Abstract
Rivers continue to be harnessed to meet humanity's growing demands for electricity, water, and flood control. While the socioecological impacts of river infrastructure projects (RIPs) have been well-documented, methodological approaches to quantify river fragmentation and flow alteration vary widely in spatiotemporal scope, required data, and interpretation. In this review, we first present a framework to visualise the effects of different kinds of RIPs on river fragmentation and flow alteration. We then review available methods to quantify connectivity and flow alteration, along with their data requirements, scale of application, advantages, and disadvantages. Finally, we present decision-making trees to help stakeholders select among these methods based on their objectives, resource availability, and the characteristics of the project(s) being evaluated. Thematic searches of peer-reviewed literature using topic-relevant keywords were conducted on Google Scholar. The bibliography of selected papers was also reviewed, resulting in the selection of 79 publications. Papers that did not define or apply a specific metric were excluded. With respect to fragmentation, we selected papers focused on instream connectivity and excluded those dealing with overland hydrologic connections. For flow alteration, we selected papers that quantified the extent of alteration and excluded those aimed at prescribing environmental flows. The expected hydrological consequences of various RIP types were 'mapped' on a conceptual fragmentation-flow alteration plot. We compiled 29 metrics of river fragmentation and 13 metrics to flow alteration, and used these to develop decision-making trees to facilitate method selection. Despite recent advances in metric development, further work is needed to better understand the relationships between and among metrics, assess their ecological significance and spatiotemporal scale of application, and develop more informative methods that can be effectively applied in data-scarce regions. These objectives are especially critical given the growing use of such metrics in basin-wide conservation and development planning.
Export citation and abstract BibTeX RIS
Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 license. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
1. Introduction
Spurred by growing human populations, rapid urbanisation, and expanding industrial and commercial activities, rivers continue to be harnessed and regulated to meet humanity's growing demands for electricity, irrigation, water supply, and flood control (Nilsson 2005, Lehner et al 2011). With more than 58 400 large dams (ICOLD 2019) and 82 891 small hydropower dams (Couto and Olden 2018) worldwide, it is estimated that humans have appropriated more than half the global accessible freshwater runoff, creating a cumulative reservoir storage capacity of about 6197 km3 (Lehner et al 2011). These dams have fragmented and affected most rivers globally, leaving only an estimated 23% of the world's large rivers (>1000 km in length) flowing uninterrupted into the ocean (Grill et al 2019). While these dams and reservoirs have significantly contributed to human development (WCD 2000), they have fundamentally altered riparian ecosystems that depend on the dynamics of streamflow and the movement of water and the materials longitudinally and laterally through the drainage network from head-waters to estuaries and deltas (Poff et al 1997).
Despite these adverse impacts, hydropower continues to be the world's largest source of renewable electricity, with a 50% expected increase in production by 2030 (IRENA 2016). Ongoing and future hydropower developments are largely concentrated in developing countries and emerging economies of Asia, South America, Africa and the Balkan region of Europe (Zarfl et al 2015, Tockner et al 2016, Winemiller et al 2016). Within these regions, subsistence communities may be especially dependent on the provisional services that aquatic ecosystems provide (Beck et al 2012). Moreover, hotspots of existing and proposed dam development often overlap with areas of high freshwater biodiversity and endemism. Examples include the Amazon, Mekong, Congo, Zambezi, Yangtze, Himalayan, and Western Ghats river basins (Tockner et al 2016, Winemiller et al 2016, Jumani et al 2018). In 2018 alone, an additional 21.8 GW of hydropower capacity was installed worldwide (Hydropower Status Report 2019). Conservative estimates suggest over 3700 hydropower dams (>1 MW) are under construction or proposed for further development across the globe (Zarfl et al 2015). This is in addition to the proliferation of other river infrastructure projects (RIPs) such as small dams, water abstraction schemes, inter-basin transfers or river interlinking projects, flood control structures, and navigation schemes that could cause major alterations in flow and sediment regimes (Grant et al 2012, Bagla 2014, Dey et al 2019). Furthermore, even within affected basins, previously untapped headwater streams, characterised by low discharge and high gradient, are increasingly being dammed by the proliferation of small dams and diversion schemes (Couto and Olden 2018).
The hydrological consequences of RIPs on riverine ecosystems are frequently framed in terms of primary effects: reduced river network connectivity (or increased river fragmentation) and flow alteration (Nilsson 2005). Physical structures such as dams, weirs, barrages, and levees fragment the river network, impeding the free movement of water, sediment, organic matter, nutrients, energy, and organisms across space and time (Pringle 2003). The disruption of these water-mediated connections further influences crucial ecosystem processes and functions within river networks (Vannote et al 1980, Wiens 2002, Hermoso et al 2011). The loss of this connectivity can be considered along a temporal dimension (seasonality of flows over time) and three spatial dimensions—longitudinal (connectivity along the length of a river channel from the source to the mouth), lateral (connectivity between the floodplain, riparian areas, and the river channel), and vertical (connectivity of stream water column with groundwater) (Ward 1989). Physical structures may also store, divert, and abstract water from the river channel, and hence alter one or more characteristics of the natural flow regime (Richter et al 2003). Flow regulation describes alteration of the natural flow regime, characterised by variability of flow magnitudes, frequencies, durations, timing, and rates of change within the year and over multi-annual periods. Streamflow directly influences stream water quality and physical habitat characteristics of the river channel and floodplain, thereby maintaining the habitat diversity required to support native biotic communities and ecosystem functions (Richter et al 1996, Poff et al 1997). Flow regulation may be caused by the active or passive management of water in rivers; some infrastructure can reduce or augment downstream discharge through specific dam operations or abstraction points, while other forms passively hold water or reduce flows based on the size of the infrastructure and the dynamics of discharge.
Whereas methods to assess connectivity in terrestrial landscapes have long been developed and applied (Tischendorf and Fahrig 2000, Calabrese and Fagan 2004, Kindlmann and Burel 2008), assessments of connectivity in riverine systems is a relatively recent topic of study (Fagan et al 2002, Wiens 2002, Cote et al 2009, Wohl 2017). Unlike terrestrial systems, where landscape connectivity is two-dimensional with numerous connectivity pathways, connectivity in river networks is water-mediated and largely driven by river flows (Pringle 2001). On the basis of their hierarchical branched structure, fragmentation in river networks can yield more variable fragment sizes compared to two-dimensional systems (Fagan 2002). Consequently, river fragmentation more severely impacts connectivity due to the existence of fewer possible pathways for water-mediated dispersal and recolonization (Fagan 2002). Furthermore, similar habitat patches that may be geographically proximate to each other in a river network, may be separated by longer stream lengths. This can significantly reduce the potential for recolonization and decrease metapopulation persistence (Fagan 2002, Fullerton et al 2010). These unique characteristics of aquatic dendritic networks and their inherent spatiotemporal complexities pose a challenge to applying measures of landscape connectivity to river networks (Fagan et al 2002, Wiens 2002, Cote et al 2009). However, being able to effectively assess and predict the impacts of RIPs is crucial to inform project-specific and basin-wide conservation, restoration, and development plans. Recognising this gap, numerous methodological advancements have been made to better assess metrics of river fragmentation and flow alteration based on several types of remotely sensed and field-based data (Nilsson 2005, Cote et al 2009, Grill et al 2014).
Understanding the suite of tools available to characterize river connectivity and flow regulation is important because these metrics can be used in a descriptive manner to quantify impacts of RIPs on both connectivity and streamflow dynamics. These tools can also be used in a prescriptive manner to develop and assess scenarios and environmental flow methodologies to aid in basin-wide conservation and development planning. In places where RIP development trajectories are tending towards proliferation of smaller projects along upstream drainage networks (Zarfl et al 2015, Couto and Olden 2018), there is a growing need to adequately assess reach- and catchment-scale fragmentation and flow regulation to account for these impacts (Athayde et al 2019). Further, recognising that countries with the most aggressive RIP development plans are often data-limited (Auerbach et al 2016), there is a need to compile relevant methods that can be applied in such data-limited environments so that stakeholders in these regions can assess the effects that RIPs might have on aquatic ecosystems and the services they provide.
Within this context, the goals of this paper are to (1) present a conceptual framework for characterizing the effects of RIPs on river fragmentation and flow alteration; (2) review published methods to assess river fragmentation and flow regulation, including metric descriptions, data requirements, output, scale of application, advantages and disadvantages; and (3) present a decision-making tree to help managers and stakeholders select the most appropriate methods based on resource availability and objectives. We conclude by identifying existing data and methodological gaps and discussing important directions for future research, in the context of current global trends of RIP development.
2. Understanding river fragmentation and flow alteration
On the basis of their branching structure, stream networks comprise functional habitats that are hierarchically nested across spatial scales (Rodríguez-Iturbe and Rinaldo 1997, Fullerton et al 2010). Consequently, the relative importance of various connectivity dimensions and drivers of ecological processes varies across spatiotemporal scales (Vannote et al 1980, Ward 1989). The effects of RIPs on connectivity and flow alteration are thus not only influenced by the extent of impact, but also on its location (i.e. headwaters versus tributaries versus the mainstem) and timing (i.e. coincident with high versus low flows) (Fagan 2002, Diebel et al 2015). RIPs can influence stream hydrology, biophysical characteristics, and ecological and functional integrity at many scales (figure 1). Together, these changes impact stream biophysical and chemical characteristics, which further influence aquatic and riparian habitat availability and quality, freshwater biodiversity, and associated ecosystem processes and functions such as nutrient cycling regimes, sediment redistribution, and ecosystem productivity (Dudgeon 2000, Rosenberg et al 2000, Vorosmarty et al 2000, Poff and Hart 2002, Pringle 2003, Nel et al 2009, Anderson et al 2015). These changes can have serious consequences on the livelihoods, food security, and the physical, cultural, and spiritual well-being of river-dependent communities (Richter et al 2010).
Figure 1. Schematic model illustrating morphological and water quality responses (orange boxes) to river infrastructure project (RIP) induced altered stream hydrology (blue boxes), and their influence on biotic integrity (green boxes) and ecosystem function (purple boxes). Arrows indicate pathways or directionality of influence. Dashed boxes represent distinct levels of impact, and solid coloured boxes within them represent the main components pertaining to that theme (adapted from Poff et al 1997).
Download figure:
Standard image High-resolution imageWhile most RIPs influence both connectivity and flow regimes, they may disproportionately affect one or the other depending on the project type and/or location (Farah-Perez et al 2020). Projects can be classified based on size (large, medium, or small based on installed capacity or dam height, though these classifications vary widely by region; Couto and Olden 2018), purpose (hydropower generation, irrigation, water supply, flood control, navigation), and design (with or without diversion/abstraction, storage capacity, and operating regimes). Nevertheless, each project can be expected to influence connectivity and the natural flow regime differently, and their impact can be visualised on a fragmentation-flow alteration plot (figure 2). Since the basin-level impact of these disturbances can be expected to vary from headwaters to the mainstem, the location of these projects will also influence their relative impact. While the specifics of each RIP dictate its actual position on this conceptual plot, it is instructive to 'map' different RIP types according to their likely impacts on these two axes (figure 2).
Figure 2. Expected location and impact of an individual project across different RIP types (medium/large storage dams, small hydropower projects, and low head dams) on river network fragmentation and flow alteration.
Download figure:
Standard image High-resolution imageMedium and large dams that aim to impound water, stabilize low flows and eliminate peak flows, such as those built for flood control, water storage, and hydropower generation, are often characterised by high barriers and substantial reservoir storage capacities. These projects are expected to significantly impact both flow regulation and network fragmentation (Grill et al 2014). When such large RIPs are coupled with water abstraction (e.g. for irrigation and water supply projects), their impact on flow alteration can be expected to increase further (figure 2). Since these projects are larger, in terms of capacity and/or size, they tend to occur on higher-order streams. Barriers located further downstream can isolate greater proportions of available upstream habitat and significantly impact metapopulation dynamics such as dispersal and recolonization abilities (Fagan et al 2002, Nilsson 2005, Fullerton et al 2010). Hence, dams farther downstream in the river network create larger fragment sizes and greater basin-wide fragmentation.
Small hydropower projects (SHPs), frequently touted as green alternatives to larger projects (Couto and Olden 2018), tend to be built across small and medium sized streams (Kibler and Tullos 2013). Usually defined by their power generation capacity, SHPs vary tremendously in definition across countries (from up to 1 MW to up to 50 MW), in size (i.e. variable dam heights, reservoir areas and storage capabilities), and in mode of operation (with or without storage and diversion) (Couto and Olden 2018). Hence, the impact of a single SHP on fragmentation and flow alteration can vary considerably based on the attributes of individual projects and their location in the river network (figure 2). Additionally, due to fewer regulations, numerous SHPs are often commissioned along a single river, leading to substantial cumulative impacts (Kibler and Tullos 2013). SHPs impede river longitudinal connectivity due to the barrier effect, which is exacerbated by the clustering of numerous SHPs on the same river channel. Although SHPs tend to have smaller storage capacities relative to large dams, their impact on the extent of flow alteration can vary based on their location, design, and operating regimes (Timpe and Kaplan 2017). In terms of design, SHPs that store and divert water from a weir to a downstream powerhouse result in the creation of dewatered river stretches, which reduce longitudinal, lateral, and vertical connectivity (Anderson et al 2006, Jumani et al 2018). Comparatively, SHPs that do not store and divert water may have a smaller impact on flow alteration. In terms of operations, continued storage and release operations (commonly employed by SHPs with storage) result in rapidly fluctuating/flashy flows downstream.
Low-head dams and other small RIPs built to facilitate infiltration or water diversion usually cluster closer to the headwater tributaries and result in smaller fragment sizes. While the impact of individual projects might be low, the cumulative fragmentation effects of numerous small RIPs can be significant (Januchowski-Hartley et al 2013). Often designed with very little active storage, these structures often allow for some movement of water and sediment and are expected to have lower individual impacts on flow alteration. Furthermore, their impact on flow regulation can be expected to vary based on the presence or absence of water abstraction (figure 2).
Figure 2 illustrates the major axes of hydrologic fragmentation and alteration, allowing us to coarsely map the expected impacts of different RIPs. However, moving from this conceptual model to a quantitative understanding of connectivity and flow regime alteration requires an understanding of the types of tools and methods available to do so, as well as their specific outputs and data requirements. In the following section, we review the metrics and tools available for quantifying river fragmentation and flow alteration, and in section 4 we provide guidance for selecting the most appropriate tool as a function of the study objective and data availability.
3. Methods to assess river fragmentation and flow alteration
We compiled key readings on the theory, concepts, and methods associated with river network connectivity and the natural flow regime. Thematic searches of published, peer-reviewed literature using topic-relevant keywords were conducted on Google Scholar. Key words used included 'river connectivity', 'river fragmentation', 'dendritic connectivity', 'hydrologic connectivity', 'dam fragmentation', 'metrics of flow alteration', 'flow regulation', and 'hydrologic alteration'. Additionally, personal reference libraries and the bibliography of selected papers were also reviewed to find related and relevant publications. This resulted in the final selection of 79 publications. Papers that did not define or apply a specific metric were excluded from the review. With respect to river fragmentation, we only selected papers focused on instream riverine connectivity and excluded those dealing with overland hydrologic connections (Pringle 2001). Similarly, for flow alteration, we selected papers that quantified the extent of alteration (descriptive metrics) and excluded those aimed at prescribing environmental flows (prescriptive methods).
3.1. Metrics of river fragmentation
Our review resulted in a compilation of 29 metrics or methods to quantify river network connectivity or fragmentation (table 1). Following the classification by Calabrese and Fagan (2004), we grouped these metrics into three categories based on whether they estimate structural, potential, or actual connectivity. Structural connectivity metrics are calculated based on the physical attributes and spatial configuration of the riverscape; potential connectivity metrics combine information describing an ecosystem process or organism dispersal abilities along with information on the structural or physical attributes of the riverscape; actual connectivity metrics are based on a measured ecosystem process or the observed movement of individuals along the spatial configuration of the river (Kindlmann and Burel 2008). Hence, potential and actual connectivity metrics will vary based on the target taxa or phenomenon being considered and the spatiotemporal scales at which they occur (Fullerton et al 2010). Table 1 summarises the description, data requirements, output, spatial scale of application, and advantages and disadvantages of each method.
Table 1. List of river connectivity or fragmentation metrics with their description, data requirements, outputs, spatial scale of application, and advantages and disadvantages.
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Structural connectivity metrics | |||||||
| Between Centrality (Freeman 1977) | Reflects the importance of each stream reach in maintaining connections between all other pairs of stream reaches in a riverscape. |
| Reaches ranked by their importance in maintaining basin-wide connectivity | Stream reach |
|
| Bodin and Saura 2010; Segurado et al 2013 |
| Lateral connectivity classes (Amoros et al 1987) | Descriptive classes of lateral connectivity (0–5) between the main channel and side channels |
| Five lateral connectivity classes (5–0 indicating completely connected to isolated) | Waterbodies/side channels |
|
| Lasne et al 2007 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Fragmentation classes (Nilsson et al 2005) | A descriptive measure based on the longest undammed length of the main river channel in relation to the entire channel length. |
| Five fragmentation classes (very low to very high) | Sub-basin to basin |
|
| Díaz et al 2019 |
| Barrier density (Park et al 2008) | A descriptive measure calculated as the total number of barriers per total river length |
| Density of barriers per length of river | River reach to basin |
|
| Jones et al 2019; Atkinson et al 2020 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Continuity index (Prato, Comoglio, and Calles 2011) | A descriptive measure calculated as the ratio of total river length to the number of obstacles |
| Ratio of total river length to the number of obstacles | River reach to river network |
|
| Prato et al 2011 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Total remaining core length (Fuller et al 2015) | The length of unaffected core habitat for a specific species or guild, calculated as the difference between the total network length and the length of river affected by fragmentation (sum of upstream and downstream matrix and edge habitats created by each barrier in the network) |
| Total remaining core length | Sub-basin to basin |
|
| Hall et al 2011 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Dam Impact Index (Latrubesse et al 2017) | An index calculated from (i) the ratio of river length affected by dams, (ii) ratio of number of major tributaries affected by dams, and (iii) number of dams per basin/sub-basin |
| Index of impact from 0 to 100 | Sub-basin to basin |
|
| Latrubesse et al 2017 |
| River channel connectivity index (Li et al 2018) | Quantifies the unobstructed degree of river flow based on the concept of time accessibility. It is calculated as the ratio of the time accessibility of a given volume of streamflow without any barriers to that with barriers from one location to another in the river channel |
| Index ranging from 0 (disconnected) to 1 (connected) | River reach to tributary |
|
| Li et al 2018 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Potential Connectivity Metrics | |||||||
| Integral index of connectivity (IIC) (Pascual-Hortal and Saura 2006) | A habitat reachability index based on habitat availability and binary connectivity values for a target taxa or guild. It assesses the possibility of dispersal between all pairs of stream reaches based on topological distances |
| Index of connectivity ranging from 0 to 1 | Subbasin to basin |
|
| Segurado et al 2013; Branco et al 2014; Lehotský et al 2018 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Probability of Connectivity (PC) (Saura and Pascual-Hortal 2007) | A habitat reachability index, like the IIC, that assesses the probabilities of dispersal between all pairs of patches or stream reaches. Connectivity is not restricted to binary values. |
| Index of connectivity ranging from 0 to 1 | Subbasin to basin |
|
| Bodin and Saura 2010; Malvadkar et al 2015 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| PCA lateral connectivity metric based on environmental variables (PCA-LC) (Paillex et al 2007) | A surrogate measure of lateral connectivity. Five environmental variables are summarized with a centred principal component analysis to produce a factorial axis that is used as the synthetic variable for the level of connectivity between the main river channel and the cut-off channels. |
| Site scores along the primary PCA factorial axis, with increasing values corresponding to increasing connectivity | Sites from which data have been gathered |
|
| Besacier-Monbertrand et al 2014 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Dendritic Connectivity Index—potadromous (Cote et al 2009) | An index of connectivity calculated from stream length, which assesses the potential of a potadromous fish to travel between two chosen points in a river network. Based on coincidence probability (Jaeger 2000) |
| An index of connectivity ranging from 0 to 100 | River reach to basin |
|
| Perkin and Gido 2012; Anderson et al 2018 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Dendritic Connectivity Index—diadromous (Cote et al 2009) | An index of connectivity calculated from stream length, which assesses the proportion of river length accessible to a diadromous fish from the mouth of a river |
| An index of connectivity ranging from 0 to 100 | River reach to basin |
|
| Buddendorf et al 2017; Choy et al 2018 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Index of longitudinal riverine connectivity (ILRC) (Crook et al 2009) | evaluate both upstream and downstream effects of dams and water withdrawals on longitudinal connectivity in tropical streams evaluate both upstream and downstream effects of dams and water withdrawals on longitudinal connectivity in tropical streams evaluate both upstream and downstream effects of dams and water withdrawals on longitudinal connectivity in tropical streams Estimates probability that an individual shrimp larva can migrate downstream to the estuary (based on proportion of median flow left in the stream after withdrawal) and return to the reach where it was released as a larva (based on proportion of days with flow over the impoundment) | stimate the probability that an
individual 'average' shrimp will be able to migrate downstream
to the estuary and return to the reach where it was released as a
larva
stimate the probability that an
individual 'average' shrimp will be able to migrate downstream
to the estuary and return to the reach where it was released as a
larva
estimate the probability that an
individual 'average' shrimp will be able to migrate downstream
to the estuary and return to the reach where it was released as a
larva.
| Index ranging between 0 and 1, split into three classes (high, moderate and low for ILRC scores of 0–0.33, 0.34–0.66, and 0.67–1 respectively) | Each water intake structure | The effect of water withdrawal on juvenile shrimps is
influenced by individual intakes in addition to all downstream
intakes. Where there are intakes in linear succession, juvenile
shrimps may have to climb past all intakes in order to reach
their ultimate habitat. In order to account for the lower
probability that an individual juvenile shrimp will successfully
scale multiple intakes, the proportion of days with flow for any
downstream intake is multiplied by the proportion of days
with flow for any upstream intake
|
| Crook et al 2009 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Between Centrality—k (Bodin and Saura 2010) | A modified BC metric that weighs each stream reach by its patch area and maximum dispersal probabilities or topological distances (based on whether PC or IIC metric is used). |
| Reaches ranked by their importance in maintaining basin-wide connectivity | River reach |
|
| Segurado et al 2013 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| DEN connectivity model (Padgham and Webb 2010) | Represents the ability of a fish to access different parts of a river network. Model based on habitat length, quality, and directional transition probabilities. |
| Matrix of transition probabilities between every pair of reaches in a network + reach scores that indicate equilibrium proportions of a population expected within each reach | River reach to network |
|
| Webb and Padgham 2013 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Barrier score (Nunn and Cowx 2012) | Each barrier is scored based on a prioritization matrix of fish stock status, passage efficiency, likelihood of access, habitat quantity and habitat quality | Scores (1–5) for:
| Barrier scores ranging from 1 to 3125 | Each barrier |
|
| Nunn and Cowx 2012 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Habitat Connectivity Index of Upstream passage (HCIUP) (Mckay et al 2013) | Assesses upstream fish passage connectivity as a habitat-weighted, cumulative passage rate. By summing across all reaches, the HCIUP is computed as the ratio of accessible to total habitat in the river network |
| Ratio of accessible habitat ranging from 0 to 1 | Sub-basin to basin |
|
| Rodeles et al 2019 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| River Connectivity Index (RCI) (Grill et al 2014) | An index of connectivity calculated from river flow volume; like DCI, it assesses the potential of a fish to travel between two chosen points in a river network. |
| An index of connectivity ranging from 0 to 100 | River reach to basin |
|
| Grill et al 2015 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Weighted River Connectivity Index (Grill et al 2014) | An index of connectivity calculated from river flow volume and weighted by ecologically meaningful variables such as river class/ecoregion (RCIclass) or species-specific migration ranges (RCIrange). |
| An index of connectivity ranging from 0 to 100 | River reach to basin |
|
| Grill et al 2014 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| C metric (Diebel et al 2015) | Defines the connectivity of a stream reach as a function of the degree of access to and from the range of seasonal habitat types that fish use. The 'C' values for all the segments in a watershed can be aggregated to describe connectivity at the watershed scale |
| Connectivity status ranging from 0 to 1 | Reach and watershed level |
|
| O'Hanley et al 2013 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Fragmentation index (Díaz et al 2019) | A fragmentation index calculated from stream length and Strahler stream order |
| Fragmentation index between 0 to 1 | Sub-basin to basin |
|
| Díaz et al 2019 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Metapopulation models of directed connectivity | A migration model for metapopulation connectivity of salmon (or any other diadromous species) |
| Diagraphs of spatially explicit populations under various scenarios of development (with population size and connectivity strength and direction illustrated) | Basin scale |
|
| Isaak et al 2007; Schick and Lindley 2007; Leibowitz and White 2009 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Actual connectivity metrics | |||||||
| Lateral connectivity parameter (Cd) (Reckendorfer et al 2006) | A connectivity parameter (Cd) defined as the average annual duration (days per year) of surface connection of floodplain waterbodies with the main river channel |
| Cd values for each waterbody | Waterbodies/side channels |
|
| Reckendorfer et al 2006 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Human observations of movement | Location of target species/taxa over the study area |
| Variable | River reach to tributary |
|
| Johnston 2000 |
| Bio-acoustic/hydroacoustic sonar | Measurement of fish locations, densities, and movement using fixed or mobile acoustic sensors |
| Variable | River reach to river network |
|
| Burwen et al 2005; Dey et al 2019 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Telemetry | Movement information of tagged individuals over space and time |
| Variable | River reach to river network |
|
| Schrank and Rahel 2004; Gosset et al 2006 |
| Direct sampling (electroshocking, seining or trapping) | In-situ fish capture | Spatially explicit information on:
| Presence-absence data, composition similarity, richness, diversity, abundance and density estimates | River reach to river network |
|
| Merritt and Wohl 2006; Alexandre and Almeida 2010; Jumani et al 2018 |
(Continued)
Table 1. (Continued).
| Connectivity/fragmentation metric | Description | Inputs/Data requirements | Outputs | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Molecular or genetic markers (such as DNA microsatellites) | Genetic material extracted from tissue samples |
| Genetic variation, diversity, differentiation or similarity | Subbasin to basin |
|
| Wofford et al 2005; Faulks et al 2011; Torterotot et al 2014 |
anot essential data requirements
3.2. Metrics of flow alteration
Methods to assess flow alteration can be descriptive or prescriptive in their application. Descriptive metrics are those that quantify or measure flow alteration (i.e. how have riverine flows been altered compared to baseline undisturbed conditions?); prescriptive methods are those aimed at determining environmental flow requirements (i.e. how much water can be extracted or used while still maintaining ecosystem processes and functions?) and usually incorporate one or more descriptive metrics. While the former is often quantified based on scientific data input, the latter is management-oriented and influenced by socio-cultural, economic, and political drivers. This review focuses only on descriptive metrics, as numerous reviews of the application of prescriptive environmental flow methodologies already exist (Jowett 1997, King et al 1999, Tharme 2003, Acreman and Dunbar 2004, Hirji and Davis 2009, Horne 2017). Table 2 summarises the description, data requirements, output, spatial scale of application, and advantages and disadvantages of the 12 main descriptive flow alteration metrics.
Table 2. List of flow alteration metrics with their description, data requirements, output, spatial scale of application, and advantages and disadvantages.
| Flow alteration metric | Description | Data requirements | Output | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Annual proportional flow deviation (APFD) (Gehrke et al 1995) | Comparison of post-impact and unimpacted baseline monthly flows, calculated as the sum of the ratios of change in monthly flow (actual—natural) to natural monthly flow | Short-term (1–5 years) monthly flow data across un-impacted and impacted spatial or temporal scales | APFD values ranging from 0 (unregulated river) to 3.46 (where there is a 100% increase or decrease in flow with no seasonal change) | River reach from which hydrological data have been gathered |
|
| Ladson et al 1999 |
(Continued)
Table 2. (Continued).
| Flow alteration metric | Description | Data requirements | Output | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Indicators of hydrologic alteration (IHA) (Richter et al 1996) | Quantifies the ecohydrological effects of flow regulation by measuring changes in 33 flow statistics, organized within the five primary components of flow regime (flow magnitude, frequency, duration, timing, and rate of change) | Time-series of daily streamflow data | Measures of central tendency and dispersion for 33 hydrologic parameters (i.e. 66 inter-annual statistics) | River reach from which hydrological data have been gathered |
|
| Mathews and Richter 2007; Timpe and Kaplan 2017 |
(Continued)
Table 2. (Continued).
| Flow alteration metric | Description | Data requirements | Output | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Range of variability approach (RVA) (Richter et al 1997) | Quantifies the change in the range of variation of 33 IHA parameters from the pre-impact period to the post-impact period. Each parameter is categorised into high, medium or low categories based on user-defined targets, and a hydrologic alteration category is calculated based on relative frequency of the RVA target range not attained | Time-series of daily streamflow data | Hydrologic alteration category for each of the 33 parameters based on the percentage of years the RVA target range is not attained, expressed as high, medium and low (with hydrologic alteration values of 68%–100%, 34%–67%, and 0%–33% respectively) | River reach from which hydrological data have been gathered |
|
| Richter et al 1998; Mittal et al 2014 |
(Continued)
Table 2. (Continued).
| Flow alteration metric | Description | Data requirements | Output | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Degree of regulation (DOR) (Lehner et al 2011) | Calculates the proportion of a river's annual flow that can be withheld by a reservoir or a cluster of reservoirs for a river reach | Reservoir storage capacities and annual discharge | A continuous index of proportions | River reach to river network |
|
| Grill et al 2019 |
(Continued)
Table 2. (Continued).
| Flow alteration metric | Description | Data requirements | Output | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Dundee Hydrological Regime Alteration Method (DHRAM) (Black et al 2005) | Applies the IHA approach to classify the risk of damage to instream ecology from streamflow alterations using a five-class scheme compatible with the requirements of the EC Water Framework Directive | Time-series of daily mean flow in un-impacted and impacted sites in relation to any type of anthropogenic hydrological impact | DHRAM scores (0–30) and DHRAM classes between 1 (Un-impacted condition) and 5 (Severely impacted condition) | River reach from which hydrological data have been gathered |
|
| Gao et al 2009 |
(Continued)
Table 2. (Continued).
| Flow alteration metric | Description | Data requirements | Output | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Hydroecological Integrity Assessment Process (HIAP) (Henriksen et al 2006) | Uses a Hydrologic Index Tool to calculate 171 streamflow statistics and a Hydrologic Assessment Tool to determine the degree of departure from baseline conditions | Time-series of daily mean flow and peak flow data | 171 biologically relevant streamflow statistics for baseline and altered condition | River reach from which hydrological data have been gathered |
|
| Kennen et al 2009 |
(Continued)
Table 2. (Continued).
| Flow alteration metric | Description | Data requirements | Output | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Environmental flow components (EFC) (Mathews and Richter 2007) | Quantifies changes in 34 flow statistics organized within five major ecologically important flow components: low flows, extreme low flows, high flow pulses, small floods, and large floods. | Time-series of daily streamflow data | Measures of central tendency and dispersion for 34 environmental flow component parameters | River reach from which hydrological data have been gathered |
|
| Morid et al 2019 |
(Continued)
Table 2. (Continued).
| Flow alteration metric | Description | Data requirements | Output | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Overall degree of hydrologic alteration (Shiau and Wu 2007) | An index of overall flow regulation based on the integration of individual degree of hydrologic alteration for each of the 33 hydrologic parameters of the IHA | Time-series of daily streamflow data | Percentage indicating overall flow regulation | River reach from which hydrological data have been gathered |
|
| Shiau and Wu 2007 |
(Continued)
Table 2. (Continued).
| Flow alteration metric | Description | Data requirements | Output | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Ecodeficit/ecosurplus concept (Vogel et al 2007) | Nondimensional metric, based on a flow duration curve (FDC), which represents the deficit or surplus streamflow resulting from flow alteration, as a fraction of the mean streamflow in a typical or median year | Unimpacted and impacted FDCs (or water resource index duration curves) for a period of record or a median annual year | Quantification of difference in the net volume of water available to meet instream flow requirements | River reach from which hydrological data have been gathered |
|
| Gao et al 2009 |
(Continued)
Table 2. (Continued).
| Flow alteration metric | Description | Data requirements | Output | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Flow-ecology response curves (part of numerous eflow assessments such as ELOHA, DRIFT) (Poff et al 2010) | Combines hydrology, channel hydraulics, ecology and social processes to build mechanistic links between hydrology and ecology through flow-ecology response curves based on river type. | Time-series of flow data to build the 'hydrologic foundation' of baseline and present-day hydrographs; ecological data and expert opinion to create flow-ecology response curves | Flow-ecology response curves for classified rivers across a broad area | River reach from which hydrological data have been gathered |
|
| Mcclain et al 2014; Cartwright et al 2017; Rosenfeld 2017 |
(Continued)
Table 2. (Continued).
| Flow alteration metric | Description | Data requirements | Output | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| River Regulation Index (RRI) (Grill et al 2014) | Quantifies how strongly a river may be affected by flow alterations from upstream dams | Reservoir storage capacities and annual discharge (measured or estimated) | Continuous index of proportions | River basin |
|
| Grill et al 2015 |
(Continued)
Table 2. (Continued).
| Flow alteration metric | Description | Data requirements | Output | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Effective Degree of Regulation (EDOR) (Ehsani et al 2017) | Ratio of volume of water that is displaced (stored or released) by the operation of a dam or a cluster of dams, to the river's naturalized flow without dams | Reservoir storage capacities and annual discharge (measured or estimated) Reservoir operation (volume of water released and stored) | Continuous index of proportions | River reach to river network |
|
| Ehsani et al 2017 |
(Continued)
Table 2. (Continued).
| Flow alteration metric | Description | Data requirements | Output | Spatial scale of application | Advantages | Disadvantages | Applications |
|---|---|---|---|---|---|---|---|
| Statistical models of the counterfactual (Valle and Kaplan 2019) | Model-based evaluation of how post-dam hydrology differs from 'what would have happened' in the absence of the impact | Time-series of hourly, daily, or monthly flow or water level data and other relevant hydroclimate data (water levels, flows, precipitation, etc) | Magnitude and statistical likelihood of difference between each post-impact observation and the expected pre-impact value | River reach from which hydrological data have been gathered |
|
| Valle and Kaplan 2019 |
4. Decision support
4.1. River connectivity metrics
Although connectivity in river networks has been less studied compared to their terrestrial counterparts, we documented 29 different methods to quantify river connectivity or fragmentation from the scientific literature (table 1). These methods vary considerably in their data requirements, spatial scale of application, and output, each having their own assumptions, advantages, and disadvantages.
Figure 3 presents a decision-making tree to help identify connectivity metrics that can be used based on the study objective, data availability, and distribution of infrastructure projects in the river basin of interest. This decision tree, when used with the information in table 1, allows users to make informed decisions when selecting among the connectivity measures available and to design impact studies with an eye toward quantifying specific outcomes. For example, when assessing the impact of fragmentation on biotic communities, in a case where little or no empirical data are available on the species/taxa of interest, the decision tree presents 16 available structural and potential connectivity metrics to choose from. Similarly, when assessing the impact of fragmentation on basin-wide processes, users can select among 11 different structural, potential, and actual measures (figure 3).
Figure 3. Decision-making tree for selection of river connectivity metrics based on objective, data availability, and distribution of infrastructure projects in the basin. Superscripts indicate barrier passability values to be binary (b) or continuous (c); The symbol indicates method designed to assess connectivity for fish communities. Colours blue, black, and orange indicate structural, potential, and actual connectivity metrics, respectively.
Download figure:
Standard image High-resolution imageWhen reviewing these methods holistically, a clear trade-off emerges between data availability and the type of connectivity that can be assessed. While actual connectivity metrics yield the most direct and reliable measure of connectivity, their application across spatial scales is often limited by the availability of field data. Nevertheless, these methods can be effectively applied at finer spatial scales to address specific objectives. For example, actual connectivity metrics are ideal to assess the efficacy of fish passes (Oldani and Baigún 2002, Knaepkens et al 2006, Naughton et al 2007), species responses to dam removals (Liermann et al 2017), or the restoration of specific migration pathways (Beasley and Hightower 2000). Among the actual connectivity metrics, only genetic or molecular techniques provide information across extended temporal scales, whereas other methods usually quantify short-term dispersal during the period of data availability.
In contrast, structural connectivity indices are not data-intensive and can be calculated with relative ease across broader spatial scales. However, they provide only a crude estimate of connectivity, which may or may not reflect actual conditions at the scale of their application (Mahlum et al 2014). Given these drawbacks, potential connectivity metrics present a more suitable choice in the absence of empirical data. These metrics can be informed by secondary information on ecological or biotic requirements (such as dispersal probabilities or habitat requirements) and can be used to calculate potential connectivity across broad spatial scales with relative ease. Often, structural connectivity metrics have been modified or adapted to suit research needs and data availability. For example, the Dendritic Connectivity Index (Cote et al 2009) has been used as the basis for other derivative connectivity metrics, such as the River Connectivity Index (Grill et al 2014) and the Fragmentation Index (Díaz et al 2019). Similarly, several structural connectivity metrics can be modified to incorporate additional information to become more ecologically meaningful. For example, river lengths can be weighted based on habitat quality or habitat preference of target taxa (Grill et al 2014, Buddendorf et al 2017).Likewise, for structural metrics that treat all river reaches as equal, increasing weights can be assigned to higher stream orders or increasing river widths based on ecological considerations and scale of analysis (Díaz et al 2019).
When assessing connectivity with respect to a target species or guild, their behaviour, life history, and resource requirements (especially directionality of movement, dispersal distances, and migration) should influence metric selection. For example, when assessing connectivity for diadromous species, the DCI-d, Weighted RCI, ILRC, DEN connectivity model, HCIUP, or the Metapopulation model of directed connectivity could be applied. Metric selection should also be informed by the distribution of RIPs in the study area. Some metrics, such as Fragmentation classes, DCI-d, BC-k, and HCIUP, may not reflect any change with the addition or removal of dams because of the way they are defined. For example, when using Fragmentation classes (Nilsson et al 2005), dammed large tributaries are assigned a fragmentation score of 2. This score remains the same irrespective of the number of dams present. Similarly, when applying the DCI-d (Cote et al 2009) at the scale of the river network with binary dam passabilities, the addition or removal of dams above the first barrier will not alter the index value. Hence, these metrics should only be used in specific instances where applicable. Another consideration for river length-dependent metrics is the presence of RIPs on seasonal headwater streams. Often such dams lie beyond the delineated stream network and are consequently excluded from the analysis (as done in Hoenke et al 2014, Anderson et al 2018). Hence, when numerous RIPs are situated in headwater streams or when connectivity in headwater reaches needs to be specifically assessed, these metrics should be used with caution.
An important application of fragmentation metrics is the optimization of barrier removal or placement to maximise connectivity for a target species or taxa (Mckay et al 2017). The reliability and ecological significance of the connectivity metric used in these applications are crucial, and hence the use of structural metrics should be avoided in these cases. When more reliable metrics are unavailable due to data limitations, all attempts should be made to validate structural metrics with empirical field data and determine their spatial scale of influence. Another point of consideration is that structural and potential metrics that rely on river network lengths are prone to non-uniform change based on the extent of the river network delineated, which itself is dependent on the resolution of the base data, delineation techniques and thresholds used (Zhou and Liu 2002, Murphy et al 2008, Ariza-Villaverde, Jiménez-Hornero, and Gutiérrez de Ravé 2015, Kumar et al 2017). It is important to note that these changes are an artefact of changing river network lengths and do not signify a change in actual connectivity.
4.2. Flow alteration metrics
A vast majority of research related to flow alteration caused by RIPs is prescriptive and mostly aimed at recommending environmental flows in regulated streams (Hirji and Davis 2009, Poff et al 2010, Horne 2017). These approaches have been well studied and reviewed in the scientific literature, but comparatively far fewer descriptive measures of flow alteration exist. Descriptive measures allow users to assess the extent of alteration of a river's natural flow regime in response to various anthropogenic influences relative to undisturbed baseline conditions. Since streamflow is a master variable influencing water quality, physical habitat characteristics, ecosystem functions and processes, and native biotic communities (Poff et al 1997), quantifying the extent of flow alteration has important implications for basin-wide conservation and development planning, and for setting suitable environmental flow recommendations.
Our review documented 13 descriptive measures of flow alteration. These methods vary in their data requirements, spatial scale of application, and output, each having their own assumptions, advantages and disadvantages (table 2). Figure 4 presents a decision-making tree to help users select a suitable method to assess flow alteration given the availability of streamflow, reservoir storage and discharge data, and specific objective. This decision tree, when used with the information in table 2, can allow users to make informed decisions about the types of flow alteration measures that can be quantified in different contexts. For example, when long-term observed or simulated streamflow data are available, we identified 10 available methods to assess flow alteration. Of these, the IHA, RVA, DHRAM, EFC, and HIAP quantify the degree to which different flow components (i.e. flow magnitude, frequency, duration, timing, and rate of change) are affected. This contrasts with the APFD, Overall Degree of Hydrologic Alteration and Ecodeficit/Ecosurplus methods which quantify the extent of flow alteration over a given time scale. Flow-ecology response curves and statistical models of the counterfactual can be used to assess both the alteration to various flow components and overall flow over a period of time (figure 4).
Figure 4. Decision-making tree for selection of flow alteration metric to be used based on data availability. Temporal scale of streamflow data required: sd = sub-daily, d = daily, m = monthly. # Incorporates ecological data of interest to build flow-ecology curves
Download figure:
Standard image High-resolution image
When long-term streamflow data are unavailable, as is the case in numerous developing countries witnessing a surge in dam development, we identified only three possible methods- DOR, RRI, and EDOR- that require data on reservoir storage capacities and annual discharge. Though these metrics are useful in data-deficit regions and are easy to calculate, they provide no insight on how various components of the flow regime are affected over time. They also do not consider the impacts of water abstraction (except for EDOR), diversion, and dewatering of river channels. This is especially problematic if the study area has numerous small or low-head dams with little or no reservoir storage. Furthermore, the impacts of flow regulation as measured by these methods can manifest differently in differently sized streams, despite having the same numerical values. To this end, reach-scale classification through characteristics such as geomorphic features may be useful to relate ecological relationships in regions with deficient streamflow records (Poff et al 2010). Complex alphanumeric classification methods (e.g. Rosgen 1994) may prove overly cumbersome to relate channel features to ecological systems (Simon et al 2007); simpler geomorphic classifications that describe variations in processes such as sediment mobility and stream power (Montgomery and Buffington 1997, Poff et al 2006) can explain ecological relationships where streamflow alteration cannot be assessed. Such classification methods could be useful for relating effects of river discharge to ungauged streams.
All but one of the above methods utilise streamflow, reservoir storage and/or discharge data to calculate various flow statistics without relating to an ecological response. If users need a metric that links flow alteration to an ecological response, the flow-ecology response curve is a versatile approach that can be used to assess the extent of change to one or more flow components or overall flow in relation to ecological responses of interest (Poff et al 2010).
Overall, the efficacy of all the connectivity and flow alteration metrics listed above will be greatly influenced by spatiotemporal extent and resolution of input data (Murphy et al 2008, Yang et al 2014, Woodrow et al 2016), uncertainties or errors associated with modelled or simulated data (Bond and Kennard 2017), and compatibility of scale of response and scale of analysis (Gaucherel 2007, Mahlum et al 2014). While no single method can be all-encompassing, the selection of appropriate connectivity and flow alteration metrics should be carefully made based on the study objective, data availability, and a thorough knowledge of the assumptions, advantages, and disadvantages of the available methods.
5. Applications and directions for furture research
The methodological advancements in characterizing river fragmentation and flow alteration described above provide a wide variety of tools for researchers and resource managers to understand the effects of RIPs on river ecosystems. A range of these metrics can be effectively applied to guide monitoring and adaptive management programs aimed at maximising riverine and ecological connectivity and restoring or maintaining the natural flow regime under various scenarios of existing and proposed RIP development. They can also be applied to identify priority reaches for the implementation of mitigation measures, and aid in the creation of basin-wide conservation and development plans not only after but also before projects are implemented. The growth and utility of such tools have coincided with widely available resources to facilitate analysis: increasing access to GIS and computational capabilities (such as FIPEX (Fisheries and Oceans Canada 2011), FIDIMO (Radinger et al 2014)), online repositories of dams (such as GRAND (Lehner et al 2011), GOODD (Mulligan, van Soesbergen, and Sáenz 2020), FHReD (Zarfl et al 2015), spatial datasets of hydrologic networks (such as HydroSHEDS (Lehner et al 2008), HydroBASINS (Lehner and Grill 2013) and streamflow data (GSCD (Beck et al 2013); GRDC (http://grdc.bafg.de), FLO1K (Barbarossa et al 2018); RiverATLAS (Linke et al 2019)) has made several fragmentation and flow alteration indices more readily applicable across larger spatial scales. Despite these advances, there remain numerous areas for further research to improve the performances of these metrics, especially given their applications in basin-wide conservation and development planning. These are briefly discussed below.
5.1. Relationships among metrics
Although river connectivity and flow alteration characterize two different types of variables, because flows control hydrologic connectivity, the two variables often interact and influence one another (Grill et al 2014). For example, dam-induced flow alterations can result in reduced wetted channel widths and/or depths, which can affect lateral and vertical connectivity (Junk et al 1989, Wiens 2002). Water abstraction and diversions can create dewatered river stretches which impede water-mediated longitudinal connectivity (Deitch, Kondolf, and Merenlender 2009). Large reservoirs can significantly alter thermal regimes, which can further act as a thermal barrier to various organisms (Caudill et al 2013).While most measures of connectivity focus on the longitudinal dimension, far fewer metrics are aimed at assessing lateral and vertical connectivity. Additionally, since river connectivity is water-mediated, the force and direction of flow exerts a strong influence on ecological connectivity and ecosystem processes such as transport of sediment, nutrients, and organisms with limited or no mobility (Fullerton et al 2010). Hence, connectivity measures that do not account for flow can be misleading in terms of ecological connectivity. In order to address these issues, future research should be aimed at developing methods that (a) measure the interactions between connectivity and flow alteration and metrics within each category, (b) measure lateral and vertical connectivity, and (c) incorporate the effects of flow within connectivity metrics,. Understanding relationships among connectivity and flow alteration metrics can provide additional insights regarding the effects of RIPs on stream ecosystems over space and time. From a management perspective, connectivity metrics could be combined with flow alteration metrics to inform prescriptive tools for maintaining environmental flows. In regions where time or resources are limited, relationships between metrics that require extensive data collection (such as actual connectivity or flow alteration methods that require streamflow data) and metrics that do not (such as structural connectivity indices and flow alteration methods that do not require streamflow data) may be useful for extrapolating actual connectivity more broadly in a region, or for understanding conditions where metrics diverge.
5.2. Ecological significance
The actual ecological relevance of most flow alteration and connectivity metrics remains largely unknown. This is especially true for structural connectivity indices that are gaining rapid popularity and widespread implementation (Perkin et al 2015, Anderson et al 2018). Despite this knowledge gap, numerous assessments and prescriptive documents use connectivity indices to prioritize barrier removal, under the assumption that an increase in connectivity (as defined by a particular index) will improve biotic communities (Bourne et al 2011, Perkin et al 2015). Similarly, the ecological relevance of most flow alteration indices has not been adequately studied. Since ecological responses are expected to be influenced not only by connectivity and flow alteration metrics, but also by other environmental factors and the behaviour and resource requirements of the target species or taxa, assessing these relationships across river classes (Dallaire et al 2019) become essential. Hence, rigorous field studies that quantify the association between these metrics and biotic communities (such as fish (Perkin and Gido 2012, Mahlum et al 2014), macroinvertebrates (Solans and Jalón 2016), and riparian vegetation (Mcmanamay et al 2013)) and/or ecosystem processes and functions (such as sediment transport and primary productivity(Yarnell et al 2015)) are an important area for further research. Such empirical studies can not only inform the ecological relevance of connectivity and flow alteration measures, but can also shed light on how behavioural components influence ecological connectivity across spatial and temporal scale (Fullerton et al 2010).
5.3. Spatial and temporal scales of application
The ecological utility of a connectivity or flow alteration index will depend its spatiotemporal scale of application and the species, assemblage or ecosystem process being considered (Crooks and Sanjayan 2006, Gaucherel 2007, Llausàs and Nogué 2012). Since different species perceive habitats at different spatial scales across their life-history stages, their response to fragmentation and flow alteration will likely be scale-dependent, and also influenced by their habitat and resource requirements (Rossi and van Halder 2010, Llausàs and Joan 2012). Generally, as spatial scales of analysis increases, other confounding landscape-level variables (such as elevation, land use, discharge) begin to influence response communities (Mahlum et al 2014). The application of spatial graph and network models across hierarchical river networks presents an opportunity to better understand factors influencing ecological connectivity across spatial scales (Erős and Lowe 2019). Similarly, due to temporal shifts in streamflow, ecosystem processes, and species life-history stages, ecological connectivity and flow alteration need to be assessed over adequate temporal (or seasonal) scales based on the ecological response being considered to avoid misrepresentation of results (Fullerton et al 2010). While it may not be feasible to quantify connectivity across all spatiotemporal scales, it is essential that further research be aimed at identifying the range of scales over which connectivity and flow alteration metrics may influence populations or processes of interest (Fullerton et al 2010).
5.4. Applications in data-scarce regions
One of the greatest challenges in understanding the effect of RIPs on connectivity and flow alteration is the effective application of informative indices in data-scarce regions. Most tropical developing countries striving to recognise their hydropower potential are characterised by high levels of freshwater biodiversity and the presence of river-dependent local communities (Auerbach et al 2016). These regions are also often limited in terms of long-term hydrologic and ecological data availability. Hence, despite there being a strong need for science-based management and decision-making, the lack of available resources precludes effective assessments of existing and proposed RIPs across spatiotemporal scales. The development of ecologically meaningful measures of connectivity and flow alteration that can be applied in such data-deficit regions to aid monitoring, restoration, and conservation development efforts remains a vital research frontier. Additionally, concerted efforts to establish partnerships and collaborations between governments, project proponents, scientists, water-managers and NGOs can go a long way in improving hydrologic data availability, which can then aid in informing water management policy and decision-making. Similar collaborations to establish a network of gaging stations and collect periodic data on river habitat variables and biotic communities can provide the foundation required to apply more sophisticated and informative methods to assess the impacts of RIPs and create basin-wide monitoring and conservation plans (Horne 2017).
Conclusion
Continued demand for non-fossil fuel-based energy and water supply to meet the needs of growing human populations and zero-emission power will likely contribute to increasing reliance on RIPs through the 21st century. While the impacts of these projects on aquatic, riparian, and terrestrial ecosystems may be profound, tools to evaluate or predict the effects of RIPs on river ecosystems can provide critical information for conservation and management to mitigate their impacts in the future. Resource managers across the globe, over a wide range of technical capacities, need to understand the tools that are available for analysing how RIPs alter connectivity and streamflow. To this end, decision support remains one of the most important contributions that hydrologists and ecologists can make to sustain aquatic ecosystems.
Our review highlights the substantial progress toward understanding the hydrological consequences of RIPs, yet significant gaps remain. The recent proliferation of research using remotely sensed metrics to evaluate river network fragmentation and flow alteration highlights the potential for remote sensing to support applications including comparisons across broad regions and predictions of future impacts, but it also underscores their limitations. Without organism-based, field-based data collection, the ecological meaning of such metrics is unsupported. Assessments of actual ecological impacts will require extensive measurement of factors such as presence and absence (and changes over time), movement, and dispersal of organisms under a range of conditions. Such studies may be complex and expensive and require multi-year study relative to remote sensing studies, but they are a necessary step for conservation and sustainability of aquatic ecosystems in the future.
Acknowledgments
We gratefully thank Mr Aldo Farah-Pérez and Mr Siddarth Machado for their invaluable suggestions on the manuscript. We express our sincere gratitude to the editor and two anonymous reviewers for their comments and suggestions, which have greatly improved the quality of this manuscript. We also acknowledge the United States Department of Agriculture Hatch Grant No. FLA-WFC-005577 and the University of Florida Soil and Water Sciences Department for their support to enable open-access publication.
Data availability statement
No new data were created or analysed in this study.