EFFECTS OF DAMS ON STREAMS OF THE CONTERMINOUS UNITED STATES:
CHARACTERIZING PATTERNS IN HABITAT FRAGMENTATION NATIONALLY AND
FLUVIAL FISH RESPONSE IN THE MIDWEST
By
Arthur Raymond Cooper

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Fisheries and Wildlife - Master of Science
2013

ABSTRACT
EFFECTS OF DAMS ON STREAMS OF THE CONTERMINOUS UNITED STATES:
CHARACTERIZING PATTERNS IN HABITAT FRAGMENTATION NATIONALLY AND
FLUVIAL FISH RESPONSE IN THE MIDWEST
by
Arthur Raymond Cooper
Dams can exert great influence on fluvial habitats through a variety of mechanisms,
however spatial measures representing dam locations throughout river networks, along with their
attributes (e.g. reservoir storage), are not available in a consistent, comparable manner for the
conterminous U.S. In this study, spatial metrics are developed that account for fragmentation
and alteration of river networks by large dams throughout the conterminous U.S., allowing for
the examination of river network fragmentation patterns by stream size and ecoregion. Results
show that streams in the conterminous U.S. have been heavily fragmented by dams, with the
greatest dam influence tending to occur in large and great rivers due to cumulative dam effects
along river networks. Using a subset of fragmentation metrics generated in this study, fish
species considered to be most sensitive to dam influences were identified for streams in
Michigan, Wisconsin, and Minnesota. Of the sensitive species identified, those that were
positively associated with greater dam effects were predominantly fishes associated with warm
water temperatures, large river habitats, and/or lentic habitats, while species negatively
associated with greater dam effects were cold and coolwater lotic species, suggesting a
combination of downstream thermal effects and upstream influences from impoundments
generated by dams. With dams representing an aging infrastructure leading to likely increases in
habitat restoration and dam management opportunities, it will be essential to further reveal the
spatial influence of dams along the river network.

Copyright by
ARTHUR RAYMOND COOPER
2013

This work is dedicated to my wife Rebecca and sons Gabriel and Maxwell, and to my parents
Marion Bradford and Raymond Cooper.

iv

ACKNOWLEDGEMENTS

I’d like to acknowledge my committee members, Drs. Dana Infante (co-advisor), Lizhu
Wang (co-advisor), Travis Brenden, and Kevin Wehrly for their guidance and feedback
throughout the development of this thesis. I’d like to thank members of the Aquatic Landscape
Ecology Lab for their support and many helpful discussions regarding analytical approaches and
for the extensive programming support provided by Jason Breck and Dr. Edward Rutherford.

v

TABLE OF CONTENTS

LIST OF TABLES ..................................................................................................................... viii
LIST OF FIGURES ..................................................................................................................... xi
CHAPTER 1
FRAGMENTATION BY DAMS: SPATIAL MEASURES OF DAM EFFECTS ON STREAMS
OF THE CONTERMINOUS U.S ....................................................................................................1
ABSTRACT ........................................................................................................................2
INTRODUCTION..............................................................................................................3
Characterizing dam effects from a landscape scale ................................................3
METHODS .........................................................................................................................6
Study area and spatial framework ...........................................................................6
Study area ....................................................................................................6
Stream network data ....................................................................................6
Dam data ......................................................................................................7
Fragmentation metrics .................................................................................8
Artificial habitat patches..............................................................................9
Metric reduction and description of fragmentation patterns .....................10
RESULTS .........................................................................................................................11
Patterns in dam characteristics .............................................................................11
Dam count and density...............................................................................11
Dam age and height ...................................................................................11
Dam storage and degree of regulation ......................................................12
Dam purpose ..............................................................................................13
Patterns in stream segment fragmentation metrics ...............................................14
Principal component analysis results ........................................................14
Select metric statistics ................................................................................14
Artificial habitat patches........................................................................................16
DISCUSSION ...................................................................................................................17
Implications for fisheries management and conservation .....................................18
Implications for restoration ...................................................................................20
Potential improvements .........................................................................................22
Conclusion .............................................................................................................23
APPENDICES ..................................................................................................................24
APPENDIX A: TABLES .......................................................................................25
APPENDIX B: FIGURES .....................................................................................39
LITERATURE CITED ...................................................................................................51

vi

CHAPTER 2
EFFECTS OF DAMS ON FLUVIAL FISH ASSEMBLAGES IN MICHIGAN, WISCONSIN,
AND MINNESOTA: CONSIDERATION OF MULTIPLE MEASURES OF
FRAGMENTATION AND DAM INFLUENCE ..........................................................................56
ABSTRACT ......................................................................................................................57
INTRODUCTION............................................................................................................58
METHODS .......................................................................................................................62
Study area and spatial framework .........................................................................62
Study area ..................................................................................................62
Spatial data ................................................................................................63
Fish data ....................................................................................................63
Site stratification ........................................................................................64
Statistical approaches ............................................................................................65
Data preparation and variable reduction ..................................................65
Detecting fish relationships with fragmentation metrics ...........................66
Canonical correspondence analysis ..........................................................67
RESULTS .........................................................................................................................69
Site landscape characteristics................................................................................69
Natural landscape characteristics .............................................................69
Non-dam anthropogenic landscape factors ...............................................69
Dam metrics ...............................................................................................70
Fish species response to fragmentation .................................................................70
Interpretation of results .............................................................................70
Cold headwaters ........................................................................................71
Warm headwaters ......................................................................................72
Cold mid-size .............................................................................................72
Warm mid-size ...........................................................................................73
Warm large size .........................................................................................74
Region-wide trends ....................................................................................74
Canonical correspondence analysis results...........................................................75
DISCUSSION ...................................................................................................................77
Potential improvements .........................................................................................82
Conclusion .............................................................................................................83
APPENDICES ..................................................................................................................84
APPENDIX A: TABLES .......................................................................................85
APPENDIX B: FIGURES ...................................................................................105
APPENDIX C: SUPPLEMENTAL TABLES .....................................................111
LITERATURE CITED .................................................................................................125

vii

LIST OF TABLES

Table 1.1: Dam metric descriptions, units, direction of metric indicating increasing levels of
fragmentation, and sign of individual species/metric relationship to indicate positive
response of species to increased fragmentation (indicated by increasing catch per unit
effort). For the principal component and canonical correspondence analyses, distancebased metrics were given the maximum value for that metric within the entire database
in cases where upstream and/or downstream mainstem dams were absence for a given
record .............................................................................................................................25
Table 1.2: Descriptive statistics (first quartile, median, and third quartile) for dam characteristics
for the conterminous U.S. (CONUS) and by ecoregion. Dam densities were calculated
per 100 km of network length within a given stratum and region. Degree of regulation
(UDOR) was calculated using cumulative upstream reservoir storage divided
estimated mean annual stream flow and had a total sample size of 34,175. Based on
the available information within the NABD dataset, a total sample size of 43,658 was
used for age, 49,356 for height, and 49,468 for storage ................................................27
Table 1.3: Main dam purpose both for the conterminous U.S. (CONUS) and by ecoregion with
values expressed as a percentage. Flood = Flood control and stormwater management;
Hydro = Hydroelectric power generation; Supply = Water supply; Fire/Farm = Fire
protection, stock, and small farm ponds; Fish/Wild = Fish and wildlife ponds; Debris =
Debris control ................................................................................................................30
Table 1.4: Principal component analysis results for the conterminous U.S. and for the NAP and
SPL ecoregions with fragmentation metric weights by axes. Weights with an absolute
value of 0.7 or greater, shown in bold, were used in the axis interpretation found at the
top of each axis column. Cumulative percentage of variation explained is located at
the bottom of each axis column.....................................................................................33
Table 1.5: Descriptive statistics (first quartile, median, and third quartile) for a subset of segmentlevel fragmentation metrics, including upstream mainstem openness (UMO) expressed
as a percentage, upstream river network density (UNDR), cumulative upstream degree
of regulation (UDOR), downstream distance to mainstem dam (DM2D), and
downstream mainstem dam density (DMD). ................................................................35
Table 1.6: Patch count, median size characteristics, and percent increase or decrease in size
measures by size strata for the conterminous U.S. when comparing patches in the
absence of dams and when considering dam locations as artificial habitat patch
boundaries .....................................................................................................................38
Table 2.1: Source datasets for the natural and non-dam anthropogenic variables summarized for
network catchments of stream reaches ..........................................................................85

viii

Table 2.2: Number of sites for each ecoregion, stream size, and temperature class grouping.
Numbers shown in bold indicate groupings used in the univariate threshold and
correlation analyses. For the multivariate CCA analysis, sites used in the univariate
analyses were grouped across ecoregions within a given size/temperature stratum
indicated by the CCA Total. A dash (-) indicates that there were no sites for a given
stratum or grouping .......................................................................................................87
Table 2.3: Number of species associations by strata. Overall association of species with greater
levels of fragmentation are summarized as positive, negative, or mixed (See Results
for definitions) ...............................................................................................................88
Table 2.4: Significant change point threshold (t) and Spearman correlation analysis (s) results
characterizing relationships between fish species catch per unit effort and selected dam
metrics (Table 1.1). For Spearman correlation, significance was determined by
p<0.05; see Methods for criteria used to determine significance of threshold results.
Overall association of species with greater levels of fragmentation are summarized as
positive (P), negative (N), or mixed (M). Asterisks (*) indicate an insufficient number
of species occurrences to perform analyses for a given metric. Results shown are for
the cold headwater strata in the EWL, LGL, and UM ecoregions ................................89
Table 2.5: Significant change point threshold (t) and Spearman correlation analysis (s) results
characterizing relationships between fish species catch per unit effort and selected dam
metrics (Table 1.1). For Spearman correlation, significance was determined by
p<0.05; see Methods for criteria used to determine significance of threshold results.
Overall association of species with greater levels of fragmentation are summarized as
positive (P), negative (N), or mixed (M). Asterisks (*) indicate an insufficient number
of species occurrences to perform analyses for a given metric. Results shown are for
the warm headwater strata in the LGL and UM ecoregions..........................................92
Table 2.6: Significant change point threshold (t) and Spearman correlation analysis (s) results
characterizing relationships between fish species catch per unit effort and selected dam
metrics (Table 1.1). For Spearman correlation, significance was determined by
p<0.05; see Methods for criteria used to determine significance of threshold results.
Overall association of species with greater levels of fragmentation are summarized as
positive (P), negative (N), or mixed (M). Asterisks (*) indicate an insufficient number
of species occurrences to perform analyses for a given metric. Results shown are for
the cold mid-size strata in the LGL and UM ecoregions ..............................................93
Table 2.7: Significant change point threshold (t) and Spearman correlation analysis (s) results
characterizing relationships between fish species catch per unit effort and selected dam
metrics (Table 1.1). For Spearman correlation, significance was determined by
p<0.05; see Methods for criteria used to determine significance of threshold results.
Overall association of species with greater levels of fragmentation are summarized as
positive (P), negative (N), or mixed (M). Asterisks (*) indicate an insufficient number
of species occurrences to perform analyses for a given metric. Results shown are for
the warm mid-size strata in the EWL, LGL, and UM ecoregions.................................96

ix

Table 2.8: Significant change point threshold (t) and Spearman correlation analysis (s) results
characterizing relationships between fish species catch per unit effort and selected dam
metrics (Table 1.1). For Spearman correlation, significance was determined by
p<0.05; see Methods for criteria used to determine significance of threshold results.
Overall association of species with greater levels of fragmentation are summarized as
positive (P), negative (N), or mixed (M). Asterisks (*) indicate an insufficient number
of species occurrences to perform analyses for a given metric. Results shown are for
the warm large size strata in the EWL, LGL, and UM ecoregions .............................100
Table 2.9: Overall set of sensitive species (highlighted in gray) within the study region with
species associations expressed as positive (P), negative (N), mixed (M), and no
association (#) by stratum. Blanks indicate that no analyses were run either due to
species absence within a strata or an insufficient number of species occurrences.
Species selected for CCA analysis are shown with numeric superscripts ..................102
Table 2.10: Percentage of variation explained in selected fish response variables by thermal/size
class using canonical correspondence analysis, subdivided into the percentage of
variation explained attributed to natural, non-dam anthropogenic, fragmentation, and
interaction components .............................................................................................104
Table C1: Descriptive statistics for natural and non-dam anthropogenic reach network catchment
variables (Table 2.1) by stratum ..................................................................................111
Table C2: Descriptive statistics for fragmentation metrics (Table 1.1) by stratum .....................118

x

LIST OF FIGURES

Figure 1.1: Distribution of the nine aggregated ecoregions (A) and dam locations (B; N = 49,468)
of the conterminous U.S. For interpretation of the references to color in this and all
other figures, the reader is referred to the electronic version of this thesis .................39
Figure 1.2: Example depiction of an NHDPlusV1 stream reach and catchment (A) and its
subdivision into stream segments with subsequent delineation of segment catchments
(B) for dams occurring greater than 100 m from an existing node in the NHDPlusV1
......................................................................................................................................40
Figure 1.3: Depiction of the traditional scales of catchment delineation (A), consisting of local
catchments (land area draining directly to a stream reach or water body) and network
catchments (cumulative upstream drainage area, including the local catchment of the
target stream reach or water body). In contrast, an alternative set of catchments for
artificial habitat patches (AHPs) are defined by the locations of dams (B).................41
Figure 1.4: Dam density distributions by size strata for the conterminous U.S (CONUS) and by
ecoregion. Dam densities were calculated per 100 km of network length within a
given stratum and region ..............................................................................................42
Figure 1.5: Number of dams constructed by decade (source NABD) ............................................43
Figure 1.6: Distribution of dam age in years for headwater (HW) and creek (CR) size strata for
th
the conterminous U.S. (CUS) and ecoregions. Box plot whiskers represent the 5
th

and 95 percentiles ......................................................................................................44
Figure 1.7: Distribution of dam age in years for small river (SR) and medium river (MR) size
strata for the conterminous U.S. (CUS) and ecoregions. Box plot whiskers represent
th
th
the 5 and 95 percentiles ..........................................................................................45
Figure 1.8: Distribution of dam age in years for large river (LR) and great river (GR) size strata
th
for the conterminous U.S. (CUS) and ecoregions. Box plot whiskers represent the 5
th

and 95 percentiles ......................................................................................................46
Figure 1.9: Stacked bar graph of overall reservoir storage showing total contribution of each size
stratum for the conterminous U.S. (CUS) and ecoregions...........................................47
Figure 1.10: Cumulative upstream degree of regulation (%) for great river segments for the
conterminous U.S. (CUS) and by ecoregion (total N = 18,752). Box plot whiskers
th
th
represent the 5 and 95 percentiles ........................................................................48

xi

Figure 1.11: Individual patches in the absence of dams (A; N = 6,037) and the resulting set of
artificial habitat patches when accounting for dams (B, N = 54,120) .......................49
Figure 1.12: Artificial habitat patches (AHPs) for the northeast U.S. with inset of AHPs for the
Saco River (basin highlighted in red in top view). For the bottom view, individual
AHPs are given a unique color, with dam locations represented as red points and
streams as blue lines ..................................................................................................50
Figure 2.1: Locations of fish survey sites (N = 2,067; A) and dam locations (N = 2,303; B) within
ecoregions of the study region ...................................................................................105
Figure 2.2: Percentage of variation explained in selected fish indicators partitioned into natural,
non-dam anthropogenic, fragmentation by dams, and interaction components using
canonical correspondence analysis for cold headwaters (A), warm headwaters (B),
cold mid-size (C), and warm mid-size (D) classes ....................................................106
Figure 2.3: Plot of canonical correspondence axis I vs. axis II for cold headwaters. Abbreviations
for natural, non-dam anthropogenic, and dam variables are found in Tables 1.1 and
2.1. Species common names are found in Table 2.9 .................................................107
Figure 2.4: Plot of canonical correspondence axis I vs. axis II for warm headwaters.
Abbreviations for natural, non-dam anthropogenic, and dam variables are found in
Tables 1.1 and 2.1. Species common names are found in Table 2.9 ........................108
Figure 2.5: Plot of canonical correspondence axis I vs. axis II for cold mid-size streams.
Abbreviations for natural, non-dam anthropogenic, and dam variables are found in
Tables 1.1 and 2.1. Species common names are found in Table 2.9 ........................109
Figure 2.6: Plot of canonical correspondence axis I vs. axis II for warm mid-size streams.
Abbreviations for natural, non-dam anthropogenic, and dam variables are found in
Tables 1.1 and 2.1. Species common names are found in Table 2.9 ........................110

xii

CHAPTER 1
FRAGMENTATION BY DAMS: SPATIAL MEASURES OF DAM EFFECTS ON STREAMS
OF THE CONTERMINOUS U.S.

1

ABSTRACT
Dams can have wide-ranging effects on fluvial habitats including fragmentation of river
networks, flow modification, and conversion of streams to lentic-like impoundments. Further,
efforts to represent the various effects of dams in river networks over very large geographic
extents, including the conterminous U.S., are lacking, highlighting the need for diverse spatial
measures to account for dam influences throughout large landscapes. In this study, multiple
types of metrics were assembled to characterize dams throughout nine large ecoregions of the
conterminous U.S using ~50,000 georeferenced dams from the 2012 National Anthropogenic
Barrier Dataset and 2.3 million stream reaches from the National Hydrography Dataset Plus
Version 1. Metrics occur in three groups: 1) individual dam characteristics (e.g. age, height), 2)
stream segment-level descriptors of fragmentation (e.g. distances-to-dams, cumulative reservoir
storage), and 3) patch-level descriptors of fragmentation, summarized by adjacent sets of stream
segments and catchments that account for dam locations. Comparison of dam characteristics and
spatial measures shows great variability in dam characteristics and fragmentation patterns by
stream size and ecoregion. Examination of segment-level spatial measures across stream size
classes suggests that most prominent dam influences occur in large and great rivers due to
cumulative dam effects along the river network. Overall, streams in the conterminous U.S. have
been heavily fragmented by dams, with the number of patches increasing between 700-1200%
depending on stream size when compared to undammed conditions. Understanding, and
accounting for, the variability in the individual, cumulative, and patch-level dam influences will
be important in national studies and assessments of fishes. Dams constitute an aging
infrastructure, further underscoring the importance of integrating multiple spatial measures in
assessing habitat restoration opportunities associated with dams.

2

INTRODUCTION
Dams constitute a complex and highly variable form of disturbance to fluvial habitats,
including altering hydrology, stream temperature, channel morphology, water chemistry, and
multiple aspects of hydrologic connectivity. In particular, dams have an enormous capacity to
affect connectivity throughout stream networks and with adjacent habitats including lateral,
vertical, temporal, and longitudinal components of connectivity (Ward and Stanford 1983, Ward
1989), with longitudinal fragmentation of river networks being one of the most commonly cited
impacts resulting from dams. This type of fragmentation has significant implications for stream
fishes that use disparate habitats for reproduction, growth, and survival (Schlosser and
Angermeier 1995, Fausch 2002), including impeding fish movement and migration and changing
species assemblage structure, genetic variation, and population abundance (e.g. Morita and
Yamamoto 2002, Guenther and Spacie 2006, Heggenes and Roed 2006, Alo and Turner 2005).
Ultimately, indirect or direct changes to the habitat factors which fish depend on (e.g., Maddock
1999) can lead to population declines, potentially leading to localized extirpation or even
extinction of species (Dunham et al. 1997, Fagan 2002, Morita and Yamamoto 2002, Perkin and
Gido 2011).
Characterizing dam effects from a landscape scale
Although dams can negatively influence the well-being of fishes, very few studies of
fishes have incorporated spatial measures that characterize dam effects across large geographic
regions (e.g., tens of thousands of square kilometers). In addition, most studies have chronicled
the effect of dams on fishes only at relatively localized scales (e.g. above or below a dam, or
before and after a single dam removal), with few having investigated the cumulative effects of
multiple dams within the river network or having addressed fragmentation across large regions

3

(Wang et al. 2011a). The paucity of studies involving spatial measures of dams along river
networks is due in part to a lack of available data on spatially consistent dam locations across
large regions and to the dendritic nature and resulting spatial complexity of river systems that
make them inherently difficult to study from a connectivity standpoint at large spatial scales
(Fullerton et al. 2010, Steel et al. 2010). As a result, most spatial measures of dams over large
geographic regions have been coarse in nature (e.g., density of dams within an entire river basin;
Graf 1999, Esselman et al. 2011), providing limited utility for understanding their effects on
stream fishes and for informing management.
Recently, a few studies have begun to incorporate spatial and temporal measures of dams
into analyses characterizing dam influences on fishes, including metrics generated through use of
geographic information systems (GIS). In general, these studies have focused on the
fragmentation of either river basins (catchment-based approach) or the stream network itself.
For example, Fukushima et al. (2007) identified catchment sub-basins isolated by dams for a
regional analysis of the presence/absence of fishes on the island of Hokkaido, Japan, finding that
fragmentation by downstream dams, and subsequent duration of isolation, had an influence on 11
of 41 species studied. Similarly, Hall et al. (2011) used dam construction dates obtained from
historical records to create a fragmentation timeline for watersheds in Maine, using this
information to describe current and historic fragmentation patterns of lake and stream habitat
accessible for two species of anadromous river herring. Lastly, Wang et al. (2011a) developed
multiple spatially-explicit measures of fragmentation by dams along stream networks for the
states of Michigan and Wisconsin, including distances from a given stream reach to the nearest
upstream and downstream dam along the mainstem of the river network, total number and
density of dams along all flow paths upstream, and total number and density of dams along the

4

downstream mainstem. These measures were used to partition the relative influence of dams
from other environmental covariates (non-dam measures that included both natural and
anthropogenic variables) using selected biotic integrity metrics and groups of fish summarized
by habitat and social preferences as response variables. Results showed that dam influences
accounted for 16% and 19% of the total variation for the two groups respectively. Findings from
this paper suggest that stream fish assemblages are responsive to a variety of dam influences,
including localized (proximity to individual dams) and cumulative (dam counts/densities within
catchments) factors originating in both an upstream and downstream direction.
Although the fragmentation of fluvial systems into discrete subsections, or patches, based
on the locations of dams has been evaluated in a number of studies pertaining to fishes, these
studies have typically only included one overall measure of stream fragmentation. These
measures have included either catchment-based approaches (e.g. Fukushima et al. 2007), lengths
of stream networks including tributaries (e.g. Bain and Wine 2010, Hall et al. 2011) or freeflowing mainstem lengths (e.g. Perkin and Gido 2011). Defining patch-based measures that
account for fragmentation of all three components – catchments, stream networks including
tributaries, and stream mainstems – would provide a more thorough examination of the
fragmentation effects of dams across large regions.
In this study, we consider effects of dams on streams of the conterminous U.S. by
characterizing stream network fragmentation and other dam influences in nine large ecoregions.
Specifically, we describe fragmentation and dam influence patterns at three different levels: 1)
individual dam characteristics (age, height, reservoir storage, and degree of regulation) for nearly
50,000 dams, 2) segment-level metrics encompassing distance-based measures to dams and
measures integrating cumulative dam effects for approximately 2.3 million stream segments, and

5

3) patch-level fragmentation metrics of river networks that account for fragmentation of
catchments, stream networks with tributaries, and stream mainstems. Lastly, we discuss the
potential uses of these dam measures in the management and conservation of fishes.
METHODS
Study area and spatial framework
Study area.—The study region includes the conterminous U.S., an area varying widely in
its physiographic, climatic, and anthropogenic settings with respect to stream environments
(Wang et al. 2011b). Due to broad regional differences in these factors, we used nine aggregated
ecoregions to stratify our analyses (Figure 1.1, Herlihy et al. 2008). The nine aggregated
ecoregions, hereafter called ecoregions, include: Northern Appalachians (NAP), South
Appalachians (SAP), Upper Midwest (UMW), Coastal Plain (CPL), Temperate Plains (TPL),
Northern Plains (NPL), Southern Plains (SPL), Western Mountains (WMT), and Xeric (XER).
In addition, a stream size stratification based on catchment area (A) was employed which
2

includes six classes (Esselman et al. 2011, Wang et al. 2011b); headwaters (HW; A ≤ 10 km ),
2

2

creeks (CR; 10 < A ≤ 100 km ), small rivers (SR; 100 < A ≤ 1,000 km ), medium rivers (MR;
2

2

1,000 < A ≤ 10,000 km ), large rivers (LR; 10,000 < A ≤ 25,000 km ), and great rivers (GR; A >
2

25,000 km ).
Stream network data.—The spatial framework used for this study is based on the
1:100,000 scale National Hydrography Dataset Plus Version 1 (NHDPlusV1; USEPA and USGS
2005), a GIS dataset that includes stream reaches, lake/reservoir polygons, and local catchment
boundaries encompassing the land area draining directly to a given stream reach. To facilitate
generation of measures of fragmentation by dams along the stream network, modifications were

6

made to the NHDPlusV1. Dam locations were used to split stream reaches (Figure 1.2) when
dam locations did not already coincide with a reach break (node) in the NHDPlusV1. Reaches
were subdivided using polyline split functions available with ArcMap GIS software (ESRI
2006). The subsequent subdivided reaches accounting for dam locations, hereafter referred to as
segments, were given a new unique identifier, and those situated immediately above or below a
dam were assigned the corresponding dam ID from the dataset used in this study (described
below). Using the elevation data available with the NHDPlusV1 and the Watershed function in
ArcMap, local reach catchments of the NHDPlusV1 were subdivided for dams that were greater
than 100 m away from an existing reach break in the NHDPlusV1, resulting in segment
catchments corresponding to dam locations (Figure 1.2).
Dam data.—Dams from the National Anthropogenic Barrier Dataset (NABD) were used
for this study (USGS 2013). The NABD consists of spatially-verified dam locations and
attributes (e.g. age, height, reservoir storage volume) derived from the 2009 National Inventory
of Dams (NID; USACE 2009) developed by the U.S. Army Corps of Engineers. Dams are
included in the NABD if they meet the following criteria: 1) the dam is classified as having a
high or significant hazard potential (dam failure would lead to a possible loss of life) or 2) the
dam is classified as a low hazard potential and either exceeds 25 ft in height and 15 acre-feet of
storage or exceeds six ft in height and 50 acre-feet of storage (USACE 2009). As a final
criterion, dams in NABD were required to be located on the NHDPlusV1 network, removing offstream dams. The spatial location (coordinates) of dams in the NABD were manually verified by
using streams from the NHDPlusV1, satellite imagery available in Google Earth(TM), and
attributes of the dams (dam name, reservoir name, etc.). As necessary, dam locations
represented in the NID were moved to align with stream reaches of the NHDPlusV1 using

7

Google Earth to create the NABD, ensuring that dams were linked with the correct stream
reaches in the NHDPlusV1. Additionally, dams greater than 25 ft in height from the USFWS
fish passage decision support system (USFWS 2008) dataset were verified against dam locations
in NABD, ensuring a full coverage of large dams within NABD. Currently, NABD includes
49,468 dams mapped to stream reaches represented by the NHDPlusV1 throughout the
conterminous United States (Figure 1.1).
Fragmentation metrics.—To generate the fragmentation metrics used in this study (Table
1.1), extensive programming using the Python programming language (Python v 2.7,
www.python.org) was developed to characterize spatial relationships between dam locations and
stream segments throughout the United States. This program managed a wide range of
conditions found within the NHDPlusV1, such as divergences, loops, and highly braided stream
networks. Using the topology of the stream network and the location of dams along the network,
the program identified total number of dams upstream of each stream segment, both along the
mainstem flow path and along all upstream flow paths in the river network. We defined the
upstream mainstem flow path as the longest navigable upstream pathway above each stream
segment. Similarly, the program also identified the number of dams along the downstream
mainstem flow path, which was defined as the shortest pathway below each segment to an ocean,
Great Lake, or terminal node in the case of disconnected stream networks. These dam counts
were used to generate both upstream (mainstem and total network) and downstream (mainstem
only) dam densities calculated using either network catchment area or mainstem/network stream
length (Table 1.1). Using the upstream and downstream mainstem pathways identified for each
segment, the program was also used to calculate the distance to the nearest mainstem dams if
they were present along the upstream and/or downstream mainstem pathways. These distance

8

values were then used to generate the total mainstem distance between dams, as well as the
proportion of upstream, downstream, and total mainstem distances free of dams for each stream
segment in the region. Lastly, the program calculated the cumulative upstream normal reservoir
storage volume (acre/ft) above each segment, which resulted in storage metrics expressed per
unit network stream length, network catchment area, or as a percentage of estimated annual
stream discharge volume (hereafter referred to as “degree of regulation” sensu Lehner et al.
2011) derived from the NHDPlusV1 (Table 1.1). Since reaches of the NHDPlusV1 were
subdivided at dam locations (including locations near the middle of a reach; Figure 1.2) and due
to occasional discrepancies in catchment areas defined between the NHDPlusV1 and the Python
program used in this study to develop the fragmentation metrics, only segments with catchments
± 25% the size of initial NHDPlusV1 reach catchment size were assigned degree of regulation
percentages.
Artificial habitat patches.—An additional unit of analysis for this study, artificial habitat
patches (AHPs), were delineated to account for the role of dams in the fragmentation of stream
networks and their catchments. AHPs are defined as an adjacent set of stream segments, and
their associated catchments, that are bounded by dams (Figure 1.3). With respect to AHPs, dams
were identified as bounding individual AHPs in either the upstream direction or downstream
direction, or in limited cases, were classified as internal to AHPs in situations where alternate
flow paths allowed for stream connectivity around dams. AHPs were assigned to a stream size
strata based on the total upstream catchment area of the most downstream stream segment, and
two measures of AHP size were calculated, total network length and catchment area. In addition,
the total mainstem length within AHPs was generated by summing the length of segments for the

9

largest size strata represented within the AHP (e.g. total length of large river segments with a
large river AHP).
Metric reduction and description of fragmentation patterns.— Due to the number of
segment-level metrics calculated (Table 1.1) and the redundancy of certain metrics, a principal
component analysis (PCA) was conducted to reduce the initial set of segment-level metrics to a
subset of five metrics used to represent and describe fragmentation patterns. For the PCA,
count-based metrics were removed from the analysis in favor of density-based metrics, as countbased metrics can be highly correlated with network catchment area and length (see Chapter 2),
resulting in a total of 14 metrics that were used in the analysis. PCAs were performed for the
conterminous U.S. and two example ecoregions, NAP and SPL, using fish community survey
locations obtained for use in the 2015 National Fish Habitat Partnership (NFHP;
http://fishhabitat.org/) river assessment for the conterminous U.S., allowing for results to inform
fragmentation metric selection for future analyses. The NAP and SPL ecoregions were selected
due to their widely varying conditions with respect to both natural conditions (climate,
hydrology, etc.) and relative dam density (Figure 1.1). Total sample sizes were 37,060 for the
conterminous U.S., 8,148 for NAP, and 2,391 for SPL, respectively. For the PCA, factors with
eigenvalues of 1 or greater were retained and a Varimax rotation was performed to aid
interpretation. SPSS software was used to run the PCAs (IBM SPSS Statistics 20 2011).
Based on the results of the PCAs (described in Results), a subset of five metrics were
chosen for summarization including upstream mainstem openness, upstream network dam
density, cumulative upstream degree of regulation, distance to downstream mainstem dam, and
downstream mainstem density. First quartile, median, and third quartile statistics were
calculated for ~ 2.3 million stream/river segments nationally, and segments representing flow

10

paths through lakes and reservoirs were removed from analysis. In characterizing fragmentation
and dam effects in this study, emphasis was placed on comparing and contrasting patterns in the
individual, segment-level, and patch-level dam measures both among ecoregions and as a
function of stream size class, utilizing statistics such as the first quartile, median, and third
quartile to characterize metric distributions. In addition, results were reported for the
conterminous U.S., allowing for comparison between ecoregions and broader conterminous U.S.
RESULTS
Patterns in dam characteristics
Dam count and density.—Among stream size classes, headwaters and creeks contained
the greatest number of dams both at conterminous U.S. and ecoregion scales, with many fewer
dams found on larger rivers (Table 1.2). Although dam counts were lower with increasing
stream size class, dam densities (as a function of stream length within each size class) had highly
variable distributions among ecoregions (Figure 1.4) showing left-skewed (e.g. UMW), rightskewed (e.g. CPL), and bi-modal (e.g. SAP) patterns. For the conterminous U.S., dam densities
ranged from a high of 1.2 dams/100 km of stream within headwaters to a low of 0.5 dams/100
km of stream within great rivers. Overall, the NAP ecoregion had the highest dam densities for
four of the six size classes, while the lowest densities were found in the XER ecoregion for
smaller size classes and the NPL/CPL ecoregions for larger size classes.
Dam age and height.—For the conterminous U.S., 56% of dams are at least 50 years old
(Figure 1.5). Median dam age increased from a low of 50 years for headwaters to high of 87
years for large rivers, then dropped for great rivers to a median age of 62 (Table 1.2; Figures 1.61.8). The headwater size class had the youngest median dam age for all ecoregions except WMT
where the creek size class was the youngest. Similarly, the oldest median age occurred in the

11

large river size class with the exception of the NAP and NPL ecoregions where the great river
size class was the oldest. Across ecoregions, the NAP ecoregion had the oldest median age
among size classes and was substantially older (by ~30-40 years) and had a larger interquartile
range than other ecoregions for headwater and creek size classes, while the youngest median
ages for dams within a given size strata tended to occur in the four plains ecoregions (CPL, TPL,
NPL and SPL). For dam height, the first quartile and median height were similar across the
headwater to large river size classes for the conterminous U.S., however the third quartile
increased across these size classes (Table 1.2). Dam height measures were largest for the great
river size class at the national scale. Dam height was the greatest in the WMT ecoregions for the
large and great river size classes with median heights exceeding 200 ft. In the NAP and TPL
ecoregions, median dam heights ranged from only 15 – 25 ft and 13 - 34 ft, respectively across
all size classes.
Dam storage and degree of regulation.—As expected, reservoir storage tended to
increase with increasing size class both for the conterminous U.S. and among ecoregions (Table
1.2). Although storage within the headwater and creek size classes was comparable across
ecoregions, there were large differences among the ecoregions in storage in the small to great
river size classes. Both for the conterminous U.S. and for a number of ecoregions, first quartile,
median, and third quartile reservoir storage increased by an order of magnitude when moving
from the large to great river size classes. In the SAP, SPL, and WMT ecoregions, median
reservoir storage exceeded 100,000 acre-feet, with third quartile storage exceeding 1,000,000
acre feet for the NPL and WMT ecoregions. Comparing the total amount of storage by size
strata, dams on great rivers collectively store the largest amount as a percentage of total storage
in the conterminous U.S., followed by medium-sized rivers (Figure 1.9). Total storage on large

12

and small rivers is comparable, while creeks and headwaters combine for a relatively small
amount (~6%) of overall storage in the conterminous U.S. For several ecoregions (SAP, UMW,
CPL, and WMT), storage on medium-sized rivers surpassed storage on great and large rivers.
Across ecoregions, storage on great rivers showed the greatest variability.
Cumulative degree of regulation at dam locations was the lowest within the small river
size class with a median value of 11% and the highest for great rivers at 40% for the
conterminous U.S (Table 1.2). The UMW ecoregion was consistently among the lowest in
median degree of regulation with values ranging from 6 - 25% across size classes. Overall, the
highest median degree of regulation at dam locations occurred in the headwater size class for the
TPL ecoregion at 295%, or nearly three years of estimated discharge volume being stored in
upstream reservoirs.
Dam purpose.—The main dam purpose varied widely across size classes and ecoregions
(Table 1.3). In the eastern ecoregions of NAP, SAP, UMW, and CPL, recreational use was the
dominant purpose among the headwater and creek size classes, ranging from 38 to 70% of dams
for these size classes, with flood control and water supply being major contributors depending
upon the ecoregion. For these regions, recreational use tended to decline, while hydroelectric
and/or navigation uses were greater with increasing stream size for the small to great river size
classes. In the more central ecoregions of TPL, NPL, and SPL, flood control and fire
protection/farm ponds were most prevalent within the headwater and creek size classes with
recreation, water supply, irrigation and flood control being main uses in the larger size classes.
Irrigation dominated (46 - 55%) for headwaters and creeks in the western ecoregions (WMT and
XER), and irrigation remained the primary purpose in the XER. In contrast, hydroelectric use
increased in prevalence with increasing size for the WMT ecoregion. For the conterminous U.S.,

13

recreation was the leading use within the headwater to small river sizes, while hydroelectric and
navigation uses were dominant among the larger size classes. Levels of irrigation and water
supply use were relatively steady across size classes ranging from 8 -15% and 6 – 12%,
respectively.
Patterns in stream segment fragmentation metrics
Principal component analysis results.—The principal component analysis of 14 segmentlevel fragmentation metrics resulted in two axes for the conterminous U.S. and NAP ecoregion
and four axes for the SPL ecoregion, with total variation explained ranging from 79-91% by
region (Table 1.4). For the conterminous U.S. and NAP ecoregion, the first axis represented all
upstream-oriented metrics and was interpreted as a combination of all upstream dam influences.
The second axis for these regions was comprised of downstream mainstem and total mainstem
metrics. Since the total mainstem metrics grouped closely with downstream mainstem metrics,
this axis was interpreted as a combined set of downstream dam influences. For the SPL
ecoregion, the first and third axes accounted for upstream dam metrics, with the first axis
representing cumulative, network-based dam influences whereas the third axis was interpreted as
representing upstream mainstem dam effects. The second and fourth axes for the SPL ecoregion
were characterized by downstream mainstem and total mainstem metrics, respectively. As
before, this combination was interpreted as representing largely downstream-oriented dam
influences, with the second axis representing downstream habitat availability, while the fourth
axis represented downstream dam density.
Select metric statistics.—Based on the PCA results, a subset of fragmentation metrics
were selected for summarization that represented a diversity of dam influences, including
upstream and downstream habitat availability (upstream mainstem openness and distance to

14

downstream mainstem dam) and cumulative effects (upstream network dam density, upstream
degree of regulation, and downstream mainstem density). For the upstream mainstem openness
metric, all ecoregions were characterized by having maximum openness values (100%) for the
first quartile, median, and third quartile statistics for headwaters and creeks, with high levels of
openness also occurring in the small river and even medium river strata for some ecoregions (e.g.
CPL and NPL; Table 1.5). Openness values typically declined with increasing size among the
strata, with great rivers having a first quartile range of only 1 - 12% and median range of 2 28%. For the conterminous U.S., median values remained high through the medium river size
strata at 91%, dropping to 30% for large rivers and 16% for great rivers, respectively.
Among headwater and creek size strata, upstream network densities were zero for
virtually all three statistics (Table 1.5). For the remaining size strata, densities tended to increase
slightly with increasing size for most ecoregions. Among the medium to great river size strata,
densities were the lowest in WMT and XER ecoregions, with median values ranging 0 - 0.3
dams/100 km and highest for the CPL and NAP ecoregions, with values ranging from 0.6 - 3.6
dams/100 km. At the national level, densities also increased slightly when moving from the small
to great river size classes, with median values ranging from 0.3 - 0.8 dams/100 km. Similarly to
upstream network densities, cumulative degree of regulation values were minimal for headwater
and creek strata within ecoregions, continuing to remain very low (median <=1%) for the small
river stratum with exception of the TPL ecoregion with a median of 8%. Within all ecoregions,
degree of regulation values increased when moving from the medium to great river strata, with
the lowest values tending to occur in the SAP and UMW ecoregions, while XER and SPL values
were among the highest. The great river strata had a wide degree of variation across ecoregions
(Figure 1.10), with median values ranging from 6 - 69%, and third quartile values exceeding

15

100% in the TPL, SPL, and XER ecoregions. For the conterminous U.S., great river values had
a large inter-quartile range, with a median value of 44%.
Although distance to downstream dam values increased with size across the first quartile,
median, and third quartile statistics for the conterminous U.S., patterns among ecoregions tended
to differ, with some having the greatest distances for the smaller strata (e.g. TPL), while others
showed the opposite trend (e.g. SPL, Table 1.5). Within some ecoregions, distances across the
various size strata were comparable, as occurred in the NAP and WMT ecoregions. Overall,
segments in the NAP and UMW ecoregions had the closest distances (generally less than 50 km)
to downstream mainstem dams, while distances to downstream dams were relatively long within
the NPL ecoregion with median values ranging 285 - 511 km across size strata. Downstream
dam densities were the highest in the NAP ecoregion for the headwater to large river size strata,
where median values were nearly double of those in the UMW ecoregion, which had the next
highest density values. For the great river stratum, the UMW and WMT had the highest median
densities at ~1 dam/100 km, while values were very low (<=0.1 dam/100 km) for several
ecoregions. The NPL and CPL had the lowest median densities among ecoregions, with values
ranging from 0 to 0.4 dams/100 km across size strata.
Artificial habitat patches
For the conterminous U.S., a total of 54,120 AHPs were identified (Table 1.6, Figure
1.11), with the highest degree of fragmentation occurring in the NAP ecoregion (Figure 1.12).
When compared to the 6,007 individual non-dam patches (Figure 1.11 top), defined by
contiguous stream networks that are not subdivided at locations of dams, this represents a total
increase of 801% in the number of patches. Across size strata, the percentage increase in the
number of patches was very high, ranging from 703 – 1188%. In comparing median network

16

length, AHP network length declined from 23 - 94%, with the lowest occurring in the creek
stratum, while influences on large and great river strata were the highest. A similar pattern
occurred for both median catchment size values and median mainstem length, with greater
declines when moving from the smaller to larger size strata. For all three size measures, there
tended to be a large decline when moving from the small to medium river size strata.
DISCUSSION
This study aimed to characterize fragmentation and dam influence patterns across the
conterminous U.S. within nine large ecoregions, specifically describing individual dam
characteristics (age, storage, etc.), segment-level distance-to-dam and cumulative dam effects,
and patch-level fragmentation of river networks. Results within all three levels show a highly
variable pattern in dam characteristics and dam effects across the U.S., both by ecoregion and
stream size. For instance, among individual dam characteristics some regions contain a high
density of dams and relatively low degree of regulation (e.g. NAP and UMW), while other
ecoregions, such as XER, had opposite characteristics, with a low dam density and a high degree
of regulation. In many ecoregions, main dam purposes changed considerably as stream size
increased. Understanding how dams vary in terms of their specific characteristics, including
their purpose and how they may be fragmenting river networks, can provide insight into their
ecological impacts (Poff and Hart 2002), particularly when accounting for influence of network
position and stream size (Ward and Stanford 1983).
Larger rivers, in particular, appear to have the greatest degree fragmentation as indicated
by both segment and patch-level measures for the large and great river size strata. This
observation is likely a function of stream network structure, as conditions in larger rivers are
associated with cumulative effect of all upstream tributaries. A study by Lehner et al. (2011)

17

suggested a level of 2% degree of regulation for flow impairment by dams, a level which would
encompass 91% of large river segments and 97% of great river segments in this study for the
conterminous U.S. This suggests an incredibly high degree of potential flow alteration to the
Nation’s rivers resulting from dams. The increase in degree of regulation with stream size in this
study is consistent with the findings of Lehner et al. (2011), which found a similar pattern of
increases by stream size in a global study of reservoir storage by dams. This pattern
demonstrates the pervasive, cumulative effects of reservoir storage along the river network,
culminating in a high degree of regulation in larger rivers. Similar to the segment-level results,
larger rivers were highly affected at the patch level, sustaining the highest amount of
fragmentation according to AHP size measures when compared to non-dam conditions. This
pattern likely arise from the “pruning” effect of dams within the stream network, as larger rivers
are influenced not only by dams on mainstems but also by dams along major tributaries. Overall,
the presence of dams in the conterminous U.S. has greatly increased the number of river patches
of all size classes.
Implications for fisheries management and conservation
As a result of the great variability in dam measures across the U.S., careful consideration
of the metrics used in future studies involving dam effects will be required. Selecting a subset
of metrics that capture a variety of influences, such as localized (resulting from effects of
individual dams), cumulative (resulting from multiple dams located throughout a river network),
and patch-level connectivity metrics will be vital in order to capture the range of dam effects.
Using a single metric (e.g. network density) would likely fail to adequately represent dam
influences in most ecoregions. The results of the PCAs suggest the selection of multiple
segment-level metrics for analyses incorporating the influence of dams on fishes, including two

18

metrics for the NAP ecoregion, one each representing upstream and downstream influences, and
four metrics in the SPL ecoregion, representing upstream cumulative, upstream mainstem,
downstream habitat availability, and downstream density dam influences. The ability to discern
between various effects (localized, cumulative, and patch-level) and reveal any changes in their
relative influences along the river network, would yield a much broader understanding of the
complex landscape-scale effects that dams have on fishes. Perhaps the development of a multimetric dam index, driven by fish taxa responses in each ecoregion, could help ensure that the
proper metrics were selected.
Data generated in this study could also be used to identify and conserve relatively
unfragmented and unimpacted river networks. For instance, due to the effects of climate change,
habitats within river networks may become unsuitable for some fish species. As a consequence
of these changes and potential movement limitations imposed by barriers, the future range of
certain species may be restricted due to localized extinctions within habitat patches (e.g.
coldwater fishes). Several studies have predicted the future distributions of individual fish
species (Rieman et al. 2007) or assemblages (Buisson et al. 2008, Buisson and Grenouillet 2009)
by factoring climate change impacts on habitat suitability (e.g. changes in both stream
temperature and precipitation), but in each case colonization of suitable habitat was allowed
regardless of the presence of dams as barriers to dispersal. While these studies can give us a
general sense of the potential range contraction or expansion of certain fish species or
assemblages under various climate change scenarios, they may offer limited practical use. By
incorporating measures of longitudinal fragmentation and dam influences along the river
network, future studies forecasting climate change impacts on fish habitat and potential for

19

species distribution would generate projections that better matched on-the-ground conditions,
providing greater utility to aquatic resource managers and decision makers.
Although dams are generally viewed as being detrimental to the existence of native
stream fishes, some have argued that dams can provide utility in certain circumstances, by
protecting native species, improving water quality (Jackson and Pringle 2010), or by offering
unique opportunities for flow and thermal regime improvements. For instance, in some
situations, dams can prevent the establishment of non-native fishes by limiting their movement
and dispersal (Novinger and Rahel 2003, Peterson et al. 2008, Fausch et al. 2009). Fausch et al.
(2009) provide two examples from the western U.S. where barriers have isolated remnant
populations of a native salmonid from introduced salmonid species, likely preventing their
extirpation. By combining fragmentation data with the known distributions of native and nonnative fishes, river networks that are free of non-native fishes can be identified and conserved.
Implications for restoration
Although dam influences can be difficult to isolate due to the complex and varied impacts
that dams can have on river systems (connectivity, hydrology, geomorphology, etc.), they also
represent a unique restoration opportunity. While other forms of human disturbance, such as
urbanization, can be mitigated to a certain extent, dam removals can offer dramatic cases of
restoration, with the initial re-establishment of connectivity after dam removal to the gradual
“resetting” of river geomorphology. Dams have limited life spans, typically resulting from
sediment accumulation that restricts water storage capabilities and a physical infrastructure that
weakens over time, presenting safety concerns (Poff and Hart 2002). With an aging set of dams
in the conterminous U.S. (nearly 75% of dams used in this study will be at least 50 years old by
2020), managers will increasingly be confronted with decisions on whether to remove a dam or

20

cope with continued repairs and maintenance. While these decisions are likely based in part on
social and economic considerations (Doyle et al. 2008), the dam measures presented here can
provide an ecological component to the decision-making process by identifying the types of
habitat which will be connected after a dam removal, determining if there will be increased
connectivity to known source populations, and providing a landscape context for the potential
dam removal project. For instance, the information gathered in this study can help guide dam
removal priorities by integrating other measures of anthropogenic disturbance (e.g. agriculture,
urbanization, point source pollutant locations; Esselman et. al 2013), allowing for the potential
to identify longitudinal connections between higher quality, relatively undisturbed habitats.
Physical habitat restoration projects are at risk for failure if the full scope of humaninduced changes to river systems are not considered (Palmer et al. 2005), with restoration
activities being potentially unsuccessful due to historical and ongoing disturbances occurring at
larger, regional scales (Bond and Lake 2003). For instance, the presence of a large dam above a
restoration site, and the associated flow regime alterations, could supersede any local habitat
enhancements, making them ineffective (Bond and Lake 2003). By considering dam effects in a
larger regional context when evaluating restoration opportunities, overriding factors related to
dams can be avoided. This is particularly true regarding stream connectivity, where
fragmentation of the stream network can impede colonization from source populations,
potentially limiting the biotic response to restoration activities (Bond and Lake 2003). A study
of six restored stream sites by Riley and Fausch (1995) found that dispersal was instrumental in
the increased abundance of three trout species, even though all three species were already present
at the restored sites, and concluded that survival and recruitment were relatively less important
factors in the abundance increases. From an ecological standpoint, the restoration of stream

21

network connectivity is a key motivation for dam removal (Bednarek 2001), emphasizing the
need for a spatial accounting of stream fragmentation and dam influences across large regions
and at multiple scales.
Potential improvements
The dams used in this study represent ~50,000 of largest dams in the U.S. that are
georeferenced to a modified 1:100,000 NHDPlusV1 stream network. As result, small dams are
underrepresented in this study. For instance, Wang et al. (2011a) identified 5,215 dams
connected to a modified version of the 1:100,000 NHD (precursor to the NHDPlusV1) for the
states of Michigan and Wisconsin, which compares to 1,612 dams found in the NABD dataset
for the two respective states. An analysis of dams in the NID (which forms the basis of the
NABD dataset) for the state of Texas found that smaller dams were greatly underrepresented
when compared to dams identified from digital orthoquads (DOQs) and the number of water
bodies found in the NHD (Chin et al. 2008). Inclusion of smaller dams from state databases and
other sources would improve connectivity-based measures, particularly in the fragmentation of
headwater and creek size classes. Development of a holistic barrier layer that includes other
potential anthropogenic sources of connectivity disruption such as road crossings (JanuchowskiHartley et al. 2013) and water diversions, along with cases of natural fragmentation (e.g.
waterfalls, endorheic basins) would provide a broader overall view of fragmentation in the
conterminous U.S.
The annual stream flow estimates used in this assessment, obtained from the
NHDPlusV1, are derived from a relatively coarse flow model. Better segment-level annual flow
estimates at national or even regional scales would likely provide improved degree of regulation
estimates. In this study, degree of regulation was calculated from normal reservoir storage

22

volume, however NABD attributes also include a maximum storage volume as well. A
comparison of the ratio between maximum and normal degree of regulation could provide a
metric related to the flood attenuation capacity of individual dams, perhaps providing a better
measure of seasonal flow impacts.
Conclusion
The spatially-explicit study of human impacts on hydrologic connectivity is an emerging
topic (Fullerton et al. 2010), one which provides many opportunities for advancing our
knowledge of the effects of dams and fragmentation on aquatic organisms. Continued research
and exploration will be needed in order to fully address and integrate the varied, multi-scale
effects of dams into the restoration, conservation, and management activities pertaining to fishes
and other aquatic organisms.

23

APPENDICES

24

APPENDIX A

TABLES
Table 1.1.— Dam metric descriptions, units, direction of metric indicating increasing levels of fragmentation, and sign of individual
species/metric relationship to indicate positive response of species to increased fragmentation (indicated by increasing catch per unit effort).
For the principal component and canonicial correspondence analyses, distance-based metrics were given the maximum value for that metric
within the entire database in cases where upstream and/or downstream mainstem dams were absence for a given record.
Higher values
Positive species
Metric
Description
Units
response:
result in:
UMCT
Upstream mainstem dam count
#
Greater fragmentation
+
1,2,3

UMD

2

Greater fragmentation

Upstream mainstem dam density per unit upstream river mainstem length #/100 km

+

UM2D

Distance to upstream mainstem dam

km

Less fragmentation

-

UMO

Proportion of open upstream mainstem

proportion

Less fragmentation

-

UNCT

Total upstream dam count

#

Greater fragmentation

+

UNDR

Upstream mainstem dam density per unit river network length

#/100 km

Greater fragmentation

+

UNDC

Upstream mainstem dam density per unit network catchment area

Greater fragmentation

+

USR

Upstream reservoir storage volume per unit river length

acre feet/100 km Greater fragmentation

+

USC

Upstream reservoir storage volume per unit catchment area

USF
DMCT

1,2

1,2,3

2

#/km

a
a

2

acre feet/km

Greater fragmentation

+

Proportion of estimated annual discharge stored in upstream reservoirs

proportion

Greater fragmentation

+

Downstream mainstem dam count

#

Greater fragmentation

+

Downstream mainstem dam density along the river network

#/100 km

Greater fragmentation

+

DM2D

Distance to downstream mainstem dam

km

Less fragmentation

-

DMO

Proportion of open downstream mainstem

proportion

Less fragmentation

-

1,2,3

DMD

1,2,3

2

25

a,b

Table 1.1 (cont'd).
Metric
TMCT

Description

Units

Total mainstem dam count

#

2

Total mainstem dam density per unit downstream river mainstem length #/100 km
Total mainstem distance between upstream and/or downstream mainstem km
TM2D
1,2,3 d
Total proportion of open mainstem
proportion
TMO
TMD

1
2
3
a
b

2

Higher values
Positive species
result in:
response:
Greater fragmentation
+
Greater fragmentation

+

Less fragmentation

-

Less fragmentation

-

Dam metrics selected for headwater and mid-size strata change point and Spearman correlation analysis.
Dam metrics selected for large size strata change point and Spearman correlation analysis.
Dam metrics selected for CCA analysis.
Normal reservoir storage volumes taken from the National Anthropogenic Barrier Dataset (NABD)
Annual flow estimates from the National Hydrography Dataset Plus Version 1 (NHDPlusV1) unit runoff method.

26

Table 1.2.—Descriptive statistics (first quartile, median, and third quartile) for dam characteristics for the conterminous U.S. (CONUS) and by
ecoregion. Dam densities were calculated per 100 km of network length within a given stratum and region. Degree of regulation (UDOR) was
calculated using cumulative upstream reservoir storage divided estimated mean annual stream flow and had a total sample size of 34,175.
Based on the available information within the NABD dataset, a total sample size of 43,658 was used for age, 49,356 for height, and 49,468 for
storage.
Age (years)
Storage (acre feet)
Height (feet)
UDOR (%)
Region Stratum Count Density Q25 Median Q75
Q25 Median Q75
Q25 Median Q75
Q25 Median Q75
HW 32351
1.2
40
50
60
18
24
31
31
58
120
7
23
81
CR
11851
0.8
44
55
78
15
24
40
60
165
554
3
12
56
SR
3259
0.6
50
75
98
14
22
47
91
479
5895
2
11
46
CONUS
MR
1411
0.7
54
82
100
15
27
67
216 1500 24943
4
16
49
LR
314
0.9
62
87
104
16
26
72
450 4588 28000
8
20
53
GR
282
0.5
46
62
78
35
66
124 4334 80000 388800
20
41
100
HW
2662
2.0
50
78
110
11
15
23
40
91
263
3
10
24
CR
1711
2.3
64
93
113
12
16
25
51
180
975
2
6
19
SR
729
3.1
72
96
113
13
19
31
60
220
2255
2
9
23
NAP
MR
280
3.7
79
97
107
16
23
39
203
892
3303
5
12
27
LR
30
2.2
92
100
107
19
25
36 1150 3939 16600
13
20
20
GR
1
0.3
188
188
188
18
18
18 10744 10744 10744
30
30
30
HW
7000
1.6
42
50
59
21
26
35
32
60
124
4
11
28
CR
1493
0.9
43
50
67
25
39
56
98
248
1071
2
5
15
SR
341
0.6
48
73
95
19
40
73
92
910 13430
1
8
25
SAP
MR
210
1.0
56
84
103
20
51
129
310 4700 97000
5
10
37
LR
68
1.8
61
87
113
21
55
104 3586 13000 119925
6
8
16
GR
52
1.2
44
57
81
53
78
110 37850 161900 420450
9
18
45
HW
644
0.6
39
46
52
11
18
28
15
45
147
2
6
17
CR
697
0.9
44
57
83
9
12
19
42
150
832
1
4
16
SR
452
1.4
74
83
113
10
15
22
120
645
3875
1
5
21
UMW
MR
195
2.0
84
94
106
18
25
40
600 2483
8958
2
9
18
LR
31
2.5
76
100
118
24
30
50 1815 8170 23250
12
15
19
GR
24
1.6
76
76
78
32
43
46 10450 52700 98600
23
25
39

27

Table 1.2 (cont'd).
Region

CPL

TPL

NPL

SPL

Stratum Count Density
HW
6004
1.6
CR
1763
0.9
SR
256
0.4
MR
79
0.2
LR
26
0.3
GR
31
0.3
HW
6116
1.8
CR
1074
0.6
SR
308
0.5
MR
216
0.9
LR
67
1.4
GR
43
0.6
HW
2750
1.4
CR
1516
1.2
SR
228
0.5
MR
30
0.2
LR
4
0.1
GR
18
0.4
HW
5313
2.2
CR
2049
1.1
SR
303
0.5
MR
167
0.6
LR
41
0.8
GR
34
0.3

Age (years)
Height (feet)
Storage (acre feet)
UDOR (%)
Q25 Median Q75
Q25 Median Q75
Q25 Median Q75
Q25 Median Q75
40
50
58
14
19
25
28
60
116
2
9
27
48
59
83
12
15
24
54
127
333
1
3
7
48
59
80
13
19
35
190
787
9100
1
4
32
43
52
75
23
43
71 1475 31500 144500
3
21
58
47
63
87
20
41
70 1300 8800 60400
11
37
53
38
45
51
49
79
108 43000 70500 212500
24
41
47
31
42
51
24
28
35
28
49
93
18
49
135
35
45
60
21
36
48
99
253
1000
7
24
105
46
74
91
12
17
39
100
469
3974
3
17
66
67
78
91
10
13
27
150
449
3807
3
13
36
74
79
98
11
16
21
231 1690
9808
10
25
63
55
75
80
14
34
48
764 55000 184000
11
18
22
49
57
64
17
21
26
30
46
72
80
295
812
50
59
73
15
20
27
40
75
161
30
118
359
50
63
76
15
20
30
55
190
626
8
33
125
53
63
75
13
20
38
135
390
7010
5
19
111
53
58
83
29
115
200
325 65250 548660
25
102
319
56
84
98
45
107
210 1710 38926 1725000
46
47
134
41
48
54
19
25
32
34
55
100
24
59
204
41
49
55
24
34
45
80
174
287
15
31
105
43
53
74
16
26
53
64
192
2085
3
18
83
44
57
79
15
28
86
112
560 39545
4
35
172
62
93
103
10
13
27
197
285
808
36
138
435
50
62
72
35
98
140
280 124067 505381
76
137
253

28

Table 1.2 (cont'd).
Region

WMT

XER

Stratum Count Density
HW
1231
0.3
CR
896
0.4
SR
315
0.6
MR
112
0.6
LR
24
1.0
GR
14
0.7
HW
631
0.1
CR
652
0.2
SR
327
0.4
MR
122
0.4
LR
23
0.4
GR
65
0.6

Age (years)
Height (feet)
Storage (acre feet)
UDOR (%)
Q25 Median Q75
Q25 Median Q75
Q25 Median Q75
Q25 Median Q75
46
60
87
17
26
36
58
130
380
9
35
104
47
59
84
21
33
58
84
347
1720
3
18
76
49
64
89
30
70
149
311 5750 38755
5
32
89
50
64
88
31
89
202
480 13984 166531
13
39
66
49
68
89
53
201
249 2150 32900 211524
13
37
84
53
58
73
125
203
330 58386 137300 1153000
30
33
48
36
52
68
22
32
49
40
115
411
23
85
317
44
58
83
20
31
50
69
200
1047
4
28
133
45
59
85
20
40
82
171
960
9010
4
28
101
46
60
94
23
60
164
222 3300 87500
26
67
156
64
96
99
23
95
225
850 4765 65540
7
31
52
45
63
80
37
87
208
500 50130 516000
33
53
108

29

Table 1.3.—Main dam purpose both for the conterminous U.S. (CONUS) and by ecoregion with values expressed as a percentage. Flood
= Flood control and stormwater management; Hydro = Hydroelectric power generation; Supply = Water supply; Fire/Farm = Fire
protection, stock, and small farm ponds; Fish/Wild = Fish and wildlife ponds; Debris = Debris control.
Region

Stratum
HW
CR
SR
CONUS
MR
LR
GR
HW
CR
SR
NAP
MR
LR
GR
HW
CR
SR
SAP
MR
LR
GR
HW
CR
SR
UMW
MR
LR
GR

Flood
19
22
12
12
7
9
6
8
13
3
3
0
13
33
17
25
6
2
22
7
2
1
0
0

Hydro
0
2
16
41
37
26
1
6
39
82
70
0
0
1
20
49
41
31
0
2
16
65
90
17

Irrigation Navigation Supply Recreation Fire/Farm Fish/Wild
8
0
9
39
18
2
12
0
12
35
9
3
15
1
13
27
4
2
9
3
12
13
2
3
12
19
6
13
0
2
10
36
9
6
0
1
1
0
16
70
2
2
1
1
16
58
2
2
0
1
10
25
1
1
0
8
2
2
1
0
0
27
0
0
0
0
0
0
0
100
0
0
4
0
6
67
5
1
1
0
17
45
0
1
0
0
27
31
0
1
0
4
11
9
0
0
0
29
1
21
0
0
0
56
2
6
0
0
1
0
0
38
11
7
4
0
2
61
1
7
2
0
2
45
0
2
0
1
2
17
0
1
0
0
0
7
0
0
0
75
4
4
0
0

30

Debris
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
3
1
0
0
0
0

Tailings
1
0
0
0
0
0
0
0
0
0
0
0
1
1
1
0
0
0
2
1
0
0
0
0

Other
4
4
9
5
4
3
3
5
8
1
0
0
3
2
3
2
1
4
15
16
29
14
3
0

Table 1.3 (cont'd).
Region Stratum
HW
CR
SR
CPL
MR
LR
GR
HW
CR
SR
TPL
MR
LR
GR
HW
CR
SR
NPL
MR
LR
GR
HW
CR
SR
SPL
MR
LR
GR

Flood
12
13
13
21
8
13
26
40
21
21
12
9
1
3
2
3
0
29
46
56
17
12
11
19

Hydro
0
0
2
15
19
0
0
0
3
10
12
5
0
0
0
0
0
47
0
0
0
3
3
19

Irrigation Navigation Supply Recreation Fire/Farm Fish/Wild
7
0
3
65
7
2
5
0
4
69
3
1
6
2
16
50
1
2
0
8
22
24
5
0
0
58
0
8
0
0
7
67
7
0
0
0
3
0
4
21
35
1
1
0
9
31
6
4
1
1
14
42
2
5
3
2
16
29
2
9
3
23
9
28
2
11
0
58
16
2
0
5
13
0
36
3
42
3
25
0
17
9
40
5
37
0
7
19
25
7
34
0
3
24
14
17
100
0
0
0
0
0
24
0
0
0
0
0
5
0
6
7
31
1
9
0
11
8
12
1
25
0
25
12
13
0
19
0
35
21
6
1
36
0
28
6
0
0
6
0
34
16
0
0

31

Debris
0
0
0
0
0
0
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0

Tailings
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0

Other
3
4
7
5
8
7
8
7
11
8
0
5
1
1
2
3
0
0
4
3
7
2
17
6

Table 1.3 (cont'd).
Region Stratum
HW
CR
SR
WMT
MR
LR
GR
HW
CR
SR
XER
MR
LR
GR

Flood
4
4
8
9
0
0
15
17
17
12
9
8

Hydro
4
7
22
46
70
93
1
1
2
17
27
37

Irrigation Navigation Supply Recreation Fire/Farm Fish/Wild
55
0
12
13
4
4
46
0
12
18
4
5
39
0
14
10
2
2
25
1
11
3
1
1
10
0
10
0
0
0
0
0
7
0
0
0
48
0
9
3
8
3
51
0
10
4
6
4
55
0
8
5
3
4
51
0
8
3
1
5
64
0
0
0
0
0
34
8
3
7
0
0

32

Debris
0
0
1
0
0
0
4
2
1
2
0
2

Tailings
2
1
0
0
0
0
4
2
0
0
0
0

Other
2
2
2
2
10
0
5
4
4
1
0
0

Table 1.4.—Principal component analysis results for the conterminous U.S. and
for the NAP and SPL ecoregions with fragmentation metric weights by axes.
Weights with an absolute value of 0.7 or greater, shown in bold, were used in
the axis interpretation found at the top of each axis column. Cumulative
percentage of variation explained is located at the bottom of each axis column.
Axis 1
Axis 2
Axis 3
Axis 4
Upstream Downstream
Combined
Combined
Region
Metric
0.89
-0.10
UMD
-0.84
0.13
UM2D
UMO
-0.87
0.11
UNDR
0.92
-0.02
UNDC
0.91
-0.01
USR
0.96
0.00
USC
0.95
0.00
CONUS
USF
0.82
0.03
DMD
-0.03
-0.89
DM2D
-0.04
0.87
DMO
0.04
0.86
TMD
0.17
-0.86
TM2D
-0.07
0.91
TMO
-0.04
0.85
C. Var. Exp.
47.19
79.15
Upstream Downstream
Combined
Combined
UMD
0.93
-0.10
UM2D
-0.90
0.12
UMO
-0.90
0.11
UNDR
0.95
-0.06
UNDC
0.95
-0.05
USR
0.97
-0.06
USC
0.97
-0.05
NAP
USF
0.85
-0.08
DMD
-0.06
-0.91
DM2D
-0.10
0.86
DMO
0.06
0.91
TMD
0.21
-0.88
TM2D
-0.20
0.91
TMO
-0.09
0.91
C. Var. Exp.
52.67
84.78

33

Table 1.4 (cont'd).

Region

Metric
UMD
UM2D
UMO
UNDR
UNDC
USR
USC
SPL
USF
DMD
DM2D
DMO
TMD
TM2D
TMO
C. Var. Exp.

Axis 1
Axis 2
Upstream Downstream
Availability
Network
0.48
0.09
-0.35
-0.01
-0.40
-0.07
0.93
0.04
0.93
0.02
0.90
0.13
0.91
0.12
0.79
0.06
-0.05
-0.28
0.07
0.92
0.02
0.93
0.03
-0.20
0.14
0.92
0.06
0.94
45.60
72.51

34

Axis 3
Axis 4
Upstream Downstream
Mainstem
Density
0.81
-0.03
-0.89
0.04
-0.86
0.01
0.21
0.00
0.18
0.01
0.35
0.00
0.33
0.01
0.37
-0.08
-0.11
0.94
0.05
-0.11
0.11
-0.16
0.05
0.97
-0.02
-0.16
0.03
-0.14
82.90
90.57

Table 1.5.—Descriptive statistics (first quartile, median, and third quartile) for a subset of segment-level fragmentation metrics, including
upstream mainstem openness (UMO) expressed as a percentage, upstream river network density (UNDR), cumulative upstream degree of
regulation (UDOR), downstream distance to mainstem dam (DM2D), and downstream mainstem dam density (DMD).
UMO (%)
UDOR (%)
DM2D (km)
DMD (#/100 km)
UNDR (#/100 km)
Region Stratum Q25 Median Q75 Q25 Median Q75 Q25 Median Q75 Q25 Median Q75 Q25 Median Q75
HW
100
100
100
0.0
0.0
0.0
0
0
0
31
93
226
0.1
0.3
0.8
CR
100
100
100
0.0
0.0
0.0
0
0
0
36
100
249
0.1
0.4
0.8
SR
100
100
100
0.0
0.3
1.3
0
0
2
42
109
264
0.1
0.4
0.8
CONUS
MR
33
91
100
0.2
0.6
1.3
1
4
20
44
119
274
0.1
0.4
0.8
LR
10
30
70
0.4
0.7
1.2
6
17
47
44
123
385
0.1
0.4
0.8
GR
5
16
38
0.3
0.8
1.1
21
44
112
56
200
544
0.1
0.3
0.6
HW
100
100
100
0.0
0.0
0.0
0
0
0
8
25
60
0.8
1.7
4.1
CR
100
100
100
0.0
0.0
3.0
0
0
1
8
25
60
0.8
1.8
4.1
SR
23
98
100
0.3
1.1
2.8
0
1
5
8
25
63
0.8
1.8
3.9
NAP
MR
4
21
60
0.7
1.2
1.9
2
6
17
6
23
57
0.8
2.1
4.2
LR
3
8
22
0.8
1.5
2.1
7
10
20
10
31
142
1.1
1.5
3.3
GR
1
3
30
3.1
3.6
3.6
30
30
30
0
0
0
0.0
0.0
0.0
HW
100
100
100
0.0
0.0
0.0
0
0
0
31
73
141
0.3
0.5
0.8
CR
100
100
100
0.0
0.0
0.0
0
0
0
36
77
144
0.3
0.5
0.8
SR
94
100
100
0.0
0.8
2.1
0
0
2
40
80
146
0.3
0.5
0.8
SAP
MR
19
62
100
0.4
0.8
1.7
1
3
10
40
90
183
0.2
0.4
0.8
LR
3
12
37
0.6
0.9
1.3
5
8
14
13
44
100
0.3
0.5
0.9
GR
2
4
9
0.8
1.0
1.2
8
9
44
19
43
78
0.3
0.4
0.7
HW
100
100
100
0.0
0.0
0.0
0
0
0
19
44
83
0.5
0.9
1.2
CR
100
100
100
0.0
0.0
0.0
0
0
0
16
39
79
0.6
1.0
1.4
SR
50
100
100
0.0
0.4
1.4
0
0
2
15
38
75
0.7
1.0
1.5
UMW
MR
9
32
72
0.5
0.9
1.4
1
4
13
12
34
80
0.8
1.1
2.0
LR
3
8
21
1.1
1.3
1.4
5
16
25
18
53
88
0.8
1.0
1.1
GR
1
2
8
0.8
0.9
0.9
22
23
25
13
24
45
0.8
1.0
1.0

35

Table 1.5 (cont'd).
UMO (%)
UNDR (#/100 km)
UDOR (%)
DM2D (km)
DMD (#/100 km)
Region Stratum Q25 Median Q75 Q25 Median Q75 Q25 Median Q75 Q25 Median Q75 Q25 Median Q75
HW
100
100
100
0.0
0.0
0.0
0
0
0
19
81
159
0.0
0.0
0.4
CR
100
100
100
0.0
0.0
0.0
0
0
0
30
90
172
0.0
0.0
0.4
SR
100
100
100
0.0
0.6
1.9
0
0
1
55
106
185
0.0
0.0
0.4
CPL
MR
71
100
100
0.5
1.0
1.9
0
1
7
53
124
230
0.0
0.0
0.4
LR
13
39
64
0.9
1.2
2.5
6
14
51
30
71
116
0.0
0.0
0.3
GR
9
18
28
1.1
1.6
2.6
24
41
83
73
167
291
0.0
0.0
0.2
HW
100
100
100
0.0
0.0
0.0
0
0
0
41
112
270
0.1
0.2
0.5
CR
100
100
100
0.0
0.0
0.0
0
0
0
49
118
288
0.1
0.2
0.6
SR
100
100
100
0.0
0.6
1.7
0
0
4
55
124
341
0.1
0.2
0.6
TPL
MR
22
71
100
0.4
0.7
1.7
1
5
16
33
96
260
0.1
0.3
0.7
LR
4
13
30
0.5
0.7
1.2
5
10
23
18
59
189
0.1
0.5
0.8
GR
2
6
24
0.8
0.9
1.2
10
32
125
29
82
486
0.0
0.1
0.4
HW
100
100
100
0.0
0.0
0.0
0
0
0
77
285
558
0.1
0.1
0.1
CR
100
100
100
0.0
0.0
0.0
0
0
0
84
307
604
0.1
0.1
0.1
SR
83
100
100
0.0
0.8
1.7
0
8
47
138
350
612
0.1
0.1
0.1
NPL
MR
64
100
100
0.5
0.9
1.4
4
19
49
156
430
692
0.1
0.1
0.1
LR
56
93
100
0.5
0.7
0.9
17
38
52
126
328
586
0.1
0.1
0.1
GR
11
28
100
0.3
0.7
0.9
20
41
76
302
511
793
0.1
0.1
0.1
HW
100
100
100
0.0
0.0
0.0
0
0
0
42
172
404
0.1
0.3
0.7
CR
100
100
100
0.0
0.0
0.0
0
0
0
51
166
384
0.1
0.4
0.7
SR
94
100
100
0.0
0.5
1.8
0
1
11
73
199
418
0.1
0.3
0.7
SPL
MR
36
88
100
0.3
0.8
1.7
1
8
31
76
200
570
0.1
0.3
0.6
LR
15
35
72
0.4
0.8
1.3
7
45
104
82
341
590
0.1
0.3
0.7
GR
12
22
56
0.5
0.8
1.1
27
59
123
179
360 1171
0.0
0.1
0.7

36

Table 1.5 (cont'd).
UMO (%)
UNDR (#/100 km)
UDOR (%)
DM2D (km)
DMD (#/100 km)
Region Stratum Q25 Median Q75 Q25 Median Q75 Q25 Median Q75 Q25 Median Q75 Q25 Median Q75
HW
100
100
100
0.0
0.0
0.0
0
0
0
39
101
226
0.2
0.5
1.0
CR
100
100
100
0.0
0.0
0.0
0
0
0
40
103
235
0.2
0.5
1.0
SR
100
100
100
0.0
0.0
0.4
0
0
0
41
99
231
0.2
0.6
1.0
WMT
MR
45
100
100
0.1
0.2
0.4
0
3
18
51
112
241
0.3
0.5
1.0
LR
11
32
72
0.1
0.2
0.4
3
15
45
62
113
164
0.3
0.7
1.3
GR
5
30
100
0.1
0.3
0.3
2
18
32
38
102
183
0.3
0.9
1.0
HW
100
100
100
0.0
0.0
0.0
0
0
0
46
150
361
0.0
0.4
0.8
CR
100
100
100
0.0
0.0
0.0
0
0
0
53
166
398
0.0
0.4
0.7
SR
100
100
100
0.0
0.0
0.2
0
0
0
56
173
403
0.0
0.5
0.8
XER
MR
45
100
100
0.1
0.2
0.4
0
7
41
68
172
417
0.2
0.5
0.8
LR
19
32
92
0.2
0.3
0.5
7
25
47
77
184
475
0.3
0.5
0.8
GR
6
18
44
0.2
0.3
0.3
42
69
146
45
152
377
0.5
0.6
1.0

37

Table 1.6.—Patch count, median size characteristics, and percent increase (Inc.)/decrease (Dec.) in metrics by siza
strata for the conterminuous U.S. when comparing patches in the absence of dams (No Dam; Figure 8 top) and
when considering dam locations as artificial habitat patch (AHP) boundaries (Dam; Figure 8 bottom).
Patch Count
Stratum No Dam Dam % Inc.
HW
3810 35017 819
CR
1535 13280 765
SR
456 3661 703
MR
147 1531 941
LR
26
335 1188
GR
33
296 797

Network Length (km)
No Dam Dam % Dec.
2.4
1.7
30
15.3 11.8
23
134.7 87.9
35
1362.3 349.6
74
9918.2 553.0
94
28049.4 2445.4
91

2

Catchment Area (km ) Mainstem Length (km)
No Dam Dam % Dec. No Dam Dam % Dec.
2.3
2.1
8
2.4 1.7
30
23.0 17.9
22
6.4 4.4
31
221.5 145.7
34
17.4 11.0
37
2206.5 691.3
69
56.8 20.9
63
13982.7 844.8
94
91.1 20.2
78
42015.7 3561.9
92
281.6 71.2
75

38

APPENDIX B

FIGURES

Figure 1.1.—Distribution of the nine aggregated ecoregions (A) and dam locations (B; N =
49,468) of the conterminous U.S. For interpretation of the references to color in this and all
other figures, the reader is referred to the electronic version of this thesis.

39

Figure 1.2.—Example depiction of an NHDPlusV1 stream reach and catchment (A) and its
subdivision into stream segments with subsequent delineation of segment catchments (B) for
dams occurring greater than 100 m from an existing node in the NHDPlusV1.

40

Figure 1.3.—Depiction of the traditional scales of catchment delineation (A), consisting of local
catchments (land area draining directly to a stream reach or water body) and network catchments
(cumulative upstream drainage area, including the local catchment of the target stream reach or
water body). In contrast, an alternative set of catchments for artificial habitat patches (AHPs) are
defined by the locations of dams (B).

41

Figure 1.4.—Dam density distributions by size strata for the conterminous U.S (CONUS) and by
ecoregion. Dam densities were calculated per 100 km of network length within a given stratum
and region.

42

Figure 1.5.—Number of dams constructed by decade (source NABD).

43

Figure 1.6.—Distribution of dam age in years for headwater (HW) and creek (CR) size strata for
th
th
the conterminous U.S. (CUS) and ecoregions. Box plot whiskers represent the 5 and 95
percentiles.

44

Figure 1.7.—Distribution of dam age in years for small river (SR) and medium river (MR) size
th
strata for the conterminous U.S. (CUS) and ecoregions. Box plot whiskers represent the 5 and
th

95 percentiles.

45

Figure 1.8.—Distribution of dam age in years for large river (LR) and great river (GR) size strata
th
th
for the conterminous U.S. (CUS) and ecoregions. Box plot whiskers represent the 5 and 95
percentiles.

46

Figure 1.9.—Stacked bar graph of overall reservoir storage showing total contribution of each
size stratum for the conterminous U.S. (CUS) and ecoregions.

47

Figure 1.10.—Cumulative upstream degree of regulation (%) for great river segments for the
conterminous U.S. (CUS) and by ecoregion (total N = 18,752). Box plot whiskers represent the
th
th
5 and 95 percentiles.

48

Figure 1.11.—Individual patches in the absence of dams (A; N = 6,037) and the resulting set of
artificial habitat patches when accounting for dams (B, N = 54,120).

49

Figure 1.12.—Artificial habitat patches (AHPs) for the northeast U.S. with inset of AHPs for the
Saco River (basin highlighted in red in top view). For the bottom view, individual AHPs are
given a unique color, with dam locations represented as red points and streams as blue lines.

50

LITERATURE CITED

51

LITERATURE CITED

Alo, D., and T. F. Turner. 2005. Effects of habitat fragmentation on effective population size in
the endangered Rio Grande silvery minnow. Conservation Biology 19(4):1138-1148.
Bain, M. B., and M. L. Wine. 2010. Testing predictions of stream landscape theory for fish
assemblages in highly fragmented watersheds. Folia Zoologica 59(3):231-239.
Bednarek, A. T. 2001. Undamming rivers: A review of the ecological impacts of dam removal.
Environmental Management 27(6):803-814.
Bond, N. R., and P. S. Lake. 2003. Local habitat restoration in streams: Constraints on the
effectiveness of restoration for stream biota. Ecological Management & Restoration
4:193-198.
Buisson, L., W. Thuiller, S. Lek, P. Lim, and G. Grenouillet. 2008. Climate change hastens the
turnover of stream fish assemblages. Global Change Biology 14: 2232-2248.
Buisson, L. and G. Grenouillet. 2009. Contrasted impacts of climate change on stream fish
assemblages along an environmental gradient. Diversity and Distributions 15:613-626.
Chin, A., L. R. Laurencio, and A. E Martinez. 2008. The hydrologic importance of small- and
medium-sized dams: Examples from Texas. The Professional Geographer 60(2):238251.
Doyle, M. W., E. H. Stanley, D. G. Havlick, M. J. Kaiser, G. Steinback, W. L. Graf, G. E.
Galloway, and J. A. Riggsbee. 2008. Aging infrastructure and ecosystem restoration.
Science 319:286-287.
Dunham, J. B., G. L. Vinyard, and B. E. Rieman. 1997. Habitat fragmentation and extinction
risk of Lahontan cutthroat trout. North American Journal of Fisheries Management
17:1126-1133.
ESRI. 2006. PC ARC/GIS Version 9.2. Redlands, California: Environmental System Research
Institute.
Esselman, P. C., D. M. Infante, L. Wang, D. Wu, A. R. Cooper, and W. W. Taylor. 2011. An
index of cumulative disturbance to river fish habitats of the conterminous United States
from anthropogenic activities in landscapes. Ecological Restoration 29(1-2):133-151.
Esselman, P., D. Infante, L. Wang, A. Cooper, D. Wieferich, Y. Tsang, D. Thornbrugh, and W.
Taylor. 2013. Regional fish community indicators of landscape disturbance to catchments
of the conterminous United States. Ecological Indicators 26:163-173.

52

Fagan, W. F. 2002. Connectivity, fragmentation, and extinction risk in dendritic
metapopulations. Ecology 83(12):3243-3249.
Fausch, K. D., C. E. Torgersen, C. V. Baxter, and H. W. Li. 2002. Landscapes to riverscapes:
Bridging the gap between research and conservation of stream fishes. BioScience
52(8):659-668.
Fausch, K. D., B. E. Rieman, J. B. Dunham, M. K. Young, and D. P. Peterson. 2009. Invasion
versus isolation: Trade-offs in managing native salmonids with barriers to upstream
movement. Conservation Biology 23(4):859-870.
Fukushima, M., S. Kameyama, M. Kaneko, K. Nakao, and E. A. Steel. 2007. Modelling the
effects of dams on freshwater fish distributions in Hokkaido, Japan. Freshwater Biology
52(8):1511-1524.
Fullerton, A. H., K. M. Burnett, E. A. Steel, R. L. Flitcroft, G. R. Pess, B. E. Feist, C. E.
Torgersen, D. J. Miller, and B. L. Sanderson. 2010. Hydrological connectivity for
riverine fish: Measurement challenges and research opportunities. Freshwater Biology
55(11):2215-2237.
Graf, W. L. 1999. Dam nation: A geographic census of American dams and their large-scale
hydrologic impacts. Water Resources Research 35(4):1305-1311.
Guenther, C. B., and A. Spacie. 2006. Changes in fish assemblage structure upstream of
impoundments within the upper Wabash River Basin, Indiana. Transactions of the
American Fisheries Society 135:570-583.
Hall, C. J., A. Jordaan, and M. G. Frisk. 2011. The historic influence of dams on diadromous
fish habitat with a focus on river herring and hydrologic longitudinal connectivity.
Landscape Ecology 26:956-107.
Heggenes, J., and K. H. Roed. 2006. Do dams increase genetic diversity in brown trout (Salmo
trutta)? Microgeographic differentiation in a fragmented river. Ecology of Freshwater
Fish 15:366-375.
Herlihy, A. T., S. G. Paulsen, J. Van Sickle, J. L. Stoddard, C. P. Hawkins, and L. L. Yuan.
2008. Striving for consistency in a national assessment: The challenges of applying a
reference-condition approach to a continental scale. Journal of the North American
Benthological Society 27(4):860-877.
IBM SPSS 20. 2011. IBM SPSS Statistics for Windows, Version 20.0. Armonk, NY.
Jackson, C. R., and C. M. Pringle. 2010. Ecological benefits of reduced hydrologic connectivity
in intensively developed landscapes. BioScience 60(1): 37-46.

53

Januchowski-Hartley, S. R., P. B. McIntyre, M. Diebel, P. J. Doran, D. M. Infante, C. Joseph,
and J. D. Allan. 2013. Restoring aquatic ecosystem connectivity requires expanding
barrier inventories. Frontiers in Ecology and the Environment (early view).
Lehner, B., C. R. Liermann, C. Revenga, C. Vorosmarty, B. Fekete, P. Crouzet, P. Doll, M.
Endejan, K. Frenken, J. Magome, C. Nilsson, J. C. Robertson, R. Rodel, N. Sindorf, and
D. Wisser. 2011. High-resolution mapping of the world’s reservoirs and dams for
sustainable river-flow management. Frontiers in Ecology and the Environment 9(9):494502.
Maddock, I. 1999. The importance of physical habitat assessment for evaluating river health.
Freshwater Biology 41(2):373-391.
Morita, K., and S. Yamamoto. 2002. Effects of habitat fragmentation by damming on the
persistence of stream-dwelling charr populations. Conservation Biology 16(5):13181323.
Novinger, D. C., and F. J. Rahel. 2003. Isolation management with artificial barriers as a
conservation strategy for cutthroat trout in headwater streams. Conservation Biology
17(3):772-781.
Palmer, M. A., E. S. Bernhardt, J. D. Allan, P. S. Lake, G. Alexander, S. Brooks, J. Carr, S.
Clayton, C. N. Dahm, J. Follstad Shah, D. L. Galat, S. G. Loss, P. Goodwin, D. D. Hart,
B. Hassett, R. Jenkinson, G. M. Kondolf, R. Lave, J. L. Meyer, T. K. O’Donnell, L.
Pagano and E. Sudduth. 2005. Standards for ecologically successful river restoration.
Journal of Applied Ecology 42:208-217.
Perkin, J. S. and K. B. Gido. 2011. Stream fragmentation thresholds for a reproductive guild of
Great Plains fishes. Fisheries 36(8):371-383.
Peterson, D. P., B. E. Rieman, J. B. Dunham, K. D. Fausch, and M. K. Young. 2008. Analysis
of trade-offs between threats of invasion by nonnative brook trout (Salvelinus fontinalis)
and intentional isolation for native westslope cutthroat trout (Oncorhynchus clarkii
lewisi). Canadian Journal of Fisheries and Aquatic Sciences 65:557-573.
Poff, N. L., and D. D. Hart. 2002. How dams vary and why it matters for the emerging science
of dam removal. BioScience 52(8):659-668.
Rieman, B. E., D. Issak, S. Adams, D. Horan, D. Nagel, C. Luce, and D. Myers. 2007.
Anticipated climate warming effects on bull trout habitats and populations across the
interior Columbia River basin. Transactions of the American Fisheries Society
136:1552-1565.
Riley, S. C., and K. D. Fausch. 1995. Trout population response to habitat enhancement in six
northern Colorado streams. Canadian Journal of Fisheries and Aquatic Sciences 52:3453.

54

Schlosser, I. J., and P. L. Angermeier. 1995. Spatial variation in demographic processes of lotic
fishes: Conceptual models, empirical evidence, and implications for conservation. Pages
392-401 in J. L. Nielsen, editor. Evolution and the aquatic ecosystem: defining unique
units in population conservation. American Fisheries Society, Symposium 17, Bethesda,
Maryland.
Steel, E. A., R. M. Hughes, A. H. Fullerton, S. Schmutz, J. A. Young, M. Fukushima, S. Muhar,
M. Poppe, B. E. Feist, and C. Trautwein. 2010. Are we meeting the challenges of
landscape-scale riverine research? A review. Living Reviews in Landscape Research 4.
USACE (U.S. Army Corps of Engineers). 2009. National inventory of dams. Unavailable
online (July 2010).
USEPA and USGS (U.S. Environmental Protection Agency and U.S. Geological Survey). 2005.
National hydrography dataset plus – NHDPlus. Edition 1.0. Available at
http://www.horizon-systems.com/nhdplus/index.php (August 2006).
USFWS (U.S. Fish and Wildlife Service). 2008. Fish Passage Decision Support System.
Available at ecos.fws.gov/geofin. (September 2008).
USGS. 2013. 2012 National Anthropogenic Barrier Dataset. Available at
https://www.sciencebase.gov/catalog/item/get/512cf142e4b0855fde669828. (September
2010).
Wang, L., D. Infante, J. Lyons, J. Stewart, and A. Cooper. 2011a. Effects of dams in river
networks on fish assemblages in non-impoundment sections of rivers in Michigan and
Wisconsin, USA. River Research and Applications 27(4):473-487.
Wang, L., D. Infante, P. Esselman, A. Cooper, D. Wu, W. Taylor, D. Beard, G. Whelan, and A.
Ostroff. 2011b. A hierarchical spatial framework and database for the national river fish
habitat condition assessment. Fisheries 36(9):436-449.
Ward, J. V., and J. A. Stanford. 1983. The serial discontinuity concept of lotic ecosystems.
Pages 29-42 in T. D. Fontaine III and S. M. Bartell, editors. The ecology of regulated
rivers. Plenum Press, New York.
Ward, J. V. 1989. The four-dimensional nature of lotic ecosystems. Journal of the North
American Benthological Society 8(1):2-8.

55

CHAPTER 2
EFFECTS OF DAMS ON FLUVIAL FISH ASSEMBLAGES IN MICHIGAN, WISCONSIN,
AND MINNESOTA: CONSIDERATION OF MULTIPLE MEASURES OF
FRAGMENTATION AND DAM INFLUENCE

56

ABSTRACT
While the site-specific and localized effects of dams on stream fish assemblages have
been relatively well-studied, little is known about dam effects on fishes across much larger
geographic scales such as entire river basins, ecoregions, and states. Uncovering patterns in
these influences would provide useful information for a variety of management activities,
including dam operation and dam removal prioritization. This study evaluated multiple networkbased dam measures representing proximity-based (distance-to-dams) and cumulative (e.g.
reservoir storage) dam influences for streams of Michigan, Wisconsin, and Minnesota,
identifying the species considered to be most sensitive to dam effects. Using change point and
correlation analyses, responses of stream fishes indicated by change in catch per unit effort in
relation to various dam metrics were analyzed using a total of ~2,000 fish survey sites stratified
by stream size, thermal regime, and ecoregion. Of the identified sensitive species, those that
were positively associated with greater dam influences were predominantly warmwater, large
river, and/or lentic species, while species negatively associated with greater dam influences were
cold and coolwater lotic species, suggesting a combination of downstream thermal effects and
upstream influences from impoundments generated by dams. A variance partitioning analysis
using sensitive species as indicators revealed a transition from upstream-dominated dam
influences in headwaters to a mixture of upstream/downstream influences in mid-sized streams.
Overall, a combination of proximity-based and cumulative metrics as well as both upstream and
downstream-oriented measures were influential in species’ responses, emphasizing the
importance of selecting a variety of dam measures when assessing the effects of dams on stream
fishes. Dams represent unique opportunities for the conservation and management of fishes,
which can be aided through the identification of large-scale dam influences on fish assemblages.

57

INTRODUCTION
Dams represent a complex form of human disturbance to riverine systems, altering
habitat both above and below the location of the dam within the stream network. Often these
alterations involve multiple facets of abiotic habitat including hydrology, stream temperature,
channel morphology, water chemistry, and multiple aspects of hydrologic connectivity
throughout river networks. In addition, individual dam influences often vary with factors such as
their position within the stream network (Ward and Stanford 1983), the way in which they are
managed (Poff et al. 1997), and the length of time that they have been in operation (Poff and
Hart 2002). Multiple dams along the stream network can have cumulative effects (e.g. Pringle
2001, Bosch 2008). The complex set of influences that dams can have on stream systems
highlights the need to account for dams both spatially and temporally in studies attempting to
characterize effects of dams on fluvial habitat and biota. Ultimately, understanding individual
and cumulative influence of sources of stream habitat degradation like dams depends on
characterizing the location and intensity of each degradation source, as well as the regional
context in which disturbances may be occurring (Utz et al. 2010).
Numerous studies have documented the effects of dams on stream fishes, typically by
sampling fish or abiotic habitat above and below dams, by comparing surveys between dammed
and undammed streams, or by sampling along individual stream networks containing a series of
dams. Such studies highlight localized effects of dams, including thermal shifts resulting in
decreased abundance of coldwater species, increased prevalence of lentic species above
reservoirs, and overall species composition changes (e.g., Lessard and Hayes 2003, Guenther and
Spacie 2006, McLaughlin et al. 2006, Slawski et al. 2008). In some cases, studies have observed
fish assemblage changes by surveying before and after dam removals (Catalano et al. 2007,

58

however see Stanley et al. 2007). Largely missing are a complimentary set of studies
incorporating spatial measures to investigate the influences of dams on fishes across large
geographic regions, such as entire river networks, ecoregions, or multi-state regions. Studies
such as these could identify patterns of dam influences not observable at smaller scales, or
alternatively, could determine if observations made in smaller field-based studies are evident
across larger regions. Information provided by large-scale studies could be utilized for
prioritizing dam management actions, benefiting fisheries conservation and management.
Recently, studies have begun investigating dam influences on fishes over large-scale
regions, often through use of geographic information systems. For example, Fukushima et al.
(2007) identified catchment sub-basins isolated by dams for a regional analysis of the
presence/absence of fishes on the island of Hokkaido, Japan. Using species-specific generalized
linear models, the authors found that fragmentation by downstream dams had an influence on 11
of 41 species studied, with eight migratory species predicted to have decreased occurrences and
three non-migratory species predicted to have increased occurrences resulting from greater
fragmentation. The authors also created a variable representing the duration of isolation using
the year when the dam was built, which was found to be a significant indicator variable in the
occurrence models for eight of the 11 impacted species. Similarly, Hall et al. (2011) used dam
construction dates obtained from historical records to create a fragmentation timeline for
watersheds in Maine. This information was used to calculate the percentage of lake and stream
habitat accessible to two species of anadromous river herring, alewife (Alosa pseudoharengus)
and blueback herring (Alosa aestivalis), at various time steps. This data allowed the authors to
describe current and historic fragmentation patterns, identifying trends in the loss of access to
lake and stream habitat through time.

59

Other studies have focused on how dams alter free-flowing stream lengths, exploring the
relationship between stream segment size and fish response. A study by Bain and Wine (2010)
analyzed the influence of stream fragment length on fish species diversity, abundance, and size
for 31 locations in the Hudson River basin, New York. As expected, species diversity increased
as a function of fragment size, however larger fragments did not contain a greater overall fish
abundance or wider range of fish sizes as the authors expected, possibly due to the confounding
influence of the stocking of the two dominant species in the study area, brown trout (Salmo
trutta) and brook trout (Salvelinus fontinalis). The authors found that fragments containing
naturally produced brown trout and brook trout were significantly larger than those that did not
support reproduction. Perkin and Gido (2011) analyzed river fragment length (largely bounded
by dams and reservoirs) for pelagic spawning fishes in the Great Plains region under multiple
population status conditions (extirpated, declining, and stable). The authors found significant
differences in effect of fragment length both for individual species and for all species combined,
with a general pattern of increased fragment length resulting in species moving from extirpated
to declining to stable population conditions. The authors also estimated fragment length
thresholds associated with the localized extirpation of each species, which together explained
67% of the variation in population persistence among the species studied.
Lastly, Wang et al. (2011a) developed multiple spatially-explicit measures of
fragmentation by dams for the states of Michigan and Wisconsin, including distances from a
given stream reach to the nearest upstream and downstream dam along the mainstem of the river
network, total number and density of dams along all flow paths upstream, and total number and
density of dams along the downstream mainstem. These measures were used in a multivariate
statistical analysis that partitioned the relative influence of dams from other environmental

60

covariates (non-dam measures that included both natural and anthropogenic variables) using
selected Index of Biotic Integrity (IBI) and habitat/social preference fish metrics as response
variables. While dam influences only accounted for 16% and 19% of the total variation
explained for groups of IBI-based metrics and habitat/social preference metrics respectively, the
authors did find that dams had a significant impact (both positive and negative) on fish metrics
evaluated in the study. For instance, overall IBI varied with multiple dam measures, including
distance-to-dam measures (upstream, downstream, and total free mainstem) and cumulative
measures (upstream dam count, downstream dam count and density). These results suggest that
key measures of stream fish assemblages are responsive to dam influences 1) in both the
upstream and downstream direction, and 2) in both a localized (proximity to individual dams)
and cumulative (dam counts/densities within catchments) context.
Despite such advancements in our understanding of dam influences on fish, measuring
fragmentation by dams, and river connectivity as whole, continues to present challenges to
stream ecologists and managers (Wang et al. 2006, Steel 2010). Although many studies have
chronicled the effect of dams on fishes at localized scales (e.g. above or below a dam or before
or after a dam removal), few studies have investigated the individual and cumulative effects of
dams within river networks across large regions. The ability to discern between localized
(resulting from effects of individual dams) and cumulative (resulting from multiple dams located
throughout a river network) dam effects, for example, and revealing any changes in the relative
influence of such factors throughout the river network would yield a much broader
understanding of the complex landscape-scale effects that dams have on fishes. Despite the
pervasive threat of dams to lotic systems, dams represent a unique restoration opportunity, one

61

that would be informed by understanding individual species-specific responses to multiple dam
measures as a function of network position.
This study addresses these needs through a detailed evaluation of individual fish species
responses to network-based dam metrics in the states of Michigan, Wisconsin, and Minnesota.
The main objectives of this study are to: 1) Determine which fish species are most sensitive to
dam influences, 2) Identify the most influential dam measures as they relate to species responses,
and 3) Examine the relative influence of dams on identified sensitive fish species as a function of
stream network position (i.e. stream order).
METHODS
Study area and spatial framework
Study area.—The study area occurs throughout the states of Michigan, Wisconsin, and
2

Minnesota, USA, a land area encompassing 514,000 km with approximately 262,000 km of
streams (as represented by the 1:100,000 NHDPlusV1; USEPA and USGS 2005). This region
has been heavily influenced by glaciation and lacustrine deposition (e.g., Farrand and Bell 1982),
leading to a diverse series of surficial geology deposits that vary widely in texture (e.g. coarse vs.
fine) and landform (e.g. moraine, outwash plain, lake plain). This geologic and topographic
complexity results in highly variable groundwater discharge regimes throughout the region,
including streams receiving large amounts of groundwater with relatively stable flow and
thermal regimes (commonly draining areas with coarse-textured land forms) to streams driven
primarily by surface runoff with “flashy” flow regimes and variable water temperatures (e.g.
silt/clay lake plain landform; Seelbach et al. 1997, Zorn et al. 2002).
In addition to highly variable natural features, this region also contains a range of
landscape-based anthropogenic disturbances, including human land use, roads, and point-source

62

pollution sites. In general, this gradient has a north/south distribution, from the relatively leastdisturbed northern areas dominated by forest and wetlands to highly disturbed areas that are
typical of the southern portions of the study region, which contain high levels of agricultural and
urban land uses (e.g., Wang et al. 2008).
Spatial data.—The 1:100,000 scale National Hydrography Dataset Plus Version 1 stream
network (NHDPlusV1; USEPA and USGS 2005) was used in this study (Chapter 1 Methods).
Network catchments (i.e. the cumulative upstream land area draining to a reach, including a
reach’s local catchment) are not represented as a spatial data layer in the NHDPlusV1, however
network catchment attributes can be generated by the upstream aggregation of local catchment
data using stream network topology information (Tsang et al. in prep.). This study utilized preexisting network catchment attributes summarized within a national spatial framework (Wang et
al. 2011b), which includes a suite of variables representing both natural and anthropogenic data
layers (Table 2.1).
In total, 2,303 dams from the National Anthropogenic Barrier Dataset (NABD; USGS
2013; Chapter 1) were identified as being connected to stream networks within the study region
(Figure 2.1). Dam spatial location along the stream network, as well as attributes of individual
dams (e.g. reservoir storage), were used to develop a set of 18 metrics describing fragmentation
by dams within the study region (Chapter1 Methods; Table 1.1).
Fish data.—Fish survey data were acquired for Michigan, Wisconsin, and Minnesota
from respective state natural resource agencies as well as a federal program (U.S. Geological
Survey, National Water-Quality Assessment Program). Survey site coordinates were verified
against the NHDPlusV1 streams using similar methodology applied for dam site verification,
ensuring that survey sites were spatially referenced to the correct stream reach. Any survey

63

locations that could not be ascertained using available site information were not included in this
study. A subset of the survey sites were then selected to produce a dataset with comparable
survey methodology which covered the time period from 1995 to 2010. These included surveys
that targeted community assemblages. We limited the fish data to surveys conducted using
either single-pass electrofishing methods (primarily boat, barge, or backpack electrofishing
gear), or used only the first pass results from multiple-pass electrofishing surveys. Distances
over which surveys were conducted allowed for calculation of individual species catch per unit
effort standardized by survey length (#/100 meters of stream length). For survey sites that were
sampled more than once, the most recent survey data were used. As a last step, survey sites that
had greater than 10% urban and/or greater than 60% agricultural land uses in their network
catchments were excluded from analysis, following a previous study of Midwestern streams that
identified high levels of anthropogenic land use related to declining fish index of biotic integrity
(IBI) scores (Wang et al. 1997). Removing these sites reduced potential influences of highly
altered upstream landscapes on stream fish assemblages, which could confound analyses
intended to detect relationships between dam metrics and fish responses (Wang et al. 2011a).
This process resulted in the initial selection of 442 survey sites for Michigan, 834 survey sites for
Wisconsin, and 835 survey sites for Minnesota. We also used Lyons et al. (2009) to characterize
the thermal preferences of individual fish species (cold vs. warm) within the study region.
Site stratification.—To account for the large degree of natural variation within the study
region and to account for large-scale biogeographic shifts in taxa distribution across the region, a
multi-level stratification approach was employed to separate fish sites into distinct groups for
analysis. The first level of the stratification consisted of freshwater ecoregions developed by
Abell et al. (2008), which were largely determined through analyzing geographic patterns in fish

64

taxa distributions. Ecoregions reflect the influence of large scale “filters” (e.g. Tonn et al. 1990,
Poff 1997), including ecological and evolutionary processes, that have shaped contemporary fish
biogeography across large regions. The study area included three primary ecoregions:
Laurentian Great Lakes (LGL), Upper Mississippi (UM), and English-Winnepeg Lakes (EWL)
(Figure 2.1). A fourth ecoregion, Middle Missouri, occurred in southwest Minnesota, however
only one fish site was located within this ecoregion after site selection screening process due to
high levels of agricultural land use in this area. As a result, this ecoregion was dropped from
analysis. The second level of the stratification was stream size, with known influences on
distributions of fish species (Lyons 1996, Zorn et al. 2002, Goldstein and Meador 2004). We
used Strahler stream order (Straher 1957) to generate three stream size groups: headwater (HW;
st

nd

1 and 2

rd

th

th

th

order streams), mid-size (MS; 3 and 4 ), and large size (LS; 5 and 6 ). Lastly,

sites were classified into two thermal groups, warmwater and coldwater, using stream
temperature estimates by Krueger et al. (in prep) for stream reaches throughout Michigan,
Wisconsin, and Minnesota that were assigned to a thermal classification first developed in
Michigan and Wisconsin by Lyons et al. (2009). Of the 16 resulting strata, 13 had sample sizes
deemed large enough for statistical analysis (sample size > 30; e.g., Utz et al., 2010; Table 2.2)
resulting in the final selection of 2,067 fish survey sites across the study region (Figure 2.1).
Statistical approaches
Data preparation and variable reduction.—Prior to evaluating the normality of the
fragmentation metrics, metrics were transformed using natural log (X + 0.01) for density and
storage metrics, arc-sine square root for proportion metrics, and square root for count metrics.
The distributions of fragmentation metrics were then evaluated by visually inspecting P-P plots
(IBM SPSS Statistics 20 2011). Some fragmentation metrics had highly skewed distributions

65

even after transformation, resulting in use of Spearman correlation (a rank-based non-parametric
correlation approach) to evaluate interrelationships between transformed variables. As a first
step in metric reduction, dam density metrics were selected over dam count metrics (UMCT,
UNCT, DMCT, and TMCT) to better control for the influence of stream size. Next, correlation
coefficients among fragmentation metrics were examined for each stratum using SPSS (IBM
SPSS Statistics 20 2011), with the least-correlated (typically < 0.8) set of upstream and
downstream-oriented metrics being retained to facilitate comparisons among size classes. Of the
14 metrics initially considered, six metrics were retained for the headwater and mid-size classes,
while 10 metrics were selected for the large size class, respectively (Table 1.1).
Detecting fish relationships with fragmentation metrics.—Change point analysis was
conducted to identify distinct, step function responses of individual fish species to dam
fragmentation metrics within each stratum using threshold indicator taxa analysis (TITAN)
methodology (Baker and King, 2010). This approach was selected due to its ability to test for
step function responses of individual taxa (as opposed to threshold approaches that are used only
for pooled, community responses), including those that occur at relatively low frequencies or
have highly variable abundance (Baker and King 2010) and for its ability to detect associations
between taxa and environmental factors. TITAN employs a rigorous multi-step significance
screening process to identify responses. This screening process includes testing for significant pvalues from indicator species analysis results (Dufrene and Legendre 1997) using random
permutations of the environmental variable. For this study, we chose to use 250 permutations
and a p-value of 0.05 for the indicator species analysis cutoff. Next, TITAN uses bootstrap
replicates (N = 500 for this study) to provide measures of purity, the proportion of replicates
matching the direction of the observed species response (positive or negative), and reliability, the

66

proportion of bootstrap-derived thresholds that result in indicator species analysis p-values that
fall below a defined probability level (Baker and King 2010). In this study, we selected species
with purity values >=0.95 (i.e., 95% or more of bootstrap replicates matched the observed
response direction) and reliability values of >=0.95 at the 0.05 probably level (i.e. 95% of
replicates had indicator values resulting in p-values <= 0.05) and >=0.5 for the 0.01 probability
level, respectively. Prior to analysis, species catch per unit effort data were log transformed (log
(x + 1)) to reduce the effect of species of with highly variable abundances on species indicator
analysis results (Baker and King, 2010). Species with less than 20 occurrences for a given
fragmentation metric (including ‘conditional’ metrics, such as distances-to-dam measures) within
a stratum were removed from the analysis to eliminate rare and underrepresented species. The
untransformed fragmentation metrics were used as the environmental variables in the threshold
analysis. TITAN threshold analysis was performed using the TITAN package available for R
statistical software (R 2.15 2012).
Because relationships between species distributions and fragmentation metrics may not
necessarily show abrupt transitions that could be characterized by point change analysis, we also
evaluated relationships of species distributions with fragmentation metrics using Spearman
correlation analysis. Correlation was performed only on individual species/metric combinations
that failed to show a threshold response using the screening process from the change point
analysis for each stratum in R statistical software (R 2.15 2012).
Canonical correspondence analysis.—To estimate the relative influences of
fragmentation metrics on selected fish species variables, we used the partial canonical
correspondence analysis (CCA; e.g. Borcard et al. 1992). We partitioned the total variance
explained in fish species indicators (identified in Results) into four variable groupings; natural,

67

non-dam anthropogenic, upstream-oriented fragmentation, downstream-oriented fragmentation
using CANOCO 4.5 software (ter Braak and Smilauer 2002). This analysis was performed to
investigate overall dam influences when compared to other variables (natural and non-dam
anthropogenic) and to identify any changes in the relative orientation (upstream vs. downstream)
of dam influences as a function of size and thermal regime across the study region. Although we
were primarily interested in assessing dam influences, we also treated natural and non-dam
anthropogenic factors as separate variable groups to compare the relative influences of dams to
other anthropogenic sources. Based on a regional set of sensitive species (identified using results
from steps above), fish indicator variables were chosen for the CCA analysis, emphasizing
species that responded similarly within the cold headwater, warm headwater, cold mid-size, and
warm mid-size thermal/size classes (see Table 2.2 for sample sizes). The warm large-size class
had only one candidate sensitive species and was subsequently dropped from the analysis. We
conducted an a priori variable reduction by identifying variable pairs with Pearson correlation
coefficients > 0.8, retaining the variable among the correlated pair that was deemed more easily
interpretable (see Table 2.1 for selected natural and non-dam anthropogenic variables). To
account for potential differences in species responses among ecoregions, ecoregion was added as
a categorical variable and included in the natural variable group. This process resulted in the
selection of six natural variables, five non-dam anthropogenic variables, and five dam metrics,
with the dam metrics separated into two upstream-oriented metrics, UMD and USR, and three
downstream-oriented fragmentation metrics, DMD, DM2D, and TMO (Tables 1.1 and 2.1).

68

RESULTS
Site landscape characteristics
Natural landscape characteristics.—Catchment area ranged from one to 93,000 km

2

across sites, with mean catchment areas tending to increase by an order a magnitude when
moving from headwaters to mid-size to large size classes (Table C1). While mean catchment
areas were similar across all three ecoregions within a given size/temperature class, catchments
in coldwater strata were smaller on average than their warmwater counterparts. Overall, sites in
the EWL ecoregion had colder mean annual air temperatures with less annual precipitation and
base flow than those in the LGL and UM ecoregions. While EWL ecoregion sites had greater
mean elevations than those in the LGL and UM ecoregions, they also had lower mean slope with
each size/temperature class. Catchment soil permeability was similar across most strata, with the
exception of coldwater headwater and mid-size coldwater sites in the UM ecoregion, which had
lower permeability than corresponding coldwater EWL and LGL sites. Within mid-size and
large size sites, the EWL ecoregion had a lower proportion of coarse-textured lithography than
the LGL and UM ecoregions. Average forest land cover varied from 38-59%, with similar
values occurring across strata within the same size/temperature class. In general, LGL and UM
sites had comparable natural catchment conditions for a given size/temperature class, including
mean annual air temperature, annual precipitation, and base flow.
Non-dam anthropogenic landscape factors.—Among non-dam anthropogenic variables,
EWL ecoregion sites had lower densities of roads and human population than the LGL and UM
ecoregions (Table C1). Average percentage of urbanization and imperviousness was low overall
(due to the site selection criteria of <10% urbanization), but tended to be highest in the LGL
ecoregion and lowest for the EWL ecoregion. For the EWL ecoregion, the average percentage of

69

agriculture increased within catchments when moving from headwater to mid-size to large size
classes. For the LGL ecoregion, coldwater classes had less overall agriculture than
corresponding warmwater size classes, however the opposite trend occurred for the UM
ecoregion.
Dam metrics.—Upstream mainstem and upstream total river network dam densities
ranged from 0.4 to 2.4 dams/100 river km across strata, with the lowest densities found in the
EWL ecoregion and highest fluctuating between the LGL and UM ecoregions by stratum (Table
C2). Headwaters and mid-size classes were characterized by a high proportion of upstream
mainstem openness (~ 0.7 and higher), which dropped for large size classes (range of 0.3-0.5).
In general, upstream reservoir storage (per unit catchment area or river network length) tended to
increase as a function of increasing size class. Average mainstem lengths (upstream,
downstream, and total) were typically longest in the EWL ecoregion for a given stratum and
comparable for the LGL and UM ecoregions.
Fish species response to fragmentation
Interpretation of results.—The number of fish species analyzed for each strata ranged
from 11 to 40 (Table 2.3) with a total of 59 species in 13 families across all strata. Species were
selected that had significant relationships with at least two of the six dam metrics in the
headwater and mid-size classes or at least three of 10 dam metrics for the large size class. This
led to the identification of a subset of species that were considered sensitive to fragmentation.
Within each stratum, relationships (hereafter referred to as “associations”) between each species
and dam measure were defined as positive, negative, or mixed through summary of individual
species’ responses to multiple dam metrics (Table 1.1 provides examples of how to interpret
species responses to each metric). For species associated with two or three metrics, we required

70

unanimous direction among metric responses to assign an association (i.e., for a given species,
all responses indicated increasing catch per unit effort with increasing or decreasing
fragmentation). For species associated with four or more fragmentation metrics, we required
75% of metric responses to be in agreement. Finally, species not meeting the conditions above
were classified as showing a mixed response to fragmentation (Table 2.3). The results below are
described for each thermal/size class, with emphasis placed on consistent species responses
among strata and the identification of key fragmentation metrics as they relate to the most
species responses.
Cold headwaters.—Overall there were 8, 19, and 23 associations identified for the EWL,
LGL, and UM ecoregions, respectively (Table 2.3). Within the EWL and LGL ecoregions, there
was a relatively even number of positive and negative species associations with few mixed
associations, whereas species associations in the UM ecoregion were predominately positive or
mixed. No species responded consistently across all three ecoregions, however five species had
positive responses to fragmentation in both the LGL and UM ecoregions, including hornyhead
chub (Nocomis biguttatus) and three centrarchids: rock bass (Ambloplites rupestris), bluegill
(Lepomis macrochirus), and largemouth bass (Micropterus salmoides) (Table 2.4). In contrast,
two species were negatively associated with fragmentation across two ecoregions, including
brook stickleback (Culaea inconstans) and pearl dace (Margariscus margarita). Three other
species, including white sucker (Catostomus commersonii), had opposite responses across two or
more ecoregions.
For individual dam metrics, a majority of species were associated with downstreamoriented metrics (DM2D, DMD, and TMO) in the LGL and UM ecoregions, compared to the
EWL ecoregion, where species’ response was mixed between upstream and downstream metrics.

71

Overall, fish were generally positively associated to upstream metrics in all three ecoregions,
while the downstream-oriented metrics tended to have more of a mixture of positive/negative
species response.
Warm headwaters.—Compared to other strata, there were relatively few associations
detected for species in warm headwaters (Table 2.3). Three species with positive associations
with fragmentation in cold headwaters (LGL & UM), bluegill, largemouth bass, and hornyhead
chub, also had positive associations in the UM warm headwaters (Table 2.5). Similarly, brook
stickleback, negatively associated to fragmentation in cold headwaters (LGL & UM), also had a
negative association in the UM warm headwaters. Due to the limited number of associations
occurring in the warm headwaters, no trends in species-metric relationships were interpreted.
Cold mid-size.—There were 17 species associations in both the LGL and UM ecoregions,
with a majority of associations being positive overall (i.e., species catch per unit effort increased
with fragmentation, Table 2.3). Several species with positive associations to fragmentation in
cold headwaters also showed positive associations for cold mid-size streams, including rock
bass, bluegill, largemouth bass, and hornyhead chub (Table 2.6). Three additional centrarchid
species, green sunfish (Lepomis cyanellus), pumpkinseed (Lepomis gibbosus), and smallmouth
bass (Micropterus dolomieu) also had positive associations with fragmentation in one or both
ecoregions. No species had a consistent negative response to fragmentation across both
ecoregions.
For dam metrics, downstream mainstem density (DMD), total mainstem openness (TMO;
LGL only) and two upstream-oriented metrics (UNDR & USR) had the greatest number of
species-metric relationships (Table 2.6). For both ecoregions, upstream species-metric
relationships were predominately positive as was also the case for the cold headwater strata.

72

Among the upstream-oriented variables, cumulative network metrics (UNDR & USR) tended to
have a higher number of associations than the upstream mainstem metric (UMD) for both
ecoregions. Overall, there were a comparable number of the upstream and downstream oriented
species-metric relationships in both ecoregions.
Warm mid-size.—Total associations identified for warm mid-size strata varied from 13 in
the EWL ecoregion to 19 in both the LGL and UM ecoregions, respectively (Table 2.3).
Similarly to cold mid-size strata, most species associations were positive across all ecoregions.
Two species, bluegill and hornyhead chub, had positive associations to fragmentation in all three
ecoregions (Table 2.7). Four additional species had positive associations among two ecoregions,
including pumpkinseed, largemouth bass, yellow perch (Perca flavescens), and bluntnose
minnow (Pimephales notatus). In total, six species had negative associations to dam metrics
across two ecoregions, which included brassy minnow (Hybognathus hankinsoni), burbot (Lota
lota), and longnose dace (Rhinichthys cataractae), the latter two species also having a negative
association in the cold mid-size strata.
The dominant dam metrics varied among ecoregions, with influential upstream-oriented
metrics varying from largely cumulative upstream metrics (UNDR & USR) in the EWL, to the
upstream mainstem metric (UMD) in the LGL ecoregion to a combination of cumulative and
mainstem metrics for the UM ecoregion (Table 2.7). In contrast to cold mid-size strata, upstream
oriented species-metric relationships in the warm mid-size strata were a mixture of positive and
negative species responses (instead of mainly positive only in cold mid-size strata). Among
downstream-oriented metrics, there were several relationships observed for the distance to
downstream mainstem dam metric (DM2D) metric in the LGL and UM ecoregions (particularly
for centrarchids), while none were identified for the EWL ecoregion. Overall, a majority of

73

species-metric relationships were linked to upstream-oriented metrics for all three ecoregions
(Table 2.7).
Warm large size.—Within this strata, the number and direction of species associations
was highly variable (Table 2.3). Species associations ranged from being entirely positive in the
EWL ecoregion to a mixture of responses in the LGL ecoregion to largely negative associations
for the UM ecoregion. Of the 18 species associations identified in the UM ecoregion, only two
had positive associations to increased fragmentation (Table 2.8). One species, shorthead
redhorse (Moxostoma macrolepidotum) had a similar response between the LGL and UM
ecoregions, while several additional species had conflicting responses among ecoregions.
Overall, fragmentation influences in the EWL and UM ecoregions were a mixture of
downstream, upstream, and total mainstem metrics, whereas in the LGL ecoregion speciesmetric relationships were associated with upstream-oriented fragmentation, namely upstream
mainstem density (UMD) and upstream reservoir storage (USR; Table 2.8). For downstreamoriented metrics, most species-metric relationships were associated with the downstream
mainstem dam density (DMD) metric and relatively few were linked to the downstream
mainstem openness (DMO) metric, particularly for the UM ecoregion.
Region-wide trends.—By evaluating species associations across strata, an overall list of
general species sensitive to fragmentation was determined by examining patterns in associations
(Table 2.9). Species were selected that had a consistent response direction (either positive or
negative) for at least two strata within a thermal/size class. While there were numerous mixed
associations identified for some species, the variable nature of their relationship with individual
dam metrics (mixture of positive and negative responses) suggests that they do not respond to
fragmentation in the same manner.

74

Across the study region, species associations for warmwater species were largely positive
in nature while coldwater species tended to respond negatively to dam influences. In particular,
hornyhead chub, yellow perch, and multiple centrarchid species had positive associations to
fragmentation across multiple strata despite differences in thermal class and/or stream size
(Table 2.9). The majority of species responding negatively to fragmentation were found in the
mid-size class, which included six species from five different families. Only two species, pearl
dace and brook stickleback, had negative associations in the headwaters while one species,
shorthead redhorse, was selected in the large size class due to the lack of consistent species
responses across strata. A few species that are among the most ubiquitous across the study
region, white sucker, common shiner (Luxilus cornutus), and creek chub (Semotilus
atromaculatus), also had the most variable responses to dams across strata (mixture of positive,
negative, mixed, and no associations; Table 2.9).
Both within and across size/thermal classes, there was variation in the key fragmentation
metrics. The orientation of these metrics (upstream vs. downstream) tended to differ between
coldwater and warmwater strata. For coldwater strata, there appeared to be a transition from
downstream-oriented metrics in headwaters to both upstream and downstream metrics for midsize streams. In contrast, there was greater upstream-oriented dam influence for warm mid-size
streams. For the large size class, there was high degree of variability in the key upstream and
downstream metrics among ecoregions.
Canonical correspondence analysis results.—Overall, CCA explained between 15 to
28% of the total variation in fish response variables within the four thermal/size classes (Table
2.10). Of the explained variation, natural factors accounted for a majority of the variation in
three of four classes, with interaction effects representing a highly variable influence across

75

classes (Figure 2.2). Fragmentation accounted for 10 to 19% of the variation explained across
classes while non-dam anthropogenic factors ranged from nine to 29% of variation explained. In
the cold headwater and cold mid-size classes, fragmentation accounted for as much or more of
the explained variation than non-dam anthropogenic factors, whereas non-dam anthropogenic
influences explained twice or more variation than fragmentation in the warm headwater and
warm mid-size classes (Table 2.10). Among the variation uniquely explained by fragmentation
effects on fish response variables, upstream-oriented influences accounted for more variation
within the cold headwater and warm headwater classes, at 63 and 64% of variation, respectively
(Table 2.10). In contrast, downstream-oriented fragmentation accounted for a majority of
variation attributed to fragmentation in the cold mid-size and warm mid-size classes at 54 and
60%, respectively.
Plots from the CCA analyses showed strong associations between individual species and
the natural, dam, and non-dam anthropogenic variables (Figures 2.3-2.6). Among dam variables,
TMO and DM2D were strongly positively associated with multiple species depending upon the
size/thermal class, including pearl dace, central mudminnow (Umbra limi), brook stickleback,
yellow perch, and brassy minnow. Conversely, bluegill and largemouth bass tended to be
negatively related to the TMO and DM2D variables indicating an association with shorter stream
mainstems and closer proximity to downstream dams. In the cold thermal classes, cumulative
upstream storage (USR) was positively associated with hornyhead chub and yellow bullhead in
the headwaters and bluegill and largemouth bass in mid-size streams (Figures 2.3 and 2.5).
Overall, bluegill tended to be associated with a mixture of dam and non-dam anthropogenic
influences.

76

DISCUSSION
This study used landscape-based dam fragmentation measures to explore influences of
dams on fish species within the states of Michigan, Wisconsin, and Minnesota, with three main
objectives; first to identify a core set of sensitive fish species with respect to stream habitat
alterations by dams, secondly to identify key fragmentation metrics related to species responses,
and lastly to investigate the relative influence of dams (including upstream vs. downstream
influences) on sensitive species when compared with other covariates such as natural and nondam human disturbance variables. There are three main findings in this study. First, among
sensitive species, there was a strong positive association between warmwater species and greater
dam effects, while cold/cool water species were negatively associated with dam measures.
Secondly, both proximity-based measures to individual dams (e.g. distance to dams, available
mainstem length) and cumulative dam measures (e.g. total upstream storage) were influential in
changes to the catch per unit effort for sensitive species. Lastly, irrespective of stream
temperature class (cold or warm), dam influences on sensitive species transitioned from
predominantly upstream-oriented influences in headwaters to greater downstream-oriented
influences in mid-size streams.
Overall, a total of 18 sensitive species (out of potential pool of 40 species) were selected
due to having consistent responses within one or more size/thermal classes. This level is similar
to those found by McLaughlin et al. (2006) (albeit with different study design and response
measures) which examined fish species sensitivity to low-head lamprey barriers in the Great
Lakes basin by matching streams with barriers to undammed reference streams. Of the 48
species analyzed, the authors considered between 8-19 species as sensitive to dams depending
upon the type of sensitivity measure being used (catch per unit effort and two types of ratio

77

measures comparing taxon abundance above and below barriers). One key difference in the
results of McLaughlin et al. (2006) and those of the current study is that the authors found a
majority of the sensitive species to be underrepresented above barriers (exhibiting a negative
response), whereas most species associations in this study were positive in nature (resulting in
catch per unit effort with greater fragmentation and dam influence). In examining the thermal
preferences for the 18 total species considered sensitive to dams in this study, many of the
species with a negative association to dams are considered cold or coolwater species (e.g.
mottled sculpin, brook stickleback, burbot, brassy minnow; Lyons et al. 2009) whereas species
with positive associations inhabit warmer streams (e.g. black bullhead, bluntnose minnow,
yellow perch, hornyhead chub, and multiple centrarchid species; Lyons et al. 2009). This result
suggests the potential for wide-scale thermal influence of dams within the study region, as sitespecific studies of the impacts of dams on fishes have also observed. For instance, Slawski et al.
(2008) detected a transition in coolwater/specialist fish species to warmwater/generalist fish
species along dammed mainstem tributaries of the Des Plaines River along the Illinois/Wisconsin
border. A study of ten surface-releasing hydropowered dams on coldwater streams in Michigan
found increased summer stream temperatures below dams, resulting in decreased densities in
three of the four coldwater species studied (Lessard and Hayes 2003). In a related study, Hayes
et al. (2008) concluded that when temperature increases substantially (> 2° C) below
hydroelectric dams, thermal alterations had a much greater impact on fish communities in
coldwater streams than the effects of habitat fragmentation by dams. A study analyzing primary
dam purposes of the NABD (dataset used in this study) found that 16% of dams on small rivers
in the upper Midwest are hydro-powered, increasing substantially to 65% and 90% for medium
and large-sized rivers, respectively (Chapter 1). The thermal influence of surface-releasing dams

78

on cold/coolwater streams could serve as an explanation for the observed increase in catch per
unit effort of warm water species, while cold and coolwater species may have reduced catch per
unit effort as observed in this study.
In addition to potential downstream thermal effects of dams, conversion of fluvial
habitats to the lentic-like impoundments above dams has known consequences on upstream
habitat (Pringle 1997) and fish assemblages (Falke and Gido 2006, Sreekantha and Ramachandra
2008). In the current study, centrarchid species were associated with relatively short distances to
downstream dams (DM2D and/or TMO metric). A study by Guenther and Spacie (2006) yielded
a similar result, finding an increase in the abundance of several centrarchids in streams located
above impoundments when compared to undammed streams in the upper Wabash River, Indiana.
Despite being commonly associated with lentic and large river habitats (Trautman 1981), the
centrarchid species in this study were indicative of fragmentation in headwater (first and second
order) systems. An analysis of dam purposes in the upper Midwest, found that a majority of
dams on headwaters, creeks, and small rivers are primarily used for recreational purposes
(Chapter 1). These headwater recreational dams could be playing a role in this observed pattern,
particularly if maintained (and potentially stocked) for warm water sport fishes such as
largemouth bass, bluegill, and other centrarchids. In addition, another species, hornyhead chub,
exhibited an overwhelmingly positive response to fragmentation across the size and thermal
classes in which it occurred. This is consistent with the findings of McLaughlin et al. (2006),
who found hornyhead chubs to be abundant above dams when compared to undammed reference
streams. Overall, these results suggest that proximity to downstream dams, and the
impoundments they form, are playing a role in the increased abundance of certain species,
including those typically associated with warm, lentic or large river environments.

79

While many previous studies have looked at the influence of individual dams on fish
assemblages by sampling above and below a dam or by comparing dammed streams to
undammed reference streams, few have explored the cumulative effects of dams. The results of
this study demonstrate the importance of network-based cumulative dam measures, which aligns
with other studies that have found cumulative dam effects on fishes in both upstream and/or
downstream directions (Cumming 2004, Slawski et al. 2008, Wang et al. 2011a; however see
Santucci et al. 2005). For instance, Cumming (2004) found that fish species richness in
headwater streams of Wisconsin was reduced significantly with an increasing number of
downstream dams, however other factors related to water volume and temperature were more
instrumental in the observed pattern. Slawksi et al. (2011) found a cumulative effect of dams in
the Des Plaines River along Illinois/Wisconsin border, as an accumulation of mainstem dams
resulted in decreased in fish diversity. Lastly, Wang et al. (2011a) found that a combination of
distance-to-dam measures and cumulative measures (e.g. upstream and downstream dam density)
were influential for numerous fish indicators, including biotic integrity and habitat/social
preference metrics. These studies, along with the current findings, underscore the importance of
considering the cumulative effects of dams on fish assemblages in addition to the influence of
individual dams, such as connectivity loss.
Results of the variance partitioning analysis showed that natural variables tended to have
a greater influence than either dams or other anthropogenic disturbances in all size/thermal
classes. This result is not unexpected given the role of natural variables, such as drainage area,
stream flow, and temperature regime, in the distribution of fishes within the study region (Lyons
et al. 1996, Zorn et al. 2002, Wehrly et al. 2003). Overall, dam influences uniquely contributed
between 10-19% of variation explained in sensitive species, which compares closely to variance

80

explained by dam measures in Wang et al. (2011a), at 16% for biotic integrity and 19% habitat
and social preference fish indicators, respectively. Although this level might appear to be
modest, it is important to view these results in regional context given the multi-scale influences
on fish assemblages, including large-scale zoogeographic factors (e.g. historic connectivity,
climate, etc.; Wang et al. 2011a). Also, outside of the unique variation attributed to dams, there
can be additional interaction effects between dams and other covariates. While the relative
influence of dams was comparable between studies, the overall amount of variation explained
was much higher for Wang et al. (2011a). There are two likely reasons for this difference. First,
a substantial amount of variation was likely accounted for a priori in the current study through
the size and thermal class stratification employed, whereas Wang et al. (2011a) conducted the
analysis across streams of all sizes and thermal regimes in Michigan and Wisconsin. Secondly,
Wang et al. (2011a) integrated more overall explanatory variables (23 vs. 16), including reachlevel flow and temperature estimates that were unavailable across the current study region. In
comparing the relative influences of dams and other anthropogenic factors across thermal
classes, dam effects tended to be more prevalent in coldwater streams than warmwater streams.
A comparison of non-dam anthropogenic variables between size classes (e.g. MS cold vs. MS
warm) shows similar levels (Table C1). As levels of non-dam anthropogenic variables are not
appreciably higher in warmwater size classes, these results provide additional support to the
importance of thermal influence of dams in coldwater streams. The finding that dams can be a
greater influence on fish assemblages than other anthropogenic disturbances in certain cases was
shown by Slawski et al. (2008), who found dams had a greater influence on fish composition
than urbanization in dammed tributaries of the Des Plaines River. Lastly, variance partitioning
showed a transition in upstream vs. downstream influences, with upstream dam measures

81

dominating in the headwaters moving to a slight downstream-oriented majority in mid-size
streams in both coldwater and warmwater systems. No other studies have explored the relative
landscape-scale effects of upstream vs. downstream dam influences within a region of this size.
These results demonstrate the importance of accounting for both upstream and downstreamoriented dam influences, and that other factors, such as the spatial position within the stream
network and thermal regime, can play important role in the relative influence of dams at a
landscape scale.
Potential improvements
The spatial measures presented in this study aim to broadly represent dam effects across
large regions, however they fail to characterize temporal variability in dam effects. For instance,
while the cumulative upstream reservoir storage metric used in this study can provide a coarse
indication of overall flow alteration, it does not capture within-year flow variability that occurs
due to the storage and managed release of water from reservoirs. In addition, measures that
incorporate the influence of dams at other spatial scales, such as the size, age, and location
within the stream network of river habitat patches formed by the locations of dams (described in
Chapter 1), could elucidate additional dam influences not accounted for in the current study.
Also, individual fish species catch per unit effort, the response variable used in this study, can
undergo a high degree of seasonal variability in studies of the impacts of dams on fishes (Gillette
et al. 2005, Buckmeier et al. 2013). Additional analyses accounting for seasonal variability in
fish responses may yield insight into the influence of dams on specific life history traits. In
addition, certain species may be responding to dams in other ways not adequately measured by
their abundance. For instance, numerous studies have documented the genetic influences of river
fragmentation by dams, resulting in intrapopulation homogeneity and interpopulation

82

heterogeneity (e.g., Yamomoto et al. 2004, Heggenes and Roed 2006). Ultimately, accounting
for the full effect of dams on stream fishes would likely involve both the use of additional types
of data (e.g. patches, seasonal sampling) and multiple response measures in addition to those
used in the current study.
Conclusion
While there have been a number of studies analyzing the effects of dams on fishes, few
have incorporated cumulative measures, particularly across large geographic areas. Landscapescale analyses of the effect of dams on fishes can complement smaller scale field studies by
determining if results extend beyond the particular stream system studied to a larger region, or
identify patterns of dam influence occurring at larger, regional scales. The findings in this study
underscore the importance of accounting for not only the downstream effect of upstream dams,
but also the upstream effect of downstream dams. In addition, a mixture of metrics capturing
both individual (proximity-based) and cumulative influences will be required in order to gain the
full breadth of the effects of dams on fishes.

83

APPENDICES

84

APPENDIX A

TABLES
Table 2.1.— Source datasets for the natural and non-dam anthropogenic variables summarized for network catchments of stream reaches.
Variable
Description
Units
Source
Dataset
Scale/Resolution Currentness
1
National Hydrography
2
AREA
Catchment area
km
1:100,000
2006
EPA/USGS
Dataset Plus v. 1
1
National Hydrography
PPT
4 km
1960-1990
Mean annual precipitation mm
EPA/USGS
Dataset Plus v. 1
Mean annual air
National Hydrography
MAAT
degrees celsius EPA/USGS
4 km
1960-1990
temperature
Dataset Plus v. 1
Base-flow Index Grid for
1
Baseflow index
%
USGS
2003
the Conterminuous United 1 km
BFIC
States
National Elevation
ELEV
Mean catchment elevation masl
EPA/USGS
30m
2006
Dataset
Derived from National
1
SLP
Catchment slope
degrees
EPA/USGS
30m
2006
Elevation Dataset
Soils Data for the
SOIL
Soil permeability
inches/hour
USGS
1:250,000
1995
Conterminuous United
States
Fine-textured surficial
FINE
%
USGS
Surficial Lithography
1 km
2004
lithography
1
Coarse-textured surficial
CRSE
%
USGS
Surficial Lithography
1 km
2004
lithography

85

Table 2.1 (cont'd).
Variable

Description

Units

Source

Developed low intensity,
developed medium intensity, %
and developed high intensity

MRLC

AG

Pasture/hay and cultivated
crops

%

MRLC

FRST

Deciduous forest, evergreen
%
forest, and mixed forest

MRLC

IMP

Imperviousness

%

MRLC

Population density

#/km

Road crossing density,

#/km

Road length density

km/km

2

URB
2

POP

2

RDC

2

2

RDL
1
2
a

Dataset

Scale/Resolution

Currentness

2

National Land Cover
Database

30 m

2001

a

National Land Cover
Database

30 m

2001

a

National Land Cover
Database

30 m

2001

a

National Land Cover
Database

30 m

2001

NOAA

2

a

U.S. Population 2000

1 km

2000

1:100,000

2000

1:100,000

2000

US Census
2

US Census

Census 2000 TIGER
Roads
Census 2000 TIGER
Roads

Variable included in natural variable grouping for CCA analysis.
Variable included in non-dam anthropogenic variable grouping for CCA analysis.
Multi-Resolution Land Characteristics Consortium

86

Table 2.2.—Number of sites for each ecoregion, stream size, and
temperature class grouping. Numbers shown in bold indicate groupings
used in the univariate threshold and correlation analyses. For the
multivariate CCA analysis, sites used in the univariate analyses were
grouped across ecoregions within a given size/temperature stratum
indicated by the CCA Total. A dash (-) indicates that there were no sites
for a given stratum or grouping.
Ecoregion
EWL
LGL
UM
Total
CCA Total

HW
Cold
Warm
68
19
348
70
582
89
998
178
998
159

Size/Thermal Class
MS
Cold
Warm
7
95
200
102
255
122
462
319
455
319

87

Cold
1
2
3
-

LS

Warm
34
50
67
151
-

Table 2.3.—Number of species associations by strata. Overall association of species with
greater levels of fragmentation are summarized as positive, negative, or mixed (See Results for
definitions).
Size/Thermal Class Ecoregion Sites
Species Assoc. Positive Negative Mixed
EWL
68
13
8
3
4
1
HW Cold
LGL
348
29
19
10
7
2
UM
582
40
23
12
3
8
LGL
70
11
2
1
1
0
HW Warm
UM
89
12
4
3
1
0
LGL
200
27
17
12
4
1
MS Cold
UM
255
32
17
12
2
3
EWL
95
29
19
12
7
0
MS Warm
LGL
102
24
13
7
6
0
UM
122
36
19
12
6
1
EWL
34
11
5
5
0
0
LS Warm
LGL
50
10
6
2
1
3
UM
67
27
18
2
13
3

88

Table 2.4.—Significant change point threshold (t) and Spearman correlation analysis (s) results characterizing relationships between fish
species catch per unit effort and selected dam metrics (Table 1.1).  For Spearman correlation, significance was determined by p<0.05;
see Methods for criteria used to determine significance of threshold results.  Overall association of species with greater levels of
fragmentation are summarized as positive (P), negative (N), or mixed (M).  Asterisks (*) indicate an insufficient number of species
occurrences to perform analyses for a given metric. Results shown are for the cold headwater strata in the EWL, LGL, and UM
ecoregions.
Fragmentation Metric
Ecoregion Family Genus and Species
Common Name
Assoc. DM2D DMD TMO UMD UNDR USR Count
Catostomidae
Catostomus commersonii white sucker
N
+s
-t
+t
3
Cyprinidae
-s
+s
+s
+s
4
Luxilus cornutus
common shiner
P
*
Margariscus margarita pearl dace
N
-t
+s
2
*
Notropis heterolepis
blacknose shiner
P
+s
+s
+s
3
*
Rhinichthys atratulus
eastern blacknose dace
P
+s
+s
+s
3
EWL
*
Percidae
Etheostoma nigrum
johnny darter
M
-t
+t
+s
+s
+s
5
*
Percina maculata
blackside darter
N
-t
+t
2
*
Umbridae
Umbra limi
central mudminnow
N
+t
-s
+t
3
Met. Cnt 2
6
5
4
4
4
Catostomidae
Catostomus commersonii white sucker
P
+t
-t
+s
+s
4
Centrarchidae
-s
+s
+t
+t
4
Ambloplites rupestris
rock bass
P
*
Lepomis cyanellus
green sunfish
P
-t
-t
+s
+s
4
LGL
Lepomis macrochirus
bluegill
P
-t
+s
+s
+s
4
*
Micropterus salmoides
largemouth bass
P
-s
+t
-t
+s
+s
5
Cottidae
Cottus bairdii
mottled sculpin
M
+t
-s
2

89

Table 2.4 (cont'd).
Ecoregion Family Genus and Species
Cyprinidae
Luxilus cornutus
Margariscus margarita
Nocomis biguttatus
Phoxinus eos
Phoxinus neogaeus
Rhinichthys cataractae
Semotilus atromaculatus
Gasterosteidae
LGL
Culaea inconstans
Ictaluridae
Ameiurus melas
Salmonidae
Oncorhynchus mykiss
Salmo trutta
Salvelinus fontinalis
Umbridae
Umbra limi

UM

Catostomidae
Catostomus commersonii
Centrarchidae
Ambloplites rupestris
Lepomis macrochirus
Micropterus salmoides
Cottidae
Cottus bairdii

Common Name

Fragmentation Metric
Assoc. DM2D DMD TMO UMD UNDR USR Count

common shiner
pearl dace
hornyhead chub
northern redbelly dace
finescale dace
longnose dace
creek chub

P
N
P
N
N
N
P

brook stickleback

+s

-t
-s

-t
+t
-t
+t
+t
+t

N

-t

+s

black bullhead

P

+t

-t

2

rainbow trout
brown trout
brook trout

N
P
N

-t
+t
-s

+t
-t
+t

2
4
2

+s
12

-t
17
+t

central mudminnow

+t
+t
*
+t
*
+t

*

M
+s
Met. Cnt 7

+s

white sucker

M

+t

rock bass
bluegill
largemouth bass

P
P
P

+t
+s
+t

mottled sculpin

N

90

+t

-t

+s

+s

+s

+t

-s

-s

-s

+s

7

+s

9

8

5

3
2

+t
+t
+s
+t

+s

4
2
4
2
2
3
2

+t
+t
+s

+t
+t
+s

4
4
4
3

Table 2.4 (cont'd).
Ecoregion Family Genus and Species
Cyprinidae
Campostoma anomalum
Clinostomus elongatus
Hybognathus hankinsoni
Luxilus cornutus
Margariscus margarita
Nocomis biguttatus
Phoxinus eos
Phoxinus neogaeus
Pimephales promelas
Semotilus atromaculatus
Gadidae
UM
Lota lota
Gasterosteidae
Culaea inconstans
Ictaluridae
Ameiurus melas
Noturus gyrinus
Percidae
Etheostoma exile
Perca flavescens
Percina caprodes
Percina maculata

Common Name

Fragmentation Metric
Assoc. DM2D DMD TMO UMD UNDR USR Count

central stoneroller
redside dace
brassy minnow
common shiner
pearl dace
hornyhead chub
northern redbelly dace
finescale dace
fathead minnow
creek chub

N
P
M
M
M
P
M
M
P
M

+s
-t
+s
+s
+s
-t
+t
+t

+s

+t

-t
+t
+t
+t
+t
+t
+t
+t
+t
+t

burbot

M

+t

+t

+t

brook stickleback

N

+t

+t

black bullhead
tadpole madtom

P
P

-t

+t
+t

P
P
P
-t
P
Met. Cnt 14

+t
+t
+t
+t
23

Iowa darter
yellow perch
logperch
blackside darter

91

+s
+t
+s
+t
+t

+s
+s

+s
+t

+s

+t

+t

+s

+t
+s

+t
+s

3
2
5
5
3
5
2
3
4
5

+s

+s

5

-s

-s

4

-t

+s

+s
+t

10

+s
+s
+t
+s
15

+s
+s
+t
+s
12

+s
+t
8

2
5
3
4
5
3

Table 2.5.—Significant change point threshold (t) and Spearman correlation analysis (s) results characterizing relationships between fish
species catch per unit effort and selected dam metrics (Table 1.1).  For Spearman correlation, significance was determined by p<0.05;
see Methods for criteria used to determine significance of threshold results.  Overall association of species with greater levels of
fragmentation are summarized as positive (P), negative (N), or mixed (M).  Asterisks (*) indicate an insufficient number of species
occurrences to perform analyses for a given metric. Results shown are for the warm headwater strata in the LGL and UM ecoregions.
Fragmentation Metric
Ecoregion Family Genus and Species
Common Name
Assoc. DM2D DMD TMO UMD UNDR USR Count
Salmonidae
Salvelinus fontinalis
brook trout
P
+t
-t
2
LGL Umbridae
Umbra limi
central mudminnow
N
-s
-s
2
Met. Cnt 0
1
1
0
1
1
Centrarchidae
Lepomis macrochirus
bluegill
P
+s
+t
+t
3
Micropterus salmoides
largemouth bass
P
-s
-t
+s
+t
+t
5
Cyprinidae
UM
Nocomis biguttatus
hornyhead chub
P
+s
+s
+s
3
Gasterosteidae
Culaea inconstans
brook stickleback
N
+t
+t
2
Met. Cnt 2
1
2
2
3
3

92

Table 2.6.—Significant change point threshold (t) and Spearman correlation analysis (s) results characterizing relationships between fish
species catch per unit effort and selected dam metrics (Table 1.1).  For Spearman correlation, significance was determined by p<0.05;
see Methods for criteria used to determine significance of threshold results.  Overall association of species with greater levels of
fragmentation are summarized as positive (P), negative (N), or mixed (M).  Asterisks (*) indicate an insufficient number of species
occurrences to perform analyses for a given metric. Results shown are for the cold mid-size strata in the LGL and UM ecoregions.
Fragmentation Metric
Common Name
Assoc. DM2D DMD TMO UMD UNDR USR Count
Ecoregion Family Genus and Species
Catostomidae
Catostomus commersonii
white sucker
P
+t
-t
+s
3
Hypentelium nigricans
northern hog sucker
P
-t
+t
2
Centrarchidae
Ambloplites rupestris
rock bass
P
-t
+t
+s
+t
4
Lepomis cyanellus
green sunfish
P
+t
-t
+s
+t
+s
5
Lepomis macrochirus
bluegill
P
+s
-t
+s
+s
+t
5
*
Micropterus dolomieu
smallmouth bass
P
+s
+s
2
*
Micropterus salmoides
largemouth bass
P
+s
-s
+t
+t
4
LGL Cyprinidae
Nocomis biguttatus
hornyhead chub
P
-t
+t
+t
+t
4
Rhinichthys cataractae
longnose dace
N
-t
+t
2
Gadidae
Lota lota
burbot
N
*
-s
+t
2
Gasterosteidae
Culaea inconstans
brook stickleback
M
+t
+t
2
Percidae
Etheostoma nigrum
johnny darter
P
-s
+s
2
Perca flavescens
yellow perch
P
+s
+s
2

93

Table 2.6 (cont'd).
Ecoregion Family Genus and Species
Salmonidae
Oncorhynchus mykiss
Salmo trutta
LGL
Salvelinus fontinalis
Umbridae
Umbra limi

UM

Catostomidae
Moxostoma macrolepidotum
Centrarchidae
Ambloplites rupestris
Lepomis gibbosus
Lepomis macrochirus
Micropterus dolomieu
Micropterus salmoides
Cottidae
Cottus bairdii
Cyprinidae
Cyprinella spiloptera
Hybopsis dorsalis
Nocomis biguttatus
Rhinichthys cataractae
Semotilus atromaculatus
Esocidae
Esox lucius

Fragmentation Metric
Assoc. DM2D DMD TMO UMD UNDR USR Count

Common Name
rainbow trout
brown trout
brook trout

N
P
N

-t
+t

+t
-t

central mudminnow

P
Met. Cnt

+t
11

-s
13

shorthead redhorse

P

rock bass
pumpkinseed
bluegill
smallmouth bass
largemouth bass

P
P
P
P
P

mottled sculpin

M

spotfin shiner
bigmouth shiner
hornyhead chub
longnose dace
creek chub

P
P
P
M
N

northern pike

P

94

1

+s

+t
-s

2

*

+t

-s
-s

-t

+t
+t

+t
+t
+t
+t

+t

+t
-t

+t

2

+s
+t
+t
+t

+s
+t
+t
+t
+t

2
3
4
3
4

+s

4

+s

+t
+t
+t

9

+t

+s
+t

10
+s

+t

6

+t
-t

2
5
2

+t

3
2
2
2
2

+t

4

-s
-s

Table 2.6 (cont'd).
Ecoregion Family Genus and Species
Ictaluridae
Noturus gyrinus
Percidae
UM
Etheostoma nigrum
Perca flavescens
Salmonidae
Salmo trutta

Fragmentation Metric
Assoc. DM2D DMD TMO UMD UNDR USR Count

Common Name
tadpole madtom

P

+t

johnny darter
yellow perch

M
P

+t
+t

brown trout

N
+t
Met. Cnt 5

95

-t
13

+s
-t

3

5

7

2

+s

2
2

10

2

Table 2.7.—Significant change point threshold (t) and Spearman correlation analysis (s) results characterizing relationships between fish
species catch per unit effort and selected dam metrics (Table 1.1).  For Spearman correlation, significance was determined by p<0.05; see
Methods for criteria used to determine significance of threshold results.  Overall association of species with greater levels of
fragmentation are summarized as positive (P), negative (N), or mixed (M).  Asterisks (*) indicate an insufficient number of species
occurrences to perform analyses for a given metric. Results shown are for the warm mid-size strata in the EWL, LGL, and UM
Fragmentation Metric
Ecoregion Family Genus and Species
Common Name
Assoc. DM2D DMD TMO UMD UNDR USR Count
Catostomidae
Catostomus commersonii
white sucker
P
+s
-s
2
Moxostoma anisurum
silver redhorse
P
*
+s
+s
2
Moxostoma erythrurum
golden redhorse
P
*
+t
-s
+t
+t
+t
5
Moxostoma macrolepidotum shorthead redhorse
P
*
+t
+t
2
Centrarchidae
Lepomis macrochirus
bluegill
P
*
+s
+s
2
Cottidae
Cottus bairdii
mottled sculpin
N
*
-t
+t
-t
3
Cyprinidae
EWL
Hybognathus hankinsoni
brassy minnow
N
*
-s
-s
-t
3
Hybopsis dorsalis
bigmouth shiner
P
*
+s
+s
+s
3
Luxilus cornutus
common shiner
P
+s
-s
2
Margariscus margarita
pearl dace
N
*
-s
-t
-t
3
Nocomis biguttatus
hornyhead chub
P
+t
-t
+t
+s
4
Phoxinus eos
northern redbelly dace
N
*
-s
-t
-t
3
Pimephales promelas
fathead minnow
P
+t
-t
2
Rhinichthys atratulus
eastern blacknose dace
P
+t
-s
2
Semotilus atromaculatus
creek chub
P
+t
-s
2
Gasterosteidae
Culaea inconstans
brook stickleback
N
-s
-s
2

96

Table 2.7 (cont'd).
Ecoregion Family Genus and Species
Percidae
Perca flavescens
Percina maculata
EWL
Umbridae
Umbra limi

LGL

Centrarchidae
Lepomis cyanellus
Lepomis gibbosus
Lepomis macrochirus
Micropterus salmoides
Cottidae
Cottus bairdii
Cyprinidae
Nocomis biguttatus
Pimephales notatus
Rhinichthys cataractae
Semotilus atromaculatus
Gadidae
Lota lota
Ictaluridae
Noturus flavus
Percidae
Etheostoma nigrum

Fragmentation Metric
Assoc. DM2D DMD TMO UMD UNDR USR Count

Common Name
yellow perch
blackside darter

P
N

central mudminnow

+s
-t

-s
+t

12

green sunfish
pumpkinseed
bluegill
largemouth bass

P
P
P
P

-t
-t
-t
-t

P
P
N
N
N

stonecat

P

johnny darter

N

-s
12

3

+s
+t
+s
+t

+t
+t
+t

+t

4
3
2
5

-t

-s

3

-t

2
2
4
2

-s

burbot

-s
12

N

hornyhead chub
bluntnose minnow
longnose dace
creek chub

10

-s
5

-t

mottled sculpin

4
2

-t

0

+s

-t

N
Met. Cnt

+s

97

+t

+s
+s
-s

+s
-s

+t

+t

-t

-t

*

-t

+t

+t

-t

-t

+s

4
+t

3
2

Table 2.7 (cont'd).
Ecoregion Family Genus and Species
Umbridae
LGL
Umbra limi

UM

Centrarchidae
Lepomis gibbosus
Lepomis macrochirus
Micropterus dolomieu
Micropterus salmoides
Pomoxis nigromaculatus
Cyprinidae
Cyprinella spiloptera
Hybognathus hankinsoni
Nocomis biguttatus
Pimephales notatus
Rhinichthys atratulus
Rhinichthys cataractae
Gadidae
Lota lota
Ictaluridae
Ameiurus melas
Ameiurus natalis
Noturus gyrinus

Fragmentation Metric
Assoc. DM2D DMD TMO UMD UNDR USR Count

Common Name
central mudminnow

N
Met. Cnt

7

pumpkinseed
bluegill
smallmouth bass
largemouth bass
black crappie

P
P
N
P
P

spotfin shiner
brassy minnow
hornyhead chub
bluntnose minnow
eastern blacknose dace
longnose dace

P
N
P
P
M
N

+t

burbot

N

+t

black bullhead
yellow bullhead
tadpole madtom

P
P
P

3

98

-s
-t

-t

-t
-t

+s
6
-t
-t
+t
-t
-s

-t
-t
-s

+s

-s
11

+t
-t
+t

+t
-t

+s

+s

+t

-t

+t

+t

-s
-s

-s

3

+t

+s

+s
+s

+t
+s

2
6
2
5
4

+s
-s
+s
+t
-s

+t
-s
+s
+t
-s
-s

4
3
5
5
3
4

-t
+s
+t
+s

7

-t
5

3

+s

+s
+t
+t

2
4
3

Table 2.7 (cont'd).
Ecoregion Family Genus and Species
Percidae
Etheostoma nigrum
Perca flavescens
UM
Percina caprodes
Percina phoxocephala

Fragmentation Metric
Assoc. DM2D DMD TMO UMD UNDR USR Count

Common Name
johnny darter
yellow perch
logperch
slenderhead darter

N
P
P
N
+t
Met. Cnt 9

99

+t

+t
-t

-s
+t

-t
7

+t
13

-t
14

-t
+t
+s
-t
13

-t
+t
+s
-t
16

4
5
2
6

Table 2.8.—Significant change point threshold (t) and Spearman correlation analysis (s) results characterizing relationships between fish species
catch per unit effort and selected dam metrics (Table 1.1).  For Spearman correlation, significance was determined by p<0.05; see Methods for
criteria used to determine significance of threshold results.  Overall association of species with greater levels of fragmentation are summarized as
positive (P), negative (N), or mixed (M).  Asterisks (*) indicate an insufficient number of species occurrences to perform analyses for a given
metric. Results shown are for the warm large size strata in the EWL, LGL, and UM ecoregions.
Fragmentation Metrics
Eco. Fam. Genus and Species
Common Name
Assoc. DM2D DMD DMO TM2D TMD TMO UM2D UMD UNDR USR Cnt
Catostomidae
Catostomus commersonii
white sucker
P
*
+t
-t
-t
+t
+t
+t 6
golden redhorse
P
Moxostoma erythrurum
*
+t
-s
*
*
*
+t
3
Centrarchidae
Ambloplites rupestris
rock bass
P
*
+s
-t
+s 3
EWL Cyprinidae
Luxilus cornutus
common shiner
P
*
+t
-t
-t
+t
4
Esocidae
Esox lucius
northern pike
P
*
-t
+t
*
+t
3
Mt. Cnt NA
2
2
1
3
3
1
2
3
2
Catostomidae
Catostomus commersonii
white sucker
P
+t
-t
-s
+s 4
Hypentelium nigricans
northern hog sucker P
*
-s
+t
+t
-t 4
Moxostoma erythrurum
golden redhorse
M
*
-s
+t
+t
-t 4
Moxostoma macrolepidotum shorthead redhorse N
+t
+t
+t
-t
+s 5
LGL Cyprinidae
Cyprinus carpio
common carp
*
-t
+t
-s
-s 5
M
+t
Percidae
Percina caprodes
logperch
M
*
+s
+s
*
-t
+t 4
Mt. Cnt 1
4
1
2
1
2
1
5
3
6

100

Table 2.8 (cont'd).

Fragmentation Metrics
Assoc. DM2D DMD DMO TM2D TMD TMO UM2D UMD UNDR USR Cnt

Eco. Fam. Genus and Species
Common Name
Catostomidae
Catostomus commersonii
white sucker
N
Hypentelium nigricans
northern hog sucker N
Moxostoma anisurum
silver redhorse
N
+s
Moxostoma erythrurum
golden redhorse
N
+t
Moxostoma macrolepidotum shorthead redhorse N
Moxostoma valenciennesi greater redhorse
N
Centrarchidae
Ambloplites rupestris
rock bass
N
Micropterus dolomieu
smallmouth bass
M
Micropterus salmoides
largemouth bass
M
Cyprinidae
UM
Cyprinella spiloptera
spotfin shiner
N
+s
Cyprinus carpio
common carp
P
Luxilus cornutus
common shiner
N
Notropis atherinoides
emerald shiner
P
Notropis stramineus
sand shiner
N
Notropis volucellus
mimic shiner
N
Percidae
Etheostoma nigrum
johnny darter
N
Perca flavescens
yellow perch
M
Percina maculata
blackside darter
N
Mt. Cnt 3

101

+t
-t
-t
-t

+t
+s
+t

+t
+t

+t

+t

+t

+s

-t
+t
-t
-t
-t
+t
+t
-t
14

+s

+t
-t
+t
-t
+t
+t
+t

3

12

-t
-t
-t
-t

-s

+t

-t
-t

-t
9

+t

+s
+t
+t

-t

+t
+t

+t

-t
-t

+t
-t
+t
-t

-t
+t
-t

+t
-t
+t

+t

-t
-t

+t
+t
+s
12

+s
7

8

-t
+t
+s
+t
+t

+t

+t
+t

+t
9

-t
-t
-t
-t
-t

5
4
6
8
6
3

-s
+t

4
3
4

-t
-s

4
6
5
5
4
6

+t
-s
11

5
5
5

Table 2.9.—Overall set of sensitive species (highlighted in gray) within the study region with species associations expressed as positive (P), negative
(N), mixed (M), and no association (#) by stratum. Blanks indicate that no analyses were run either due to species absence within a strata or an
insufficient number of species occurrences. Species selected for CCA analysis are shown with numeric superscripts.
HW Cold
HW Warm MS Cold
MS Warm
LS Warm
Family Genus and Species
Common Name
EWL LGL UM LGL UM LGL UM EWL LGL UM EWL LGL UM
Catostomidae
Catostomus commersonii
white sucker
N
P
M
#
#
P
#
P
#
#
P
P
N
Moxostoma macrolepidotum shorthead redhorse
P
P
#
#
N
N
Centrarchidae

Lepomis gibbosus

1,3

rock bass

P

P

P

P

pumpkinseed

Ambloplites rupestris

#

#

#

P

1,2,3,4

bluegill

P

P

P

P

3

smallmouth bass

P

P

4

Lepomis macrochirus

Micropterus dolomieu

Micropterus salmoides
Cottidae
Cottus bairdii
Cyprinidae

1,2,3,4 largemouth bass

4

Nocomis biguttatus
Pimephales notatus

4

4

Margariscus margarita

P

#

#

N

P

P

N

#

M

N

#

#

#

M

N

#

N

M
P

P

#

#

N

M

N

N

#

P

M

longnose dace
creek chub

P

P

hornyhead chub

4

Rhinichthys cataractae
Semotilus atromaculatus

P

P

M

pearl dace

P

P

P

1

P

#

brassy minnow
common shiner

#

P

bluntnose minnow

1,2,3,4

#

P

mottled sculpin

Hybognathus hankinsoni
Luxilus cornutus

P

#

N
#

102

#

P

#

N

#
#

#

M
M

N

#

#

#

P

#

#

P

P

P

P

P

P

#

#

#

P

P

#

N
#
#

N

M

#

N

N

#

#

N

P

N

#

P

N

Table 2.9 (cont'd).
Family Genus and Species
Gadidae
Lota lota
Gasterosteidae

Common Name

4

Ameiurus melas
Percidae

1,2

1

Etheostoma nigrum
Perca flavescens
Umbridae
1
2
3
4

#

2,4

3,4

brook stickleback

#

black bullhead
4

johnny darter

M

N

N

P

burbot

Culaea inconstans
Ictaluridae

Umbra limi

MS Warm
HW Cold
HW Warm MS Cold
LS Warm
EWL LGL UM LGL UM LGL UM EWL LGL UM EWL LGL UM

P

N

#

yellow perch

#
M

#

#

N

M

#

N
N

N

#

N

Species selected as an indicator variable in CCA analysis for the warm headwater class.
Species selected as an indicator variable in CCA analysis for the cold mid-size class.
Species selected as an indicator variable in CCA analysis for the warm mid-size class.

#

#

#
P

P

M

#

N

N

P

#

Species selected as an indicator variable in CCA analysis for the cold headwater class.

103

#

#

P

central mudminnow N

N

P

P

#

P

P

#

N

N

#

#

N
#

M

Table 2.10.—Percentage of variation explained in selected fish response variables by thermal/size class using CCA, subdivided into the
percentage of variation explained attributed to natural, non-dam anthropogenic, fragmentation, and interaction components. Fragmentation
effects are further separated into the relative upstream and downstream variation explained (Up Frag./Down Frag.) and percentage of overall
dam influence (% Up Frag/% Down Frag).
Class
Sites Total Var. Exp. Natural Non-dam Anthro. Fragmentation Interaction Up Frag. Down Frag. % Up Frag. % Down Frag.
HW Cold 998
15.4
42.5
9.3
15.6
32.6
9.9
5.7
63.5
36.5
HW Warm 178
28.2
38.2
29.0
15.2
17.6
9.6
5.6
63.1
36.9
MS Cold 462
18.7
54.9
23.2
19.1
2.8
8.7
10.4
45.7
54.3
MS Warm 319
23.2
27.8
25.7
9.6
36.9
3.9
5.7
40.4
59.6

104

APPENDIX B

FIGURES

Figure 2.1.—Locations of fish survey sites (N = 2,067; A) and dam locations (N = 2,303; B)
within ecoregions of the study region.
105

Figure 2.2.—Percentage of variation explained in selected fish indicators partitioned into natural,
non-dam anthropogenic, fragmentation by dams, and interaction components using canonical
correspondence analysis for cold headwaters (A), warm headwaters (B), cold mid-size (C), and
warm mid-size (D) classes.

106

Figure 2.3.—Plot of canonical correspondence axis I vs. axis II for cold headwaters.
Abbreviations for natural, non-dam anthropogenic, and dam variables are found in Tables 1.1
and 2.1. Species common names are found in Table 2.9.

107

Figure 2.4.—Plot of canonical correspondence axis I vs. axis II for warm headwaters.
Abbreviations for natural, non-dam anthropogenic, and dam variables are found in Tables 1.1
and 2.1. Species common names are found in Table 2.9.

108

Figure 2.5.—Plot of canonical correspondence axis I vs. axis II for cold mid-size streams.
Abbreviations for natural, non-dam anthropogenic, and dam variables are found in Tables 1.1
and 2.1. Species common names are found in Table 2.9.

109

Figure 2.6.—Plot of canonical correspondence axis I vs. axis II for warm mid-size streams.
Abbreviations for natural, non-dam anthropogenic, and dam variables are found in Tables 1.1
and 2.1. Species common names are found in Table 2.9.

110

APPENDIX C

SUPPLEMENTAL TABLES
Table C1.—Descriptive statistics for natural and non-dam
anthropogenic reach network catchment variables (Table
2.1) by stratum.
Stratum
Variable Minimum Maximum Mean
58.8
446.3
2.6
AREA
760.8 654.7
478.5
PPT
3.3
5.1
2.2
MAAT
52.2
69.5
27.5
BFIC
566.5 431.1
305.5
ELEV
5.2
1.2
0.0
SLP
11.0
5.1
1.3
SOIL
14.3
100.0
0.0
FINE
EWL HW Cold
76.0
100.0
0.0
CRSE
4.5
0.3
0.0
URB
59.4
12.4
0.0
AG
95.6
58.3
9.8
FRST
0.2
0.0
3.0
IMP
1.9
0.0
19.2
POP
0.2
0.0
0.6
RDC
8.6
0.0
2.4
RDL
217.9
37.9
1.7
AREA
643.2
997.7 801.8
PPT
5.5
2.1
9.7
MAAT
43.6
87.6
62.6
BFIC
601.0 340.9
193.7
ELEV
1.8
0.0
7.7
SLP
0.9
13.0
6.0
SOIL
14.3
0.0
100.0
FINE
LGL HW Cold
81.1
0.0
100.1
CRSE
1.3
0.0
9.9
URB
59.7
13.4
0.0
AG
55.9
6.7
99.1
FRST
0.7
0.0
5.2
IMP
10.9
0.0
200.5
POP
1.3
0.3
0.0
RDC
1.3
0.0
2.9
RDL

111

Table C1 (cont'd).
Stratum
Variable Minimum Maximum Mean
1.3
247.9
26.0
AREA
615.2
896.4 794.0
PPT
8.3
6.2
3.1
MAAT
30.8
74.9
57.5
BFIC
256.2
533.9 337.4
ELEV
4.4
0.0
12.5
SLP
0.8
12.4
3.4
SOIL
0.0
100.0
1.8
FINE
UM HW Cold
0.0
100.0
57.8
CRSE
0.0
10.0
0.9
URB
36.0
0.0
59.9
AG
10.1
94.2
48.5
FRST
0.0
7.3
0.6
IMP
0.2
198.0
7.6
POP
0.0
2.3
0.4
RDC
0.0
3.4
1.2
RDL
5.5 3254.7
78.7
AREA
687.6
870.6 790.4
PPT
9.0
6.5
2.8
MAAT
35.4
74.1
57.3
BFIC
186.5
495.4 308.2
ELEV
0.0
3.1
1.3
SLP
0.6
10.5
5.1
SOIL
0.0
100.0
19.0
FINE
LGL HW Warm
0.0
100.0
78.8
CRSE
0.0
8.2
1.8
URB
0.0
59.9
34.1
AG
2.8
81.7
35.6
FRST
0.1
3.6
0.9
IMP
0.2
194.2
20.2
POP
0.0
1.0
0.3
RDC
0.5
2.8
1.5
RDL

112

Table C1 (cont'd).
Stratum
Variable Minimum Maximum Mean
4.9
607.1
51.5
AREA
608.6
868.0 805.5
PPT
8.3
5.8
3.4
MAAT
33.7
71.0
56.8
BFIC
246.5
527.2 378.9
ELEV
1.8
0.0
11.3
SLP
0.6
12.4
5.3
SOIL
0.0
100.0
6.8
FINE
UM HW Warm
0.0
100.0
76.7
CRSE
0.0
9.5
1.2
URB
22.0
0.0
58.3
AG
8.4
75.9
45.8
FRST
0.0
4.5
0.7
IMP
0.3
154.7
12.1
POP
0.0
0.8
0.2
RDC
0.4
3.3
1.3
RDL
17.1 2696.7 227.0
AREA
647.4
990.2 806.7
PPT
9.2
5.3
2.1
MAAT
43.8
87.6
64.9
BFIC
205.4
583.0 366.5
ELEV
0.2
6.3
1.9
SLP
1.2
12.5
6.4
SOIL
0.0
99.2
8.5
FINE
LGL MS Cold
0.8
100.0
89.4
CRSE
0.0
7.9
1.3
URB
0.0
59.9
10.0
AG
11.6
97.1
58.6
FRST
0.0
3.8
0.7
IMP
0.0
57.6
7.4
POP
0.0
1.6
0.2
RDC
0.2
3.1
1.3
RDL

113

Table C1 (cont'd).
Stratum
Variable Minimum Maximum Mean
8.9
766.2
96.4
AREA
683.1
847.4 808.5
PPT
8.4
6.7
4.2
MAAT
30.0
73.9
60.3
BFIC
265.8
443.3 323.3
ELEV
5.8
0.1
12.0
SLP
0.8
11.0
3.3
SOIL
1.6
0.0
65.6
FINE
UM MS Cold
0.0
100.0
38.3
CRSE
0.0
8.1
1.2
URB
42.9
0.0
59.9
AG
8.4
89.7
44.6
FRST
0.1
4.1
0.7
IMP
0.7
101.8
9.1
POP
0.0
0.9
0.4
RDC
0.5
2.4
1.3
RDL
51.2 4382.1 558.3
AREA
489.1
750.1 643.7
PPT
4.8
3.3
2.4
MAAT
28.1
65.8
52.3
BFIC
327.2
560.8 408.7
ELEV
0.0
2.9
0.9
SLP
1.4
8.0
4.4
SOIL
0.0
80.4
18.4
FINE
EWL MS Warm
0.7
100.0
63.2
CRSE
0.0
1.2
0.3
URB
0.0
59.9
13.6
AG
15.1
87.4
48.9
FRST
0.0
0.9
0.3
IMP
0.0
24.7
2.2
POP
0.0
0.5
0.1
RDC
0.1
1.6
0.7
RDL

114

Table C1 (cont'd).
Stratum
Variable Minimum Maximum Mean
29.8 1935.6 358.0
AREA
647.0
995.7 782.1
PPT
9.7
5.9
2.7
MAAT
35.7
78.3
56.3
BFIC
202.1
539.2 347.4
ELEV
1.1
0.1
2.5
SLP
1.7
10.3
4.6
SOIL
0.0
100.0
17.3
FINE
LGL MS Warm
0.0
100.0
75.7
CRSE
0.0
9.1
2.0
URB
26.5
0.0
59.2
AG
3.4
93.0
42.8
FRST
0.0
4.4
1.0
IMP
0.2
87.0
15.9
POP
0.0
1.4
0.3
RDC
0.2
2.4
1.3
RDL
21.0 91043.8 1132.7
AREA
600.3
917.9 751.2
PPT
8.7
5.1
3.3
MAAT
32.7
71.6
54.9
BFIC
242.6
531.9 382.8
ELEV
0.2
4.7
1.1
SLP
0.6
11.4
3.8
SOIL
0.0
100.0
2.8
FINE
UM MS Warm
0.0
100.0
87.6
CRSE
0.0
9.5
0.9
URB
0.0
59.8
23.6
AG
1.2
82.3
48.1
FRST
0.0
4.7
0.5
IMP
0.5
93.6
7.4
POP
0.0
0.7
0.2
RDC
0.2
3.3
1.1
RDL

115

Table C1 (cont'd).
Stratum
Variable Minimum Maximum Mean
529.5 48662.7 9996.6
AREA
527.8
713.9 626.2
PPT
4.2
3.4
2.7
MAAT
31.2
56.5
50.0
BFIC
337.9
430.5 392.2
ELEV
0.7
0.1
2.3
SLP
2.7
7.7
4.3
SOIL
16.3
0.7
41.8
FINE
EWL LS Warm
24.8
99.3
59.6
CRSE
0.1
4.5
0.8
URB
21.4
0.1
58.1
AG
22.8
67.7
38.4
FRST
0.1
3.9
0.7
IMP
0.7
2.7
1.6
POP
0.0
0.2
0.1
RDC
0.1
1.2
0.7
RDL
700.2 15755.8 4681.3
AREA
694.8
917.5 784.4
PPT
9.1
6.0
2.9
MAAT
36.9
84.2
61.6
BFIC
230.8
510.5 339.8
ELEV
0.4
2.2
1.1
SLP
2.5
11.2
5.8
SOIL
0.0
51.4
16.8
FINE
LGL LS Warm
36.9
100.0
77.9
CRSE
0.1
9.9
3.0
URB
0.1
60.0
20.7
AG
9.8
86.4
43.7
FRST
0.1
5.0
1.5
IMP
0.8
87.7
21.8
POP
0.1
0.6
0.3
RDC
0.6
2.2
1.5
RDL

116

Table C1 (cont'd).
Stratum
Variable Minimum Maximum Mean
748.8 34020.2 11284.7
AREA
640.1
823.6
745.0
PPT
3.4
7.1
4.7
MAAT
48.5
69.2
58.1
BFIC
335.8
480.1
375.7
ELEV
0.5
8.6
1.3
SLP
1.7
7.4
4.9
SOIL
0.0
14.8
4.0
FINE
UM LS Warm
0.0
99.7
86.1
CRSE
0.4
1.6
0.9
URB
3.5
58.3
18.6
AG
26.0
70.4
52.8
FRST
0.3
0.9
0.5
IMP
2.7
12.7
5.7
POP
0.1
0.5
0.2
RDC
0.9
1.4
1.2
RDL

117

Table C2.—Descriptive statistics for fragmentation metrics (Table
1.1) by stratum.
Stratum
Variable N Minimum Maximum Mean
68
0.0
3.0
0.1
UMCT
68
0.0
8.0
0.4
UMD
4
0.0
8.1
2.0
UM2D
68
0.0
1.0
0.9
UMO
68
0.0
3.0
0.1
UNCT
68
0.0
7.6
0.4
UNDR
68
0.0
3.3
0.1
UNDC
68
0.0
506.5
8.2
USR
68
0.0
146.9
2.4
USC
EWL HW Cold
68
0.0
2.2
0.0
USF
68
0.0
24.0
2.8
DMCT
66
0.0
2.7
0.5
DMD
36
3.7
294.8
82.6
DM2D
68
0.0
1.0
0.6
DMO
68
0.0
27.0
2.9
TMCT
68
0.0
2.9
0.5
TMD
38
15.3
423.8 107.9
TM2D
68
0.0
1.0
0.6
TMO
348
0.0
4.0
0.1
UMCT
348
0.0
49.7
1.0
UMD
25
0.0
16.3
4.5
UM2D
348
0.0
1.0
0.9
UMO
348
0.0
4.0
0.1
UNCT
348
0.0
49.7
1.1
UNDR
348
0.0
20.3
0.4
UNDC
348
0.0 7711.6
26.5
USR
348
0.0 1395.8
5.7
USC
LGL HW Cold
348
0.0
5.1
0.0
USF
348
0.0
16.0
2.7
DMCT
327
0.0
21.5
2.3
DMD
221
0.0
167.1
44.9
DM2D
348
0.0
1.0
0.6
DMO
348
0.0
16.0
2.8
TMCT
348
0.0
17.1
1.9
TMD
226
4.7
191.1
59.2
TM2D
348
0.0
1.0
0.6
TMO

118

Table C2 (cont'd).
Stratum
Variable
UMCT
UMD
UM2D
UMO
UNCT
UNDR
UNDC
USR
USC
UM HW Cold
USF
DMCT
DMD
DM2D
DMO
TMCT
TMD
TM2D
TMO
UMCT
UMD
UM2D
UMO
UNCT
UNDR
UNDC
USR
USC
LGL HW Warm
USF
DMCT
DMD
DM2D
DMO
TMCT
TMD
TM2D
TMO

N Minimum Maximum Mean
582
0.0
2.0
0.1
582
0.0
40.6
0.6
29
0.0
19.6
3.3
582
0.0
1.0
1.0
582
0.0
4.0
0.1
582
0.0
40.6
0.7
582
0.0
34.1
0.5
582
0.0
566.5
3.9
582
0.0
253.8
1.9
582
0.0
1.5
0.0
582
0.0
46.0
28.3
579
0.0
1.4
1.0
578
0.0
278.0
57.9
582
0.0
1.0
0.0
582
0.0
46.0
28.3
582
0.0
1.4
1.0
578
4.1
285.1
69.3
582
0.0
1.0
0.0
70
0.0
5.0
0.2
70
0.0
19.9
0.9
6
0.0
3.7
0.8
70
0.0
1.0
0.9
70
0.0
25.0
0.5
70
0.0
19.9
0.8
70
0.0
8.8
0.3
70
0.0
173.0
7.2
70
0.0
52.1
2.5
70
0.0
20.3
0.3
70
0.0
13.0
6.2
67
0.0
8.6
3.8
58
0.2
155.1
33.4
70
0.0
1.0
0.3
70
0.0
13.0
6.3
70
0.0
7.2
3.4
60
1.7
172.0
47.3
70
0.0
1.0
0.4

119

Table C2 (cont'd).
Stratum
Variable
UMCT
UMD
UM2D
UMO
UNCT
UNDR
UNDC
USR
USC
UM HW Warm
USF
DMCT
DMD
DM2D
DMO
TMCT
TMD
TM2D
TMO
UMCT
UMD
UM2D
UMO
UNCT
UNDR
UNDC
USR
USC
LGL MS Cold
USF
DMCT
DMD
DM2D
DMO
TMCT
TMD
TM2D
TMO

N Minimum Maximum Mean
89
0.0
2.0
0.3
89
0.0
30.1
2.0
18
0.0
9.3
2.7
89
0.0
1.0
0.8
89
0.0
4.0
0.4
89
0.0
30.1
2.2
89
0.0
13.5
1.0
89
0.0 2996.0
69.4
89
0.0 1049.1
24.5
89
0.0
3.8
0.1
89
0.0
46.0
31.7
87
0.4
1.4
1.1
87
0.0
138.2
33.0
89
0.0
1.0
0.0
89
0.0
47.0
32.0
89
0.0
1.4
1.1
87
7.0
157.9
46.1
89
0.0
1.0
0.0
200
0.0
4.0
0.3
200
0.0
11.1
0.7
39
0.0
63.7
12.6
200
0.0
1.0
0.9
200
0.0
9.0
0.7
200
0.0
13.1
0.7
200
0.0
6.7
0.3
200
0.0
475.6
16.6
200
0.0
246.0
8.0
200
0.0
1.1
0.0
200
0.0
14.0
2.0
186
0.0
11.1
1.8
111
0.2
164.5
49.3
200
0.0
1.0
0.7
200
0.0
14.0
2.3
200
0.0
5.7
1.4
123
6.2
231.7
74.9
200
0.0
1.0
0.7

120

Table C2 (cont'd).
Stratum
Variable
UMCT
UMD
UM2D
UMO
UNCT
UNDR
UNDC
USR
USC
UM MS Cold
USF
DMCT
DMD
DM2D
DMO
TMCT
TMD
TM2D
TMO
UMCT
UMD
UM2D
UMO
UNCT
UNDR
UNDC
USR
USC
EWL MS Warm
USF
DMCT
DMD
DM2D
DMO
TMCT
TMD
TM2D
TMO

N Minimum Maximum Mean
255
0.0
4.0
0.2
255
0.0
22.6
1.3
43
0.0
73.0
8.3
255
0.0
1.0
0.9
255
0.0
11.0
0.7
255
0.0
10.1
1.0
255
0.0
9.8
0.8
255
0.0 1330.5
10.5
255
0.0
470.2
5.5
255
0.0
5.2
0.0
255
0.0
40.0
25.2
253
0.0
1.3
0.9
249
0.0
239.8
53.6
255
0.0
1.0
0.0
255
0.0
42.0
25.5
255
0.0
1.3
0.9
249
4.3
243.5
72.4
255
0.0
1.0
0.0
95
0.0
20.0
0.6
95
0.0
7.3
0.4
18
0.0
105.1
38.0
95
0.0
1.0
0.9
95
0.0
38.0
1.5
95
0.0
6.6
0.6
95
0.0
2.9
0.2
95
0.0 1188.0
71.7
95
0.0
296.4
24.7
95
0.0
4.4
0.3
95
0.0
17.0
1.5
95
0.0
2.1
0.3
39
6.5
282.5 100.9
95
0.0
1.0
0.7
95
0.0
27.0
2.0
95
0.0
2.9
0.3
41
23.6
353.5 155.4
95
0.0
1.0
0.7

121

Table C2 (cont'd).
Stratum
Variable
UMCT
UMD
UM2D
UMO
UNCT
UNDR
UNDC
USR
USC
LGL MS Warm
USF
DMCT
DMD
DM2D
DMO
TMCT
TMD
TM2D
TMO
UMCT
UMD
UM2D
UMO
UNCT
UNDR
UNDC
USR
USC
UM MS Warm
USF
DMCT
DMD
DM2D
DMO
TMCT
TMD
TM2D
TMO

N Minimum Maximum Mean
102
0.0
6.0
0.9
102
0.0
24.2
2.4
41
0.0
82.4
13.1
102
0.0
1.0
0.7
102
0.0
15.0
2.0
102
0.0
18.1
1.7
102
0.0
13.4
0.9
102
0.0
691.4
41.5
102
0.0
258.1
19.1
102
0.0
0.9
0.1
102
0.0
12.0
4.2
101
0.0
14.5
3.0
87
0.0
171.2
46.2
102
0.0
1.0
0.4
102
0.0
13.0
5.1
102
0.0
8.1
2.7
88
2.8
187.4
80.4
102
0.0
1.0
0.5
122
0.0
15.0
0.8
122
0.0
20.1
1.6
46
0.0
68.3
12.0
122
0.0
1.0
0.7
122
0.0
293.0
4.1
122
0.0
8.4
1.3
122
0.0
4.8
0.5
122
0.0 1444.1
86.5
122
0.0
479.4
32.0
122
0.0
48.2
0.6
122
10.0
46.0
33.1
122
0.4
1.4
1.1
122
0.4
217.9
61.0
122
0.0
0.1
0.0
122
11.0
47.0
33.9
122
0.4
1.5
1.1
122
11.7
234.1
96.0
122
0.0
0.1
0.0

122

Table C2 (cont'd).
Stratum
Variable
UMCT
UMD
UM2D
UMO
UNCT
UNDR
UNDC
USR
USC
EWL LS Warm
USF
DMCT
DMD
DM2D
DMO
TMCT
TMD
TM2D
TMO
UMCT
UMD
UM2D
UMO
UNCT
UNDR
UNDC
USR
USC
LGL LS Warm
USF
DMCT
DMD
DM2D
DMO
TMCT
TMD
TM2D
TMO

N Minimum Maximum Mean
34
0.0
4.0
1.6
34
0.0
2.3
0.6
25
0.0
149.4
71.2
34
0.0
1.0
0.5
34
2.0
27.0
8.7
34
0.1
1.2
0.4
34
0.0
0.5
0.2
34
1.6 2704.9 391.9
34
0.8
627.0
99.8
34
0.0
5.0
0.9
34
0.0
6.0
1.4
34
0.0
1.0
0.3
17
2.2
214.0 135.6
34
0.0
1.0
0.7
34
0.0
6.0
3.0
34
0.0
1.1
0.5
26
88.0
301.6 202.2
34
0.1
1.0
0.5
50
0.0
15.0
4.6
50
0.0
7.9
2.1
42
0.0
135.4
39.3
50
0.0
1.0
0.3
50
0.0
98.0
23.2
50
0.0
3.0
1.0
50
0.0
1.6
0.5
50
0.0
428.7
71.3
50
0.0
177.7
34.1
50
0.0
0.5
0.1
50
0.0
8.0
2.0
48
0.0
6.9
2.1
28
0.1
125.6
39.2
50
0.0
1.0
0.7
50
0.0
15.0
6.5
50
0.0
4.3
2.1
46
8.2
192.2
87.3
50
0.0
1.0
0.4

123

Table C2 (cont'd).
Stratum
Variable
UMCT
UMD
UM2D
UMO
UNCT
UNDR
UNDC
USR
USC
UM LS Warm
USF
DMCT
DMD
DM2D
DMO
TMCT
TMD
TM2D
TMO

N Minimum Maximum Mean
67
0.0
26.0
5.3
67
0.0
5.5
1.6
57
0.0
231.2
51.1
67
0.0
1.0
0.3
67
3.0
160.0
54.4
67
0.3
1.8
1.1
67
0.1
0.9
0.5
67
1.1
855.4 172.2
67
1.0
248.3
64.0
67
0.0
654.5
14.3
67
21.0
42.0
32.5
67
0.8
1.3
1.1
67
0.0
274.7
70.0
67
0.0
0.1
0.0
67
21.0
47.0
37.7
67
0.8
1.4
1.1
67
2.4
274.9 134.1
67
0.0
0.1
0.0

124

LITERATURE CITED

125

LITERATURE CITED

Abell, R., M. L. Thieme, C. Revenga, M. Bryer, M. Kottelat, N. Bogutskaya, B. Coad, N.
Mandrak, S. Contreras Balderas, W. Bussing, M. L. J. Stiassny, P. Skelton, G. R. Allen,
P. Unmack, A. Naseka, R. Ng, N. Sindorf, J. Robertson, E. Armijo, J. V. Higgins, T. J.
Heibel, E. Wikramanayake, D. Olson, H. L. Lopez, R. E. Reis, J. G. Lundberg, M. H.
Sabaj Perez, and P. Petry. 2008. Freshwater ecoregions of the world: A new map of
biogeographic units for freshwater biodiversity conservation. BioScience 58(5):403:414.
Bain, M. B., and M. L. Wine. 2010. Testing predictions of stream landscape theory for fish
assemblages in highly fragmented watersheds. Folia Zoologica 59(3):231-239.
Baker, M. E., and R. S. King. 2010. A new method for detecting and interpreting biodiversity
and ecological community thresholds. Methods in Ecology and Evolution 1:25-37.
Borcard, D., P. Legendre, and P. Drapeau. 1992. Partialling out the spatial component of
ecological variation. Ecology 73(3):1045-1055.
Bosch, N. S. 2008. The influence of impoundments on riverine nutrient transport: An evaluation
using the Soil and Water Assessment Tool. Journal of Hydrology 355:131-147.
Buckmeier, D. L., N. G. Smith, B. P. Fleming, and K. A. Bodine. 2013. Intra-annual variation
in river-reservoir interface fish assemblages: Implications for fish conservation and
management in regulated rivers. River Research and Applications (early online view).
Catalano, M. J., M. A. Bozek, and T. D. Pellett. 2007. Effects of dam removal on fish
assemblage structure and spatial distributions in the Baraboo River, Wisconsin. North
American Journal of Fisheries Management 27:519-530.
Cumming, G. S. 2004. The impact of low-head dams on fish species richness in Wisconsin,
USA. Ecological Applications 14(5):1495-1506.
Dufrene, M., and P. Legendre. 1997. Species assemblages and indicator species: The need for a
flexible asymmetrical approach. Ecological Monographs 67(3):345-366.
Falke, J. A., and K. B. Gido. 2006. Effects of reservoir connectivity on stream fish assemblages
in the Great Plains. Canadian Journal of Fisheries and Aquatic Sciences 63:480-493.
Farrand, W. R. and D. L. Bell. 1982. Quaternary Geology of Michigan. State of Michigan
Department of Natural Resources, Geology Survey Map.
Fukushima, M., S. Kameyama, M. Kaneko, K. Nakao, and E. A. Steel. 2007. Modelling the
effects of dams on freshwater fish distributions in Hokkaido, Japan. Freshwater Biology
52(8):1511-1524.

126

Gillette, D. P., J. S. Tiemann, D. R. Edds, and M. L. Wildhaber. 2005. Spatiotemporal patterns
of fish assemblage structure in a river impounded by low-head dams. Copeia
2005(3):539-549.
Goldstein, R. M., and M. R. Meador. 2004. Comparisons of fish species traits from small
streams to large rivers. Transactions of the American Fisheries Society 133:971-983.
Guenther, C. B., and A. Spacie. 2006. Changes in fish assemblage structure upstream of
impoundments within the upper Wabash River Basin, Indiana. Transactions of the
American Fisheries Society 135:570-583.
Hall, C. J., A. Jordaan, and M. G. Frisk. 2011. The historic influence of dams on diadromous
fish habitat with a focus on river herring and hydrologic longitudinal connectivity.
Landscape Ecology 26:956-107.
Hayes, D. B., H. Dodd, and J. Lessard. 2008. Conservation considerations for small dams on
cold-water streams. Proceedings of the 4th World Fisheries Congress, May 2-6, 2004,
Vancouver, BC.
Heggenes, J., and K. H. Roed. 2006. Do dams increase genetic diversity in brown trout (Salmo
trutta)? Microgeographic differentiation in a fragmented river. Ecology of Freshwater
Fish 15:366-375.
IBM SPSS 20. 2011. IBM SPSS Statistics for Windows, Version 20.0. Armonk, NY.
Krueger, D, L. Wang, D. Infante, K. Wehrly, J. Stewart, S. Westenbroek, Y. Tsang, and D.
Wieferich. (in prep). How will streams respond to climate changes in the midwestern
US? Development of a neural network modeling framework to assess thermal stream
habitat.
Lessard, J. L., and D. B. Hayes. 2003. Effects of elevated water temperature on fish and
macroinvertebrate communities below small dams. River Research and Applications
19(7):721-732.
Lyons, J. 1996. Patterns in the species composition of fish assemblages among Wisconsin
streams. Environmental Biology of Fishes 45:329-341.
Lyons, J., T. Zorn, J. Stewart, P. Seelbach, K. Wehrly, and L. Wang. 2009. Defining and
characterizing coolwater streams and their fish assemblages in Michigan and Wisconsin,
USA. North American Journal of Fisheries Management 29:1130-1151.
McLaughlin, R. L., L. Porto, D. L. G. Noakes, J. R. Baylis, L. M. Carl, H. R. Dodd, J. D.
Goldstein, D. B. Hayes, and R. G. Randall. 2006. Effects of low-head barriers on stream
fishes: taxonomic affiliations and morphological correlates of sensitive species.
Canadian Journal of Fisheries and Aquatic Sciences 63:766-779.

127

Perkin, J. S. and K. B. Gido. 2011. Stream fragmentation thresholds for a reproductive guild of
Great Plains fishes. Fisheries 36(8):371-383.
Poff, L. 1997. Landscape filters and species traits: towards mechanistic understanding and
prediction in stream ecology. Journal of the North American Benthological Society
16(2):391-409.
Poff, N. L., J. D. Allan, M. B. Bain, J. R. Karr, K. L. Prestegaard, B. D. Ricter, R. E. Sparks, and
J. C. Stromburg. 1997. The natural flow regime. BioScience 47(11):769-784.
Poff, N. L., and D. D. Hart. 2002. How dams vary and why it matters for the emerging science
of dam removal. BioScience 52(8):659-668.
Pringle, C. M. 1997. Exploring how disturbance is transmitted upstream: Going against the
flow. Journal of the North American Benthological Society 16(2):425-438.
Pringle, C. M. 2001. Hydrological connectivity and the management of biological reserves: A
global perspective. Ecological Applications 11(4):981-998.
R 2.15. 2012. R Development Core Team (2012). R: A language and environment for
statistical computing. R Foundation for Statistical Computing, Vienna, Austria.
Santucci Jr., V. J., S. R. Gephard, and S. M. Pescitelli. 2005. Effects of multiple low-head dams
on fish, macroinvertebrates, habitat, and water quality in the Fox River, Illinois. North
American Journal of Fisheries Management 25:975-992.
Seelbach, P. W., M. J. Wiley, J. C. Kotanchik, and M. E. Baker. 1997. A landscape-based
ecological classification for river valley segments in lower Michigan. Michigan
Department of Natural Resources, Fisheries Research Report 2036, Ann Arbor.
Slawski, T. M., F. M. Veraldi, S. M. Pescitelli, and M. J. Pauers. 2008. Effects of tributary
spatial position, urbanization, and multiple low-head dams on warmwater fish community
structure in a Midwestern stream. North American Journal of Fisheries Management
28(4):1020-1035.
Sreekantha, G. K. V., and T. V. Ramachandra. 2008. Nestedness pattern in freshwater fishes of
the Western Ghats: An indication of stream islands along riverscapes. Current Science
95(12):1707-1714.
Stanley, E. H., M. J. Catalano, N. Mercado-Silva, and C. H. Orr. 2007. Effects of dam removal
on brook trout in a Wisconsin stream. River Research and Applications 23:792-798.
Steel, E. A., R. M. Hughes, A. H. Fullerton, S. Schmutz, J. A. Young, M. Fukushima, S. Muhar,
M. Poppe, B. E. Feist, and C. Trautwein. 2010. Are we meeting the challenges of
landscape-scale riverine research? A review. Living Reviews in Landscape Research 4.

128

Straher, A. N. 1957. Quantitative analysis of watershed geomorphology. Transactions of the
American Geophysical Union 38(6):913-920.
ter Braak, C. J. F., and P. Smilauer. 2002. CANOCO reference manual and user’s guide to
Canoco for windows: Software for canonical community ordination (version 4.5).
Microcomputer Power, Ithaca, NY.
Tonn, W. M., J. J. Magnuson, M. Rask, and J. Toivonen. 1990. Intercontinental comparison of
small-lake fish assemblages: The balance between local and regional processes. The
American Naturalist 136(3):345-375.
Trautman, M. B. 1981. The fishes of Ohio. Ohio State University Press. 782 pp.
Tsang, Y-P., D. Wieferich, K. Fang, and D. M. Infante. (in prep). An approach of aggregating
upstream catchment information to support research and management of fluvial systems
across large landscapes.
USEPA and USGS (U.S. Environmental Protection Agency and U.S. Geological Survey). 2005.
National hydrography dataset plus – NHDPlus. Edition 1.0. Available at
http://www.horizon-systems.com/nhdplus/index.php (August 2006).
USGS. 2013. 2012 National Anthropogenic Barrier Dataset. Available at
https://www.sciencebase.gov/catalog/item/get/512cf142e4b0855fde669828. (September
2010).
Utz, R. M., R. H. Hilderbrand, and R. L. Raesly. 2010. Regional differences in patterns of fish
species loss with changing land use. Biological Conservation 143(3):688-699.
Wang, L., J. Lyons, P. Kanehl, and R. Gatti. 1997. Influences of watershed land use on habitat
quality and biotic integrity in Wisconsin streams. Fisheries 22(6):6-12.
Wang, L., P. W. Seelbach, and R. M. Hughes. 2006. Introduction to landscape influences on
stream habitats and biological assemblages. Pages 1-23 in R. M. Hughes, L. Wang, and
P. W. Seelbach, editors. American Fisheries Society, Symposium 48, Bethesda,
Maryland.
Wang, L., T. Brenden, P. Seelbach, A. Cooper, D. Allan, R. Clark Jr., and M. Wiley. 2008.
Landscape based identification of human disturbance gradients and reference conditions
for Michigan streams. Environmental Monitoring and Assessment 141:1-17.
Wang, L., D. Infante, J. Lyons, J. Stewart, and A. Cooper. 2011a. Effects of dams in river
networks on fish assemblages in non-impoundment sections of rivers in Michigan and
Wisconsin, USA. River Research and Applications 27(4):473-487.

129

Wang, L., D. Infante, P. Esselman, A. Cooper, D. Wu, W. Taylor, D. Beard, G. Whelan, and A.
Ostroff. 2011b. A hierarchical spatial framework and database for the national river fish
habitat condition assessment. Fisheries 36(9):436-449.
Ward, J. V., and J. A. Stanford. 1983. The serial discontinuity concept of lotic ecosystems.
Pages 29-42 in T. D. Fontaine III and S. M. Bartell, editors. The ecology of regulated
rivers. Plenum Press, New York.
Wehrly, K. E., M. J. Wiley, and P. W. Seelbach. 2003. Classifying regional variation in thermal
regime based on stream fish community patterns. Transactions of the American Fisheries
Society 132:18-38.
Yamamoto, S, K., Morita, I. Koizumi, and K. Maekawa. 2004. Genetic differentiation of whitespotted charr (Salvelinus leucomaenis) populations after habitat fragmentation: Spatialtemporal changes in gene frequencies. Conservation Genetics 5:529-538.
Zorn, T. G, P. W. Seelbach, and M. J. Wiley. 2002. Distributions of stream fishes and their
relationship to stream size and hydrology in Michigan’s lower peninsula. Transactions of
the American Fisheries Society 131:70-85.

130