Integrative modeling tools for the assessment of fish habitat and population dynamics in rivers

In recent times, habitat models at the meso-scale are becoming increasingly accepted techniques to quantify the impact of hydro-morphological pressures on rivers, to help guide water resource planning through the design of environmental flows, or renaturalization of rivers through restoration. Despite their potential, broader applicability of meso-scale habitat models is limited by the difficulty to conduct field-based mesohabitat mapping in large streams or at high flows, at which wadeability decreases. Furthermore, construction of reliable habitat-streamflow rating curves can be a highly time consuming process, as it requires mapping over a wide range of streamflow values. An important gap in the use of habitat models for ecological river management is also represented by the poor understanding of biological communities' response to habitat improvement. The present PhD thesis aims to address some of these fundamental gaps in habitat modeling and its links with biological communities' response through an integrative set of different modeling tools. The scene is set by a non specialist preface (Chapter 1), followed by a detailed analysis of the scientific state-of-art that highlights the fundamental research questions of the thesis (Chapter 2). Chapter 3 and 4 address some of the key challenges in the integration of 2D hydraulic modeling and habitat modeling at the meso-scale, namely the reproduction of an accurate description of the meso-scale distributions of water depth and velocities in gravel bed-rivers, and the spatial mapping of hydro-morphological units (HMUs). Two study cases were selected in the North-East Italian alpine region, in the province of South Tyrol, which were chosen for their contrasting characteristics, in terms of channel size, hydro-morphology and river type. The study reach in the Mareta River was recently restored to a braided morphology, while in the Aurino reach a meander bend was restored to improve fish habitats. In Chapter 3, three different commonly used approaches were compared in terms of their ability to accurately describe the meso-scale hydraulics of a river reach of the Mareta River, and of their effects on the description of meso-habitat suitabilities for two fish species, the marble trout (Salmo marmoratus) and the European bullhead (Cottus gobio). Two approaches were based on 2D hydraulic modeling, which were constructed on computational meshes with varying resolution and quality: (1) high resolution meshes derived from topographical data obtained from Airborne LiDAR Bathymetry; and (2) a mesh extrapolated from topographical cross-sectional profiles. The third approach used for the comparison was based on in-stream surveys. Results from the comparison suggest that decreasing resolutions and mesh quality negatively affects the accuracy of the results, with the largest residual differences found for the cross-sectional based modeling and the surveys. Amongst the analyzed HMU types, the highest sensitivities to the choice of approach were recorded for backwaters, followed by morphologically complex units such as pools, steps and rapids. The least sensitive units were riffles and glides. While similar effects were also observed for the estimated habitat suitabilities, errors in suitability estimates were minimized when deriving habitat-streamflow rating curves at the reach scale. In Chapter 4, a novel approach for the delineation of habitats at the meso-scale, based on the outputs of two-dimensional hydraulic modeling was developed. The approach uses a four-step strategy: 1) hydraulic data at the micro-scale is classified into groups by means of an unsupervised clustering algorithm (k-means); 2) homogeneous flow patches are identified and polygonized in the channel; 3) a region growing process ensures that all hydraulic patches reach a scale that is equivalent to the river meso-scale; and 4) an optimal segmentation is selected by minimizing a Global Score. Applications of the model were tested on the two case studies of the Mareta and Aurino rivers, and habitat suitability was assessed for three fish species: marble trout; grayling (Thymallus thymallus); European bullhead. A high level of agreement was found when comparing model- and survey-based habitat suitability estimates. Compared to existing approaches, the developed methodology is unsupervised, and does not require a river- and site-specific calibration. While habitat models are largely used in the context of river rehabilitation, in recent times the assumption that optimizing habitat quality and quantity also improves ecological health of rivers has been questioned. A need for predictive process-based modeling has been recognized, which is able to quantitatively assess how spatial and temporal habitat dynamics affects ecologically relevant measures, such as recruitment and population potential. Within this context, a tool in the form of an age-structured population model was developed (Chapter 5), that is based on the quantification of the suitable habitat area in a river reach, and can be used for the assessment and comparison of river restoration scenarios. The model uses a Bayesian approach, which was parameterized for the common Barbel (Barbus barbus), a cyprinid litophilic fish, and was used to study the effects of habitat limitations on the population dynamics. By testing various scenarios of habitat availabilities for the common Barbel, it was hypothesized that improvements in the fish stock can be reached only when a well specified ratio of spawning habitat to fry habitat exists, and even substantial improvements of only either spawning or juvenile habitats will result in little or no increase of abundance. A synthesis of the key findings of the thesis is finally presented in Chapter 6, together with their broader ecohydraulic and management implications.