Doctor of Philosophy / Department of Biology / Keith B. Gido / Habitat fragmentation and loss threaten global biodiversity, but organism responses to changing habitat availability are mediated by structural properties of their habitats. In particular, organisms inhabiting dendritic landscapes with hierarchically arranged branches of habitat tend to have limited access to some patches even in the absence of fragmentation. Consequently, organisms inhabiting dendritic landscapes such as streams respond strongly to fragmentation. Using a combination of meta-analysis, field observations, and ecological network modeling I show that stream fishes respond to fragmentation in predictable ways. First, I addressed how dams and stream dewatering have created a mosaic of large river fragments throughout the Great Plains. Using a geographic information system and literature accounts of population status (i.e., stable, declining, extirpated) for eight “pelagic-spawning” fishes, I found stream fragment length predicted population status (ANOVA, F2,21 = 30.14, P < 0.01) and explained 71% of reported extirpations. In a second study, I applied a new measure of habitat connectivity (the Dendritic Connectivity Index; DCI) to 12 stream networks in Kansas to test the DCI as a predictor of fish response to fragmentation by road crossings. Results indicated fish communities in stream segments isolated by road crossings had reduced species richness (alpha diversity) and greater dissimilarity (beta diversity) to segments that maintained connectivity with the network, and the DCI predicted patterns in community similarity among networks (n = 12; F1,10 = 19.05, r2 = 0.66, P < 0.01). Finally, I modeled fish distributions in theoretical riverscapes to test for mechanistic linkages between fragmentation and local extirpations. Results suggested the number of small fragments predicted declines in patch occupancy, and the magnitude of change in occupancy varied with dispersal ability (“high” dispersers responded more strongly than “low” dispersers). Taken together, these works show context-dependencies in fish responses to fragmentation, but a unifying theme is that small fragments contribute to attenuated biodiversity. Moreover, the predictable manner in which stream fish react to fragmentation will aid in biodiversity conservation by revealing potential responses to future scenarios regarding changes to habitat connectivity.

The health of freshwater aquatic systems, particularly stream networks, is mainly influenced by water temperature, which controls biological processes and influences species distributions and aquatic biodiversity. Thermal regimes of rivers are likely to change in the future, due to climate change and other anthropogenic impacts, and our ability to predict stream temperatures will be critical in understanding distribution shifts of aquatic biota. Spatial statistical network models take into account spatial relationships but have drawbacks, including high computation times and data pre-processing requirements. Machine learning techniques and generalized additive models (GAM) are promising alternatives to the SSN model. Two machine learning methods, gradient boosting machines (GBM) and Random Forests (RF), are computationally efficient and can automatically model complex data structures. However, a study comparing the predictive accuracy among a variety of widely-used statistical modeling techniques has not yet been conducted. My objectives for this study were to 1) compare the accuracy among linear models (LM), SSN, GAM, RF, and GBM in predicting stream temperature over two stream networks and 2) provide guidelines in choosing a prediction method for practitioners and ecologists. Stream temperature prediction accuracies were compared with the test-set root mean square error (RMSE) for all methods. For the actual data, SSN had the highest predictive accuracy overall, which was followed closely by GBM and GAM. LM had the poorest performance overall. This study shows that although SSN appears to be the most accurate method for stream temperature prediction, machine learning methods and GAM may be suitable alternatives.

Species diversity and coexistence have long been central foci of ecology, but field studies are often limited to describing diversity patterns, while theory frequently ignores environmental variation. Scale transition theory is an ideal framework in which to study species diversity, as it explicitly accounts for this environmental variability and allows for the quantification of coexistence mechanisms. Each coexistence mechanism arises from specific types of biotic and abiotic interactions. Moreover, mechanism magnitudes provide information about how these interactions contribute to coexistence. By studying how the natural history of a community determines these biotic and abiotic interactions, insight can be gained into how that natural history influences coexistence. Environmental variation is a central hypothesis for the maintenance of diversity in both desert annual plants and streams. This dissertation is broadly interested in the way differences in the environmental responses of species interact with the structure of the environmental conditions to affect coexistence. I use scale transition theory to develop theoretical understanding of how life history and environmental structure in these communities influence coexistence mechanisms and diversity. In desert annual plants, the focus is on the environmental response itself: how germination depends on environmental conditions. I analyze how this life history interacts with variation in the environment to affect coexistence. The germination responses of desert annual plants to an unstudied type of environmental variation, duration of soil moisture after rainfall, generate species-specific but highly structured patterns of germination variation. Although this germination variation is one-dimensional, the nonlinearities that arise due to germination biology generate sufficient germination variation to promote coexistence by the temporal storage effect. In stream communities, I examine how the physical structure of stream environments affects coexistence given that species’ performance is environmentally dependent. This dissertation demonstrates that patterns of diversity along the stream are related to the strength of coexistence. The downstream drift of organisms has relatively minor effects on coexistence despite asymmetric shifts in the distribution of organism in the stream. This study identifies conditions that eliminate the effects of the branched structure of stream networks on coexistence. Branching has no effect on community dynamics if (a) tributaries have identical environmental conditions, (b) habitat size increases additively at confluences, and (c) demographic stochasticity is unimportant. Any effects of branching on coexistence caused by violating the environmental condition are asymptotically eliminated as streams increase in size. These studies provide a theoretical, mechanistic foundation for the study of stream communities that addresses environmental and life history factors long recognized as important by empirical stream ecologists.

In February of 2014 NASA has launched the core observatory of The Global Precipitation Measurement Mission (GPM). Since then, the mission has been providing a wealth of observation data collected by the core observatory along with other satellites belonging to the mission space constellation. One of the most important data products that GPM provides is the Level 4 (L4) rainfall data product called Integrated Multi-satellitE Retrievals for GPM (IMERG). IMERG is constructed using the raw data collected by the Microwave (MW) sensors on board the constellation satellites along with the Infrared (IR) sensors on board geostationary satellites and the advance Dual-frequency Precipitation Radar (DPR) on board the GPM core satellite. The IMERG product is available globally for all interested researchers to use. In this dissertation, I focus on the applicability of IMERG in hydrologic applications, and specifically in flood peak modeling. In order to conduct a comprehensive evaluation of IMERG that is oriented towards hydrologic modeling. I have explored multiple hydrologic models which can be used to produce stream flow estimates using IMERG without the need of parameter calibration based on the model’s inputs. The calibration free capability is essential since model parameter calibration obscures the effect of the errors associated with the rainfall input on the estimated discharges, which in turn will limit our understanding about the distribution of the errors in IMERG over space and time. The two hydrologic models we used in this study are both physically based distributed models and were setup over the domain of the state of Iowa which is located in the United States’ Midwest. I also explored the performance of one of the hydrologic models’ component, which is the runoff-routing component, in order to estimate an additional portion of the errors in the discharge estimates that is not attributed to the model’s input but rather to the hydrologic model itself. A significant portion of my dissertation is concerned with identifying and using accurate methods to evaluate both IMERG and the hydrologic models’ outputs in a hydrologic context that is useful for flood modeling. Several studies have evaluated other satellite rainfall products using methods that vary in complexity. Some studies used the simplest methods of evaluation, such as, mean aerial differences and standard deviation of the differences (additive or multiplicative) compared to a benchmark rainfall product. This is done without taking the spatial dependency of the errors in space into consideration. Other studies modeled the spatial dependency (correlation) between the errors in the rainfall product, however, using Euclidean distance based approaches that do not account for the hydrologic basins’ shape and size. Nevertheless, it is important to realize that hydrologic models will eventually aggregate the rainfall values, along with the errors associated with them, through a stream network that is dichotomous in nature and does not comply with Euclidean distance. Thus, we employed a stream based evaluation framework, called the Spatial Stream Network (SSN) approaches, to characterize the errors in IMERG taking into account the stream distances and the stream connectivity information between evaluation sites. Although previously used in applications such as modeling water temperatures and pollutant transport, to the best of my knowledge this approach has not been used in rainfall product evaluation before this study. The SSN analysis of IMERG allowed me to answer the question, “What is the proper basin scale which is capable of filtering out the correlated errors in IMERG by accumulating the rainfall values through the stream network?” Finally, in order to add value to the current methods of evaluating model simulated stream flows. I proposed a time based evaluation that is capable of detecting peaks in both the observed and simulated flows and estimating the lag time of the simulated peaks. Typically, previous studies have used simple skill scores such as Root Mean Squared Errors (RMSE), correlation coefficient, and Nash-Sutcliff Efficiency (NSE) to evaluate hydrograph performance as a whole, or the difference in time to peak which involves primitive peak detection method (e.g., a moving or a defined time window). In this dissertation I propose a Continuous Wavelet Transform (CWT) based method to evaluate the peak times and shapes produced by the hydrologic model. The method is based on filtering the frequencies in the hydrograph by treating it as a signal and detecting sharp features in both the observed and time series and the phase difference between them. We also emphasized on the importance of the choice of wavelet shape used in the evaluation, and how different wavelet shapes can affect the inference about the time series.

Through distributed stream network routing, it has quantitatively been shown that the relationship between flow travel time and discharge varies strongly nonlinearly with stream stage and with catchment-specific properties. Physically derived distributions of water travel times through a stream network were successfully used to parameterise the streamflow response function of a compartmental hydrological model. Predictions were found to improve compared to conventional statistically based parameterisation schemes, for most of the modelled scenarios, particularly for peakflow conditions. A Fourier spectral analysis of 55-110 years of daily discharge time series from 79 unregulated catchments in Sweden revealed that the discharge power spectral slope has gradually increased over time, with significant increases for 58 catchments. The results indicated that the catchment scaling function power spectrum had steepened in most of the catchments for which historical precipitation series were available. These results suggest that (local) land-use changes within the catchments may affect the discharge power spectra more significantly than changes in precipitation (climate change). A case study from an agriculturally intense catchment using historical (from the 1880s) and modern stream network maps revealed that the average stream network flow distance as well as average water levels were substantially diminished over the past century, while average bottom slopes increased. The study verifies the hypothesis that anthropogenic changes (determined through scenario modelling using a 1D distributed routing model) of stream network properties can have a substantial influence on the travel times through the stream networks and thus on the discharge hydrographs. The findings stress the need for a more hydrodynamically based approach to adequately describe the variation of streamflow response, especially for predictions of higher discharges. An increased physical basis of response functions can be beneficial in improving discharge predictions during conditions in which conventional parameterisation based on historical flow patterns may not be possible - for example, for extreme peak flows and during periods of nonstationary conditions, such as during periods of climate and/or land use change. / <p>QC 20150903</p>

Accurate models to reliably predict sediment and chemical transport in watershed water systems enhance the ability of environmental scientists, engineers and decision makers to analyze the impact of contamination problems and to evaluate the efficacy of alternative remediation techniques and management strategies prior to incurring expense in the field. This dissertation presents the conceptual and mathematical development of a general numerical model simulating (1) sediment and reactive chemical transport in river/stream networks of watershed systems; (2) sediment and reactive chemical transport in overland shallow water of watershed systems; and (3) reactive chemical transport in three-dimensional subsurface systems. Through the decomposition of the system of species transport equations via Gauss-Jordan column reduction of the reaction network, fast reactions and slow reactions are decoupled, which enables robust numerical integrations. Species reactive transport equations are transformed into two sets: nonlinear algebraic equations representing equilibrium reactions and transport equations of kinetic-variables in terms of kinetically controlled reaction rates. As a result, the model uses kinetic-variables instead of biogeochemical species as primary dependent variables, which reduces the number of transport equations and simplifies reaction terms in these equations. For each time step, we first solve the advective-dispersive transport of kinetic-variables. We then solve the reactive chemical system node by node to yield concentrations of all species. In order to obtain accurate, efficient and robust computations, five numerical options are provided to solve the advective-dispersive transport equations; and three coupling strategies are given to deal with the reactive chemistry. Verification examples are compared with analytical solutions to demonstrate the numerical accuracy of the code and to emphasize the need of implementing various numerical options and coupling strategies to deal with different types of problems for different application circumstances. Validation examples are presented to evaluate the ability of the model to replicate behavior observed in real systems. Hypothetical examples with complex reaction networks are employed to demonstrate the design capability of the model to handle field-scale problems involving both kinetic and equilibrium reactions. The deficiency of current practices in the water quality modeling is discussed and potential improvements over current practices using this model are addressed.