Author(s): P. S. Kemp; C. Katopodis
Linked Author(s): Christos Katopodis, Paul Kemp
Keywords: No Keywords
Abstract: In an article published in Fisheries entitled, “Why it is time to put PHABSIM out to pasture”, Steven Railsback (2016) criticizes the use of a tool widely adopted to help assign environmental flows and presents a case for why it is not fit for purpose in the current era of environmental modelling and management. Those that are familiar with PHABSIM (Physical Habitat Simulation) will be aware that it has been used by regulatory agencies for decades, particularly in the United States, to manage flows in modified fluvial systems, and today continues to be adopted in many other regions. The title of this editorial, combining and extending archaic nautical phrases, is based on an assumption that Railsback (2016) is correct, and that the field of environmental flows is in an uncertain position and at risk of becoming, or has for a long time been, lost in the face of troublesome and difficult times ahead. Here we summarize some of the concerns raised by Railsback (2016) and others and highlight potential alternative approaches that may represent a positive change in direction. Our vision is that the Journal of Ecohydraulics will play an important role in facilitating the debate in this area and publishing the evidence needed to better develop policy and navigate a route through the challenges of environmental flow management. Models such as PHABSIM quantify a measure of usable area of physical habitat available to a specific life stage of a target aquatic species, such as a trout, for a selected flow (Nestler et al. 1989; Johnson and Law 1995). These models are employed within an overarching conceptual framework referred to as the Instream Flow Incremental Methodology (IFIM; Bovee 1982), first developed by the US Fish and Wildlife Service in the 1970s. PHABSIM is composed of two modelling elements. The first quantifies hydraulic characteristics of a stream reach based on surveys, e.g. of depth and water velocities measured in the field. The second provides ecological input, using information on the preference for physical habitat variables, usually depth, velocity and substrate, for the species of interest. Railsback (2016) describes several weaknesses and failings of the approach and challenges some of the basic assumptions that underpin it, arguing they violate important conventions of modern modelling. Criticisms of PHABSIM are not new, however, with those first published relatively shortly after its introduction (e.g. Scott and Shirvell 1987) and several thereafter (e.g. Railsback 1999; Armstrong and Nislow 2012). “Incremental” in this context appears an apt description of a process of proposition, criticism, refinement and defence, followed by further criticism. Ultimately, an inability to fend off the fundamental concerns repeatedly raised appears to indicate that an alternative approach has long been overdue. Why has PHABSIM been so often criticized? There is no single problem that can be easily fixed; concerns raised range from those related to fundamental assumptions that underpin the physical and ecological models through to basic criticisms of transferability of indices and of how the data are collected. So let us focus on the big ones. Railsback (2016) highlights the lack of accommodation of discharge variability, while inclusion of dynamic regimes is now considered an essential component of modern flow modelling. Indeed, a bias towards considerations of ecological response as a function of flow magnitude, rather than dynamics, is a common problem in the environmental flows literature more generally. In a review of published ecological data used to assess ecosystem response to flow alteration, three-quarters of studies related to magnitude as opposed to variability (Poff and Zimmerman 2010). Moreover, a review by Costa et al. (2017) suggests that fish do respond to natural or modified dynamic flow regimes, and variability in flow can be a stressor, particularly in extreme flow conditions (e.g. floods, droughts, dewatering and pulsed flows). However, there are strong limitations to scientifically establishing thresholds which may be inherently stressful rather than fish habituation and compensation. Railsback (2016) also highlights the use of inappropriate scales, driven more by hydraulic modelling convenience than a justification based on an understanding of the ecology of the target species (note that the focus is on target species rather than communities may itself be criticized). Decisions on what scale to model at should depend on the appropriateness to the species and the purpose of the model. Furthermore, PHABSIM tends to combine the results of hydraulic modelling conducted at one scale with habitat preference data obtained at a different, usually considerably finer, resolution; an observation Railsback (2016) highlights as a fundamental modelling mistake. Further, Railsback (2016) criticizes PHABSIM’s production of outputs that lack clear meaning and are impossible to validate, with an inability to generate testable predictions of responses at a population level, echoing concerns previously raised some time ago by others (e.g. Scott and Shirvell 1987; Gore and Hamilton 1996). One of the most frequent criticisms levelled against PHABSIM, and similar models, is the lack of biological realism, particularly with respect to assumptions related to habitat suitability based on inferences of preference. Preference, calculated as the quotient of a physical habitat variable selected and its availability within the area modelled using fine-scale observation data of fish position (e.g. snout velocity obtained by snorkelling surveys, Strakosh et al. 2003), is often presented as curves. Here the underlying assumption is that the target species exhibit ideal free distribution (Fretwell and Lucas 1970), i.e. they have ideal knowledge of the environment in which they live based on holistic sampling, and are free to select optimal habitat when they identify it. This is unlikely to be the case for many species. Consider a stream dwelling juvenile salmonid; a dominant individual defends a territory, possibly representing optimal habitat, thus preventing fish access to sample and use space that it occupies (Hakoyama and Iguchi 2001). As a result, more subordinate fish are likely displaced to lower quality habitat in greater numbers than dominant fish. Even if the researcher could account for such despotic behaviour, an assumption that habitat use bestows advantages in terms of fitness is difficult to validate. Thus, frequency of occurrence or fish density is not a good predictor of habitat quality. A further problematic assumption is that preference remains constant, and that as flow changes the organism will track its preferred habitat, ignoring the possibility that some life stages for some species, including juvenile Atlantic salmon (Salmo salar), may exhibit high site fidelity (Kemp et al. 2003). Indeed, concerns have long been raised over the use of overly simplistic descriptors of habitat that considers only physical variables, such as depth, velocity and substrate, while ignoring density-dependent effects that influence behaviour (e.g. Heggenes 1996), and other key drivers, such as demographic rates (Lancaster and Downes 2009). On the other hand, some engineers and scientists, as well as habitat modelling practitioners, do recognize such scientific challenges and assumptions with PHABSIM, including limited consideration of river morphodynamics and ice regimes. It has been suggested that, although rather challenging, new paradigms need to emerge from scientific advances to validate more realistic estimates of habitat quantity and quality, better understanding of biological and ecological mechanisms, as well as biota interactions and movements, including population dynamics and a multiplicity of inter- and intra-specific interactions (Katopodis 2005). In recommending environmental flow regimes, several factors other than depth, velocity and substrate are considered, estimates are supplemented with historical data and observations (Annear et al. 2004), and although resource limitations rarely allow it, some base more emphasis on scientifically based validation studies and monitoring (e.g. Hardy et al. 2006; Souchon et al. 2008). Nevertheless, if we assume Railsback (2016) and others are right, and a proportion of the environmental flows community should recant their belief in PHABSIM, then where do we go next? This is not an easy question to answer, as many apparent alternatives have similar assumptions at their heart, and can be considered “PHABSIM-like” in their essence. From a regulatory perspective, there remains a need for a methodology to dispassionately inform the prescription of appropriate environmental flows and management of water resources, before taking into account the perspectives and requirements of different stakeholders and the potential for conflict. Indeed, for some it is easier to continue to rely on a tool, even with understanding that it is not perfect, than to wait for a validated alternative. We suggest that from the perspective of the research community we should resist such an impulse. An inappropriate tool will always be a bad tool and increase the probability of poor decision-making, and, in light of the threats and rates of degradation that freshwater ecosystems currently face (Vörösmarty et al. 2010), we do not have the luxury to be able to ignore the evidence presented. Instead, we may conclude that current environmental flow practices are based on traditional approaches that represent the legacy established by the use of earlier methods, and that we now need to take a step back and re-evaluate. But if we do, what are the alternatives? For some, the answer is to combine the traditional hydraulic–habitat models, such as PHABSIM discussed, that poorly accommodate temporal variation in flow, with hydrological approaches that do a better job, but are poorly suited to spatial analysis (Poff et al. 2010; Shenton et al. 2012). Others propose a move towards a more biologically relevant energetics performance-based approach by relating discharge to growth and survival (Armstrong and Nislow 2012). Railsback (2016) suggests that a broader ecological perspective is needed in which the fitness of the individual should be central to the modelling process, and that PHABSIM might be replaced by more powerful modern habitat selection models in some instances, or individual-based models if the expertise is available to develop them. It is clear that there continues to be much debate around future directions in this area, and a continued lack of consensus; this is not negative. It is an important part of a process of scientific endeavour in which researchers strive to improve their understanding. How might the Journal of Ecohydraulics help? We propose to provide a forum for advancing this subject area within an atmosphere of honest and constructive review. In so doing, we are providing this editorial with the intention to provoke debate and generate discussion. We invite members of the environmental flows community to respond by submitting articles related to this topic.