Overcoming persistent challenges in putting environmental flow policy into practice: a systematic review and bibliometric analysis

The implementation of environmental flows (e-flows) aims to reduce the negative impacts of hydrological alteration on freshwater ecosystems. Despite the growing attention to the importance of e-flows since the 1970s, actual implementation has lagged. Therefore, we explore the limitations in e-flows implementation, their systemic reasons, and solutions. We conducted a systematic review and a bibliometric analysis to identify peer-reviewed articles published on the topic of e-flows implementation research in the last two decades, resulting in 68 research and review papers. Co-occurrence of terms, and geographic and temporal trends were analyzed to identify the gaps in environmental water management and propose recommendations to address limitations on e-flows implementation. We identify the underlying causes and potential solutions to such challenges in environmental water management. The limitations to e-flow implementation identified were categorized into 21 classes. The most recognized limitation was the competing priorities of human uses of water (n = 29). Many secondary limitations, generally co-occurring in co-causation, were identified as limiting factors, especially for implementing more nuanced and sophisticated e-flows. The lack of adequate hydrological data (n = 24) and ecological data (n = 28) were among the most mentioned, and ultimately lead to difficulties in starting or continuing monitoring/adaptive management (n = 28) efforts. The lack of resource/capacity (n = 21), experimentation (n = 19), regulatory enforcement (n = 17), and differing authorities involved (n = 18) were also recurrent problems, driven by the deficiencies in the relative importance given to e-flows when facing other human priorities. In order to provide a clearer path for successful e-flow implementation, system mapping can be used as a starting point and general-purpose resource for understanding the sociohydrological problems, interactions, and inherited complexity of river systems. Secondly, we recommend a system analysis approach to address competing demands, especially with the use of coupled water-energy modeling tools to support decision-making when hydropower generation is involved. Such approaches can better assess the complex interactions among the hydrologic, ecological, socioeconomic, and engineering dimensions of water resource systems and their effective management. Lastly, given the complexities in environmental water allocation, implementation requires both scientific rigor and proven utility. Consequently, and where possible, we recommend a move from simplistic flow allocations to a more holistic approach informed by hydroecological principles. To ease conflicts between competing water demands, water managers can realize more ‘pop per drop’ by supporting key components of a flow regime that include functional attributes and processes that enhance biogeochemical cycling, structural habitat formation, and ecosystem maintenance.


Introduction
The concept of environmental flows (e-flows) emerged from the need to recognize the needs of specific species, such as economically important salmonid fisheries (Tharme 2003), with infrequent consideration of the water needs of entire river ecosystems and the people who directly depend on them (Matthews et al 2014). E-flows then evolved to wholecommunity and ecosystem perspectives to mitigate the undesirable hydrological impacts of dams and water diversions (Poff and Matthews 2013), and protect or restore the benefits of naturally flowing rivers, in regulated systems (Owusu et al 2021). In response to well documented global degradation of freshwater ecosystems, several efforts have emerged over the previous four decades to document and support e-flow development and implementation (Arthington et al 2018b). At the policy level, several instruments have been adopted, such as the Brisbane Declaration (2007) which formalized the e-flow paradigm as 'the quantity, timing, and quality of freshwater flows and levels necessary to sustain aquatic ecosystems which, in turn, support human cultures, economies, sustainable livelihoods, and well-being' . This approach was reiterated in 2018 with additional guidelines for practitioners of different regions and disciplines, intending to set a common vision and direction for e-flows globally (Arthington et al 2018a).
The actual implementation of e-flows has remained limited to date despite the abundance of theories and concepts for e-flows (Owusu et al 2022). Several factors have contributed to this disparity in implementation, including a lack of research on e-flow implementation and trade-off analysis with other uses (Pahl-Wostl et al 2013), uncertainty over method choice (Opperman et al 2018), and physical and policy constraints that limit flow releases from dams (Aldous et al 2011, Pittock and Hartmann 2011). Delays in implementation can also occur due to differing technical term definitions or other incongruences among implementing and regulatory authorities, thus leading to misunderstanding among stakeholders and managers (Pahl-Wostl et al 2013).
E-flow implementation efforts can also miss the systematic and integrated conceptualization of a river system as one complex socio-hydrological system (Madani and Shafiee-Jood 2020). This can lead to fragmentation of effort, failure to take advantage of local adaptive management learnings, and poor public understanding of why decisions are being made and where responsibility lies . Indeed, there is a pressing need for a more committed effort to protect and restore freshwater ecosystems as resilient human-water systems through the implementation and adaptation of e-flows (Arthington et al 2018a). Poff et al 1997 introduced a paradigm shift in environmental water management when presenting the natural flow concept, in which the natural streamflow dynamics supports native habitats and species assemblages. Yet, the role of the natural flow regime in creating the spatiotemporal variation in biogeographic patterns and processes has been neglected (Meitzen et al 2013). A disregard for the natural system complexity of river ecosystems persists despite substantial progress in understanding how natural flow variation maintains river health (Sofi et al 2020). Given the complexities in water allocation for e-flows, implementation requires both scientific rigor and proven utility.
Therefore, a crucial priority for freshwater conservation is to accelerate the implementation of effective e-flows and their potential benefits such as improvements in water quality, critical habitat maintenance, and hydrologic connectivity (Tickner et al 2020). Consequently, our review addresses the gaps in e-flow implementation to help guide water management decisions and better meet ecosystem needs while satisfying human demands (Viers 2017). Within this context, the goals of this paper are to (1) explore the limitations in e-flow implementation by emphasizing current and future challenges of environmental water management and their implications; (2) identify systemic reasons for the lack of implementation and solutions for overcoming them; and (3) present a conceptual framework as the basis for decision-making to help managers and stakeholders select the most appropriate methods based on their resource availability, physical and legal constraints and objectives. We conclude by identifying existing data and conceptual gaps and discussing important recommendations for the effective implementation of e-flows.

Background
The objective of setting e-flows is (or should be) to modify water abstraction from water bodies or flow releases from water infrastructure to restore natural or normative flow regimes that benefit river and riparian ecosystems downstream (Poff and Matthews 2013). The 'natural flow regime' determines the geomorphic processes that shape river channels, floodplains, and other riverine habitats, consequently governing the ecological processes and the composition of flora and fauna (Poff et al 1997, Taniguchi-Quan et al 2022. The maintenance of natural flow patterns (i.e. magnitude, frequency, duration, timing, predictability, and rate of change) allows lateral and longitudinal habitat connectivity, a major determinant of biotic diversity, and controls invasive species while triggering life-history strategies of native species that are adapted to the natural flow regime (Bunn and Arthington 2002, Poff et al 2007, Poff and Matthews 2013, Koster and Crook 2017. Conversely, disruptions to natural flow regimes can have long-ranging impacts on adapted species. For instance, the Balbina Dam in the Brazilian Amazon removed the periods of extended low flow that allowed floodplain forests to establish, causing the death of tree species in waterlogged areas for over 100 km downstream (Assahira et al 2017).
The application of e-flows emerged in the midtwentieth century in developed countries within Europe and in the US in response to the biodiversity impacts of flow regulation and diversion of surface waters (Matthews et al 2014). E-flow assessments began in the late 1940s in the western US to establish minimum flows required for the protection of valuable cold-water fisheries in snow-dominated environments (Poff et al 2017). Since then, international policy landmarks that favored the implementation of e-flows have been created, such as the UK Water Resources Act of 1963, which required minimum acceptable flows to maintain natural beauty and fisheries (Overton et al 2014, Neachell andPetts 2017). After decades, environmental water science and assessment have advanced with the development of many approaches and tools in response to changing societal objectives and values, paradigms, and increasing knowledge base and modeling capabilities (Poff et al 2017). Still, one of the methods widely used to 'preserve' river flows is to set a minimum flow below which any water abstraction must be reduced or ceased (Acreman 2005).
Water allocation for the environment also involves trade-offs with other competing needs, such as hydropower and urban/agricultural water supply, and can be limited by different objectives or strategies occurring under different jurisdictional boundaries and institutional settings . Simply reallocating water from human uses to the environment often faces uncertainties due to overallocation (Loch et al 2011, Stein et al 2021 primarily hampered by competition between human and environmental needs, and lack of political will (Overton et al 2014). Trade-offs among contrasting goals need to be identified so that the appropriate strategies can be prioritized, especially in hydropower and multiplepurpose water projects which are prone to conflicting interests at different scales (figure 1). Trade-off analyses are especially important when considering the need for the implementation of more nuanced e-flows, which necessitate a better understanding of ecological needs and seasonal variability . In that sense, system analysis in water operations is increasingly needed due to the more intense competition for limited supplies, necessitating efficient allocation among conflicting objectives (Brown et al 2015). Based on that, technical processes can be developed to better guide practitioners in the development of e-flow standards for rivers and streams focusing on the habitat needs of native species, to deliver broad benefits for people and nature (Grantham et al 2020, Null et al 2021. In the last 20 years, many reviews on e-flows have been specific to certain policies or regions. For instance, Adams (2014) assessed the environmental water requirements of estuaries, Hayes et al (2018) focused on the advances in functional eflows for temperate floodplain rivers, and O'Brien et al (2021) assessed good e-flows practice for the small hydropower sector in Uganda. In addition, other reviews considered the influence of e-flows on the abundance of native riparian vegetation on lowland rivers (Miller et al 2012), the methodologies and application in the Qianhe River in China (Hao et al 2016), the socioeconomic values of restoring e-flows (Jorda-Capdevila and Rodríguez-Labajos 2017), e-flows within the process of Water Framework Directive (WFD) implementation in Europe (Ramos et al 2018), gaps between the science and implementation of e-flows in China (Chen et al 2019), practical experiences of dam reoperation (i.e. change in the operational schedule for storing and releasing water to different us and volumes) (Owusu et al 2021), and challenges on e-flow implementation in waterlimited systems (Wineland et al 2022). More generalized studies involve the review of global trends in the development and application of e-flow methodologies by Tharme (2003), the review and categorization of e-flow methods and requirements by Acreman and Dunbar (2004), and predicting ecological responses to e-flows (Webb et al 2015).
Consequently, in the last two decades, many countries have recognized the importance of e-flows in water management and have incorporated e-flow provisions in updated water policy (Harwood et al 2018). However, despite efforts, aquatic ecosystems continue to degrade at alarming rates, mainly due to habitat loss, direct overexploitation of resources (i.e. species, ecosystems, and water), and hydrological alteration (Salinas-Rodríguez et al 2021). Policy does not always translate into practice. A good example of this shortfall is the adoption of the Water Resources Act, enacted by the State of Washington in 1971, which firmly established the need for instream flows to preserve fish, wildlife, and other environmental values (Hurst 2015). Still, e-flows are frequently unmet during at least part of the year in the state's watersheds, even when instream flow rules are followed, as they do not prevent senior water rights holders (those with earlier priority dates) from using the water downstream (Hurst 2015). Shortcomings happen elsewhere too, such as the lack of enforcement of California's Fish and Game Code statutes intended to maintain fish populations in 'good condition' below dams (Grantham et al 2014). These cases illustrate the underlying difficulty in shifting water away from human uses to streamflow because of economic and political resistance.
As suggested by Arthington et al (2018b), increased flow alteration and less water dedicated to the environment are expected in coming decades as human demands (e.g. flood and drought protection, electricity generation, urban/agricultural water supply, recreation) increase; and the consequences of that growing demand initiate a cascade of biophysical changes to ecosystems (Viers 2017). The management of competing demands with imperfect knowledge and constraints of existing governance structure requires a systems approach, challenging the more linear thinking often applied to policy development and implementation . Systems thinking can be a useful tool to better understand the various processes and interrelationships of complex systems, to provide effective decisionmaking strategies towards a more sustainable water resources management (Ram andIrfan 2021, Zhang et al 2021). In this study, we apply systems thinking concepts described hereafter to explore complex system interdependences in the implementation of e-flows and their implications for environmental water management.

Systems thinking
Systems thinking provides a structured approach to understanding complex problems by viewing the overall systems, their components, interdependencies, and purpose (Mijic 2021). System feedbacks can exacerbate existing problems and result in ineffective policy interventions when they are neglected (Refulio-Coronado et al 2021). Consequently, systems thinking aspects have been applied to understand and address a wide range of issues in different settings, including environmental policy (Castro 2022). Tasca et al (2020) identified a need to bring systems thinking more generally into water resources planning and management because of the increasing complexity, scope, and urgency of environmental issues. The authors illustrate how a river system can be represented as a sociohydrological system with hierarchically organized sub-systems at successively lower levels (e.g. stream segments, reaches, pool riffle sequences, and microhabitat subsystems, as well as governance, consisting of institutions, networks, bureaucracies, and policies). Riverine ecosystems form a complex system of human and natural biotic and abiotic feedbacks, thus identifying clear ecological responses-either positive or negative-to flow alteration can become a significant challenge due to the inherited complexity of river systems (Arthington et al 2006, Wu andChen 2018). As stated by Pahl-Wostl et al (2013), implementation of e-flows requires a more systematic and integrated approach in order to capture the nuanced interaction between sociopolitical and environmental systems. According to the authors, the combined effect of poor governance and unrecognized complex feedbacks can lead to ineffective management, overexploitation of resources, and the ultimate long-term degradation of ecological integrity. Any remedy will require better and more explicit ways of acknowledging the enabling conditions and underlying drivers of conflict, explicit recognition and incorporation of systems interactions, and transparent accountability in water allocation decision-making and resulting trade-offs (Hjorth and Madani 2023). Understanding the complexity of each sub-system and connectedness to overall system behavior allows scientists or managers to identify appropriate points of intervention to meet management objectives (figure 2). In that way, systems thinking can help evolve policy-making from narrow, sectoral, and little coordinated, or even overlapping and conflicting, towards more integrated decision-making (Voulvoulis et al 2022).

Systematic review and bibliometric analysis
We gathered key studies on the implementation of e-flows, including theory, concepts, and applications associated. For that purpose, thematic searches of published, peer-reviewed literature using topicrelevant keywords were conducted on Web of Science (WoS) (www.webofknowledge.com, 8 August 2022) search engine. Keywords used for this systematic review include 'environmental flow' , and 'environmental flows' , and 'functional flows' , or 'implementation of environmental flow' , or 'environmental flow implementation' , or 'dam reoperation' , or 'implementation of environmental flows' , or 'environmental flows implementation' . The search for articles published since 2001 resulted in 68 research and review articles, retrieved as Bibtex for further bibliometric analysis. The articles were analyzed to identify general limitations to the implementation of e-flows, when mentioned, and their co-occurrence.
A bibliometric analysis was conducted to identify the state of the intellectual structure and emerging trends in e-flows research. The WoS Bibtex file dataset was analyzed in RStudio (R version 4.0) (R Core Team 2022) using the Bibliometrix R package (version 3.2.1) and its web application counterpart called Biblioshiny (Aria and Cuccurullo 2017). Bibliometrix calculates frequency statistics and performs data visualization of leading authors, conceptual and intellectual maps, collaboration and co-citation networks, and overall trends of e-flows science (sensu Hao et al 2021). Herein, a limitation is that e-flow implementation is not necessarily a scientific process that is being captured and reported in the peer-reviewed literature. Therefore, this analysis reflects findings on the research around implementation when reported and may not reflect all findings in e-flow practice. Additionally, bibliographies of selected papers were reviewed to find related and relevant publications for broadening the discussion below.

Results and discussion
Our review identified 21 obstacles in the implementation of e-flows (table 1), and their co-occurrence in 59 out of the 68 studies analyzed (figure 3). The limitations found in the literature are either specific local barriers or broadly recognized obstacles mentioned by the authors. In general, a combination of these factors reinforces these impediments, such as insufficient political will, institutional roadblocks, limited scientific methods, conflicting interests, and lack of stakeholder support, capacity and resources (Barchiesi 2018).
We identified the difficulty in shifting water from competing human uses (n = 29) as the primary factor and leading challenge to overcome. Traditional approaches to water management have mostly focused on basin productivity, as indicated by Overton et al (2014), and thus these measures of economic development skew analysis toward valuing human benefits over environmental needs (Pahl-Wostl et al 2013). Shinozaki and Shirakawa (2021) illustrate this problem in Japan, where even though e-flows can be reassessed during relicensing of hydropower projects every 10 years, conventional water Table 1. List of challenges in e-flow implementation mentioned in the literature.

Limitations (n) Country (Publications)
Differing e-flow methods (13) China ( withdrawals for consumptive use by rice paddies tend to be prioritized. Richter (2009) states that the degree of 'sustainability' achieved in a water system is directly proportional to the degree to which stakeholders are satisfied with water allocation and management. Human use objectives can impose direct or indirect system constraints that cannot be countered without a prior change in system configuration. For instance, flood flows releases from a dam might be restricted due to downstream urban development, while at the same time water quality objectives might require elevated flows to dilute pollution (Aldous et al 2011). In this case, human objectives prevent the implementation of high and low (natural) flow levels.
The remaining limitations identified form a host of other problems associated with implementing more sophisticated e-flows. Two of the most recognized and often co-occurring limitations identified were the lack of adequate hydrological data (n = 24) and ecological data (n = 28). These, for instance, may ultimately lead to difficulties in starting or continuing monitoring/adaptive management (n = 28) efforts.
The lack of resource/capacity (n = 21), experimentation (n = 19), regulatory enforcement (n = 17), and differing authorities involved (n = 18) were also recurrent problems, generally driven by the absence of funding and deficiencies in the relative importance given to e-flows when facing competing human priorities (n = 29).
Major conflicts in water allocation are expected between hydropower generation and irrigation, drinking water and irrigation, and/or between conventional energy and agricultural purposes (Sharma and Kumar 2020). Provisioning services arguably are often perceived to provide the most direct socio-economic benefits, and therefore, guide governance and management (Pahl-Wostl et al 2013). Consequently, when faced with other competing demands e-flows are generally given low priority in allocation systems and are limited to low or 'minimum' flows (Richter 2009). The prioritization of human uses is generally a result of politics and power differentials among competing interests, where economic and political power have precedence and resist changes in allocation (Sharma and Kumar 2020). Misaligned purposes and inappropriate resource allocations are common governance problems, in addition to institutional fragmentation, unclear roles and responsibilities, poorly drafted legislation, and lack of long-term strategic planning (Hjorth and Madani 2023).
Governance that fails to recognize the systems nature of decision-making  produces fragmentation and duplication of authority, policy inconsistencies and high transaction costs (Folke et al 2005). Action, then, is often compartmentalized and fragmented, where the bigger, integrated picture is lost (Tasca et al 2020). Several common patterns have emerged that encompass the body of persistent challenges now facing the practice in e-flows. The challenges identified in this review are classified into data, institutional and regulatory, sociohydrological, and political problems, and are further discussed below. Therefore, we explore examples to show their occurrence and cocausality within the core system's problems illustrated in figure 2, and points in a system's structure where interventions and action can produce more effective results.

Data problems
Two of the most recognizably mentioned obstacles to e-flow implementation were the lack of hydrological data (n = 24) and lack of ecological data (n = 28), generally co-occurring (figure 3). The significant data needs, and consequently, financial and technical resources usually required to apply the ecological limits of hydrologic alteration framework (Poff et al 2010) for developing regional eflow standards was a specific example, as discussed by Richter et al (2012). References to the lack of consideration to geomorphology (n = 4) and lack of information on riparian condition (n = 1) are generally considered as data needs about the riverine environment to be maintained or restored, as these are mediating factors that can alter flowecology relationships (Taniguchi-Quan et al 2022).
Although not common limitations, these characteristics of the river channel and its surroundings can help policymakers implement a broader set of management options, such as land-use restrictions to slow development (Giacomoni et al 2013), pest control, grazing management, and riparian restoration . Consequently, these data problems lead to practical gaps, i.e. lack of experimentation (n = 18), as actions at a dam cannot be predicted to produce specific results downstream with certainty, and many times it is unclear which dam releases will provide the desired results (Owusu et al 2021). As stated by Meadows (2008), decision-makers cannot respond to information they do not have, cannot respond accurately to inaccurate information nor in a timely way to late information. For instance, a program aimed to create a 'sustainable' balance between human and environmental water uses in the UK, yet resulting in the maintenance of the status quo, due to the vague information flows that obscured inequities in water rights and constraints . The ambiguous goal of 'sustainability' was unsuccessful. This also demonstrates the relevance of adaptive management/monitoring (n = 28) ecological benefits produced by e-flows and their adequacy to achieve the environmental goals defined (Ramos et al 2018).
For instance, monitoring of fish populations in China showed they have been impacted, in part by water infrastructure, with a decline of approximately 90% in the total number of fish fry for the four economically-important Chinese carp species (Cheng et al 2018). To counter this problem, shifts in policy priorities have been promoted with a greater focus on river restoration and e-flow implementation (Cheng et al 2018). Likewise, Mexico adopted measures to implement e-flows nationwide. E-flows are determined based on the Mexican Environmental Flows Norm, established at a river basin scale through a presidential decree for 50 years; and prioritization of basins is based on information on water availability and demand, biological richness, and conservation values (Salinas-Rodríguez et al 2018).
Our bibliometric analysis revealed a potential lack of collaboration on e-flows research that corroborates to information asymmetries, with little representation from developing and emerging countries, where it is unknown to the degree that e-flows are being implemented and, even if present, unlikely to be well represented in the scientific literature (figure 4). Brazil is a remarkable example, which despite recent widespread dam-building (Aledo Tur et al 2018, Arias et al 2020), was not detected in this analysis. Unfortunately, international collaboration has been mostly limited among the countries with the most scientific production, namely the USA (particularly California), Australia and the UK, followed by the Netherlands, Canada, Germany and South Africa. A lack of cross-collaboration in scientific research can also be a limiting factor to e-flow implementation, as most river research still operates within local paradigms (Tasca et al 2020), especially when considering the bias in the literature with a prevalence of regions with a Mediterranean-montane climate (notably California, Australia, Chile, and South Africa).

Institutional and regulatory problems
Other limitations identified were the institutional and regulatory problems. Notably, differing authorities involved (n = 19) constitute barriers, at times with overlapping roles. This unclear jurisdiction can also lead to the use of conflicting or ambiguous definitions in policy goals resulting in a lack of standard definitions (n = 14). Thus, policies with unclear goals, such as to achieve a 'sustainable' balance between water users , have 'beneficial uses' of water (Hurst 2015), maintain a 'dry weather flow' (Neachell and Petts 2017), or allow water abstractions within 'reasonable limits' (MacPherson and Salazar 2020) are found in the literature. Numerous competing definitions for e-flows, such as 'ecological flow' , 'ecological minimum flow' , and 'minimum acceptable flow' (Ramos et al 2018), can also be found in addition to differing e-flow methods (n = 13). For instance, according to Wu et al (2020), e-flow implementation in hydropower projects is regulated by the National Energy Administration of China, meanwhile water projects with other main purposes are regulated by the Ministry of Water Resources of the People's Republic of China. Both entities together use seven different terms to refer to e-flows, with inconsistent definitions, and both recommend various differing standard methodologies to assess e-flow requirements, with no specific explanation on the methods selection principle. The lack of integrated calculation methods in China is also highlighted by Wang et al (2009), as flow prescriptions are not easily transferable across a country with a such diverse geography. Therefore, the lack of regulatory enforcement (n = 17) was also identified as a recurrent problem in e-flow implementation. On the other hand, flow releases that comply with the law can also be difficult to implement due to their specific standards and objectives, and onerous enforcement (Owusu et al 2021). Although such regulations pave the way for the implementation of e-flows in Europe, legislations at the national and regional levels and obligations under the WFD, Habitats Directive, other European Directives, and international commitments need to be considered (European Commission 2016).
Similarly, another limitation we identified was climate change (n = 8), as traditional approaches assume stationarity (Overton et al 2014). For example, in the state of Washington, instream flow rules do not require scientifically-grounded standards, by definition are not meeting desired 'maximum net benefits' and cannot be modified to changing conditions except through additional notice-andcomment rulemaking (Hurst 2015). Acknowledging potential climate change impacts into planning can be an opportunity to re-examine policies and management procedures for rivers and infrastructure (Watts et al 2011, as the increasing uncertainty and conflicts will require constant updates to system rules to adapt to nonstationary conditions (Brown et al 2013). Pittock and Hartmann (2011) suggest that opportunistic policy windows for reoperation (i.e. change in the operational schedule for storing and releasing water from reservoirs) of dams are safety reviews, utility management, systems operations, and relicensing. However, physical, financial, and legal constraints can limit the implementation of e-flows when a change to infrastructure (n = 11) and system reoperation (n = 10) are needed. Changes in infrastructure may necessitate reconsideration to operational design and/or siting of infrastructure (Opperman et al 2019), often leading to retrofitting, or perhaps decommissioning (Arthington et al 2018b) in cases of infrastructure without specific e-flow release devices (Ramos et al 2018).
Another complication that emerges from institutional and regulatory problems is the rigid structures that lack adaptability. Due to the complex and relatively uncertain feedback responses among water demands, land uses, hydrological variability, biodiversity, and aquatic ecosystem services, the governance systems that manage e-flows must be adaptive, flexible, and capable of learning from experience (Pahl-Wostl et al 2013). A response to feedback systems is the creation of feedback policies. Static policies cannot respond to system dynamics and are more likely to produce temporary solutions and a greater number of escalating problems (Grigg 2016). In this way, adaptive management is a promising approach necessary to implement longterm strategies to maintain riverine ecosystems; however, this approach is difficult to implement under rigid regulatory and institutional governance (Folke et al 2005, Bruno andSiviglia 2012). Thus, management objectives and physical constraints (e.g. hydropower generation, type and size of dam outlets), can directly negate consideration of dam reoperation (Wang et al 2009).
For instance, the US Federal Energy Regulatory Commission (FERC) is responsible for the licensing process of non-federal hydropower projects in the country. FERC has been disregarding the potential impacts of climate change on hydropower operations, stating that 'although there is consensus that climate change is occurring, we are not aware of any climate change models that are known to have the accuracy that would be needed to predict the degree of specific resource impacts and serve as the basis for informing license conditions' (Federal Energy Regulatory Commission 2009). Viers (2011) argues that the issuance of FERC licenses will ensure a series of fixed operating rules based on stationary hydrology for the life of the license, typically 30-50 years in length. Viers and Nover (2018) suggest the inclusion of formal environmental impact studies, more academic sensitivity analyses, or the development of climate-informed 'worst case' scenario planning into their licensing process. According to the authors, FERC should make licenses adaptive (adaptively alter operations based on new information) to offset that oversight. However, the agency continues to dismiss the need for adaptation and flexibility in operations to support environmental water needs under a changing climate. According to FERC, predicting future flow scenarios in climate change studies is 'too speculative given the state of the science at this time' (Ulibarri and Scott 2019). With the rapid changes in climate science as well as the evolving body of e-flows work, these cases illustrate the existing challenges of e-flow implementation.

Sociohydrological problems
Securing e-flows is also limited by challenges in sociohydrology (Pande and Sivapalan 2017), such as the lack of awareness of the multiple human impacts (n = 8) on the environment caused by river regulation. This is linked to the ignorance of the multiple ecological and social benefits provided by river systems, their related ecosystem (e.g. wetlands and floodplains) and water needs, as well as the exclusion of specific stakeholders from central decision-making and priority-setting in river basins (Barchiesi 2018). Stakeholder engagement in these cases works as an information-dissemination exercise for government departments or implementing agencies, with an opportunity to comment (Acreman and Ferguson 2010). Ignorance on the importance and functioning of natural systems leads to disregard for human impacts on the environment and to how people benefit from the different components of a river system. Another limitation we identified was the lack of stakeholder engagement (n = 15), reflected as indifference toward or ignorance of the problem. Conflicts regarding what stakeholders want from water allocation and flow patterns persist when decision-making processes are opaque and/or unbalanced in representation (Carr 2015, Mehrparvar et al 2020).
Comparing successful and unsuccessful cases of eflow implementation, Owusu et al (2022) found that the key difference between them was stakeholders' involvement, especially the support of scientists, who increase the odds of successful dam reoperation. Yet, the authors emphasize that scientists should play a supportive role rather than drive the process. O'Keeffe (2018) indicates that the change in mindset (n = 1) of all levels of stakeholders, including water policymakers, managers, and scientists, is the most intractable limiting factor. The mindset is determined by their paradigm (i.e. the pattern of values, beliefs, and assumptions) that sees rivers as resources to be used to maximum benefit, which ultimately drives planning, policies, and decisions.
The minimum flow paradigm became a widely accepted form of e-flows also as a reflection of sociohydrological problems. Despite advances in e-flow science and policy, minimum flows remains the most implemented approach (Owusu et al 2021). For example, in China the minimum flow demand is widely accepted and usually adopted as 10% of the annual average flow as an empirical rule (Wu and Chen 2018). Minimum flows are also the most widely legal provisioning form of e-flows adopted in the US. Schramm et al (2016) analyzed 300 licenses of hydropower projects in the US and found that most plants throughout the country are required to release a static minimum flow or the natural inflow, whichever is less, in either the facility tailrace or bypass. The authors mention that the California hydrologic region is the only area where the majority of minimum flow releases change by season or annual water conditions.
Although minimum flows mimic some hydrological flow characteristics, they are not designed to capture more nuanced and critical aspects of the flow regime throughout the year (Grantham et al 2020). And even when implemented, e-flow allocations with lesser seniority or priority are among the first to be sacrificed when water is in short supply; the complete drying of rivers by water extractions, particularly in arid and semi-arid regions, is not uncommon (Richter 2009). Therefore, the water governance systems need to be designed to send feedback about the consequences of decisionmaking directly, quickly, and compellingly to the decision-makers.
A classic example is California's 'first in time, first in right' system of water rights, combined with the overallocation of many river systems (Grantham and Viers 2014). The state's water system is an enigma of interconnections of geographic, sociopolitical, infrastructure, and environmental factors. As a result, human and natural systems form a complex web of competing demands for freshwater, which has been the focus of continuous political, legislative, and legal battles (Stewart et al 2020). The unrealistic allocation of water creates 'paper water' , as the water proposed for transfer does not translate into the natural system's capacity of producing water for human supply (Chong and Sunding 2006).
Similarly, unfeasible water allocations have happened in Australia, where such rights are termed 'sleeper rights' , which can be later activated for larger use of water than in previous years (Chong and Sunding 2006). Overallocation in the Hawkesbury-Nepean River in Australia has reduced e-flows from the recommended 80% to around 3%; a scenario similar to the Durance River in France, in which 97.5% of river flows are diverted for hydropower production . One approach to this type of problem is the regulation of the commons enforced by policing and penalties, to create the feedback link from the condition of the resource through regulators to users. When there is a commonly shared resource, every user benefits directly from its use but also shares the costs of its abuse with everyone else.
After California's 2012-2015 drought, in which low flows and high temperatures reduced water quality and impaired habitat for native fish species and supported expansions of invasive species, the discussion of an environmental water right began (Lund et al 2018). Agricultural demands for irrigation supply under drought-induced water scarcity have resulted in widespread groundwater overdraft, resulting in decreased base flow and localized subsidence (Pinter et al 2019), leading government agencies to expand e-flow regulations (Lund et al 2018). Likewise, in the UK, water rights were perpetual, as established by the Water Resources Act in 1963, therefore, new licenses could only be issued if they did not impact existing rights . To change that, time-limited licenses started being issued in the 1990s, together with curtailments to protect environmental features on a case-by-case basis, in which the Environment Agency can, for instance, ban on spray irrigation or non-essential water use during drought .

Political problems
We identified that the abovementioned limitations are generally and ultimately driven by the absence of funding and deficiencies in the relative importance given to e-flows when facing competing priorities (n = 29), such as pollution (Wang et  . The controversies among these limitations are generally caused by political problems, i.e. the lack of political willingness (n = 12) to recover water for the environment, especially in over-allocated systems with conflicts between economic development and conservation. This results in a lack of resource/capacity (n = 21), for instance, due to changes in funding cycles or priorities within government agencies, and short-term commitments to environmental water management and monitoring (Conallin et al 2018b). That, in turn, leads to a lack of initiative (n = 12) in reallocating water resources for protecting riverine environments, as human uses are prioritized (Shinozaki and Shirakawa 2021). Ultimately, political decisions determine the level of acceptable compromise of human uses in face of environmental water requirements .
As stated by Meadows (2008), 'If a government proclaims its interest in protecting the environment but allocates little money or effort toward that goal, environmental protection is not, in fact, the government's purpose. Purposes are deduced from behavior, not from rhetoric or stated goals' . For instance, in 2007, e-flows were required to maintain the normal function and state of streams in South Korea through the River Act (Kim et al 2022). In addition, e-flows to conserve the health of aquatic ecosystems have been endorsed in the Water Environment Conservation Act of 2017; however, implementation of e-flows is still in its early stages due to the lack of established criteria for the selection of target sites (n = 4) (Kim et al 2022). Similarly, in Chile, e-flows have been applied in a discretionary and ad hoc manner, as safeguarding e-flows may be costly and politically unpalatable, potentially requiring the redirection of water away from consumptive, economic purposes (MacPherson and Salazar 2020). Therefore, policy goals can also be eroded due to political and economic pressures.
The systems nature of decision-making and the need for information sharing can lead to the fragmentation of effort, and a failure to take advantage of local adaptive management learnings . This surfaces the need for policy coherence, in order to reduce conflicts and strengthen interactions, coordination, and effects of governmental actions to achieve a desirable goal. Assessing policy coherence and the potential implications of one sectoral policy across the system is key to minimizing trade-offs and establishing compromises (Pereira Ramos et al 2021). Policy coherence might include policy harmonization, i.e. making the regulatory requirements, laws, or governmental policies of different jurisdictions identical or at least more similar, or by assigning decisions to a common political authority, as defined by Majone (2014). Polycentric governance may be of significance in responding to ecosystem dynamics at different scales (Folke et al 2005) and provide better results when involving all stakeholders and therefore addressing all the economic activities within a resource system (Refulio-Coronado et al 2021). For instance, in 2000 the WFD was the first legislation ruling in the European Union to use ecological conditions as the benchmark for the management of 'ecological flows' (Wu and Chen 2018). The WFD sets a common definition and understanding of how ecological flows should be calculated to facilitate their integration into river basin management plans in Europe (European Commission 2016).
Representative of these political challenges are efforts for restoring fish population. Salmonid populations in the Yuba River, California illustrate a case in which e-flows alone are unable to achieve the purpose of maintaining native species. As stated by Viers (2012), salmon 'cannot go beyond the dam because there is no water, and there is no water because they cannot go beyond the dam' . Restoration in this case would require not only e-flow releases in proper timing, quantity, and quality from an upstream hydropower project but also fish passage downstream for the reintroduction of salmonids in the system. Fish hatcheries are another example of efforts that try to balance the effects of dysconnectivity, and hydrologic alteration caused by river regulation and inappropriate e-flows. According to Sturrock et al (2019), fish hatcheries in California often eliminate the entire migratory corridor by trucking fish directly to the estuary to prevent in-river mortality, particularly during droughts. However, the authors state that this practice has inadvertently caused excessive straying rates due to genetic homogenization and increasingly synchronized population dynamics. Similarly, Brown et al (2013) found that restoration projects of Atlantic salmon have not yielded self-sustaining populations in any eastern US river, despite hundreds of millions of dollars spent in hatcheries, although a complete extinction was avoided in a few rivers, albeit at the expense of genetic integrity. In addition, the authors mention the poor performance of fish ladders by portraying the mean passage efficiency of <3% from the first dam up to the spawning grounds for American shad. The authors indicate that the systemic cause of fish declines (i.e. main stem dams and overfishing) were not properly addressed, exerting a greater pressure on natural resources management agencies to restore fisheries. Consequently, more funding is applied to the problem, more agency personnel are hired, and it becomes difficult to dismantle ineffective programs.
Likewise, flow restoration projects that neglect how sediment availability influences abiotic and biotic responses to flow by reshaping channel morphology and creating habitat (Wohl and Brian 2015) can be unsuccessful or even deleterious. For instance, channel incision in sediment-deprived reaches can be aggravated, and inevitably floodplain connectivity is further jeopardized (Consoli et al 2022). Gravel augmentation projects have been implemented in California to improve anadromous salmonid spawning habitat, however beneficial results tend to be temporary, as placed gravels were usually scoured and transported downstream by subsequent high flows (Harvey et al 2005). Moreover, ecosystem restoration with poor consideration of the influence of hydrological alteration on freshwater biodiversity across spatial scales creates a paradoxical situation where even e-flows may inadvertently contribute to further biodiversity declines (Rolls et al 2018). Restoration considering hydrology (i.e. e-flows) in isolation addresses symptoms, meanwhile process-based restoration also accounting for geomorphology, connectivity and biology addresses the causes of degradation (Beechie et al 2010).

Recommendations
The limitations identified in this study, including lack of financial resources, organizational capacity, and regulatory enforcement, are a reflection of the primary factor identified in the review, which is the prioritization of human uses over environmental needs. Below, we provide a set of recommendations based on the reviewed literature and the systemic problems identified to address this overarching factor. Although these recommendations do not consider resource limitations and other practical on-the-ground considerations (e.g. data-poor river basins), we discuss several examples from the literature to provide a clearer path for successful e-flow implementation.
6.1. System mapping: understand the system to be managed and the management system A systemic map or conceptual model can express insights about the purpose, processes, and structures governing the system and producing its behavior (Mingers andWhite 2010, Tasca et al 2020). A systems perspective allows the integration of the subsystem goals, as it can bring information not only about existing problems (system unintended outcomes) but also on elements (resources), their arrangement (hierarchy), rules, and consequences. System mapping can be produced in terms of causalities (causal loop diagrams) and flows (flow charts), that ultimately reveal effective intervention points (Haraldsson and Sverdrup 2021), by surfacing the system problems (Mijic 2021), or areas where reliable quantitative information is not available (Mingers and White 2010). It is also suggested that system mapping be an interactive participatory process of managers, governments, infrastructure operators, farmers, and other stakeholders to integrate multiple general perspectives (Mijic 2021, Ram andIrfan 2021).
Understanding the system to be managed may involve mapping physical, natural, and/or human components, meanwhile, the management system may involve mapping people, resources, and procedures (Grigg 2016). If the system to be managed is a water supply reservoir, the management system involves data, operational rules based on legal requirements, and a decision support system (Grigg 2016). In this way, we can identify the level of flexibility of the systems involved, i.e. the case-specific context, limits, and barriers that can or cannot be adapted or overcome (System accommodation to constraint vs. Constraint accommodation to the system). For instance, Mijic (2021) used system mapping to identify that most infrastructure and technological solutions to improve water quality at Lake Windermere in the UK, would fail unless implemented across the system as a whole.
As described in section 5.2, information on system deficiencies might not be enough for triggering action due to institutional constraints, requiring therefore an accommodation to the system constraint. For instance, FERC can reject mitigation measures due to 'high' costs, if their cost represents more than 10% of a project's annual power benefits (Black et al 1998). To counter that the economic analysis could consider the recreational fishing benefits, which could outweigh the costs to implement it (Black et al 1998). However, the construction of a fish ladder to minimize the impact of damming on migratory species is not always carried out due to the costs involved. This constraint can be accommodated to the system needs in specific windows of opportunity that are open during the relicensing process. In the relicensing of existing hydropower projects, other US Federal agencies have a 'conditioning authority' through which they can issue conditions that FERC must incorporate into a license, including requirements to change a project's design or its operation, e.g. a retrofit to include a fish passage structure or change its e-flow release schedule (Opperman et al 2019).
The main operating purpose of a dam influences dam reoperation strategies and reoperation might require integration across sectors or involve multiple dams to simultaneously achieve human and environmental objectives (Vonk et al 2014). Systems thinking can assist different bodies to work together with a shared view to develop more coherent management options and policies, to provide multiple outcomes while considering and preventing unintended consequences (Mijic 2021). Considering that, we recommend that the first step for e-flow implementation be system mapping to surface the problems in effect and the limitations in place, as discussed in the examples above. This intends to fix information flows and avoid delays in response by identifying mechanisms that impose conditions and constraints that can limit success. A system map allows the first considerations of alternative practices and policy options, as well as inflexible frameworks to be reconsidered, countered, and overcome. The identification and extraction of relationships among and within social and ecological systems can also later be used as input variables into empirical models (Bouchet et al 2022), which can be used in the next step, the system analysis.

System analysis: support decision-making using suitable modeling tools
The conceptualization in the system mapping works as an actual model development (Haraldsson and Sverdrup 2021), and building a model can change paradigms as the builder is forced to see the system as a whole (Meadows 2008). The delineation and quantification of system interconnections and influences allow for building and testing a computer model (Mingers and White 2010). That allows the learning process during planning and implementation, by identifying the problems to be solved and the questions to be answered based on the information available and the understanding provided by it.
For instance, water supply systems require careful simulation as their outcomes need to provide very high reliability, with sensitivity restricted to critical periods (e.g. droughts) (Marchau et al 2019). Consequently, a detailed representation of temporal and spatial variability (i.e. demands, inflows, outflows, competing needs) is required to accurately assess these systems (Marchau et al 2019). The model outcomes can then be effective communication tools to engage stakeholders in technical decision-making. In hydropower systems, energy generation tends to follow electricity price signals and can also reflect constraints imposed on a facility (e.g. multi-objective reservoir) that can limit the timing, period, and intensity of power generation (Stoll et al 2017). Therefore, adding the economic-driven factor of hydropower in water systems can produce more realistic information for better decisions. In that way, adverse effects of hydropower can be better assessed to provide alternatives that minimize them by restoring vital features of the natural flow regime and/or avoiding hydropower-induced habitat bottlenecks, such as through the adoption of restricted ramping rates (Freeman et al 2001).
Hydropower system design and planning that fully integrates environmental and social resources remain relatively rare, although this integration can provide relevant information to energy planners and operators, and provide better opportunities to achieve climate and energy goals while also supporting the ecological integrity of rivers (Opperman et al 2023, Rheinheimer et al 2023. Sector integration, aided by system analysis is needed to address the main challenge of competing priorities by promoting the transparent assessment of needs, allocations and inherent trade-offs . Better modeling, including forecasted energy prices and/or hydrological conditions, can also help explore alternative flood control rules for flexibility in the operation and management of reservoirs, resulting in 'extra' water that could be used for other purposes, including e-flows (Lee et al 2006, Zarei et al 2021. In addition, stress testing can assess the system's performance to meet the desired objectives under critical periods (namely, the driest or wettest season, the highest recorded flood) (Nagy et al 2013). Modeling of e-flow requirements has been employed for planning flow standards in Texas, USA, to advance statewide implementation efforts (Wurbs and Hoffpauir 2017). Similarly, modeling studies have allowed the analysis of trade-offs among water uses to allow preliminary discussions on the implementation of functional flows by water agencies in Brazil (Dalcin et al 2022), and the initial implementation stages of the California Environmental Flows Framework on a small subset of watersheds in California (California Environmental Flows Working Group (CEFWG) 2021).
Hydropower planning and operations, due to the significant infrastructure investment and high value as a source of renewable energy, is perhaps the most viable area for continued improvement of systems analysis tools in water systems operations (Brown et al 2015). They also state that especially in hydropower-dominated systems economic incentive justifies optimizing the use of operational flexibility for maximizing hydropower revenue subject to nonpower constraints and objectives. Water and energy systems can be represented separately in modeling frameworks; their interdependency is indirectly considered through the input variables employed in each model (Voisin et al 2016). For instance, Jager and Martinez (2012) used an energy model to estimate relative electricity value and one to estimate relative salmon production to optimize e-flows for hydropower generation and the environment in the Tuolumne River, California. However, as stated by Stevanato et al (2021), model integration can consider a joint optimization according to one objective function that allocates the resources in the two systems. In that way, coupled water-energy models allow the assessment of their feedbacks without coarse approximations in their dynamic behavior, and therefore, a more scientifically solid energy and hydrological planning (Stevanato et al 2021). In that sense, coupled water-energy models can help integrate the management of the complex interactions among the hydrologic, environmental, engineering, and socioeconomic dimensions of water resource systems when hydropower systems are present.
Simulation models can be used to assess the performance of alternative water management system configurations, plans, or policies, including economic and environmental performance indicators and trade-off analysis (Brown et al 2015, Loucks andvan Beek 2017). For instance, Rheinheimer et al (2016) considered climate-adaptive instream flow requirements in hydropower systems using wholesale electricity prices to better inform release decisions. Similarly, Willis et al (2022) and Maskey et al (2022) used a multi-objective water system simulation model built in Pywr written in Python (van Rossum 1995), which simulates customizable water allocation and operation rules throughout complex managed water systems (Tomlinson et al 2020). The modeling framework simulates a daily time step basin-scale water resources system with optimization of discretionary hydropower based on day-ahead market, a key methodological advancement over typical water system models . In this coupled water-energy modeling effort, forecasted energy prices and hydrological conditions drive generation, as water is allocated based on a monthly planning optimization model with partial foresight and a long (multi-month) planning horizon, and a daily scheduling model. In that way, competing demands for human uses are accounted for, meanwhile the influence of current or alternative scenarios can be assessed, including climate change impacts (Maskey et al 2022) and the implementation of alternative e-flow schedules .
Therefore, we recommend the employment of modeling as a useful tool for identifying more realistic, reliable, and robust infrastructure designs and operating policy rule sets for a given hydrological input, aiming at historical or future scenarios. Models have had an increasingly important role in providing a common way for planners and managers to predict the behavior of any proposed water resources system design or management policy before it is implemented (Loucks and van Beek 2017). Thus, a system analysis with coupled water-energy models can help better identify and assess different scenarios by allowing the modeler to test different rules and goals, producing new information flows for decisionmaking. Based on the problems assessed, trade-offs among conflicting goals are identified and the appropriate strategies can be prioritized.

System resilience: move from simplistic flow designs to ecosystem process-based approaches
Resilience is a system's ability to survive and persist within a variable environment. As defined by Meadows (2008), resiliency is not achieved by static or constant states over time, as systems can be very dynamic, with short-term oscillation, periodic outbreaks, long cycles of succession, climax, and collapse within their nature. Comparably, the resilience of a river system is in its dynamic behavior. Therefore, building resilience into e-flows involves setting flow targets necessary to achieve ecosystem goals by mimicking key components of a natural river's flow.
Discrepancies between the desired and actual state of a system compel managers to intervene, however a lack of understanding of system structure and functions results in ineffective interventions (Endreny 2020). Flow designs based on institutional and technological panaceas without long-term monitoring of their performance and effectiveness produce no revision and critical reflection on practice (Pahl-Wostl et al 2013). As noted by Viers and Nover (2018), despite the extraordinary effort and money generally directed toward hydropower relicensing, the requirements for post-licensure performance monitoring are comparatively negligible. The authors state that, even though requirements for monitoring in hydropower licenses are commonplace, they are not used in an adaptive manner as licenses typically do not specify any consequences based on unfavorable outcomes of monitoring. For instance, large efforts such as the WFD in Europe propose the implementation of e-flows to counter the negative ecological impacts caused by river regulation in many countries. Yet, a minimum flow paradigm that disregards the different flow regimes for normal or dry years is still largely adopted in the participating countries (Acreman and Ferguson 2010, Ramos et al 2018). Ultimately, the unachievable goals can lead to more problems such as when only 'paper water' is available to the environment due to overexploitation (Chong and Sunding 2006), or when restoration addresses the only symptoms of hydrologic alteration (Sturrock et al 2019).
Protecting the dynamism of river systems involves setting dynamic boundaries in water withdrawals from/releases to rivers, reflecting changing societal needs and values over time as well as new scientific knowledge (Richter 2009). The Building Block Methodology was created to reproduce the dynamic flow regime components consisting of different 'blocks' of flow in South Africa (King et al 1998). The method establishes monthly volumes of low flows, and the duration, timing, and magnitude of floods, for both maintenance and drought years, based on data, global literature, expert opinion, and local knowledge (King 2016). More recently, progress in e-flows has focused on specific functional flows that support natural disturbances that promote the physical dynamics and drive ecosystem functions (Yarnell et al 2015, Taniguchi-Quan et al 2022. The functional flows approach emphasizes process-based hydrograph components that are key to specific environmental outcomes and can guide management of regulated river systems (Viers 2017). Developing inherent relationships between ecological responses to flow alteration is necessary to enhance the scientific credibility of flow designs (Poff and Matthews 2013).
E-flows that fail to support ecological functions can result in inefficiencies in water allocation that foster conflicts between competing water demands (Stein et al 2021). As noted by O'Keeffe (2018), misunderstanding and resistance to implementing eflows are not uncommon. Moyle et al (2018) address a vivid example revolving the controversies around the Delta smelt, an 'economically insignificant' fish protected by state and federal conservation mandates. Complicated by the intersection of declining water quality, aging infrastructure, climate change, and multi-institutional governance, the deficiencies of single species management have been laid bare (Luoma et al 2015). A species recovery plan remains in limbo as the state negotiates with local water authorities on how to interpret the science and how best to implement water allocations for the environment (Hanemann and Dyckman 2009). Although rigor and ecologically comprehensive processes in setting e-flows targets can increase the challenge of implementing e-flows (e.g. the need for better data, capacity, and resources) and exacerbate competing demands, they are indeed required if e-flows are to be effectively adopted.
In California, functional flow components such as peak flows, dry-season low baseflows, wet season initiation flows, spring recession flows, and interannual variability can be identified in all rivers although their dimensions (timing, magnitude, frequency, and duration) vary regionally (Grantham et al 2020). These flow regime components are key for focal native species such as salmonids to thrive in the State, for instance, providing cues for migration and spawning (Yarnell et al 2015). Functional flows rest on the assumption that reservoirs releases that reproduce these key flow components will produce the necessary hydrologic signals that trigger biophysical processes upon which native biological communities depend (Yarnell et al 2020). Viers et al (2017) applied this framework to connect natural flow regime components to specific biophysical riverine functions in various hydroclimatic systems, to help inform water resource management strategies and functional flow requirements globally. The authors consider the functional flow approach a plausible management option to maintain ecosystem services and biodiversity while also continuing conventional reservoir operations. For instance, in a system analysis, Willis et al (2022) modeled the adoption of functional flows in the central Sierra Nevada, California. The authors found that generally functional flows can provide enhanced ecological functions and reduce uncontrolled spills, although at the expense of lower agricultural deliveries, with the extent of tradeoffs depending on the river basin and water year.
However, institutional and regulatory structures need to be flexible enough so that water releases can be timed to accommodate the needs of target species or ecosystems. Thus, rather than focusing on specific species or life-stage requirements through static minimum instream flow requirements that do not enhance ecological integrity, broadly, the implementation of e-flows should focus on the functional attributes and processes that enhance biogeochemical cycling, structural habitat formation, and ecosystem maintenance (Grantham et al 2020, Yarnell et al 2020). In addition, knowledge-driven decisions to solve problems based on research and analysis are key to adapting water management and reservoir operations. Therefore, we recommend that essential hydrograph components be identified for guiding the implementation of ecosystem process-based e-flows. The focus on elements of the natural flow regime should replace generalized e-flows based on simplistic thresholds that work against ecosystem health.

Conclusion
E-flow implementation is primarily constrained by the unbalanced competing human priorities having precedence over environmental needs, as the result of poor environmental water governance. When governance fails to address the entrenched interests and legitimate expectations of water allocation from stakeholders, other limitations represent only incidental hurdles for implementing e-flows. For instance, the need for education, stakeholder engagement, or better data and science has been recognized in the literature (Harwood et al 2018, Mezger et al 2019, Tasca et al 2020, Owusu et al 2022. Technological fixes alone, however, are unlikely to overcome historical and structural impediments to cooperation. Implementation in the absence of cooperation, therefore, is likely to be limited to those portions of an e-flow regime that do not conflict with other purposes, and thus reduced to minimum flow targets. Consequently, the smallest amount of water that can maintain a wetted channel is allocated for the environment, limiting functional attributes and processes that enhance biogeochemical cycling, structural habitat formation, and ecosystem maintenance. Many other secondary challenges, generally cooccurring in causality, limit the implementation of more nuanced and sophisticated e-flows. The implementation of e-flows might require regulatory and even institutional changes that allow for their preexistence, stakeholder engagement at all levels, setting achievable flow designs based on the available data, resources, and human-driven limiting factors (e.g. financial and infrastructural constraints, other demands) to guide decision-making. Implementation also requires timely, reliable, and available data, not only on the historical and current environmental states of river systems but also on potential future scenarios. These observations are even more pressing as hydroclimatic non-stationarity requires the acknowledgment that climate change impacts are imposing limitations on the natural system's ability to provide water for allocation systems (Milly et al 2008).
Ecologically functioning systems require the presence of functional flow components. However, as the most common controlling factor of e-flow implementation is regulation (Owusu et al 2021), reoperating reservoirs to accommodate more sustainable e-flow strategies requires flexibility and policy changes. Although water resources management is not reduced to computer-based models, water-energy modeling is an important part of decision-making in these processes when hydropower projects are present. Simulation models have been widely adopted for evaluating the impacts of changes in supply and demand under different scenarios when planning and managing water resources systems. To achieve the goal of sustainable water management, the investigation of water resources system design and management policies establishes the foundation for assessing system performance. Modeling tools can represent important interactions among the various control structures (i.e. different designs and policies) and users of a water resource system, and therefore, can help inform planners and managers on the tradeoffs when allocating water resources (Loucks and van Beek 2017). Assessment of management actions and frameworks is needed to find sustainable compromises between the different values, or to at least allow trade-offs to be explicit (Conallin et al 2018b).
Although different levels of impacts on human uses are expected for different locations, they may not be greatly affected by the environmental water allocation (Owusu et al 2021. Therefore, modeling studies constitute a powerful tool to balance these often-conflicting interests towards sustainable solutions to environmental water management problems. The growing demands of human water uses associated with the uncertain water supplies caused by global changes, most notably climate change, will likely affect current management and operations and further stress already impaired and threatened riverine ecosystems and processes (Overton et al 2014). Consequently, planning and management decision-making processes increasingly need sciencebased approaches to guide e-flow implementation. Further studies involving facility-specific to system re-operation can better inform decision-makers and managers to maintain the health of freshwater ecosystems, building long-term resilience rather than short-term survival. Future research on e-flows should assess water system performance of conflicting objectives at different scales, to identify efficient trade-offs by using combinations of different scenarios of system designs, operating policies, and stressors. Some challenges for the future include the encouragement of stakeholder participation and awareness about the human impacts on the environment as well as the importance and benefits of eflows. In addition, a greater focus on habitat formation by considering improvements on floodplain reconnection, sediment transport, geomorphology, temperature gradients, and restoration of riparian ecosystems, for example, can lead to greater environmental benefits than improvements in hydrologic conditions alone.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).