Ecohydraulic modelling of anabranching rivers

In this paper we provide the first quantitative evidence of the spatial complexity of habitat diversity across the flow regime for locally anabranching channels and their potential increased biodiversity value in comparison to managed single‐thread rivers. Ecohydraulic modelling is used to provide evidence for the potential ecological value of anabranching channels. Hydraulic habitat (biotopes) of an anabranched reach of the River Wear at Wolsingham, UK, is compared with an adjacent artificially straightened single‐thread reach downstream. Two‐dimensional hydraulic modelling was undertaken across the flow regime. Simulated depth and velocity data were used to calculate Froude number index, known to be closely associated with biotope type, allowing biotope maps to be produced for each flow simulation using published Froude number limits. The gross morphology of the anabranched reach appears to be controlling flow hydraulics, creating a complex and diverse biotope distribution at low and intermediate flows. This contrasts markedly with the near uniform biotope pattern modelled for the heavily modified single‐thread reach. As discharge increases the pattern of biotopes altered to reflect a generally higher energy system, interestingly however, a number of low energy biotopes were activated through the anabranched reach as new subchannels became inundated and this process creates valuable refugia for macroinvertebrates and fish, during times of flood. In contrast, these low energy areas were not seen in the straightened single‐thread reach. Model results suggest that anabranched channels have a vital role to play in regulating flood energy on river systems and in creating and maintaining hydraulic habitat diversity.

developing on a meandering or wandering morphological template through vegetative succession and subsequent stabilization, or through lateral extension, where channel widening and bar complex development have been initiated.
Although there have been numerous studies on the geomorphology of anabranching rivers (e.g. Carling, Jansen, & Meshkova, 2014;Kleinens, de Haas, Lavooi, & Makaske, 2012;Knighton & Nanson, 1993;Nanson & knighton, 1996;Makaske, 2001), few of these have documented processes in any detail (with the exception of Harwood & Brown, 1993), and none have examined their ecohydraulics over the flow regime (sensu Maddock, Harby, Kemp, & Wood, 2013), despite their high biodiversity value (Puckridge, Walker, & Costelloe, 2000), ecotone provision (Naiman, Décamps, Pastor, & Johnston, 1988), and their significance as potential refugia (Sedell, Reeves, Hauer, Stanford, & Hawkins, 1990). With recent drives towards renaturalization of floodplains for natural flood management and heightened interest in restoring rivers and floodplains (Dixon, Sear, Odoni, Sykes, & Lane, 2016), practitioners of river management require evidence to demonstrate the potential ecological value of anabranching systems at the reach-scale. In a companion paper Entwistle et al. (2018) use 2D hydraulic modelling to demonstrate stage-dependent contrasts in hydraulics between anabranching and managed single-thread channels; demonstrating how locally anabranched channels may be important for dissipating flood flow energy and reducing morphological destabilization further downstream. This study uses the 2D hydraulic modelling outputs from Entwistle et al. (2018) to explore stagedependent variations of instream habitat (biotopes) for the same anabranched reach of the River Wear, UK, with the aim of (a) quantifying spatial and temporal biotope availability and patterns over the flow regime and (b) highlight the ecological significance of anabranching channels through a comparison with an adjacent heavily modified single-thread reach situated downstream.
The most widely utilized flow variable for characterizing biotopes is the Froude number (Fr; Jowett, 1993;Wadeson, 1994;Padmore, 1998) that defines the ratio of the inertial to gravity forces in the flow: (1) where V is the local flow velocity, g is the gravitational acceleration, and d is the local flow depth.
At values below 1, gravitational forces are dominant and flow is subcritical; where Fr exceeds 1, internal forces dominate, and flow is supercritical. Despite describing flow based on the water column rather than at the bed, Fr has been shown to be associated with the distribution of benthic macroinvertebrates (Demars et al., 2012;Jowett, 2003;Hill et al., 2008;Reid & Thoms, 2008) and has been used as a hydraulic delimiter to support the existence and ecological relevance of biotopes (Wadeson & Rowntree, 1998;Padmore, 1998;Newson et al., 1998;Newson & Newson, 2000;Clifford et al., 2006;Harvey, Clifford, & Gurnell, 2008). It is clear from these studies that biotopes exist on a continuum across the range of Fr conditions experienced and distinct biotope types have been associated with a characteristic range of Fr values ( Figure 1).

| STUDY LOCATION
This study focused on a 1.5-km reach of the upper River Wear at Wolsingham, County Durham, situated at around 140-m A.O.D.
( Figure 2). The catchment drains impermeable Carboniferous Limestone, overlain by peat in the headwaters and till and alluvium in the middle reaches. The river has been impounded in its upper reaches by Burnhope reservoir, since 1937. The river valley at Wolsingham is dominated by two late glacial and three Holocene terraces (Moore, 1994). The river bed is composed of coarse gravels and cobbles

| METHODS
Tonina and Jorde's (2013) review of hydraulic modelling for ecohydraulic studies note that there is no general rule as to which modelling approach to apply and why when simulating river flow; however, they do state that 2D models are appropriate for scales ranging from geomorphic unit to reach and that 2D modelling is becoming a preferred approach for ecohydraulic studies concerned with developing a strong spatial understanding of fundamental hydraulic parameters such as depth, velocity, and shear stress. Tonina and Jorde (2013) also note that generally 2D hydraulic models are applied at the morphologic unit to reach scale (10-50 channel widths), which is appropriate to this study; however, longer multikilometre reach models have been published using advanced computer processing (Alabyan & Lebedeva, 2018) with progress being made in quantifying stream mesohabitats (Demarchi, Bizzi, & Piégay, 2016).
The model simulated depth-averaged hydraulics on a 1-m digital terrain model of the study reach using bare-earth LiDAR, sourced from the EA Geomatics group (Figure 2), a resolution reported as suitable for fish micro-habitat simulations (Pasternack & Senter, 2011

| Biotope mapping
CAESAR-Lisflood FP predictions of velocity and depth were computed at a 1-m 2 grid resolution across the study reach including both the heavily modified single-thread and anabranched channel network, and these were used to compute Fr maps using Equation (1). The Fr maps were then classified into biotopes using the delimeters shown in  Overall these statistics suggest a more complex biotope patch structure in the anabranched reach, which is most pronounced around

| DISCUSSION
High-resolution morphological data permit detailed 2D hydraulic modelling and mapping over the flow regime. When the hydraulic variables are converted into a meaningful habitat metric (e.g., Fr), it is possible to map spatial and temporal patterns of instream habitat across flow regime. This clearly has advantages over reach-average approaches of habitat classification (e.g., Lamouroux & Jowett, 2005) that fail to adequately predict hydraulic habitat distribution at a representative scale. The subsequent hydraulic outputs were converted to biotope maps for each channel type using Fr as a discriminator. The results of the biotope mapping provide a detailed habitat scale appraisal of conditions across the flow regime and the patterns of FIGURE 6 Area-weighted mean patch fractal dimension, defining the complexity of each patch (biotope) shape for the anabranched and single-thread reaches on the study river FIGURE 7 Patch (biotope) number for the anabranched and singlethread reaches on the study river biotope distribution and dominance offer insights into hydraulic habitat character rarely, if ever, measured in nature.
The gross morphology of the anabranched reach appears to be controlling the flow hydraulics, creating a complex and diverse biotope distribution across the site, most notably at low and intermediate flows (Figure 4). This contrasts markedly with the near uniform biotope pattern predicted for the heavily modified single-thread reach ( Figure 4). As flow discharge increased the pattern of hydraulic habitats alters to reflect a generally higher energy system; interestingly, however, a number of low energy biotope areas were activated through the anabranched reach as new subchannels were inundated.
These biotopes are likely to create valuable ecological refugia during times of flood, which are unavailable along the single-thread channel.
The anabranched reach exhibits a more diverse range of biotopes over the flow regime and hence will also show the most variability in flow structure (Harvey & Clifford, 2009). This heterogeneity in hydraulic habitat has been recognized in river system structure and habitat since the pioneering work of Hynes (1970) and Vannote, Minshall, Cummins, Sedell, and Cushing (1980) and reinforced by Rinaldi et al. (2015) with many species occupying different habitats at different stages of their life cycle (Hynes, 1970). Pringle et al.
(1988) described how environmental heterogeneity influences the dynamics of virtually all ecological processes within rivers. The greater morphological diversity displayed by the anabranched reach is also likely to increase the range of niches available for different species, and this has been shown by Shmida and Wilson (1985) to reduce the likelihood of competitive exclusion, thereby increasing the likelihood of a more diverse biotic community compared with the singlethread reach.

| Implications for fish species
Although little is known about the movements of different species into anabranching channels during floods, good knowledge exists concerning the velocity and depth preferences of certain species.
Few studies report Fr number preferences for freshwater fish species.
Numerous studies suggest low-velocity preferences for juvenile fish that are likely to be found in the anabranched channels as these become inundated with increasing discharge.  (Maitland, 2003). Although, Hardisty (1986) has recorded velocities of 0.08-0.10 m/s over FIGURE 8 Patch (biotope) size coefficient of variation for the anabranched and single-thread reaches on the study river, providing a measure of patch uniformity lamprey burrows. These lower velocity ranges are increasingly likely to be located in the anabranched channel network as this becomes inundated with increasing discharge.

| Habitat complexity
Change in patch complexity across the flow regime was highlighted for both reach types, although complexity decreased at higher flows in the single-thread reach. As such, competitive exclusion processes would be less in the anabranched reach as flow change induced disturbances open new habitat patches for colonization by inferior competitors before they can be completely excluded from the landscape by superior competitors (see early work by Hutchinson, 1951).
The value of biotope patch complexity discussed above may be contrasted with the work of Naiman et al. (1988), who demonstrated that some species prefer large unbroken habitat patches to thrive and hence may favour the biotope character shown in the singlethread reach. They contrasted this with other species, which were found to exploit the interface between patches, as a result, river reaches displaying biotope assemblages and patterns that are too patchy (the anabranched reach), or insufficiently patchy (the singlethread reach), may be detrimental to certain species. However, Downes, Lake, Schreiber, and Glaister (1998) suggest that a patchier watercourse configuration displaying a high diversity of habitats at large, intermediate, and local spatial scales supports increased abundance and species richness of benthic invertebrates.
Other studies have considered both the configuration and persistence of hydraulic habitat in influencing biotic diversity and resilience. Townsend (1989) emphasized the important roles of disturbance refugia, with the value of patches as refugia shown to be dependent upon their size and arrangement (Lancaster & Hildrew, 1993), and frequency of disturbance (Silver, Wooster, & Palmer, 2004), which impacts on their recolonization potential (Gjerløv, Hildrew, & Jones, 2003;Matthaei et al., 2004). Again in this study the spatial and temporal character of the anabranched channel type (complex, diverse, and quite resilient with refugia patches present) appears to offer greater potential for species diversity over the more uniform spatial and temporal biotope assemblage modelled for the single-thread reach. This uniformity of patch type has also been shown to hinder the formation of refugia by conveying disturbances across the network (Hanski, 1999). The impact of such a conclusion is strengthened by studies that demonstrate the use of multiple habitats by many species which move from one biotope to another seeking flood refuge associated with the presence of slower moving water and more stable substrates (Rempel, Richardson, & Healey, 1999). Highly connected patches, such as those seen in the single thread reach, may thus lead to a reduced range of species in the river.
Recolonization following flood disturbance has been shown to occur in larger stable patches (Holyoak, Leibold, & Holt, 2005), a condition more prevalent in the single-thread reach, and also from adjacent patch populations (Roughgarden, Gaines, & Pacala, 1987), such as those found in the anabranched reach. This feature is particularly evident where some biotope patches remain during a flood event forming undisturbed locations to recolonize disturbed areas and promoting resilience (Labbe & Fausch, 2000). Persistance has been shown to be highest for the anabranched reach at Wolsingham; however, larger floods do see a change to higher energy hydraulic habitats.

| CONCLUSIONS
Anabranched channels provide a morphological template for the development of complex, diverse and resilient biotopes. Anabranched channels provide refugia during high flows and are likely to be both more biologically diverse and ecologically resilient compared with single-thread reaches, although it is acknowledged that certain species are well adapted to the more uniform but temporally less stable environment present along the single-thread reach. For river managers, river rehabilitation back towards an anabranching planform, where appropriate, may provide a means of protecting species sensitive to increases in flood magnitude, resulting from climate change or urbanization.
We argue that anabranched reaches, increasingly seen on unmanaged temperate rivers , provide a more diverse range of hydraulic conditions in both time and space, which, as a consequence, promotes greater ecological diversity. Where possible, river managers should encourage renaturalization processes leading to the development of such systems and this could be as simple as promoting naturalization through vegetative succession. Fuller, Passmore, Heritage, Large, Milan, and Brewer (2002) noted that prevention of grazing in riparian zones and on bars across some multithread wandering gravel-bed channels found in United Kingdom allowed vegetation succession to stabilize bars promoting transition towards an anabranching system.
On a practical level, the availability of high-quality morphological data from LiDAR and the ease with which a 2D flow model may be constructed results in high-quality hydraulic outputs that may be used to provide spatial and temporal habitat information, linked to river management targets (Logan, McDonald, Nelson, Kinzel, & Barton, 2011). This makes this an excellent tool for use in predicting changes to reach hydromorphology, a process that is critical to achieving the pan-European Water Framework Directive objectives. In addition, modelling results can help to restore rivers in a sustainable and ecologically meaningful way and provide a usable measure to monitor instream habitat health and issues triggered as a result of human intervention.