing are more predictable. Third, I conducted a long-term whole-ecosystem experiment to assess phytoplankton responses to changes in water column thermal gradients which are predicted to increasingly occur under global change. I found that phytoplankton depth distributions responded similarly to thermal gradient disturbance over multiple years, and changes in depth distributions were related to changes in community composition. Fourth, I produced weekly hindcasts of phytoplankton density in a lake for two years to determine the dominant sources of uncertainty in phytoplankton density predictions. I found that better estimation of current phytoplankton density improved representation of error in phytoplankton models, and incorporation of additional life history stages to model structure may improve phytoplankton predictions. Overall, my dissertation chapters demonstrate that the vertical distribution and community structure of phytoplankton are linked, and that the interaction of phytoplankton community structure with environmental heterogeneity is more predictable over longer-term (e.g., months to years) than shorter-term (e.g., days to weeks) scales. My research emphasizes that consideration of phytoplankton community dynamics and the uncertainty associated with phytoplankton predictions are needed for freshwater management under global change. / Doctor of Philosophy / Freshwater phytoplankton, which are microscopic primary producers, are experiencing many environmental changes in lakes and reservoirs due to global change. This includes changes in water temperature, which affects phytoplankton growth and the types of phytoplankton that are present in the water. As a result, phytoplankton communities are changing in ways that affect water quality. For example, phytoplankton may grow rapidly and form blooms which cause unsightly surface scums, clog filters at water treatment plants, or release toxins. My dissertation research uses ecosystem experiments, computer modeling, and large datasets from many lakes to study how the interactions between phytoplankton and their environment might change due to human activities. I found that it is difficult to predict how phytoplankton will respond to changes in water temperature over the short term (days to weeks), but that longer-term (months to years) responses to water temperature changes are more predictable. I also found that the types of phytoplankton present in the water vary across depth in response to light, temperature, and predation. Since the species of phytoplankton that are present determine a waterbody's water quality, my results indicate that water quality can vary substantially among different depths. Finally, I found that the greatest sources of uncertainty in predicting phytoplankton are due to the challenges in accurately measuring the amount of phytoplankton that are present in a lake and representing complex phytoplankton processes in computer models. My research demonstrates that it is important to think about multiple types of phytoplankton and how they interact with the environment, not just the total amount of phytoplankton present, when predicting how water quality will change due to global change. In addition, it is important to consider the uncertainty associated with predictions of phytoplankton when we make decisions about how to manage water quality.

Accelerating rates of change in ecosystems globally heighten the need for improved predictions of future ecological conditions. Freshwater lakes and reservoirs, which provide numerous ecosystem services, are particularly threatened by global change stressors and have already exhibited substantial changes to their physical, chemical, and biological functioning. Thus, to provide useful predictive tools for managing freshwater resources in the face of global change, we must improve our ability to build, apply, and communicate understanding of lake and reservoir ecosystem dynamics. To address this, I first built ecosystem understanding by conducting multiple whole-ecosystem surveys to quantify the spatial and temporal variability of biogeochemistry in two reservoirs over a year. We found that temporal heterogeneity was higher than spatial heterogeneity for most biogeochemical variables, with the stream-reservoir interface as a consistent hotspot of biogeochemical processing. Second, I applied ecosystem understanding by producing ecological forecasts of physical (water temperature), chemical (dissolved oxygen), and biological (chlorophyll-a) variables across three waterbodies using diverse modeling methods. I developed daily, weekly, and fortnightly forecasts of chlorophyll-a at two drinking water reservoirs using a Bayesian linear model, and found process uncertainty dominated total forecast uncertainty. Additionally, I produced forecasts of water temperature and dissolved oxygen in an oligotrophic lake using a hydrodynamic-ecosystem model and found that water temperature was more predictable than oxygen despite variable performance over depth and between years. Across these two forecasting studies, forecast skill relative to a null model varied among water quality metrics: water temperature forecasts outperformed the null model up to 11 days ahead, oxygen forecasts outperformed the null model up to 2 days ahead, and chlorophyll-a forecasts outperformed the null model up to 14 days ahead. Third, to communicate forecasts for decision-making, I developed an educational module for undergraduate ecology students which taught important concepts on visualization and decision science. Following completion of the module, students' ability to identify methods for uncertainty communication increased significantly, as well as their understanding of the benefits of ecological forecasting. Overall, my dissertation provides insight into how reservoirs function in global biogeochemical cycles, the predictability of multiple water quality variables, and deepens our understanding of how to communicate ecosystem science for improved management and protection of ecosystems. / Doctor of Philosophy / Human activities such as changing land use and climate have altered ecosystems around the world. As a result, many ecosystems no longer exhibit the same patterns from year to year as they have in the past, motivating the need for new tools to predict their future conditions. Among all ecosystems, freshwater lakes and reservoirs have been especially vulnerable due to human activities, threatening the critical services they provide to society, including drinking water and food, opportunities for recreational and cultural activities, and flood protection. Thus, because freshwater ecosystems are rapidly changing, forecasts of important water quality variables are needed. However, in order to use these forecasts effectively, a better understanding of how lakes and reservoirs change over space and time is needed, in addition to thoughtful communication of scientific findings for users. In my dissertation, I first addressed these needs by monitoring two reservoirs to examine how water quality varies over time and space. I found that water quality was more different over time (i.e., seasons) than over space (i.e., from one location to another within the reservoir). Second, I made ecological forecasts of water quality variables over space and time at different lakes. I forecasted phytoplankton levels in two drinking water reservoirs using a relatively simple model and water temperature and dissolved oxygen in a large clear-water lake using a more complex model. Comparing across forecasts in the three waterbodies, I found that my temperature forecasts provided valuable information above a baseline model up to 11 days ahead, oxygen forecasts up to 2 days ahead, and phytoplankton forecasts up to 14 days ahead. Third, because forecasts must be communicated effectively to be used as decision-making tools, I designed an educational module to teach forecast visualization and decision-making concepts. I found that students who completed the module better understood the benefits of forecasting for decision-making, and identified more ways to communicate forecasts after they completed the module. Overall, my dissertation shows the importance of measuring water quality over time and space to develop lake and reservoir forecasts, and demonstrates how forecast communication can help build students' understanding of the importance of forecasts for protecting and managing ecosystems.

Freshwater reservoirs store, process, and emit to the atmosphere large quantities of carbon (C). Despite the important role of reservoirs in the global carbon cycle, it remains unknown how human activities are altering their carbon cycling. Climate change and land use are resulting in lower dissolved oxygen (DO) concentrations in freshwater ecosystems, yet more frequent, powerful storms are occurring that temporarily increase DO availability. The net effect of these opposing forces results in anoxia (DO < 0.5 mg L-1) punctuated by short-term increases in DO. The availability of DO controls alternate redox reactions in freshwaters, thereby determining the rate and end products of organic C mineralization, which include two greenhouse gases, carbon dioxide (CO2) and methane (CH4). I performed ecosystem-level DO manipulations and evaluated how changing DO conditions affected redox reactions and the production and emission of CO2 and CH4. I also explored how the magnitude and drivers of CH4 emissions changed spatio-temporarily in a eutrophic reservoir using time series models. Finally, I used a coupled data-modeling approach to forecast future emissions of CH4 from the same reservoir. I found that the depletion of DO results in the rapid onset of alternate redox reactions in freshwater reservoirs for organic C mineralization and greater production of CH4. When the anoxia occurred in the water column (vs. at the sediments), diffusive CO2 and CH4 efflux phenology was affected, and resulted in degassing occurring during storms before fall turnover. I observed that the magnitude of CH4 emissions varied along a longitudinal gradient of a small reservoir and that the environmental drivers of ebullition and diffusion can change substantially both over space (within one hundred meters) and time (within a few weeks). Finally, I developed a forecasting workflow that successfully predicted future CH4 ebullition rates during one summer season. My research provides insight to how changing DO conditions will alter redox reactions in the water column and greenhouse gas emissions, as well as provides a new technique for improving future predictions of CH4 emissions from freshwater reservoirs. Althogether, this work improves our understanding of how freshwater lake and reservoir carbon cycling will change in the future. / Doctor of Philosophy / Freshwater reservoirs store a lot of carbon in their sediments and emit a lot of carbon as greenhouse gases (carbon dioxide and methane) to the atmosphere. However, climate change, land use, and water quality management can change the chemical reactions that are responsible for the production of carbon dioxide and methane, which could have substantial effects on the global carbon budget. Here, I manipulated the oxygen conditions of a freshwater reservoir and monitored the chemistry and greenhouse gas emissions in the experimental reservoir relative to an upstream reference reservoir. I then estimated the methane emissions from the reservoir to understand how the chemistry and greenhouse gas emissions in freshwater reservoirs may change in the future. I found that reservoir oxygen availability controls the magnitude and timing of the chemical reactions that produce carbon dioxide and methane, which in turn alters greenhouse gas emissions. Additionally, I developed models that showed how the magnitude and drivers of methane emissions changed within a small reservoir over time. Finally, I was able to predict the timing and magnitude of methane bubbling from the sediments. Altogether, this work provides evidence how climate change, land use change, and water quality management will affect future water chemistry and greenhouse gas emissions from reservoirs.