Content uploaded by Vincent Lecours
Author content
All content in this area was uploaded by Vincent Lecours on Feb 17, 2022
Content may be subject to copyright.
Ecological Indicators 136 (2022) 108680
1470-160X/© 2022 Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
A new index to quantify longitudinal river fragmentation: Conservation and
management implications
Suman Jumani
a
,
*
, Matthew J. Deitch:
b
, Denis Valle
c
, Siddarth Machado
c
, Vincent Lecours
c
,
David Kaplan
d
, Jagdish Krishnaswamy
e
, Jeanette Howard
f
a
Soil and Water Sciences Department, University of Florida, Gainesville, FL, USA
b
Soil and Water Sciences Department, University of Florida, IFAS West Florida Research and Education Center, Milton, USA
c
School of Forest, Fisheries, and Geomatics Sciences, University of Florida, Gainesville, FL, USA
d
Engineering School of Sustainable Infrastructure & Environment, University of Florida, Gainesville, FL, USA
e
School of Environment and Sustainability, Indian Institute for Human Settlements, Bengaluru, India
f
The Nature Conservancy, San Francisco, CA, USA
ARTICLE INFO
Keywords:
River connectivity
Dams
Fragmentation index
Conservation
River network
Aquatic connectivity
ABSTRACT
The proliferation of river infrastructure projects has altered aquatic longitudinal connectivity, posing a growing
threat to riverine biodiversity and ecosystem processes worldwide. Effective methods to quantify loss of river
connectivity across spatiotemporal scales and in data-limited landscapes are important to understand and inform
basin-wide conservation and development planning. Here we introduce a Catchment Area-based Fragmentation
Index (CAFI) and its derivative, the Catchment Area- and Rainfall-based Fragmentation Index (CARFI) as new
metrics to quantify river fragmentation. These indices use catchment area as a proxy for riverine habitat
availability, avoiding the drawbacks of existing metrics that rely on river length and associated derivatives.
CAFI/CARFI can be computed across spatiotemporal scales, incorporate barrier passability values, assesses the
cumulative impact of multiple barriers, and be applied even in data-limited environments.
We rst applied CAFI and CARFI to a simulated network to illustrate their properties with respect to the
number and location of barriers and compared these results to the widely applied Dendritic Connectivity Index
(DCI). While all indices varied with barrier addition, CAFI and CARFI were more sensitive to both barrier number
and location. Next, we illustrated the utility of CAFI and CARFI through case studies in two contrasting settings:
the Klamath River in California, where dam building has ceased (and dam removals are being considered) and
the Netravathi River in India, where dam building is ongoing, with 65 dams proposed for future development.
Results indicate that CAFI and CARFI can effectively quantify trends in fragmentation across spatial scales and
temporal scenarios of dam development (i.e. descriptive applications) and can aid the prioritization of sites for
dam removal, restoration, or conservation (i.e. prescriptive applications). Overall, these indices can quantify the
impacts of individual dams and assess a range of development scenarios to inform basin-wide conservation and
development planning.
1. Introduction
The loss of connectivity is a ubiquitous threat facing rivers world-
wide (Grill et al., 2015; Nilsson, 2005). In addition to the approximately
140,000 large and small dams across the world (McCully, 1996; Nilsson,
2005), tens of thousands of additional undocumented river infrastruc-
ture projects (RIPs) exist worldwide (Belletti et al., 2020). Furthermore,
thousands of dams and other RIPs continue to be commissioned to meet
humanity’s growing demands for hydropower, ood control, and water
supply (Zar et al., 2015). Not surprisingly, freshwater ecosystems are
among the most altered and threatened globally. Existing dams regulate
over half the world’s major river systems (Nilsson, 2005) and allow only
23% of large rivers (>1000 km in length) to ow uninterrupted into the
ocean (Grill et al. 2019).
The primary impact of RIPs is the loss of river network connectivity
through the construction of a physical barrier. This can impede the
movement of sediments, nutrients, water, and aquatic organisms along
the river network, thereby altering riverine habitat structure and
* Corresponding author at: Soil and Water Sciences Department, University of Florida, Gainesville, FL 32611, USA.
E-mail address: sumanjumani@u.edu (S. Jumani).
Contents lists available at ScienceDirect
Ecological Indicators
journal homepage: www.elsevier.com/locate/ecolind
https://doi.org/10.1016/j.ecolind.2022.108680
Received 16 September 2021; Received in revised form 9 February 2022; Accepted 10 February 2022
Ecological Indicators 136 (2022) 108680
2
ecosystem processes and functions (Nel et al., 2009; Poff and Hart, 2002;
Pringle, 2003; Richter et al., 1996). Dams and other RIPs also pose direct
barriers to the movement of aquatic biological communities, most
notably on sh species that migrate along dendritic networks to access
feeding or spawning grounds (Poff and Zimmerman, 2010). This loss of
connectivity can lead to isolation of sh populations (Schick and Lind-
ley, 2007), reduced potential for recolonization and metapopulation
persistence (Fagan, 2002; Fullerton et al., 2010), decreased access to
feeding, spawning and/or nursery habitats (Godinho et al., 2007; Hu
et al., 2015), change in species composition (Consuegra et al., 2021;
Jumani et al., 2018), and even extirpation of isolated sh populations
(Hamilton et al., 2005; Winston et al., 1991).
To better understand and quantify the impacts of RIPs on river
connectivity, several metrics of river fragmentation or connectivity have
been proposed (see Jumani et al. 2020 for a review). These methods can
be categorised into actual, structural, and potential connectivity metrics
(Calabrese and Fagan, 2004). While numerous methods describe actual
connectivity (i.e. based on observed or measured processes), their
application across spatiotemporal scales can be constrained by data and
resource limitations and analytical complexity. On the other hand, many
structural metrics (such as barrier densities and longest free-owing
river length), though easy to compute, are descriptive, spatially inex-
plicit, often insensitive to the addition or removal of barriers, and
incapable of quantifying the individual and cumulative impact of every
dam (Jumani et al., 2020; Kemp & O’Hanley, 2010). As a middle
ground, potential connectivity metrics are spatially explicit and can be
used to robustly characterize connectivity with minimal data and
resource requirements.
Habitat-weighted structural or potential connectivity indices such as
the Dendritic Connectivity Index (DCI) (Cote et al., 2009) and indices
derived from it, like the River Connectivity Index (Grill et al. 2014), can
quantify the cumulative effects of multiple barriers on river connectivity
across spatiotemporal scales. Such indices are also used to inform basin-
wide conservation, restoration, and development plans (Cote et al. 2009;
Bourne et al. 2011; Grill et al. 2014). These indices not only consider the
extent of habitat available (as measured by river length or volume), but
also the spatial conguration of barriers across the river network.
Furthermore, increasing access to GIS capabilities and spatial datasets
make these indices readily applicable across large spatiotemporal scales
and data-decit landscapes. Consequently, such metrics are gaining
rapid popularity and widespread implementation.
Although such potential connectivity metrics can quantify structural
river fragmentation, most use river lengths as a measure of habitat
availability. This reliance on river lengths and the implicit treatment of
river reaches across longitudinal gradients as ecologically equivalent
poses serious drawbacks (detailed in Section 2). Examining and
addressing these issues is important since numerous assessments use
connectivity indices like the DCI to quantify river fragmentation in
response to dam development (Anderson et al., 2018; Atkinson et al.,
2020; Buddendorf et al., 2017; Choy, et al., 2018; Edge et al., 2017;
Jaeger et al., 2014; McManamay et al., 2015) and prioritize barrier
removal, under the assumption that an increase in structural connec-
tivity (as quantied by an index) will improve biotic communities
(Bourne et al., 2011; Kemp & O’Hanley, 2010; Perkin et al., 2015).
Here, we discuss the strengths and drawbacks of existing connec-
tivity metrics (Section 2). We adopt a similar habitat-weighted approach
to introduce a new index of longitudinal river fragmentation – the
Catchment Area-based Fragmentation Index or CAFI (Section 3) – that
addresses some of the drawbacks associated with existing metrics while
retaining their advantages. We then examine the characteristics of the
CAFI through simulations by varying the location and number of bar-
riers in a hypothetical watershed and compare these results to those
obtained using the DCI (Section 4). Finally, we apply our index to the
Klamath River basin (USA) and the Netravati River basin (India) to
illustrate its application in quantifying river fragmentation across
spatiotemporal scales, and in spatial prioritization analyses to inform
basin-wide conservation and development planning (Section 5).
2. Strengths and drawbacks of existing potential connectivity
metrics
Longitudinal connectivity is crucial in determining habitat avail-
ability for river-dependent fauna and hence determines the composition
and distribution of aquatic biological communities (Cote et al., 2009;
Fagan et al., 2002). Since connectivity in rivers is water-mediated
through dendritic pathways, barriers cannot be considered in isolation
of other barriers on the river network. Consequently, graph-theoretic
approaches are an emerging method to quantify the cumulative
impact of several barriers on network connectivity (Grill et al., 2014).
Such approaches consider not just the number of barriers, but also their
spatial conguration relative to each other and the river network by
representing the river as a network of links and nodes. Data on habitat
availability and other variables of interest can also be assigned to such
networks. Various connectivity indices have been described using these
techniques, the majority of which use river length as a measure of
available habitat (Cote et al., 2009; Diebel et al., 2015; McKay et al.,
2013; Segurado et al., 2013), though others use river-length dependent
variables like river volume (Grill et al. 2014) and stream order (Díaz
et al., 2019). The DCI by Cote et al. (2009) is among the most widely
used connectivity metrics within this family of metrics. It is an index of
river connectivity calculated from stream length, which assesses the
probability that a sh may move between two points in a river network
(Cote et al. 2009). The DCI can be calculated for both potamodromous
(DCIp) and diadromous (DCId) life histories, and values range from
0 (no connectivity) to 100 (fully connected). Assuming each barrier is
impassable and splits the river into distinct segments, the DCIp and DCId
can be calculated as:
DCIp =∑
n
i=1
l2
i
L2*100 (1)
DCId =lm
L*100 (2)
where, ‘n’ is the number of segments; ‘l
i
’ is the river length of the
segment ‘i’ that is disconnected by one or more dams; ‘l
m
’ is the length of
the segment closest to the mouth of the system; ‘L’ is the total length of
the entire river network.
Furthermore, the DCI can incorporate directional barrier pass-
abilities. Since these metrics can be readily computed across spatio-
temporal scales, even in data-decit environments, they have gained
rapid popularity and widespread implementation (Jaeger et al., 2014).
Despite these strengths, river-length dependent indices have a few
drawbacks. First, when river lengths are used without other weights, the
underlying assumption is that stream reaches across a longitudinal
gradient are ecologically equivalent. For example, the DCIp can produce
the same value for dams located upstream or downstream in a watershed
as long as the lengths of the resultant river fragments are the same,
despite these scenarios presenting different ecological impacts (Grill
et al. 2014). This can be particularly problematic when applied to the
prioritization of dam mitigation or removal. Second, headwater dams
that lie beyond the delineated river network are often excluded from the
analysis when river-length dependent metrics are used. For example,
Anderson et al. (2018; Supplementary Material) excluded about 70 dams
from their analysis that were beyond the mapped river network. This is
further illustrated in Supplementary Figure S1, which shows how vari-
able ow accumulation thresholds can include or exclude dams on low-
order reaches. Since headwater streams tend to harbour greater number
of endemic species (Colvin et al., 2019; Meyer et al., 2007) and a large
number of ongoing and proposed RIPs are being commissioned on
headwater streams (Couto and Olden, 2018), omission of such dams
from connectivity analyses poses a signicant problem. Lastly, river-
S. Jumani et al.
Ecological Indicators 136 (2022) 108680
3
length dependent indices can vary in value based on the spatial extent of
the delineated river network (Baker et al., 2007). This, in turn, depends
on the threshold set for ow accumulation to river lines, and to a lesser
extent, the resolution of the underlying digital elevation model (DEM)
(Dark and Bram, 2007; Murphy et al., 2008), and the ow direction and
accumulation algorithm used (Erskine et al., 2006). The area threshold
to convert ow accumulation rasters to river polylines can drastically
change the extent of river branching and river length, which yields
inconsistent and non-uniform changes in connectivity index values. To
illustrate this, we examined DCIp and DCId values for a river network
delineated across three different ow accumulation thresholds (Fig S1 in
Supplementary Material). Resultant inconsistencies in values and
number of dams excluded from the analysis render DCI values incom-
parable across the three networks. In terms of the underlying DEM
resolution, ne-scale elevation data more accurately represent the
contours of the landscape (Dark and Bram 2007), yielding more accurate
hydrologic derivatives compared to coarser resolution DEMs (Murphy
et al., 2008). Errors from coarse resolution DEMs are enhanced in
headwater reaches and smaller catchments; smaller study regions thus
need to be characterised by input data that appropriately capture
landscape heterogeneity.
3. Catchment Area-based fragmentation index (CAFI) as a
fragmentation metric
To address the above issues, we propose the Catchment Area-based
Fragmentation Index (CAFI) as a new metric of river network frag-
mentation. Building from the DCI (Cote et al., 2009), this spatially
explicit index of fragmentation replaces river length with cumulative
catchment area as a measure of habitat quantity. Notably, upstream
catchment area is an excellent predictor of discharge, a measure of
habitat availability (Ziv et al., 2012), with a unit increase in contributing
area generally yielding a unit increase in water volume under the
simplifying assumption that all parts of the catchment contribute the
same volume of water (Galster et al. 2006; Deitch and Kondolf 2012;
Burgers et al. 2014). Assuming each barrier is impassable, the CAFI is
calculated as the sum of the ratio of the catchment area of each dam to
the total catchment area of the entire river network (Eq. (3)). In cases
where barrier passability is known, the CAFI can be weighed by barrier
passabilities as shown in Eq. (4).
CAFI =∑
n
i=1
ai
A*100 (3)
CAFI =∑
n
i=1
ai.ci
A*100 (4)
where, ‘n’ is the number of dams; ‘a
i
’ is the total catchment area of dam
‘i’; ‘c
i
’ is the barrier impassability score ranging from 1 (impassable) to
0 (completely passable); ‘A’ is the catchment area of the entire river
network
In catchments characterised by uniform spatial distribution of rain-
fall, drainage area is an adequate predictor of discharge. However, for
rivers in mountainous terrains having strong orographic or latitudinal
rainfall trends, incorporating spatially explicit information about pre-
cipitation may improve metric performance. In such cases, we propose
the Catchment Area- and Rainfall-based Fragmentation Index or CARFI
(Eq. (5))
CARFI =∑
n
i=1
ai.ri
A.R
×100 (5)
where ‘r
i
’ is the average annual rainfall intensity in ‘a
i
’; ‘R’ is the average
annual rainfall intensity in the entire catchment area ‘A’.
Like the DCI, the CAFI and CARFI also use network analysis to
quantify the cumulative impact of each barrier relative to its location on
river fragmentation. It is important to note that, while the DCI is a
measure of connectivity, CAFI is a measure of fragmentation, with higher
values indicating greater fragmentation. Since contributing areas in-
creases from upstream to downstream, a dam’s impact on fragmentation
will be greater the further downstream it is located. Based on the species
or ecosystem process being considered, this may better reect expected
conditions since barriers located further downstream isolate greater
proportions of available upstream habitat compared to headwater dams
(Fagan et al., 2002; Nilsson, 2005). The CAFI has a lower limit of
0 (indicating complete connectivity), and for a basin with a single dam,
the CAFI can range between 1 and 100 (low to high levels of fragmen-
tation). However, as the number of barriers increase, the CAFI can
surpass 100, with increasing values corresponding to increasing levels of
fragmentation.
The CAFI and CARFI are relatively easy to compute even in data-
decit regions since catchment area can be delineated with any sur-
face elevation model on a GIS platform and rainfall can be ascertained
through global datasets such as WorldClim (Fick and Hijmans, 2017).
Since these metrics rely on catchment area, all dams can be incorporated
in the analysis, including those on headwater streams beyond delineated
river networks. Furthermore, although the extent of contributing areas
varies based on DEM resolution, the magnitude of error associated with
areal measures tends to be lower when compared to errors in drainage
network length (Ghaffari, 2011; Sukumaran and Sahoo, 2020; Tan et al.,
2018). While CAFI/CARFI addresses some of the drawbacks of the DCI,
they are limited in not being restricted to a maximum value.
4. Application of the CAFI and CARFI in a simulated drainage
network
To examine the properties of these metrics, we calculated the CAFI
and CARFI across a simulated second-order catchment (similar to Cote
et al. 2009) and examined the areas of convergence and divergence
between the DCIp, DCId, CAFI and CARFI. To enable better comparison
with CAFI/CARFI values, DCIp and DCId were visualised as their inverse
(100-DCI). The simulated network was characterised by a total drainage
length of 1000 km (Fig. 1a), a catchment area of 10,000 square kilo-
metres, and total annual rainfall of 2000 mm (Fig. 1b) with 20
impassable barriers distributed across its longitudinal length (Table S1
in Supplementary Material). Rainfall was simulated to be orographic
such that rainfall intensity decreased from headwater to downstream
reaches. We calculated index values under three scenarios: (1) a single
barrier at decreasing distances from the river mouth, (2) directional
increase in the number of barriers, and (3) varying the spatial congu-
ration and number of barriers from 0 to 20. All simulations were con-
ducted in R 4.0.5 (R Core Team, 2021). R code and associated data are
provided in the Supplementary Information.
4.1. Simulation 1- Variation in metric based on the location of a single
barrier
Here index values were computed for a single barrier (barriers A to T
in Fig. 1) to evaluate the effect of barrier location on fragmentation.
Since the DCIp treats all stream reaches as functionally equivalent,
regardless of their position on the river network, dams located upstream
and downstream at similar distances from the headwater or river mouth,
respectively, produce the same connectivity values, yielding a bell-
shaped distribution (Fig. 2). On the other hand, DCId inverse, CAFI,
and CARFI values show a monotonic decline with increasing distance of
barrier to the mouth (Fig. 2). The DCId, CAFI and CARFI indicate that
fragmentation caused by a single dam is highest near the mouth of the
river and decreases as distance to mouth increases. For basin-wide
processes, this reects the expected trend where a downstream barrier
isolates larger portions of the upstream habitat compared to a headwater
dam (Fagan et al., 2002; Fullerton et al., 2010). On the other hand, DCIp
generates maximum fragmentation when a dam splits a river network
S. Jumani et al.
Ecological Indicators 136 (2022) 108680
4
into equal fragments. This may be better suited to measure fragmenta-
tion for potamodromous species and other non-directional processes.
CARFI values are generally lower than CAFI. This difference increases
from upstream to downstream dams because in instances of orographic
rainfall, catchments of headwater dams have higher annual rainfall in-
tensities compared to those of downstream dams.
4.2. Simulation 2- Variation in metric based on directional increase in the
number of dams
Here, index values were calculated for an increasing number of
barriers cumulatively added to the network. This increase was done in
two ways, rst with successive barriers added in the upstream to
downstream direction (i.e. starting with barrier A in Fig. 1) and next
with barriers added to the network in the reverse order (i.e. starting with
barrier T in Fig. 1). While DCIp inverse increased with an increase in the
number of barriers, similar trends of change were exhibited in the up-
stream and downstream direction of dam addition, with the extent of
change (as indicated by the slope of the line) decreasing after the
addition of the rst 10–12 barriers (Fig. 3). Although DCId inverse
showed differential trends of change based on the direction of barrier
addition, values do not change with the addition or removal of dams
above the most downstream dam. Consequently, the addition of dams in
the downstream to upstream direction did not show any change after the
rst dam ‘T’ (orange line, Fig. 3). On the other hand, for CAFI and
CARFI, the direction and location of dam addition were reected in the
index value. When moving from upstream to downstream (blue lines,
Fig. 3), the index increases gradually at rst and then more rapidly as
successive dams are added further toward the basin outlet; the opposite
is true for dam addition from downstream to upstream (orange lines,
Fig. 3), with the largest impacts for those rst few dams farthest
downstream. Overall CAFI and CARFI yielded smaller increases in
fragmentation for barriers added further upstream, irrespective of the
direction of increment. This trend reects the increase in basin-wide
fragmentation caused by the addition of headwater dams versus dams
on the mainstem.
Fig. 1. Simulated drainage network and location of barriers along its path. Numbers in left panel correspond to the length of each river segment. Numbers in the right
panel refer to the catchment area of each barrier, as illustrated with the dashed lines.
S. Jumani et al.
Ecological Indicators 136 (2022) 108680
5
4.3. Simulation 3 - Variation in metric based on the number and spatial
conguration of barriers
Here, index values were calculated for an increasing number of
barriers (ranging from 1 to 20) across 100 random iterations of barrier
locations on the simulated river network. This simulation illustrates the
sensitivity of these indices to the incremental addition of a dam across
varying spatial congurations (Fig. 4). DCIp and DCId values were most
sensitive to the addition of the rst 5–10 and 3–5 barriers respectively.
CAFI and CARFI, on the other hand, retained sensitivity to the incre-
mental addition of more barriers. This difference may make the CAFI/
CARFI better suited to measure changes in fragmentation, especially in
previously dammed basins.
In our second simulation, we addressed the potential for thresholds
in connectivity associated with the number of barriers and changes in
watershed topology. Passage rate was held constant at 0.5, while
watershed topology, the number of barriers, and the locations of the
barriers were varied. As anticipated, we observed a strong non-linear
Fig. 2. 100-DCIp, 100-DCId, CAFI and CARFI values for a single dam across varying distances from the mouth of a simulated network. The sudden change at the
point close to 750 m corresponds to dam ‘E’ located below the conuence of two tributaries, where fragment length and catchment area vary by a higher margin.
Fig. 3. 100-DCIp, 100-DCId, CAFI, and CARFI values
for an increasing number of dams, added in the up-
stream to downstream (blue lines) and downstream to
upstream (orange lines) directions. The sudden
change at the point close to 750 m corresponds to dam
‘E’ located downstream of the conuence of two
tributaries, where fragment length and catchment
area vary by a higher margin. (For interpretation of
the references to colour in this gure legend, the
reader is referred to the web version of this article.)
S. Jumani et al.
Ecological Indicators 136 (2022) 108680
6
decline in connectivity associated with the number of barriers. For
instance, the median connectivity (HCIU value) over 500 watershed
topologies and 20 dam congurations reduces from 1.00 at 0 barriers to
0.52 at 3 barriers, 0.36 at 5 barriers, and 0.17 at 10 barriers. Fig. 5
demonstrates this nonlinearity, stratied by network diameter, where
three classes of diameter were identied to contain nearly equal sample
size (34.0% of samples have diameters of 6–15, 31.7% of samples have
diameters of 16–21, and 34.3% of samples have diameters of 22–39).
This gure shows that watersheds with low branching topology and high
diameter are more susceptible to disconnection by lower numbers of
barriers In our second simulation, we addressed the potential for
thresholds in connectivity associated with the number of barriers and
changes in watershed topology.
The above simulations illustrate the properties of the CAFI, CARFI,
DCIp and DCId in relation to barrier location, number of barriers, and
sensitivity to additional barriers in a river network. The DCIp, CAFI, and
CARFI share common trends of curvilinear responses to barrier addition
(Fig. 3). While the DCId is sensitive to the location of a single barrier
(Fig. 2), CAFI and CARFI are more sensitive to both barrier location
(Fig. 2) and a varying number of dams (Fig. 3). While DCI p and DCId
values begin to plateau after the addition of the rst few dams, CAFI and
CARFI retain sensitivity to the incremental addition of barriers (Fig. 4).
5. Case study applications of the CAFI and CARFI
We applied the CAFI to the Klamath River Basin in the USA and the
CARFI to the Netravathi River Basin in India to illustrate the application
of these indices in different settings. The Klamath is a large river origi-
nating in southern Oregon, USA that traverses over 425 km before
reaching the Pacic Ocean off northern California. The basin spans an
area of ~ 31,340 km
2
and supports several endangered and threatened
species such as the Deltistes luxatus (Cope, 1879) and Chasmistes brevir-
ostris (Cope, 1879) (USFWS, 2012), and extensive anadromous sh runs
for salmonids such as Oncorhynchus kisutch (Walbaum, 1792),
O. tshawytscha (Walbaum, 1792), and O. mykiss (Walbaum, 1792). Up
until the year 2000, the main river network of the Klamath had 62 dams
along its course; eight dams have been removed since that time. Four
dams have been identied for further removal under the Klamath Hy-
droelectric Settlement Agreement (Gosnell and Kelly, 2010). Here, we
compute the extent of longitudinal fragmentation under three scenarios:
the past scenario based on the presence of all anthropogenic barriers (n
=62), the present scenario based on the presence of existing dams (n =
54), and the future scenario considering the removal of four large dams
(n =50). These scenarios were examined at the spatial scales of the basin
and sub-basin (dened by the 8-digit hydrologic code of the USGS
Watershed Boundary Dataset (Simley & Carswell, 2009)) under the
assumption that every barrier is impassable. We also use CAFI to identify
priority dam removal sites to best improve longitudinal connectivity.
The Netravathi River is a small west-owing river originating in the
mountains of the Western Ghats of Karnataka State in India. The basin
receives an average rainfall of 4063 mm and experiences strong
orographic rainfall trends, with rainfall intensity varying between 6433
mm and 2780 mm from upstream to downstream reaches (Fig S2 in
Supplementary Material). The basin encompasses an area of approxi-
mately 3470 km
2
and the river ows over 104 km before emptying out
into the Arabian Sea. This region is part of the Western Ghats global
biodiversity hotspot (Myers et al. 2000) and UNESCO world heritage
site. This river is also identied to be a potential freshwater key biodi-
versity area (Molur et al. 2011), characterised by exceptionally high
levels of species richness and endemism and intense anthropogenic
pressures. The river has 24 current dams along its network, and 65 small
hydropower dams (those that produce <25 MW) have been proposed
for further development. Hence, CARFI values were examined across
two scenarios – the present scenario characterised by all existing dams
(n =24), and the future scenario characterised by the presence of all
existing and proposed dams (n =89). These scenarios were examined at
the spatial scales of the basin and sub-basin (dened by level 8 of the
HydroBASINS dataset (Lehner et al., 2008)) under the assumption that
every barrier is impassable. Additionally, ‘optimised’ strategies of dam
development were identied to minimize fragmentation while max-
imising human gain with respect to hydropower capacity.
Fig. 4. Simulation of 100-DCIp, 100-DCId, CAFI, and CARFI values for an increasing number of dams across 100 random iterations of barrier locations. The error bars
denote standard deviation.
S. Jumani et al.
Ecological Indicators 136 (2022) 108680
7
5.1. Quantifying longitudinal fragmentation across spatial and temporal
scales
Klamath River Basin: Change in basin-scale CAFI scores between
1800 and 2025 is shown in Fig. 5a. Dams constructed between 1840 and
1910 contributed to low levels of basin-wide fragmentation as they were
located on headwater streams. Fragmentation increased steeply in the
1920 s and 1960 s. The last dam was constructed in 1991, after which
defragmentation occurred. The removal of eight dams between 2002
and 2012 marginally decreased basin-wide CAFI from 281 to 271 (red
solid line, Fig. 5). The future removal of four large dams (Iron Gate,
Copco 1, Copco 2, and JC Boyle) will decrease basin-wide CAFI by more
than half to 94.8 (red dotted line, Fig. 5). Comparing this chronose-
quence to the inverse DCIp and DCId (Fig. 5b and 5c) reveals some
similarities, such as the rst peak of fragmentation around the 1920 s.
However, some important dissimilarities remain. First, the rate of in-
crease in fragmentation for the rst few dams (1800–1900) is greater for
DCIp and DCId. Additionally, after 1925, DCIp and DCId values tend to
plateau despite the subsequent building of over 40 dams, although DCId
values varied more than those of DCIp. Hence, the second peak of dam
building (as seen in case of CAFI around the 1960 s) is not as apparent.
Furthermore, the proposed removal of four large dams, which will free
up ~ 400 km of upstream river length, has a smaller impact on
improving connectivity as per DCIp and DCId, in stark contrast to CAFI.
At the sub-basin scale, the removal of eight dams between 2002 and
2012 improved sub-basin level connectivity along the mid- and down-
stream sections of the river (Fig. 6). Specically, fragmentation
decreased in the Shasta, Trinity, Salmon, and Sycan river sub-basins
(Table 1). This decrease was greatest in the Shasta river sub-basin,
where the removal of just two dams decreased the CAFI from 143 to
19 (Table 1). The removal of four dams in the Upper Klamath will
drastically improve connectivity in the future scenario.
Netravathi River Basin: Dam building on the Netravathi began in
1990, and a steep increase in fragmentation was seen in the years
following 2010 due to the construction of ve dams along mainstem
channels (Fig. 7a). The present scenario, characterised by 24 dams,
resulted in a basin-level CARFI score of 281.8; the addition of 65 pro-
posed dams will increase basin-level CARFI to 958.9. A similar
increasing trend is apparent with the inverse DCIp (Fig. 7b). However,
the DCId, which is sensitive to the most downstream dam, exhibited a
much smaller extent of change with the addition of 65 proposed dams.
Both, DCIp and DCId indicated sharper increases in fragmentation after
the addition of the rst few dams and a smaller response to the future
addition of 65 dams as compared to the CARFI.
At the sub-basin scale, all sub-basins show an increase in fragmen-
tation from the present to the future scenario, except for the Gowri river
sub-basin (Fig. 8; Table 2). Currently undammed sub-basins of middle
and lower Kumaradhara will be subjected to severe increases in frag-
mentation if proposed dams are built. For the Yettinahole sub-basin,
CARFI levels before the construction of the controversial Yettinahole
Diversion Project (YDP) was 94. After the construction of eight dams of
the YDP in 2019, the CARFI increased to 158.
5.2. Sensitivity analysis to prioritize dam mitigation/removal
Klamath River Basin: To prioritize dams for removal or mitigation
efforts, a sensitivity analysis of existing dams was conducted. Using the
rst scenario with 116 dams as the starting point, all projects were
iteratively removed to determine the individual effect of removing a
single dam (Branco et al., 2014). At each step, we rst identied the dam
that caused the greatest decrease in fragmentation upon its removal,
after which it was permanently dropped. This process was repeated until
all the dams were ranked (Fig. 9). The removal of just seven dams can
reduce basin-level fragmentation to pre-1920 levels, from 252 to 44.5.
The four dams (Iron Gate, Copco 1, Copco 2, and JC Boyle) identied for
removal are also ranked the highest in the sensitivity analysis and will
result in the highest possible decrease in basin-level fragmentation. On
the other hand, the eight dams removed between 2002 and 2012 have
had a signicantly lower impact on decreasing fragmentation (Fig. 9).
Netravati River Basin: To aid the planning and prioritization of
proposed dams, a sensitivity analysis was conducted using the existing
scenario as the starting point. Each individual proposed project was
iteratively added to the river network with all existing dams to deter-
mine the individual effect of adding a single dam on CARFI. At each step,
the dam with the lowest impact was added to the river network and the
process repeated until all the dams were ranked in increasing order of
impact. Fig. 10 represents the range of CARFI values for each added dam
across all simulations. The lower end of the boxplot represents the
“better” option that causes the lowest possible increase in basin-level
CARFI, while the upper end represents the “worst” option that causes
the greatest increase in basin-level CARFI.
The ranked proposed dams were overlaid with their hydropower
capacity to assess the utility of each dam. Dams were grouped into four
categories based on their impact (high versus low fragmentation) and
output (high versus low utility) (Fig. 11a). Dams with a low impact but
high utility are generally considered ‘better’, while those with high
Fig. 5. Chronosequence of basin-level CAFI (a), 100-DCIp (b), and 100-DCId (c) in the Klamath. Numbers indicate the number of dams built or decommissioned on
select years. Blue and red lines respectively indicate increases and decreases in fragmentation; solid and dashed lines represent present and future scenarios
respectively. (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)
S. Jumani et al.
Ecological Indicators 136 (2022) 108680
8
impact and low utility are poor options recommended to be avoided at
all costs.
We also overlaid the ranked dams’ CARFI scores with their hydro-
power capacity to determine the least fragmenting combination of
barriers to be built to achieve a given hydropower production goal
(Fig. 12). For example, the existing 24 dams on the Netravathi have a
cumulative installed capacity of 100 MW. Considering a hypothetical
goal to double the hydropower generating capacity to 200 MW (dashed
line, Fig. 12), the construction of 18 selected dams would achieve that
goal whilst causing the lowest cumulative impact on basin-wide frag-
mentation as measured by the CARFI. It is important to note that this is
an illustrative exercise; we intend for such applications to be used along
with other socio-ecological considerations and site-specic characteris-
tics to inform dam siting.
6. Discussion
Given the continued global development of river networks, there is a
need to develop robust and exible metrics of river fragmentation that
have the ability to: (a) assess the individual and cumulative impacts of
multiple barriers; (b) be applied across spatial scales and scenarios; (c)
incorporate dam passabilities; and (d) be easily and efciently computed
in data decit regions with (e) computational ease and efciency. The
CAFI and CARFI are metrics of longitudinal river fragmentation that
meet the above requirements.
As catchment areas increase from upstream to downstream, the
CAFI/CARFI yields a higher impact for dams located further down-
stream (Figs. 2 and 3), incorporating an implicit sense of ‘directionality’.
This differential impact along a longitudinal gradient of a river better
reects expected trends of impeded basin-wide ecosystem processes and
the movement of diadromous species as barriers located further down-
stream isolate greater proportions of available upstream habitat
compared to headwater dams, and impact the generally higher diversity
of larger rivers (Fagan et al., 2002; Fullerton et al., 2010; Grill et al.,
2014). Although the DCId tends to produce declining connectivity
values for downstream dams, their values do not change with the
addition or removal of barriers above the most downstream dam since it
is calculated from the perspective of a diadromous sh arriving from the
sea. This trend, while suitable for diadromous species, could pose a
drawback for other applications.
The CAFI/CARFI is easy to compute, especially in data-decit envi-
ronments. This attribute is crucial since ongoing and future dam de-
velopments are mainly concentrated in tropical developing countries
(Zar et al., 2015) that are often limited in hydro-ecological data
availability and tend to overlap with areas of high freshwater biodi-
versity (Pandit and Grumbine, 2012; Tockner et al., 2016). In these
scenarios, the CAFI/CARFI can be particularly useful in aiding science-
based decision-making pertaining to basin-wide conservation and RIP
management. The lack of reliance on river fragment lengths and their
derivatives also makes the CAFI/CARFI more robust and easier to apply,
especially with respect to dams on low-order tributaries. This is
important since most proposed dam developments are geared towards
small dams on headwater streams (Couto and Olden, 2018) where the
derivation of channel length and stream order measures can be subjec-
tive and relative to thresholds for channel inception or DEM resolution.
Further, unlike river-length dependent measures (such as river length to
next barrier), contributing area values do not change with the removal
or addition of dams. This improves the computational efciency of these
metrics, allowing users to evaluate a large number of scenarios and
barrier combinations more quickly compared to recursive analyses in
Fig. 6. Sub-basin level change in CAFI in the Klamath Basin (shaded area in (a)) across three scenarios: past scenario with 62 dams (b), present scenario with 54 dams
(c) and future scenario involving the removal of four dams (c). The last two digits of the HUC8 sub-basin code are illustrated in panel b. ‘Unassessed’ areas include
parts of the river that do not connect to the main network via surface connections.
Table 1
Sub-basin level CAFI scores across three scenarios for the Klamath River Basin.
Numbers in parentheses represent the number of barriers.
Sub-basin HUC 8 code CAFI – past
scenario (62)
CAFI – present
scenario (54)
CAFI – future
scenario (50)
Miller Creek 18,010,201 0.0 (0) 0.0 (0) 0.0 (0)
Sycan River 18,010,202 120 (18) 104.3 (16) 104.3 (16)
Fall Creek 18,010,203 31.9 (4) 31.9 (4) 31.9 (4)
Lost River 18,010,204 0.1 (2) 0.1 (2) 0.1 (64)
Upper
Klamath
River
18,010,206 115 (10) 115 (8) 5.6 (4)
Shasta River 18,010,207 143 (15) 19.6 (15) 19.6 (15)
Scott River 18,010,208 0.23 (3) 0.23 (3) 0.23 (3)
Lower
Klamath
River
18,010,209 0.11 (1) 0.11 (1) 0.11 (1)
Salmon
River
18,010,210 3.09 (2) 0.0 (0) 0.0 (0)
Trinity River 18,010,211 71.7 (6) 69.7 (4) 69.7 (4)
South Fork
Trinity
18,010,212 0.08 (1) 0.08 (1) 0.08 (1)
S. Jumani et al.
Ecological Indicators 136 (2022) 108680
9
graph-theoretic river routing models.
The CAFI/CARFI is adaptable in that it can be weighted by user-
dened variables of importance such as species richness, unique river
classes, and habitat quality as illustrated by Grill et al. (2014) and
Rodeles et al. (2021; 2020; 2019) based on study objectives or site-
specic considerations. It can also account for differences between
impermeable and partially permeable barriers through the barrier
impassability score, although this does not implicitly account for the
spatial relationship between dams. Given that these indices are not
species- or taxa-specic, we expect them to represent fragmentation
more holistically for biotic communities (particularly diadromous or
migratory species) and connectivity-dependent river dynamics such as
sediment redistribution. Since CAFI weighs barriers with smaller
catchments as having a smaller impact, we also propose the CARFI for
rain-fed catchments experiencing orographic rainfall trends. In such
cases, the impact of a barrier with a small catchment area but a higher
rainfall intensity (as compared to that of the total basin) will be
weighted higher.
The lack of an upper bound to CAFI/CARFI values poses certain
advantages and disadvantages. This feature allows the index to retain
sensitivity to the addition of new barriers, even in a previously dammed
basin (Figs 4, 5a, 7a), making them particularly suited to analysing and
visualizing spatiotemporal trends in fragmentation in response to
increasing dam densities. Since the DCIp and DCId begin to plateau after
the addition of the rst few dams, in basins with high-dam densities,
they do not adequately reect changes in connectivity with additional
Fig. 7. Chronosequence of basin-level CARFI (a), 100-DCIp (b), and 100-DCId (c) in the Netravathi. Numbers indicate number of dams built each year. Solid and
dashed lines represent present and future scenarios respectively.
Fig. 8. Sub-basin level change in CARFI scores in the Netravathi Basin (shaded area in (a)) over two scenarios: present scenario with 24 built dams (b) and future
scenario with 89 dams (c). The last digit of the sub-basin code is illustrated in panel b.
S. Jumani et al.
Ecological Indicators 136 (2022) 108680
10
dam building (Figs 4, 5b and c, 7b and c). However, the linear rela-
tionship between CAFI and additional dams may overestimate the
impact of new barriers or their removal in heavily dammed basins. In
other words, the potentially innite increase in CAFI/CARFI values
within a nite basin loses functional relevance beyond a threshold in
basins with very high dam densities. The extent of loss of functional
relevance will depend on the species or process being considered.
Additionally, while CAFI/CARFI values can be compared across simi-
larly sized basins or across various scenarios of barrier placement for a
given basin, the lack of a mathematical upper limit makes it difcult to
compare CAFI values across differently sized basins. Hence, we recom-
mend that index values generally be interpreted relative to each other or
within a basin, with increasing values corresponding to increasing levels
of fragmentation.
Through the applied case studies, we also illustrate the descriptive
(Section 5.1) and prescriptive (Section 5.2) applications of CAFI/CARFI
to study trends in fragmentation and inform management and conser-
vation planning. We caution against the deterministic interpretation of
these results, and rather use these as an illustrative exercise. The index
does not replace ground-level studies or project-specic impact assess-
ments. Instead, we suggest the index be used in conjunction with site-
specic considerations to holistically inform river conservation and
development planning. In terms of descriptive applications, the index
can demonstrate the increases or decreases in fragmentation caused by
varying number of dams across spatial scales due to its sensitivity to the
addition or removal of barriers (Figs. 5 and 7). The choice of scale should
depend on the size of the basin and the process or subject of study. A
combination of basin and sub-basin scale analyses is expected to produce
the most comprehensive results, allowing users to analyse a range of
scenarios and determine sub-basins of restoration or conservation
interest.
The indices can also be used in optimization or sensitivity analyses
(McKay et al., 2017) to better plan dam removal, mitigation action, or
management. Ranking of dams based on their impact on fragmentation
can help identify good and bad options for dam removal (Fig. 9) or
placement (Fig. 10). The impact of individual and cumulative dams on
fragmentation can be viewed against dam contributions to determine
relatively “better” (low fragmentation and high benet) and “worse”
(high fragmentation and low benet) projects (Fig. 11a). However, due
to the way the index is dened, CAFI/CARFI emphasises the impacts of
dams having larger catchment areas. Hence, when using CAFI/CARFI to
inform dam placement (as in the case of the Netravathi), sites with
smaller contributing areas tend to be prioritized (Fig. 11b). This could be
problematic given the importance of headwater streams in providing
breeding/nursery grounds and habitat for numerous endemic species of
conservation value (Colvin et al., 2019; Meyer et al., 2007). Undammed
tributaries also provide critical buffer to riverine ecosystem processes
and function in dammed basins (Atkore et al., 2017). Hence, we caution
against the use of this (or any) index as the sole determinant in
Table 2
Sub-basin level CARFI scores for present and future scenarios for the Netravathi
River Basin. Numbers in parentheses represent the number of barriers.
Sub-basin Sub-basin
code
CARFI – present
scenario (24)
CARFI – future
scenario (89)
Shishila River NET1 184.38 (8) 366.48 (33)
Kumarahalli NET2 0 (0) 10.19 (1)
Yettinahole NET3 158.37 (12) 313.21 (24)
Upper
Kumaradhara
NET4 4.19 (1) 339.18 (16)
Addahole NET5 297.26 (1) 1077.14 (4)
Middle
Kumaradhara
NET6 0 (0) 2165.79 (3)
Gowri River NET7 0 (0) 0 (0)
Lower
Kumaradhara
NET8 0 (0) 3260.95 (2)
Netravati River NET9 1355.28 (2) 3429.67 (6)
Fig. 9. Dam removal ranking based on CAFI scores for 62 barriers on the Klamath. The red dots represent the eight dams removed between 2002 and 2012. (For
interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)
Fig. 10. Dam placement sensitivity analysis based on CARFI scores for 65
proposed barriers on the Netravati River basin.
S. Jumani et al.
Ecological Indicators 136 (2022) 108680
11
prioritizing dam placement. In these cases, users are advised to incor-
porate additional rules such as retaining a xed number of undammed
tributaries and maintaining some minimum inter-dam distance. Given
the complexities regarding the spatiotemporal scale of application and
the analyses used, careful interpretation of results is also required. Since
these indices measure only structural fragmentation, they are intended
to guide conservation and management decisions along ground-level
studies or impact assessments and social and ecological considerations.
An underlying assumption behind the application of all structural
connectivity metrics is that an increase in fragmentation (as measured
by a given metric) will correspond to a loss of functional connectivity
(with respect to biotic communities or ecosystem processes). However,
the ecological relevance of most indices remains unknown, presenting
an important frontier for further research. Since connectivity is depen-
dent on the point of view of the species or process being considered,
different indices may be better suited for different applications. For
instance, potamodromous species may respond better to fragmentation
measured by DCIp, while diadromous species or basin-wide connectivity
dependent processes may correlate better with CAFI/CARFI. Hence,
empirical, eld-based studies are required to ecologically validate CAFI/
CARFI and other connectivity metrics. Since different species and
ecosystem processes perceive habitats and operate at different
a) b)
Fig. 11. (a) Proposed dams categorised based on their impact on fragmentation and hydropower capacity (b) The ten ‘better’ (green triangle) and ‘worse’ (red
triangle) dams as selected by CARFI. (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)
Fig. 12. Addition of proposed dams in order of increasing impact on fragmentation (blue line) overlaid with the cumulative increase in installed hydropower (orange
line). The dashed line represents the number of dams to be considered to achieve a hypothetical goal of 200 MW generating capacity. (For interpretation of the
references to colour in this gure legend, the reader is referred to the web version of this article.)
S. Jumani et al.
Ecological Indicators 136 (2022) 108680
12
spatiotemporal scales (Gaucherel, 2007; Llaus`
as and Nogu´
e, 2012), their
response to fragmentation is also expected to be scale-dependent. Thus,
research is also needed to identify the range of spatial scales over which
the CAFI/CARFI can be meaningfully applied (Fullerton et al., 2010;
Jumani et al., 2020). Since different metrics vary in their properties and
assumptions, research on comparative assessments between various
river connectivity or fragmentation indices can shed light on areas of
convergence and divergence in metric performance. Coupled with in-
formation on the ecological relevance of different metrics, this can
further inform metric selection given river habitat characteristics, extent
of damming, and study objectives. Finally, research is also needed to
better quantify barrier passability (Kemp & O’Hanley, 2010), incorpo-
rate spatial interdependence in barrier passabilities (Cote et al., 2009),
and test the integration of the index with ecological information such as
species diversity or habitat quality.
7. Conclusion
Metrics that evaluate the impacts of RIPs on fragmentation can play
an important role in the conservation and management of riverine
ecosystems. The widespread use of fragmentation metrics highlights the
need for easily derived and ecologically relevant tools to make such
evaluations. Though such metrics cannot substitute for empirically
derived data, they can be effectively used by stakeholders to assess po-
tential impacts of specic dams and provide a means to assess a range of
conditions as a rst step towards basin-wide conservation planning.
Given their widespread utility, such metrics need to be sufciently
robust with respect to their underlying assumptions, properties, and
ecological relevance to be meaningfully applied.
The results presented here demonstrate the descriptive and pre-
scriptive utility of the CAFI/CARFI for conservation and management
planning, as well as potential drawbacks that may constrain their value.
Despite their limitations, these indices overcome some of the disad-
vantages associated with existing metrics. These improvements make
the CAFI/CARFI a useful metric that can be applied across scenarios to
quantify the individual and cumulative impacts of barriers on a river
network. Their applications in quantifying fragmentation and in iden-
tifying ‘good’ and ‘bad’ dams, priority sites for dam removal or miti-
gation, and project locations that can have lower impacts on
fragmentation make CAFI/CARFI a useful tool to guide conservation and
restoration of rivers and the biodiversity they support.
CRediT authorship contribution statement
Suman Jumani: Conceptualization, Funding acquisition, Writing –
original draft. Matthew J. Deitch:: Conceptualization, Writing – review
& editing. Denis Valle: Conceptualization, Writing – review & editing.
Siddarth Machado: Software, Writing – review & editing. Vincent
Lecours: Writing – review & editing. David Kaplan: Writing – review &
editing. Jagdish Krishnaswamy: Writing – review & editing. Jeanette
Howard: Writing – review & editing.
Declaration of Competing Interest
The authors declare that they have no known competing nancial
interests or personal relationships that could have appeared to inuence
the work reported in this paper.
Acknowledgements
We thank Kurt Fesenmyer for proving spatial layers for the Klamath
River Basin, as well as feedback on this paper. We also thank the editor
and two anonymous reviewers for their comments. Funding for this
study was provided in part by the Rufford Foundation (31454-B).
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.ecolind.2022.108680.
References
Anderson, E.P., Jenkins, C.N., Heilpern, S., Maldonado-Ocampo, J.A., Carvajal-
Vallejos, F.M., Encalada, A.C., Rivadeneira, J.F., Hidalgo, M., Ca˜
nas, C.M.,
Ortega, H., Salcedo, N., Maldonado, M., Tedesco, P.A., 2018. Fragmentation of
Andes-to-Amazon connectivity by hydropower dams. Sci. Adv. 4, eaao1642. https://
doi.org/10.1126/sciadv.aao1642.
Atkinson, S., Bruen, M., O’ Sullivan, J.J., Turner, J.N., Ball, B., Carlsson, J., Bullock, C.,
Casserly, C.M., Kelly-Quinn, M., 2020. An inspection-based assessment of obstacles
to salmon, trout, eel and lamprey migration and river channel connectivity in
Ireland. Sci. Total Environ. 719, 137215. https://doi.org/10.1016/j.
scitotenv.2020.137215.
Atkore, V., Kelkar, N., Krishnaswamy, J., 2017. Assessing the recovery of sh
assemblages downstream of hydrological barriers in India’s Western Ghats. River
Res. Appl. 33 (7), 1026–1035. https://doi.org/10.1002/rra.v33.710.1002/rra.3163.
Baker, M.E., Weller, D.E., Jordan, T.E., 2007. Effects of stream map resolution on
measures of riparian buffer distribution and nutrient retention potential. Landscape
Ecol. 22 (7), 973–992. https://doi.org/10.1007/s10980-007-9080-z.
Belletti, B., Garcia de Leaniz, C., Jones, J., Bizzi, S., B¨
orger, L., Segura, G., Castelletti, A.,
van de Bund, W., Aarestrup, K., Barry, J., Belka, K., Berkhuysen, A., Birnie-
Gauvin, K., Bussettini, M., Carolli, M., Consuegra, S., Dopico, E., Feierfeil, T.,
Fern´
andez, S., Fernandez Garrido, P., Garcia-Vazquez, E., Garrido, S., Giannico, G.,
Gough, P., Jepsen, N., Jones, P.E., Kemp, P., Kerr, J., King, J., Łapi´
nska, M.,
L´
azaro, G., Lucas, M.C., Marcello, L., Martin, P., McGinnity, P., O’Hanley, J., Olivo
del Amo, R., Parasiewicz, P., Pusch, M., Rincon, G., Rodriguez, C., Royte, J.,
Schneider, C.T., Tummers, J.S., Vallesi, S., Vowles, A., Verspoor, E., Wanningen, H.,
Wantzen, K.M., Wildman, L., Zalewski, M., 2020. More than one million barriers
fragment Europe’s rivers. Nature 588 (7838), 436–441. https://doi.org/10.1038/
s41586-020-3005-2.
Bourne, C.M., Kehler, D.G., Wiersma, Y.F., Cote, D., 2011. Barriers to sh passage and
barriers to sh passage assessments: the impact of assessment methods and
assumptions on barrier identication and quantication of watershed connectivity.
Aquat. Ecol. 45 (3), 389–403. https://doi.org/10.1007/s10452-011-9362-z.
Branco, P., Segurado, P., Santos, J.M., Ferreira, M.T., Strecker, A., 2014. Prioritizing
barrier removal to improve functional connectivity of rivers. J. Appl. Ecol. 51 (5),
1197–1206. https://doi.org/10.1111/1365-2664.12317.
Buddendorf, W.B., Malcolm, I.A., Geris, J., Wilkinson, M.E., Soulsby, C., 2017. Metrics to
assess how longitudinal channel network connectivity and in-stream Atlantic salmon
habitats are impacted by hydropower regulation. Hydrol. Processes 31 (12),
2132–2142. https://doi.org/10.1002/hyp.v31.1210.1002/hyp.11159.
Burgers, H.E.R., Schipper, A.M., Jan Hendriks, A., 2014. Size relationships of water
discharge in rivers: scaling of discharge with catchment area, main-stem length and
precipitation. Hydrol. Processes 28 (23), 5769–5775. https://doi.org/10.1002/
hyp.10087.
Calabrese, J.M., Fagan, W.F., 2004. A comparison-shopper’s guide to connectivity
metrics. Front. Ecol. Environ. 2, 529–536. https://doi.org/10.2307/3868383.
Choy, M., Lawrie, D., Edge, C.B., 2018. Measuring 30 years of improvements to aquatic
connectivity in the Greater Toronto Area. Aquat. Ecosyst. Health Manag. 21,
342–351. https://doi.org/10.1080/14634988.2018.1497400.
Colvin, S.A.R., Sullivan, S.M.P., Shirey, P.D., Colvin, R.W., Winemiller, K.O., Hughes, R.
M., Fausch, K.D., Infante, D.M., Olden, J.D., Bestgen, K.R., Danehy, R.J., Eby, L.,
2019. Headwater streams and wetlands are critical for sustaining sh, sheries, and
ecosystem services. Fisheries 44, 73–91. https://doi.org/10.1002/fsh.10229.
Consuegra, S., O’Rorke, R., Rodriguez-Barreto, D., Fernandez, S., Jones, J., Garcia de
Leaniz, C., 2021. Impacts of large and small barriers on sh assemblage composition
assessed using environmental DNA metabarcoding. Sci. Total Environ. 790, 148054.
https://doi.org/10.1016/j.scitotenv.2021.148054.
Cote, D., Kehler, D.G., Bourne, C., Wiersma, Y.F., 2009. A new measure of longitudinal
connectivity for stream networks. Landscape Ecol. 24 (1), 101–113. https://doi.org/
10.1007/s10980-008-9283-y.
Couto, T.BA., Olden, J.D., 2018. Global proliferation of small hydropower plants –
science and policy. Front. Ecol. Environ. 16 (2), 91–100. https://doi.org/10.1002/
fee.1746.
Dark, S.J., Bram, D., 2007. The modiable areal unit problem (MAUP) in physical
geography. Prog. Phys. Geogr.: Earth Environ. 31 (5), 471–479. https://doi.org/
10.1177/0309133307083294.
Deitch, M.J., Kondolf, G.M., 2012. Consequences of variations in magnitude and duration
of an instream environmental ow threshold across a longitudinal gradient.
J. Hydrol. 420–421, 17–24. https://doi.org/10.1016/j.jhydrol.2011.11.003.
Díaz, G., Arriagada, P., G´
orski, K., Link, O., Karelovic, B., Gonzalez, J., Habit, E., 2019.
Fragmentation of Chilean Andean rivers: expected effects of hydropower
development. Revista Chilena de Historia Natural 92, 1. https://doi.org/10.1186/
s40693-019-0081-5.
Diebel, M.W., Fedora, M., Cogswell, S., O’Hanley, J.R., 2015. Effects of road crossings on
habitat connectivity for stream-resident sh. River Res. Appl. 31 (10), 1251–1261.
https://doi.org/10.1002/rra.2822.
Edge, C.B., Fortin, M.-J., Jackson, D.A., Lawrie, D., Staneld, L., Shrestha, N., 2017.
Habitat alteration and habitat fragmentation differentially affect beta diversity of
S. Jumani et al.
Ecological Indicators 136 (2022) 108680
13
stream sh communities. Landscape Ecol. 32 (3), 647–662. https://doi.org/
10.1007/s10980-016-0472-9.
Erskine, R.H., Green, T.R., Ramirez, J.A., MacDonald, L.H., 2006. Comparison of grid-
based algorithms for computing upslope contributing area. Water Resour. Res. 42 (9)
https://doi.org/10.1029/2005WR004648.
Fagan, W.F., 2002. Connectivity, fragmentation, and extinction risk in dendritic
metapopulations. Ecology 83, 3243–3249. https://doi.org/10.1890/0012-9658
(2002)083[3243:CFAERI]2.0.CO;2.
Fagan, W.F., Unmack, P.J., Burgess, C., Minckley, W.L., 2002. RARITY,
FRAGMENTATION, AND EXTINCTION RISK IN DESERT FISHES. Ecology 83,
3250–3256. https://doi.org/10.1890/0012-9658(2002)083[3250:RFAERI]2.0.CO;
2.
Fick, S.E., Hijmans, R.J., 2017. WorldClim 2: new 1-km spatial resolution climate
surfaces for global land areas. Int. J. Climatol. 37 (12), 4302–4315. https://doi.org/
10.1002/joc.5086.
Fullerton, A.H., Burnett, K.M., Steel, E.A., Flitcroft, R.L., Pess, G.R., Feist, B.E.,
Torgersen, C.E., Miller, D.J., Sanderson, B.L., 2010. Hydrological connectivity for
riverine sh: measurement challenges and research opportunities. Freshw. Biol. 55
(11), 2215–2237. https://doi.org/10.1111/j.1365-2427.2010.02448.x.
Galster, J.C., Pazzaglia, F.J., Hargreaves, B.R., Morris, D.P., Peters, S.C., Weisman, R.N.,
2006. Effects of urbanization on watershed hydrology: The scaling of discharge with
drainage area. Geology 34 (9), 713. https://doi.org/10.1130/G22633.110.1130/
2006149.
Gaucherel, C., 2007. Multiscale heterogeneity map and associated scaling prole for
landscape analysis. Landscape Urban Planning 82 (3), 95–102. https://doi.org/
10.1016/j.landurbplan.2007.01.022.
Ghaffari, G., 2011. The impact of DEM resolution on runoff and sediment modelling
results. Res. J. Environ. Sci. 5 (8), 691–702. https://doi.org/10.3923/
rjes.2011.691.702.
Godinho, A.L., Kynard, B., Godinho, H.P., 2007. Migration and spawning of female
surubim (Pseudoplatystoma corruscans, Pimelodidae) in the S˜
ao Francisco river,
Brazil. Environ. Biol. Fish. 80 (4), 421–433. https://doi.org/10.1007/s10641-006-
9141-1.
Gosnell, H., Kelly, E.C., 2010. Peace on the River? Social-ecological restoration and large
dam removal in the Klamath Basin, USA. Water Alternatives 3, 361–383.
Grill, G., Lehner, B., Lumsdon, A.E., MacDonald, G.K., Zar, C., Reidy Liermann, C.,
2015. An index-based framework for assessing patterns and trends in river
fragmentation and ow regulation by global dams at multiple scales. Environ. Res.
Lett. 10 (1), 015001. https://doi.org/10.1088/1748-9326/10/1/015001.
Grill, G., Lehner, B., Thieme, M., Geenen, B., Tickner, D., Antonelli, F., Babu, S.,
Borrelli, P., Cheng, L., Crochetiere, H., Ehalt Macedo, H., Filgueiras, R., Goichot, M.,
Higgins, J., Hogan, Z., Lip, B., McClain, M.E., Meng, J., Mulligan, M., Nilsson, C.,
Olden, J.D., Opperman, J.J., Petry, P., Reidy Liermann, C., S´
aenz, L., Salinas-
Rodríguez, S., Schelle, P., Schmitt, R.J.P., Snider, J., Tan, F., Tockner, K., Valdujo, P.
H., van Soesbergen, A., Zar, C., 2019. Mapping the world’s free-owing rivers.
Nature 569 (7755), 215–221. https://doi.org/10.1038/s41586-019-1111-9.
Grill, G., Ouellet Dallaire, C., Fluet Chouinard, E., Sindorf, N., Lehner, B., 2014.
Development of new indicators to evaluate river fragmentation and ow regulation
at large scales: a case study for the Mekong River Basin. Ecol. Ind. 45, 148–159.
https://doi.org/10.1016/j.ecolind.2014.03.026.
Hamilton, J.B., Curtis, G.L., Snedaker, S.M., White, D.K., 2005. Distribution of
anadromous shes in the upper klamath river watershed prior to hydropower
dams—A synthesis of the historical evidence. Fisheries 30, 10–20. https://doi.org/
10.1577/1548-8446(2005)30[10:DOAFIT]2.0.CO;2.
Hu, M., Hua, Q.i., Zhou, H., Wu, Z., Wu, X., 2015. The effect of dams on the larval
abundance and composition of four carp species in key river systems in China.
Environ. Biol. Fish. 98 (4), 1201–1205. https://doi.org/10.1007/s10641-014-0342-
8.
Jaeger, K.L., Olden, J.D., Pelland, N.A., 2014. Climate change poised to threaten
hydrologic connectivity and endemic shes in dryland streams. Proc. Natl. Acad. Sci.
111 (38), 13894–13899. https://doi.org/10.1073/pnas.1320890111.
Jumani, S., Deitch, M.J., Kaplan, D., Anderson, E.P., Krishnaswamy, J., Lecours, V.,
Whiles, M.R., 2020. River fragmentation and ow alteration metrics: a review of
methods and directions for future research. Environ. Res. Lett. 15 (12), 123009.
https://doi.org/10.1088/1748-9326/abcb37.
Jumani, S., Rao, S., Kelkar, N., Machado, S., Krishnaswamy, J., Vaidyanathan, S., 2018.
Fish community responses to stream ow alterations and habitat modications by
small hydropower projects in the Western Ghats biodiversity hotspot, India. Aquat.
Conserv.: Mar. Freshw. Ecosyst. 28 (4), 979–993. https://doi.org/10.1002/
aqc.2904.
Kemp, P.S., O’Hanley, J.R., 2010. Procedures for evaluating and prioritising the removal
of sh passage barriers: a synthesis: EVALUATION OF FISH PASSAGE BARRIERS.
Fisheries Management and Ecology no-no. https://doi.org/10.1111/j.1365-
2400.2010.00751.x.
Lehner, B., Verdin, K., Jarvis, A., 2008. New global hydrography derived from
Spaceborne elevation data. Eos, Trans. Am. Geophys. Union 89 (10), 93. https://doi.
org/10.1029/2008EO100001.
Llaus`
as, A., Nogu´
e, J., 2012. Indicators of landscape fragmentation: the case for
combining ecological indices and the perceptive approach. Ecol. Ind. 15 (1), 85–91.
https://doi.org/10.1016/j.ecolind.2011.08.016.
McCully, P., 1996. Silenced Rivers: the Ecology and Politics of Large Dams. Zed Books,
London, Atlantic Highlands, N.J., USA.
McKay, S.K., Cooper, A.R., Diebel, M.W., Elkins, D., Oldford, G., Roghair, C.,
Wieferich, D., 2017. Informing watershed connectivity barrier prioritization
decisions: a synthesis. River Res. Appl. 33 (6), 847–862. https://doi.org/10.1002/
rra.3021.
McKay, S.K., Schramski, J.R., Conyngham, J.N., Fischenich, J.C., 2013. Assessing
upstream sh passage connectivity with network analysis. Ecol. Appl. 23 (6),
1396–1409. https://doi.org/10.1890/12-1564.1.
McManamay, R.A., Samu, N., Kao, S.-C., Bevelhimer, M.S., Hetrick, S.C., 2015. A multi-
scale spatial approach to address environmental effects of small hydropower
development. Environ. Manage. 55 (1), 217–243. https://doi.org/10.1007/s00267-
014-0371-2.
Meyer, J.L., Strayer, D.L., Wallace, J.B., Eggert, S.L., Helfman, G.S., Leonard, N.E., 2007.
The contribution of headwater streams to biodiversity in river networks1. JAWRA J.
Am. Water Resour. Assoc. 43, 86–103. https://doi.org/10.1111/j.1752-
1688.2007.00008.x.
Murphy, P.N.C., Ogilvie, J., Meng, F.-R., Arp, P., 2008. Stream network modelling using
lidar and photogrammetric digital elevation models: a comparison and eld
verication. Hydrol. Processes 22 (12), 1747–1754. https://doi.org/10.1002/
hyp.6770.
Nel, J.L., Roux, D.J., Abell, R., Ashton, P.J., Cowling, R.M., Higgins, J.V., Thieme, M.,
Viers, J.H., 2009. Progress and challenges in freshwater conservation planning.
Aquat. Conserv.: Mar. Freshw. Ecosyst. 19 (4), 474–485. https://doi.org/10.1002/
aqc.1010.
Nilsson, C., 2005. Fragmentation and ow regulation of the world’s large river systems.
Science 308 (5720), 405–408. https://doi.org/10.1126/science:1107887.
Pandit, M.K., Grumbine, R.E., 2012. Potential effects of ongoing and proposed
hydropower development on terrestrial biological diversity in the Indian Himalaya:
hydropower and Himalayan biodiversity. Conserv. Biol. 26 (6), 1061–1071. https://
doi.org/10.1111/j.1523-1739.2012.01918.x.
Perkin, J.S., Gido, K.B., Cooper, A.R., Turner, T.F., Osborne, M.J., Johnson, E.R.,
Mayes, K.B., 2015. Fragmentation and dewatering transform Great Plains stream sh
communities. Ecol. Monogr. 85 (1), 73–92. https://doi.org/10.1890/14-
0121.110.1890/14-0121.1.sm.
Poff, N.L., Hart, D.D., 2002. How dams vary and why it matters for the emerging science
of dam removal. BioScience 52, 659. https://doi.org/10.1641/0006-3568(2002)052
[0659:HDVAWI]2.0.CO;2.
Poff, N.L., Zimmerman, J.K.H., 2010. Ecological responses to altered ow regimes: a
literature review to inform the science and management of environmental ows:
review of altered ow regimes. Freshw. Biol. 55, 194–205. https://doi.org/10.1111/
j.1365-2427.2009.02272.x.
Pringle, C., 2003. What is hydrologic connectivity and why is it ecologically important?
Hydrol. Processes 17 (13), 2685–2689. https://doi.org/10.1002/hyp.5145.
Richter, B.D., Baumgartner, J.V., Powell, J., Braun, D.P., 1996. A method for assessing
hydrologic alteration within ecosystems. Conserv. Biol. 10 (4), 1163–1174. https://
doi.org/10.1046/j.1523-1739.1996.10041163.x.
Rodeles, A.A., Galicia, D., Miranda, R., 2021. A simple method to assess the
fragmentation of freshwater sh meta-populations: implications for river
management and conservation. Ecol. Ind. 125, 107557. https://doi.org/10.1016/j.
ecolind.2021.107557.
Rodeles, A.A., Galicia, D., Miranda, R., 2020. A new method to include sh biodiversity
in river connectivity indices with applications in dam impact assessments. Ecol. Ind.
117, 106605. https://doi.org/10.1016/j.ecolind.2020.106605.
Rodeles, A.A., Leunda, P.M., Elso, J., Ardaiz, J., Galicia, D., Miranda, R., 2019.
Consideration of habitat quality in a river connectivity index for anadromous shes.
Inland Waters 9 (3), 278–288. https://doi.org/10.1080/20442041.2018.1544817.
Schick, R.S., Lindley, S.T., 2007. Directed connectivity among sh populations in a
riverine network. J. Appl. Ecol. 44, 1116–1126. https://doi.org/10.1111/j.1365-
2664.2007.01383.x.
Segurado, P., Branco, P., Ferreira, M.T., 2013. Prioritizing restoration of structural
connectivity in rivers: a graph based approach. Landscape Ecol. 28 (7), 1231–1238.
https://doi.org/10.1007/s10980-013-9883-z.
Simley, J.D., Carswell Jr, W.J., 2009. The national map—hydrography. US Geological
Survey Fact Sheet 3054.
Sukumaran, H., Sahoo, S.N., 2020. A methodological framework for identication of
baseline scenario and assessing the impact of DEM scenarios on SWAT model
outputs. Water Resour. Manage. 34 (15), 4795–4814. https://doi.org/10.1007/
s11269-020-02691-5.
Tan, M.L., Ramli, H.P., Tam, T.H., 2018. Effect of DEM resolution, source, resampling
technique and area threshold on SWAT outputs. Water Resour. Manage. 32 (14),
4591–4606. https://doi.org/10.1007/s11269-018-2072-8.
Tockner, K., Bernhardt, E.S., Koska, A., Zar, C., 2016. A Global View on Future Major
Water Engineering Projects. In: Hüttl, R.F., Bens, O., Bismuth, C., Hoechstetter, S.
(Eds.), Society - Water - Technology. Springer International Publishing, Cham,
pp. 47–64. https://doi.org/10.1007/978-3-319-18971-0_4.
USFWS, 2012. Revised recovery plan for the Lost River sucker (Deltistes luxatus) and
shortnose sucker r (Chasmistes brevirostris). U.S. Fish and Wildlife Service, Pacic
Southwest Region, Sacramento, California.
Winston, M.R., Taylor, C.M., Pigg, J., 1991. Upstream extirpation of four minnow species
due to damming of a prairie stream. Trans. Am. Fish. Soc. 120, 98–105. https://doi.
org/10.1577/1548-8659(1991)120<0098:UEOFMS>2.3.CO;2.
Zar, C., Lumsdon, A.E., Berlekamp, J., Tydecks, L., Tockner, K., 2015. A global boom in
hydropower dam construction. Aquat. Sci. 77 (1), 161–170. https://doi.org/
10.1007/s00027-014-0377-0.
Ziv, G., Baran, E., Nam, S., Rodriguez-Iturbe, I., Levin, S.A., 2012. Trading-off sh
biodiversity, food security, and hydropower in the Mekong River Basin. Proc. Natl.
Acad. Sci. 109 (15), 5609–5614. https://doi.org/10.1073/pnas.1201423109.
S. Jumani et al.
