REGIONAL INFLUENCE OF LANDSCAPE FEATURES AND PROCESSES ON FLUVIAL
FISH ASSEMBLAGES
By
Darren Jay Thornbrugh

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Fisheries and Wildlife - Doctor of Philosophy
2014

ABSTRACT
REGIONAL INFLUENCE OF LANDSCAPE FEATURES AND PROCESSES ON
FLUVIAL FISH ASSEMBLAGES
By
Darren Jay Thornbrugh
Habitat fragmentation, degradation and loss are dominant reasons for global declines in
biodiversity of fishes in stream systems, and humans have drastically modified landscapes
drained by streams due to activities including urbanization and agriculture. Such human land
uses are known to change stream habitats through inputs of excess nutrients, sediments, or toxics
and through changes in stream flow and thermal regimes, and human land uses have been shown
in many studies to negatively affect stream habitats and the fishes they support. Despite this
understanding, degradation of stream habitats and fishes continues globally, and freshwater
fishes remain one of the most threatened groups of organisms on the planet. Less understood are
the specific mechanisms by which land uses affect stream habitats and how these can vary by
region, and how additional landscape-scale characteristics may alter effects of human land uses,
resulting in regionally-specific responses in stream fishes to stressors. Such differences across
regions may render one locale more sensitive to biodiversity loss or fish assemblage change from
the same magnitude of anthropogenic disturbance in the landscape and confound efforts to
develop and apply specific actions to conserve biodiversity of stream fishes. The goal of this
study is to help address these limitations in understanding. In these chapters, I characterize
important natural landscape factors and human land uses influencing distributions and
abundances of stream fishes across large regions in both Michigan and within five freshwater
ecoregions in the eastern portion of the United States. The results show major regional patterns
of natural landscape factors and human stressors that affect fluvial fishes and how these factors

vary in influence across regions. This more in-depth understanding of landscape influences on
fluvial fish assemblages will allow managers to better account for this regional variability when
working to protect and conserve freshwater fisheries and biodiversity of fluvial fish assemblages
from both current and future threats.

Copyright by
DARREN JAY THORNBRUGH
2014

I dedicate my dissertation to my mother and father. I want to thank you for your endless support,
in life and throughout my education. You have always helped me to feel close to home while
living far away.

v

ACKNOWLEDGEMENTS

This dissertation was completed with the financial, intellectual, and emotional
contribution of many people of the last five and a half years. The following is my humble
attempt to express the appreciation I feel for all of the support I have received to complete this
body of work and helped me grow as a person.
I am grateful to live in a country that invests public funds in science. This research was
conducted with support from the USGS National Aquatic GAP Program, FWS, National Fish
Habitat Partnership, Michigan Department of Natural Resources, and grants from Michigan State
University, Dr. Howard A. Tanner Fisheries Excellence Fellowship, Robert C. and Betty A. Ball
Department of Fisheries and Wildlife Fellowship; including the College of Agriculture and
Natural Resources, Bridge Fellowship, Summer Doctoral Recruitment Fellowship, Dissertation
Completion Fellowship, and other scholarships, the Red Cedar Fly Fishers Graduate Fellowship
in Fisheries Management and the Austrian Marshall Plan Foundation Scholarship. These funds
supported me to complete my education, research, study abroad and attend conferences.
My major advisor and committee chair Dana Infante, I am happy to call her my boss,
supervisor, mentor, colleague, personal friend and family, without your constant support,
guidance, your kind and caring nature and your friendship this dissertation would not have been
possible. I would also like to thanks my graduate guidance committee members, Dan Hayes,
Gary Roloff, R. Jan Stevenson, and Lizhu Wang, who provided guidance in the process of
completing this dissertation. I am thankful for their time, technical help, and mentoring.

vi

I want to thank the support staff from the Department of Fisheries and Wildlife especially
Marcia Baar, Jill Cruth, Julie Traver and Sharon Reasoner for all that they you do to make travel,
field work and graduate school easier to navigate.
I would like to thank recruitment fellowship that allowed me to come early to start my
education in Michigan in the summer of 2009. It allowed me an invaluable opportunity to be
part of the “dream team” a summer stream sampling field crew over the summer of 2009 that
gave me a first-hand account of the variety of streams in Michigan. This was one of my best
field seasons and my best summer in Michigan. It also allowed me the ability to get to know to
of my best mates throughout all of my time at MSU, Ralph Tingley and Jared Ross a.k.a. “kid
poncho”. Their friendship is rare to find in life and has helped me complete this dissertation,
thanks guys.
I would like to thank my 2010 field crew Molly Good and Ryan Young that endured the
long days on the road, camping, and in the field. Your positive attitudes and hard work made the
field season go very smooth. Thanks also for your help with data entry. This was a really great
field season.
I want to say a special thanks to the two labs that I have been part over MIRTH and ALE
labs, for graduated school is over 50% about the cohort of students that you go through the
process with, that you learn just as much from as you do from your classes. The sharing of
knowledge between your fellow graduate students maybe the biggest teacher, thank you all for
your help along the way.
My fellow grad students and friends provided camaraderie, advice, encouragement and
liveliness. Grad school would have been a lonelier place without Ralph Tingley III, Jared Ross,

vii

Dana Infante, Daniel Wieferich, Arthur Cooper, Yin-Phan Tsang, Wesley Daniel, Kyle
Herreman, Marta Jarzynam, Jim Stagge, Ben Schmitt, Brenda Pracheil, Andrea Bowling, Tucker
Wilkinson, Eric MacMillan, Carson Pritchard, Kyle Molton, Shikha Singh, Kiira Sitarii, Kara
Stevens, Abigail Lynch, Chiara Zuccarino-Crowe, Clint and Tammey Otto, and Corrine Higley.
I would like to thank the Austrian Marshall Plan Foundation for their Scholarship
program that allowed me the best opportunity of my whole graduate experience. I am especially
grateful to the Austrian Marshall Plan Foundation who provided the support for this exchange,
research, work and provided a unique opportunity to exchange ideas and build future
collaboration between the United States and Austria. Their support allowed me the ability to
study at the Institute of Hydrobiology and Aquatic Ecosystem Management at the University of
Natural Resources and Applied Life Sciences and live in Vienna Austria for a semester.
I would like to express my deepest gratitude to Dr. Rafaela Schinegger and Dr. Andreas
Melcher for supervising my work, sharing your professional expertise, knowledge and pleasant
collaboration, it was a great and unique opportunity to learn and grow as a person. Mostly I
would like to thank Rafaela and Andreas for your friendship, your mentorship, your guidance at
BOKU, your inviting nature and openness to include me into your lives and families, your
kindness, your patience to help me with any problems or issues I had, and just being great friends
and office mates. I truly thank you for your continued friendship and look forward to finding
ways to collaborate on projects in the future with you two.
I’m very grateful to Dr. Stefan Schmutz for showing interest in my research proposal,
offering to supervise and mentor my research during my exchange at BOKU and making
available office space for me at the Institute of Hydrobiology and Aquatic Ecosystem

viii

Management during my exchange at BOKU and stay in Vienna. I’m also grateful for the
opportunities that you provided to me. Lastly, thank you for your friendship, your mentorship
and your openness to invite me into you home and family.
I’m also very thankful to the Institute of Hydrobiology and Aquatic Ecosystem
Management at BOKU and all the wonderful people that work there, who welcomed me into
your research family, your lives, became my friends, supported me, and helped me with
everything from daily life stuff to technical issues; it was truly wonderful to have met each of
you.
Thanks to the Center for International Relations at BOKU, all of the international
students studying at BOKU and the European Erasmus Program for making this exchange
experience so international. I met students from over 20 nations and got to experience, exchange
ideas and points of view on many different topics and issues. This exchange experience has
broadened my global prospective of the world around me, the extent that the European Union
(EU) utilizes natural resources and how conservation and restoration efforts are conducted in the
EU. It has also enlightened me on the politics, issues and funding framework established in the
EU associated countries and surrounding countries to dispense funds and tackle large
multinational projects of conservation and restoration efforts. By studying at the Institute of
Hydrobiology and Aquatic Ecosystem Management it has exposed me to and educated me on the
pressing aquatic ecosystem conservation issues in Austria but also in Europe and as far stretching
as the countries of Burkina Faso, Ethiopia in Africa, Iran in the Middle East and Thailand in
Southeast Asia. This has emphasized and reminded me that most issues of conservation interest
related to the degradation of aquatic ecosystems are global in nature and are prevalent in almost
all nations.
ix

Finally, I would like to thank Juliana Baker, I am grateful to have found you during this
winding road, and I am thankful for your support and smiling face especially during the last year
and hardest part, you have made my life happier, more enjoyable, and fuller.

x

PREFACE

The major research chapters (i.e. chapter 1, chapter 2, and chapter 3) are prepared as
standalone manuscripts to be submitted for publication. For this reason there is some repetition
between chapters in the study site descriptions and methods.

xi

TABLE OF CONTENTS

LIST OF TABLES ...................................................................................................................... xv
LIST OF FIGURES ................................................................................................................. xviii
INTRODUCTION......................................................................................................................... 1
REFERENCES .......................................................................................................................... 6
CHAPTER ONE ......................................................................................................................... 11
MULTIPLE FACETS OF A STREAMS THERMAL REGIME: EFFECTS OF STREAM
TEMPERATURE ON MICHIGAN'S FLUVIAL FISH ASSEMBLAGES. ................................ 11
Abstract .................................................................................................................................... 11
Introduction ............................................................................................................................. 12
Methods .................................................................................................................................... 15
Study area ............................................................................................................................. 15
Stream layer, stream catchment and landscape data ........................................................... 16
Fish data ............................................................................................................................... 18
Instream temperature data .................................................................................................... 18
Instream temperature metrics ............................................................................................... 19
Analysis ................................................................................................................................. 19
Indicatior species analysis (ISA) .......................................................................................... 20
Covariance structure analysis .............................................................................................. 21
Results ...................................................................................................................................... 24
Patterns in fish species diversity and distribution ................................................................ 24
Patterns in SGCN fish species diversity and distribution ..................................................... 24
Patterns in stream temperature metrics ................................................................................ 25
Indicator species for key stream temperature metrics .......................................................... 26
Covariance structure analysis (CSA) model fit and significance ......................................... 28
Effects of landscape features on stream temperature ........................................................... 28
Effects of landscape controls and stream temperature on fish species ................................ 28
Discussion................................................................................................................................. 29
Dominant patterns in thermal regimes throughout Michigan streams ................................. 31
Landscape–scale influences on prominent thermal characteristics ..................................... 32
Fish response to thermal characteristics and regional differences in response .................. 33
Landscape influences on stream fishes via effects on temperature ...................................... 35
Management implications ..................................................................................................... 37
Acknowledgments ................................................................................................................... 40
APPENDICES ......................................................................................................................... 41
APPENDIX 1.A: TABLES................................................................................................... 42
APPENDIX 1.B: FIGURES ................................................................................................. 62
APPENDIX 1.C: SUPPLEMENTAL TABLES ................................................................... 64
REFERENCES ........................................................................................................................ 76

xii

CHAPTER TWO ........................................................................................................................ 84
LANDSCAPE EFFECTS ON FLUVIAL FISH ASSEMPLAGE STRUCTURE: REGIONAL
RESPONSES TO HUMAN LAND USES. .................................................................................. 84
Abstract .................................................................................................................................... 84
Introduction ............................................................................................................................. 85
Methods .................................................................................................................................... 88
Study area ............................................................................................................................. 88
Stream layer .......................................................................................................................... 89
Natural landscape variables ................................................................................................. 90
Human land uses ................................................................................................................... 91
Fish data ............................................................................................................................... 91
Data analysis ............................................................................................................................ 92
Landscape metric selection ................................................................................................... 92
Fish metric selection ............................................................................................................. 93
Redundancy analysis (RDA) ................................................................................................. 93
Developing a composite variable to characterize natural landscape influences ................. 94
Cascade multivariate regression tree (CMRT) ..................................................................... 94
Results ...................................................................................................................................... 96
Study area ............................................................................................................................. 96
Variance partitioning with (RDA), differences across study ecoregions ............................. 98
Developing a composite variable to characterize natural landscape influences ................. 98
Variance partitioning with cascade multivariate regression trees (CMRT) ........................ 99
Regional influences and differing levels of human land use variables for the second wave of
CMRT .................................................................................................................................. 100
Discussion............................................................................................................................... 101
Variance in fish metrics explained by human and natural landscape factors across five
ecoregions ........................................................................................................................... 102
Regionally-specific effects of urban land use on stream fish .............................................. 103
Two levels of urban land use in Appalachian Piedmont..................................................... 104
Regionally-specific effects of agricultural land use on stream fish .................................... 105
Density of dams in the catchment ....................................................................................... 106
Stream-road crossing in the catchment .............................................................................. 107
Spatial extent of study and number of sampling sites ......................................................... 108
Management implications and conclusions ........................................................................ 108
APPENDICES ....................................................................................................................... 111
APPENDIX 2.A: TABLES................................................................................................. 112
APPENDIX 2.B: FIGURES ............................................................................................... 123
APPENDIX 2.C: SUPPLEMENTAL TABLES ................................................................. 125
REFERENCES ...................................................................................................................... 129
CHAPTER THREE .................................................................................................................. 138
REGIONAL TRENDS OF BIODIVERSITY INDICES: NATURAL LANDSCAPE AND
HUMAN LAND USE CONTROLS ON STREAM FISH ASSEMBLAGES ........................... 138
Abstract .................................................................................................................................. 138
Introduction ........................................................................................................................... 139
Methods .................................................................................................................................. 142

xiii

Study area ........................................................................................................................... 142
Stream layer ........................................................................................................................ 143
Natural landscape variables ............................................................................................... 143
Human land uses ................................................................................................................. 144
Fish data ............................................................................................................................. 145
Biodiversity indices ............................................................................................................. 145
Data analysis .......................................................................................................................... 147
Pearson correlation among fish biodiversity indices ......................................................... 147
Regression analysis ............................................................................................................. 147
Results .................................................................................................................................... 148
Study area ........................................................................................................................... 148
Regional patterns in biodiversity indices ............................................................................ 149
Regional patterns in correlation of biodiversity indices..................................................... 151
Predicting biodiversity indices from landscape factors with linear regression models ..... 151
Trends in predicting biodiversity indices from landscape factors ...................................... 152
Regional differences in predicting biodiversity indices from landscape factors ................ 152
Discussion............................................................................................................................... 153
Correlation among biodiversity indices.............................................................................. 154
Natural landscape factors predicting biodiversity indices ................................................. 155
Human land uses predicting biodiversity indices ............................................................... 157
Spatial extent of study and scope of natural and human environmental gradients captured
............................................................................................................................................. 159
Management implications ................................................................................................... 159
APPENDICES ....................................................................................................................... 161
APPENDIX 3.A: TABLES................................................................................................. 162
APPENDIX 3.B: FIGURES ............................................................................................... 170
APPENDIX 3.C: SUPPLEMENTAL TABLES ................................................................. 173
REFERENCES ...................................................................................................................... 180
CHAPTER FOUR ..................................................................................................................... 188
CONCLUSIONS: REGIONAL INFLUENCES ON FISH ASSEMBLAGES–SUMMARY OF
FINDINGS AND IMPLICATIONS FOR MANAGEMENT .................................................... 188
Principal Findings ................................................................................................................. 188
Chapter one: multiple facets of a streams thermal regime: effects of stream temperature on
Michigan’s fluvial fish assemblages ................................................................................... 188
Chapter two: landscape effects on fluvial fish assemblage structure: regional responses to
human land uses .................................................................................................................. 189
Chapter three: regional trends of biodiversity indices: natural landscape and human land
use controls on stream fish assemblages ............................................................................ 190
Management Implications .................................................................................................... 191
Chapter one: management implications ............................................................................. 191
Chapter two: management implications ............................................................................. 192
Chapter three: management implications ........................................................................... 194
REFERENCES ...................................................................................................................... 195

xiv

LIST OF TABLES
Table 1.1 Average, maximum, and minimum values of landscape variables for 233 study sites,
for the state of Michigan and by three regions, Upper Peninsula (UP), Northern Lower
Peninsula (NLP) and Southern Lower Peninsula (SLP). Variables marked with (*) indicate
those used in Covariance Structure Analysis.. ...................................................................... 42
Table 1.2 Fish species collected at study sites (n=233), species codes, percent site occurrence,
median and quantiles of catch, and percent site occurrence by 3 Michigan regions, Upper
Peninsula (UP), Northern Lower Peninsula (NLP) and Southern Lower Peninsula. Species
marked with (*) indicate those used in CSA. Species marked with (†) indicate Species of
Greatest Conservation Need (SGCN) in Michigan ............................................................... 44
Table 1.3 Taxonomic summaries statewide and by regions... ...................................................... 48
Table 1.4 Instream temperature metric description, codes, mean, maximums and minimums for
variables at 233 study sites, for the state of Michigan and by three regions, Upper Peninsula
(UP), Northern Lower Peninsula (NLP) and Southern Lower Peninsula (SLP). Temperature
metrics marked with (*) indicate those used in CSA. Temperature units in (°C)................ 49
Table 1.5 Principle component analysis (PCA) results for 22 instream temperature variables,
including interpretations of axes, loading assigned to each variable for each axes, and
percentage of variation in data explained by each axis. Instream temperature variable codes
are described in Table 1.3. Total amount of variation explained was 89.24%. Temperature
metrics marked with (*) indicate those used in CSA. ........................................................... 53
Table 1.6 Pearson product moment correlations between landscape variables for covariance
structure analysis (CSA). Mean July air temperature is at the local catchment scale, coarse
and fine lithology, forest land cover and catchment area are at the network catchment scale.
............................................................................................................................................... 54
Table 1.7 ISA values for species associations with instream temperature variables, Statewide
(e.g., ISA value > 0.65) and 3 Michigan regions (e.g., ISA value>0.70). Statewide; n=233,
Upper Peninsula; n=97, Northern Lower Peninsula; n=62, Southern Lower Peninsula; n=74.
Species codes are described in Table 1.2 and instream temperature variable codes are
described in Table 3 with .h and .l depicting high and low species associations with the
variable, respectively... ......................................................................................................... 55
Table 1.8 Fit statistics for CSA models predicting fish descriptor metrics directly for landscape
factors and indirectly through metrics of a streams thermal regime. Fit equals yes if X2 prob
> 0.05, RMSEA ≤ 0.05, TLI > 0.9, and NFI > 0.9 ............................................................... 59
Table 1.9 Standardized total effects of landscape variable generated through CSA directly and
indirectly on individual fish descriptors and indirectly through metrics of the thermal regime
(e.g., MaxT = the Maximum of 30 days moving average of daily maximum in stream
temperature and ADRT= Maximum of 30 days moving average of daily range is stream
xv

temperature). Landscape variables are: Regions= three study regions in Michigan (Figure
1), percent of coarse or fine lithology in the network catchment, network catchment area,
percent of forested land cover in the network catchment, and mean July air temperature
summarized at the network scale. Variables without values were models that had non–
significant fit statistics .......................................................................................................... 60
Table 1.10 Standardized total effects coefficients of landscape variables on stream temperature
metrics were analyzed for statistical significance (t distribution, α = 0.05) using Monte
Carlo bootstrapped SEs. Table includes squared multiple correlation coefficients (R2)
describing variance explained in stream temperature metrics. MaxT = the Maximum of 30
days moving average of daily maximum in stream temperature and ADRT= Maximum of
30 days moving average of daily range is stream temperature. ............................................ 61
Table C1.1 List of instream temperature metric descriptions calculated for 233 sites with
matching temperature and fish assemblage data collected from study reaches. Averages
(mean), maximums (max) and minimums (min) for each temperature variable, statewide
and for each of three regions of Michigan, Upper Peninsula (UP), Northern Lower
Peninsula (NLP), and Southern Lower Peninsula (SLP).. .................................................... 64
Table C1.2 List of sites with temperature and fish assemblage data collected from study reaches.
Each site’s stream name, reach code, and region code in which it is located are listed, along
with the latitude and longitude of the point marking the upstream end of the sample reach..
............................................................................................................................................... 68
Table 2.1 List of 52 landscape variables initially evaluated for regional comparison. .............. 112
Table 2.2 Fish metrics calculated and evaluated for a regional landscape comparison. ............ 114
Table 2.3 Descriptive statistics of landscape variables for the five freshwater ecoregion with
landscape variable descriptions (i.e., mean, minimum, maximum, and the 10th and 90th
percentiles).... ...................................................................................................................... 115
Table 2.4 Descriptive statistics (i.e., mean, range, 10th and 90th percentiles) of fish metrics in
five freshwater ecoregions with number of sample sites per region. .................................. 120
Table 2.5 Principle component analysis (PCA) results for six natural landscape variables for all
study sites across the entire study area. Variable descriptions and axes weightings are
provided. Total amount of variation explained was 59.91%. ............................................ 121
Table 2.6 Variable, value and percent variance explained from the main (natural) and (human)
influences on fish assemblage metrics across five freshwater ecoregions from cascade
multivariate regression tree analysis (CMRT). ................................................................... 122
Table C2.1 Results of Pearson correlation among landscape metrics; only results with r values
≥.0.70 or ≤ -0.70 are displayed.. ......................................................................................... 125

xvi

Table C2.2 PCA results for landscape variables including variance explained by each axis and
axis interpretation................................................................................................................ 126
Table C2.3 Pearson Correlation of fish metrics, only results with r values ≥.0.70 or ≤ -0.70 are
displayed. ............................................................................................................................ 127
Table C2.4 PCA results for fish metrics ..................................................................................... 128
Table 3.1 Descriptive statistics of landscape variables for the five freshwater ecoregion with
landscape variable descriptions (i.e., mean, minimum, maximum, and the 10th and 90th
percentiles).. ........................................................................................................................ 162
Table 3.2 Descriptive statistics (i.e., mean, range, 10th and 90th percentiles) of fish biodiversity
indices in five freshwater ecoregions. ................................................................................. 167
Table 3.3 Pearson’s correlations between the five biodiversity indices for the five study
freshwater ecoregions ......................................................................................................... 168
Table 3.4 Results of multiple linear regression models for relationships of the biodiversity
indices to natural landscape variables and human land use variables, with degrees of
freedom, adjusted R2, and standardized regression coefficient (β) describing the variable
strength in the model and direction of influence on biodiversity indices. All variables
selected for inclusion in best models as predictors for biodiversity indices were significant
at p-value ≤ 0.01.................................................................................................................. 169
Table C3.1 List of species use in biodiversity calculations. (n=312) ......................................... 173

xvii

LIST OF FIGURES
Figure 1.1 Locations of study sites in Michigan having both fish community and instream
temperature data collected from 1990 to 2011 (n= 233)....................................................... 62
Figure 1.2 Model developed for covariance structure analysis to evaluate hypothesized
relationships among landscape variables and fish assemblage descriptors, with landscape
effects modeled directly and indirectly through stream temperature variables (straight
arrows), based on theorized influences. Curved arrows among landscape variables indicate
correlations supported by Pearson pairwise correlations (Table 1.5). .................................. 63
Figure 2.1 Five freshwater ecoregions of the United States that comprised the study area (Abell
et al. 2008). ......................................................................................................................... 123
Figure 2.2 Results from RDA partitioned variation explained in fish community metrics by
natural and human variables across the five freshwater ecoregions in our study area. ...... 124
Figure 3.1 Five freshwater ecoregions of the United States that comprised the study area (Abell
et al. 2008) .......................................................................................................................... 170
Figure 3.2 Part of a taxonomic cladogram, showing examples of path length weights {wij} used
to define taxonomic diversity and taxonomic distinctness (example from Clark and
Warwick 1998).................................................................................................................... 171
Figure 3.3 Range in Adjusted R2 values in biodiversity indices across five study ecoregions from
the multiple linear regressions. The dots are mean values with the whiskers maximum and
minimum values.. ................................................................................................................ 172

xviii

INTRODUCTION

If you have ever drifted along the course of a river in a canoe, paddling around its bends
and meanders, you may be fixed on the water or what’s below, but the wide vistas your eyes
capture are encompassed within the rivers’ valley. Although the river may be the life blood of
that valley, it is in part an extension of that valley as a whole. Hynes (1975) described that
characteristics of the valley including climate, lithology and topology determine many in-stream
processes like flow, sediment load, ion concentrations and nutrient loads; based on this, streams
derive nearly all of their energy from their valleys. The type of vegetation and soils available in
a watershed are partially defined by the climate and lithology of the valley, in turn influencing
the amount of allochthonous inputs and types of ions released into the stream. Water chemistry,
along with the type and amounts of organic material within a stream, determines rates of decay
and establishes the food web that defines the distribution and structure of in-stream biota. These
ideas by Hynes (1975) led in part to the proposal of the River Continuum Concept (RCC) by
Vannote et al. (1980), a conceptual model or framework linking longitudinal changes in resource
gradients of a stream to structural and functional changes in aquatic communities from
headwater streams down to mainstem rivers.
The organization of biological communities and physical habitat of stream systems is
largely determined by characteristics of the natural and anthropogenic landscapes through which
they flow (Hynes 1975; Allan 2004). Localized river conditions, including communities and
habitat, change longitudinally from headwaters to river mouths (Vannote et al. 1980), and
localized river conditions are also constrained by their spatial arrangement within the topological
1

architecture of a river network (Benda et al. 2004). Punctuated breaks along these river
continuums, at river reach confluences, define unique habitat units influenced by their
watersheds (Poole 2002; Fausch et al. 2002; Benda et al. 2004). The hierarchical arrangement of
river systems within watersheds constrains regional influences like geology, topography,
lithology and climactic variables like precipitation and temperature, helping to define the flow
and temperature regimes of rivers (Frissell et al. 1986). This hierarchical nature of river systems
constrains the influence of environmental factors, defines local biotic processes, limiting the
available community species pool and controls distributions and abundances of individual
species and community structure (Schlosser 1991; Tonn 1990; Poff 1997; Fausch et al. 2002).
Using these qualities of nested river systems and landscape-scale factors on local factors to
describe physical habitat, water quality, species abundances and distributions has been described
as a landscape perspective of hierarchical riverine networks (Schlosser 1991; Osborne and Wiley
1992; Ward 1998; Fausch et al. 2002; Benda et al. 2004; Grant et al. 2007).
Building on this idea, some of the greatest threats to freshwater ecosystems globally are
modifications caused by anthropogenic activities of fluvial systems and their catchments (Allan
2004; Dudgeon et al. 2006; Vörösmarty et al. 2010). Examples of such changes to stream
systems include direct modification of stream flows and fluvial habitats from construction of
dams and stream-road crossings, water withdrawals, diversions, and channelization of stream
habitats and indirect threats that originate from human perturbations in adjacent landscapes that
lead to inputs of excess nutrients, sediments, or toxics, altered water chemistry, altered
hydrology, and altered thermal regimes (Allan and Flecker 1993; Sala et al. 2000; Duncan and
Lockwood 2001; Caissie 2006; Dudgeon et al. 2006; Helfman 2007; Jelks et al. 2008; Olden and
Naiman 2010; Carpenter et al. 2011; Esselman et al. 2011). Landscape-scale human

2

disturbances have commonly been described in the literature as having negative effect on stream
habitats and fish assemblages include urbanization and agriculture (Wang et al. 1997; Wang et
al. 2000; Wang et al. 2001; Walsh et al. 2005), with barriers to fluvial ecosystems like dams and
stream road-crossings also being considered a landscape-scale disturbance in more recent
investigations (Cooke et al. 2012; Perkin and Gido 2012; Januchowski–Hartley et al. 2013).
Despite understanding of landscape influences on streams, trends in degradation of
aquatic ecosystems and loss of freshwater fish biodiversity continue to occur, as both indirect
and direct human impacts on freshwater ecosystems continue throughout the United States (U.S.)
driven in part by large-scale changes in human land use. A factor that complicates protection of
streams is that land use influences on streams may vary regionally due to influences of cooccurring natural landscape factors such as lithology and topography. This can result in
regionally-specific trends and/or differences in the types and levels of human land uses affecting
stream fishes across regions. There is inherent natural variability across different regions of the
U.S., and landscape-level anthropogenic disturbances and their impacts also vary widely
throughout the Nation. With regionally-different underlying natural landscape variables like
lithology, land cover and climate, fish assemblages with similar membership in one region may
respond to anthropogenic disturbances in a different manner than fish assemblages in another
region. These differences across regions may render one locale more sensitive to biodiversity
loss or fish assemblage compositional change from the same magnitude of anthropogenic
disturbance in the landscape. Substantial differences have been seen in fish diversity and
abundance across physiographic regions due to changes of human land use (Utz et al. 2010), and
differential responses of stream conditions have been quantified with different levels of urban
and agricultural land use in different regions (e.g., Walsh et al. 2005, Meador et al. 2005). These

3

studies emphasize species-specific inter-regional variability in fish response to similar
anthropogenic stressors (Hugget 2005, Meador et al. 2005, Utz et al. 2010, Esselman et al. 2013).
Few studies have attempted to characterize regionally-distinct fish assemblage response to
gradients of landscape human disturbance, yet this understanding is essential to develop largescale policies and practices to protect and conserve fish habitats, fish diversity and fish
population abundances in stream ecosystems.
In this dissertation, my aim is to provide new ecological insights into the large scale
regional variability of influences that natural landscape factors and human land uses have on
structuring fluvial fish assemblages. Fluvial fish assemblages have been utilized as response
indicators to fluvial ecosystem condition because fauna are sensitive to environmental
disturbance and degradation in their surrounding landscapes (Pont et al. 2006; Stoddard et al.
2008; Esselman et al. 2013). I used fish species and assemblage metrics describing community
structure, trophic structure, levels of tolerance, and life history to quantify interregional
variability in their response to human stressors and natural landscape factors through a
landscape-based approach. This enhanced understanding of the variability of regional responses
of fish assemblage metrics to human stressors should inform the development of a multimetric
index of biotic condition (MMIs) for fish, for large regional studies assessing stream condition
and integrity. This will provide insight into specific types and locations of threats to regional
fish assemblage distinctness. This knowledge will aid managers in improvement of regional
prescriptive management action to protect fish assemblage biodiversity into the future.
I begin by considering regional differences in stream thermal regimes across Michigan
and the association of specific fish species to multiple facets of a thermal regime to assess which
attributes of stream temperature are affecting fish assemblages and how this varies across the
4

state. I use a covariance structure analysis to consider how landscape factors mediate effects on
specific fish species and community structure through dominant characteristics of stream
temperature. I next consider interregional differences in fish metric response, like feeding
guilds, levels of tolerance and life history characteristics, to the effects of human land uses across
five freshwater ecoregions in the eastern U.S. Finally, I evaluate dominant natural landscape
factors and human land uses influencing five biodiversity indices and how these influences vary
across five freshwater ecoregions. My research provides a framework and new methods to
elucidate regional differences in major human stressors and their effects on fluvial fish
assemblage structure. Further, findings from my study will inform landscape-scale conservation
measures across my large study regions and provide prescriptive regional management actions to
preserve regional fish assemblage distinctness.

5

REFERENCES

6

REFERENCES

Allan, J. D., 2004. Landscapes and riverscapes: the influence of land use on stream ecosystems.
Annual Review of Ecology, Evolution, and Systematics 35: 257–284.
Allan, J. D. and A. S. Flecker. 1993. Biodiversity Conservation in Running Waters. BioScience
43:1 32–43.
Benda, L., N. L. Poff, D. Miller, T. Dunne, G. Reeves, G. Pess and M. Pollock. 2004. The
network dynamics hypothesis: how channel networks structure riverine habitats. BioScience
54(5): 413–427.
Caissie, C. 2006. The thermal regime of rivers: a review. Freshwater Biology 51:1389–1406.
Carpenter, S. R., E. H. Stanley, and M. J. V. Zanden. 2011. State of the world’s freshwater
ecosystems: physical, chemical, and biological changes. The Annual Review of
Environment and Resources 36:75–99.
Cooke, S. J., C. Paukert, and Z. Hogan. 2012. Endangered river fish: factors hindering
conservation and restoration. Endangered Species Research 17:179–191.
Dudgeon, D., A. H. Arthington, M. O. Gessner, Z. Kawabata, D. J. Knowler, C. Lévêque, R. J.
Naiman, A. Prieur-Richard, D. Soto, M. L. J. Stiassny, and C. A. Sullivan. 2006.
Freshwater biodiversity: importance, threats, status and conservation challenges.
Biological Review 81:163–182.
Duncan, J. R., and J. L. Lockwood. 2001. Extinction in a field of bullets: a search for causes in
the decline of the world's freshwater fishes Biological Conservation 102:97-105.
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
landscape anthropogenic activities. Ecological Restoration 29:133-151.
Esselman, P. C., D. M. Infante, L. Wang, A. R. Cooper, D. Wieferich, Y. Tsang, D. J.
Thornbrugh, and W. W. Taylor. 2013. Regional fish community indicators of landscape
disturbance to catchments of conterminous United States. Ecological Indicators 26:163–
173.
Fausch, K. D., C. E. Torgersen, C. V. Baxter and W. L. Hiram. 2002. Landscapes to riverscapes:
bridging the gap between research and conservation of stream fishes. BioScience 52(6): 483–
498.
Frissell, C. A., W. J. Liss, C. E. Warren and M. D. Hurley. 1986. A hierarchical framework for
stream habitat classification: viewing streams in a watershed context. Environmental
Management, 10(2): 199–214.
7

Grant, E. H. C., W. H. Lowe, and W. F. Fagan. 2007. Living in the branches: population
dynamics and ecological processes in dendritic networks. Ecology Letters 10(2):165–175.
Helfman, G. S. 2007. Fish conservation: A guide to understanding and restoring global aquatic
biodiversity and fishery resources. Island Press, Washington, D.C.
Huggett, A. J. 2005. The concept and utility of ‘ecological thresholds’ in biodiversity
conservation. Biological Conservation 124:301–310.
Hynes, H. B. N. 1975. The stream and its valley, in Proceedings of the 19th Congress of the
International Association of Theoretical and Applied Limnology, Winnipeg, Canada,
August 1974, Verhandlungen IVL, 19:1–15, Stuttgart (Schweizerbart).
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 11(4):211-217.
Jelks, H. L., S. J. Walsh, N. M. Burkhead, S. Contreras-Balderas, E. Díaz-Pardo, D. A.
Hendrickson, J. Lyons, N. E. Mandrak, F. McCormick, J. S. Nelson, S. P. Platania, B. A.
Porter, C. B. Renaud, J. J. Schmitter-Soto, E. B. Taylor, and M. L. Warren Jr., 2008.
Conservation status of imperiled North American freshwater and diadromous fishes.
Fisheries 33 (8), 372–407.
Meador, M. R., J. F. Coles, and H. Zappia, 2005. Fish assemblage responses to urban intensity
gradients in contrasting metropolitan areas: Birmingham, Alabama, and Boston,
Massachusetts. Pages 409–422 in L. R. Brown, R. H. Gray, R. M. Hughes, M. R.
Meador, editors, Effects of Urbanization on Stream Ecosystems. American Fisheries
Society, Bethesda.
Olden, J. D., and R. J. Naiman. 2010. Incorporating thermal regimes into environmental flows
assessments: modifying dam operations to restore freshwater ecosystem integrity.
Freshwater Biology 55:86–107.
Osborne, L. L. and M. J. Wiley. 1992. Influence of tributary spatial position on the structure of
warmwater fish communities. Canadian Journal of Fisheries and Aquatic Sciences. 49:
671-681.
Perkin, J. S. and K. B. Gido. 2012. Fragmentation alters stream fish community structure in
dendritic ecological networks. Ecological Applications 22:2176–2187.
Poff, N. 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.

8

Pont, D., B. Hugueny, U. Beier, D. Goffaux, A. Melcher, R. Noble, C. Rogers, N. Roset and S.
Schmutz. 2006. Assessing river biotic condition at the continental scale: a European
approach using functional metrics and fish assemblages. Journal of Applied Ecology, 43, 70–
80.
Poole, G. C. 2002. Fluvial landscape ecology: addressing uniqueness within the river
discontinuum. Freshwater Biology, 47, 641–660.
Sala, O. E., F. S. Chapin, J. J. Armesto, E. Berlow, J. Bloomfield, R. Dirzo, E. Huber-Sanwald,
L. F. Huenneke, R. B. Jackson, A Kinzig, R. Leemans, D. M. Lodge, H. A. Mooney, M.
Oesterheld, N. L. Poff, M. T. Sykes, B. H. Walker, M. Walker, D. H. Wall, 2000.
Biodiversity – global biodiversity scenarios for the year 2100. Science 287:1770–1774.
Schlosser, I. J. 1991. Stream fish ecology: a landscape perspective. BioScience 41(10): 704–712.
Stoddard, J. L., A. T. Herlihy, D. V. Peck, R. M. Hughes, T. R. Whittier and E. Tarquinio. 2008.
A process for creating multimetric indices for large-scale aquatic surveys. The North
American Benthological Society 27(4):878-891.
Tonn, W. M. 1990. Climate change and fish communities: a conceptual framework. Transactions
of the American Fisheries Society 119: 337–352.
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 143:688–699.
Vannote, R. L., G. W Minshall, K. W. Cummins, J. R. Sedell and C. E. Cushing 1980. The river
continuum concept. Canadian Journal of Fisheries and Aquatic Science 37:130–137.
Vörösmarty, C.J., P. B. McIntyre., M. O. Gessner, D. Dudgeon, A. Prusevich, P. Green, S.
Glidden, S. E. Bunn, C. A. Sullivan, C. R. Liermann, P. M. Davies. 2010. Global threats
to human water security and river biodiversity. Nature 467:555–561.
Walsh, C. J., A. H. Roy, J. W. Feminella, P. D. Cottingham, P. M. Groffman, R. P. I. Morgan,
2005. The urban stream syndrome: current knowledge and the search for a cure. Journal of
the North American Benthological Society 24:706-723.
Wang, L., J. Lyons and P. Kanehl, 2001. Impacts of urbanization on stream habitat and fish
across multiple spatial scales. Environmental Management 28:255–266.
Wang, L., J. Lyons, P. Kanehl, R. Bannerman, and E. Emmons. 2000. Watershed urbanization
and changes in fish communities in southeastern Wisconsin streams. Journal of the
American Water Resources Association 36(5):1173–1189.
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.

9

Ward, J. V. 1998. Riverine landscapes: biodiversity patterns, disturbance regimes and aquatic
conservation. Biological Conservation 83(3): 269–278.

10

CHAPTER ONE

MULTIPLE FACETS OF A STREAMS THERMAL REGIME: EFFECTS OF STREAM
TEMPERATURE ON MICHIGAN’S FLUVIAL FISH ASSEMBLAGES.

Abstract
Stream thermal regimes are one of the most important factors characterizing distributions
and abundances of fishes, and thermal characteristics are themselves influenced by landscape
factors. Understanding the role of landscape factors on thermal characteristics and indirectly on
fishes is important for managing current stressors to aquatic habitats and to plan for changes that
can occur in landscape conditions resulting from changes in land use and climate. The goal of
this study was to better understand how landscape factors and stream thermal regimes structure
fluvial fish assemblages in Michigan. Objectives include: 1) characterize how thermal regimes
vary throughout stream reaches in Michigan, describing prominent characteristics of thermal
regimes and characterize major landscape–scale factors controlling these thermal characteristics,
2) test which thermal characteristics stream fish are responding most strongly to and describe if
there are regional differences in response across Michigan, and 3) evaluate the way in which
landscape factors affect fish through major facets of thermal characteristics. Indicator species
analysis was used to evaluate relationships among various metrics characterizing stream thermal
regimes and fish measures included taxa richness, Shannon’s diversity, and Pielou’s evenness.
Results showed region–specific patterns in fish species association with different thermal
metrics. Covariance structure analysis (CSA) indicated that fish species responses were
regulated by maximums and variability in summer stream temperatures which were determined

11

by percent forest cover in the catchment and mean July air temperature in the catchment. . Our
results provide new insights on regional differences in controls on stream fish assemblages and
species–specific responses by indirect effects of landscape factors and direct effects of multiple
prominent stream thermal metrics. This offers opportunities for improving conservation and
management of stream systems and fish resources in the face of current and future changes in
land use and climate.
Introduction
A thermal regime is one of the most important characteristics of stream habitat,
influencing distributions and abundances of fluvial fishes (Magnuson et al. 1979; Brazner et al.
2005; Buission et al. 2008). Absolute magnitudes of water temperatures and temperature
variation directly and differentially affect individual fish species by triggering key events such as
spawning, hatching as well as growth throughout ontogeny (Schlosser 1991; Matthews 1998;
Helfman 2007). Due to the ectothermic physiology of fish, a thermal regime plays a large role in
streams, controlling population density (Inoue and Nakano 2001); determining range of species
(Quist et al. 2004; Buission et al. 2008), influencing metabolism, growth and fecundity (Lobon–
Cervia et al. 1996), and affecting behavior (Taniguchi et al. 1998). Temperature may also act as a
physicochemical habitat filter (Poff et al. 1997), limiting availability of specific thermal habitats
as well as fish movements to those habitats. In addition to direct influences on fish, stream
temperatures may also have a myriad of indirect effects on particular fish species as they control
predators, prey, competitors, and food web interactions.
Many different facets characterize thermal regimes of streams. Stream thermal regimes
include average, maximum, and minimum temperatures within a particular system expressed

12

over days, seasons, and/or years. Thermal regimes, however, also include variability in
temperatures, expressed as daily, seasonal, and/or annual fluctuations (Caissie 2006; Olden and
Naiman 2010). Additionally, other elements of stream thermal regimes include timing of
thermal events, frequency and duration of critical temperatures, and the rate of change between
temperature extremes (Poff et al. 1997; Olden and Poff 2003; Chu et al. 2010). Together these
multiple facets characterize the thermal regime of a stream and also constrain available physical
habitat for fishes.
Stream thermal regimes are influenced by numerous natural factors, including climate,
which is a primary determinant of river temperatures (Poole and Berman 2001; Caissie 2006).
Climate, along with other landscape–scale drivers, exerts spatially–explicit controls on thermal
characteristics across climatic and environmental gradients (Wehrly et al. 2003). Typically,
streams warm from headwaters to river mouths, with potentially wide variability in diversity and
quality of thermal habitats (Vannote et al. 1980). That variability can be influenced by local
conditions, or by spatial topology affecting the access of fish to thermal refugia (Matthews and
Berg 1997; Martin and Petty 2011). At broad to fine geographic scales, latitude, elevation,
surficial geology, and soils also contribute to differences in thermal characteristics within rivers,
while site–specific variation in stream temperature can be influenced by localized conditions
such as groundwater inputs, stream channel dimensions, and shading (Wehrly et al. 1997; Poole
and Berman 2001; Caissie 2006; Wehrly et al. 2006).
The organization of local stream communities is largely determined from the natural and
anthropogenic landscapes in their surrounding catchments in a hierarchical fashion from
landscape factors influencing physical habitat that in turn effect local community structure
(Hynes 1975; Allan 2004). Anthropogenic activities, including urbanization, forest management,
13

agriculture, and groundwater withdrawal, alter stream temperatures and thermal variability
(Wang et al. 1997; Allan 2004; Caissie 2006; Wehrly et al. 2006). Barriers like dams and road
crossings also affect stream thermal regimes and fluvial fish assemblages by warming or cooling
streams , fragmenting habitats that inhibit movement, and altering stream flow (Lessard and
Hayes 2003; Hayes et al. 2008; Wang et al. 2011b; Januchowski–Hartley et al. 2013).
The thermal regime of a stream has varied influences on thermal habitats and stream
fishes operating at both local and catchment spatial scales. Influences on stream temperature,
operating over multiple spatial extents, affect the thermal regime of a stream in a hierarchical
fashion. This is known as a landscape perspective (Schlosser 1991; Allan and Johnson 1997;
Johnson and Gage 1997; Ward 1998; Wiens 2002). This nested hierarchy of morphological units
is constrained by processes operating at higher levels within the catchments like regional
climatic variables such as precipitation and temperature, but it is also influenced by local
processes that help to define the thermal regime of a river (Frissell et al. 1986; Benda et al. 2004;
Parsons and Thoms 2007). This hierarchical arrangement of dendritic river ecosystems
constrains the influence of environmental factors, defines local biotic processes, limiting the
available community species pool and controlling distribution and abundance of individual
species and community structure (Tonn 1990; Schlosser 1991; Fausch et al. 2002). This
complex suite of hierarchical factors help to shape the thermal regimes of river systems across
multiple biogeographic regions, providing the necessary thermal characteristics that support and
structure cold, cool and warm water fish assemblages (Jackson et al. 2001; Wehrly et al. 2003).
Proactive conservation of fish assemblages that protect common species, preserve regional
distinctiveness of rare fishes like species of greatest conservation need (SGCN), and
management to ensure sustainable fisheries in the face of both current and future threats to steam

14

systems is a daunting challenge. The hierarchical relationships between landscapes, physical
habitat and how the structure of fish assemblages is affected are difficult to test and require a
better understanding of how landscapes affect fish and how they are mediated by physical habitat
like stream temperature.
The goal of this study is to characterize multiple facets of thermal regimes of streams and
quantify how landscape-level factors affect stream fish in a hierarchical structure. We address
four specific objectives. First, we characterize how thermal regimes vary among stream reaches
in Michigan, describing prominent characteristics of thermal regimes and characterizing major
landscape–scale factors controlling these thermal characteristics. Second, we test which thermal
characteristics stream fishes are associated with and describe if there were regional differences in
these associations across Michigan. Third, we evaluate the hierarchical process by which
landscape factors affected fish, mediated by major facets of the thermal characteristics of the
streams. Finally, we describe regional patterns in SGCN fish species diversity, distributions, and
relationships between temperature metrics and rare fish species.
Methods
Study area.—This study was conducted in the state of Michigan which covers an area of
149,000 km2 (Figure 1.1). We divided our study area into three regions to better account for
natural landscape patterns and to account for regional differences in distributions of fish species.
Study regions were delineated by aggregated ecological drainage units (EDUs) for Michigan
(MDNR, unpublished); these regions are described as Upper Peninsula (UP), Northern Lower
Peninsula (NLP), and Southern Lower Peninsula (SLP, Figure 1.1). For this study, data from
233 stream reaches were evaluated, (see details for reach definition below), and each reach had

15

both stream fish data and stream temperature data. Ninety-seven reaches were located in the UP,
62 in the NLP, and 74 in the SLP (Figure 1.1).
Natural land cover in Michigan is dominated by forests (forest land cover in catchments
of study sites averaged 46% and ranged from 0.53% to 98%) based on the 2001 National Land
Cover Dataset (NLCD, Homer et al. 2004, Table 1.1). The two major human land uses include
agriculture and developed land cover classes. The agriculture category was comprised of
pasture/hay and cultivated crops land use classes and had a mean of 19% in study reach
catchments. The category for developed lands included land use classes of low, medium, high
intensity developed land, and open developed lands with an average of 7.78% in study
catchments, (Table 1.1). Land cover patterns vary throughout the state, with more agriculture
and urban land use occurring in the SLP and more forests in the NLP and UP. Surficial lithology
is an important control on stream thermal regimes in Michigan by indirectly regulating the
potential of groundwater contribution to stream base-flow depending on lithology material size.
Surficial lithology varies widely across the state and these landforms in Michigan are primarily
comprised of coarse–textured materials (mean = 85% in catchments of study sites), with fine–
textured materials being less common (Soller and Reheis 2004; Cress et al. 2010, Table 1.1).
Geology also varies, generally in a north to south gradient over the state, with more coarse
geology occurring in the UP and NLP and more fine geology occurring in the SLP.
Stream layer, stream catchments and landscape data.—The stream layer used was the
1:100,000 National Hydrography Dataset Plus Version 1 (NHDPlusV1) national streams layer
(USEPA and USGS 2005). We defined a stream reach as a section of a stream that extends 1)
from the stream origin to the first downstream confluence or junction with a lake or reservoir, 2)
from an upstream confluence or lake/reservoir outflow to the next downstream confluence or
16

lake/reservoir junction, or 3) from an upstream confluence or lake/reservoir outflow to the river
mouth where it meets the Great Lakes (Brenden et al. 2006; Esselman et al. 2011; Wang et al.
2011a). Two types of catchments associated with stream reaches were used in this study: local
catchments and entire upstream network catchments. The local catchment is defined as all land
that drains directly into an individual stream reach without being transported via other fluvial
pathways represented in the NHDPlusV1 and referred to from here after as a local catchment.
The network catchment is all land upstream of and draining into a given reach including the local
catchment.
Landscape data and catchment area were summarized at the local and network catchments
for corresponding NHDPlusV1 stream reaches. Land cover types from the NLCD (Homer et al.
2004) were grouped into classes and summarized, but only the 4 highest proportions of total land
cover types were considered for analyses (e.g., urban, forest, agriculture, and wetlands; Table
1.1). Percentage of catchment in surficial lithology classes (e.g., coarse, fine) were grouped and
summarized (Soller and Reheis 2004; Cress et al. 2010, Table 1.1). Reach elevation and gradient
were calculated from the National Elevation Data (NED, Gesch 2007). Base–flow index, which
represented hydraulic potential characterizing the percentage of groundwater contribution to
stream flow, was modeled by the U.S. Geological Survey (USGS, Wolock 2003, Table 1.1).
Average mean July air temperature was summarized for the climatological period from 1981 to
2010 by the PRISM Climate Group and attributed to the stream reach (Daly et al. 2008; PRISM
2013, Table 1.1). Variables were chosen based on known prominent landscape-scale factors that
have strong natural and human influences on stream physical condition, stream habitats (e.g.
Esselman et al. 2011) and fluvial fish (e.g. Esselman et al. 2013)

17

Fish data.—Fish data for study sites were sampled following MDNR Streams Status and
Trends sampling protocols (Wills et al. 2008) using single pass electrofishing during the summer
months of years between 1990 to 2010.. A stream backpack shocker or tow barge shocker was
used for sampling depending on stream size and water depths (Wills et al. 2008). Lengths of
streams sampled were determined by catchment area and by site specific mean stream widths.
Streams with catchment area < 105 km2 were defined as small streams and streams with
catchment area between 105 km2 and 465 km2 were defined as medium streams. For a small
stream < 4.6 m wide, a stream length of 152 m was sampled. For a small stream ≥ 4.6 m wide a
stream length of 244 m was sampled. A length of 366 m was sampled for medium stream (Wills
et al. 2008). Large and very large stream sites were excluded from analyses because data
collected using gear for sampling these systems (e.g., boomshocker) were deemed incompatible
with data collected using small to medium stream sampling methods (e.g., backpack and barge
shockers). All fish were identified to species except for some cottids, centrarchids, and juvenile
petromyzontids which were identified to genus or family level. Fish that were only identified to
genus or family level were used to calculate relative abundance of species at sites but were not
included in taxon–specific summaries. We calculated the relative abundance (number of total
individuals collected/100m stream length) for each species at each sampling location (Table 1.2).
Taxa richness, Shannon’s diversity, and Pielou’s evenness indices were also calculated for all
study sites (Shannon 1948; Pielou 1975, Table 1.3).
Instream temperature data.—Temperature data were collected at study sites following
MDNR Stream Status and Trend sampling protocols (Wills et al. 2008) between 1990 and 2011.
Sample sites that were within 10 km downstream of a major dam were not included in this study
(U.S. Geological Survey 2012). All instream temperature data were collected every hour using
18

temperature data loggers that collected data at least from May through September. Temperature
data loggers were deployed at each site in the thalweg of the stream or in an area likely to remain
wetted through the sampling period. Once temperature data were downloaded, records were
graphed and inspected visually to identify periods when the logger may have been out of the
water or missing data points. Any site with less than 5 days consecutive collection of stream
temperature or high degree of variability (e.g., recording dry periods) were excluded from
analyses. All of the 233 sites evaluated in this study had corresponding instream temperature
data for the entire months of June, July, and August.
Instream temperature metrics.—Temperature data were summarized into 69 different
metrics characterizing magnitude and variability in stream temperature conditions (Table C1.1,
Chu et al. 2010; Arismendi et al. 2013). From these, a subset of metrics was selected for further
analysis based on ecological interpretability and their availability for all study sites. For
example, metrics calculated for the months of May and September were excluded from analyses
because not all study sites had data collected during these time periods. Also, moving day
averages calculated over 1 and 3 day time periods were found to be highly correlated with
moving day averages over longer time periods and excluded from analyses because longer time
periods are assumed to capture a broader range in conditions. Following these selection steps, 22
temperature metrics were used for further analysis (Table 1.4). Before analysis, temperature
metrics were natural log transformed to help meet the assumption of linearity.
Analysis.—To reduce the dimensionality of the temperature data and to identify major
patterns in temperature for the study area, principal component analysis (PCA, Legendre and
Legendre 2012) was performed on a correlation matrix of the 22 temperature metrics from our
study sites (Table 1.5). Resulting factors with eigenvalues greater than 1 were retained for
19

interpretation. To aid in interpretation, Varimax rotation was applied to resulting axes, and
individual variables weighted by an absolute value of 0.6 or higher were considered for
interpretation of the PCA axes (Table 1.5)
Indicator species analysis (ISA).—We used Indicator Species Analysis (ISA) to calculate
an indicator index (IndVal) that identifies associations between specific fish taxa (Table 1.2) and
temperature metrics (Dufrene and Legendre 1997). ISA determines the strength of associations
between pairs of variables, with one variable characterizing species distributions across a range
of sites and a second variable characterizing an environmental factor that may control differences
in species distributions. The indicator index is a measure of specificity for the relative
abundance of a fish species for a particular temperature metric group or site-grouping
(DeCaceres and Legendre 2009; DeCaceres et al. 2010). To be considered a good indicator, a
species must have a high relative abundance at a majority of sites and a high or low association
at sites with the environmental characteristic of interest. In this way, results of ISA reveal
indicator species along with environmental characteristics potentially important in controlling
species distributions across sites.
To describe state-wide and regional trends in associations between fish and temperature
metrics we used widely distributed fish species. Fish that did not occur at more than 10% of
study sites were not included in the ISA analysis to reduce the influences of species with small
ranges, rarity in occurrence or detection. Fish and temperature metrics that were outside of three
standard deviations from the mean were excluded before running the ISA to preserve large
regional trends and patterns seen in the datasets. To run ISA, we first identified values indicating
the top and bottom 50% of each of the 22 temperature metrics, characterizing site groups. Next,
the indicator analysis was run using the function “IndVal.g” in the “indicspecies” package in R
20

(DeCaceres and Legendre 2009; R Core Team 2011) to calculate the indicator value index. The
indicator value index is composed of two metrics: the probability of a site belonging to a sitegroup combination when the species has been found at that site and how frequently the species is
found at sites belonging to the site-group (Dufrene and Legendre 1997; DeCaceres et al. 2010).
The indicator value ranges from 0 to 1, with higher values representing a greater association with
a particular site-group. Statistical significance of the association was evaluated with a
permutation test that uses the maximum Sqrt IndVal for the test value. We ran 100 random
permutations. Species with p-values < 0.05 were considered significant indicators of a particular
site-group. We performed ISA on our statewide data set and within each of the three regions to
identify statewide and regional differences in relationships between species and temperature
metrics (Figure 1.1).
Covariance structure analysis.—Covariance Structure Analysis (CSA) is a multivariate
statistical technique that quantifies sources of variance in complex systems and allows for the
testing of hierarchical relationships among variables structured by a priori hypotheses of
interrelationships (Bollen 1989; Maruyama 1997). In this study, we used CSA to investigate
relationships between fish metrics, stream thermal characteristics, and landscape features. With
CSA, correlations among landscape variables can be accounted for, while also accounting for
direct influence of landscape variables on fish metrics and indirect influence via thermal
characteristics of the stream (Maruyama 1997; Riseng et al. 2004; Infante et al. 2006). CSA
cannot be used to prove causality, but it can be used to compare hypothesized interrelationships
among variables and test if those models are consistent with the sampled data (Bollen 1989;
Maruyama 1997).

21

Two inputs are required to perform a CSA, a model of hypothesized relationships among
all variables of interest (Figure 1.2) and the actual sampled data. CSA results include estimates
of the overall fit of the hypothesized model to the actual sampled data as well as estimations of
the strengths of relationships among variables (Maruyama 1997). Estimates provide information
on the direct effects that one variable has on another and on the indirect effects that one variable
has on another as it acts through one or more variables as constrained by relationships depicted
in the hypothesized model (Maruyama 1997).
A single CSA model was developed for this study, with paths representing hypothesized
relationships among variables (Figure 1.2). The model developed for this study compares the
indirect influence of landscape features including study region, coarse and fine lithology,
catchment area, percent forest land cover and mean July air temperature on fish through effects
of the two most dominant metrics of the thermal regime in a stream, as informed from PCA (i.e.,
summer maximums and summer ranges); while also accounting for direct effects of coarse and
fine lithology and catchment area on fish taxa and community metrics. We developed this model
based on current understanding of dominant controls on stream fish and thermal regimes. Fine
and course surficial lithology were chosen for the CSA model to represent differences in
lithology across the study sites but also as a surrogate of hydraulic potential (Wolock 2003). For
example, surficial lithology has been shown to influence instream temperature in Michigan due
to the relationship between lithology, hydraulic potential and the percentage of groundwater
contribution to base–flow in glaciated regions (Baker et al. 2003). Modeling the direct effects of
fine and coarse surficial lithology on fish as well as indirect effects via metrics of the thermal
regime allowed us to estimate the relative importance of other influences of surficial lithology
compared to their effects through stream temperature. It is well recognized in stream systems
22

that as metrics of stream size increase, like catchment area and correlates (e.g., stream order,
stream volume, stream width), there is an increase in area and diversity of available habitat and a
propensity for greater fish species diversity and abundance (Angermeier and Schlosser 1989
Goldstein and Meador 2004; Infante et al. 2006; Thornbrugh and Gido 2010). Forest land cover
and mean July air temperature have been recognized as influential on stream temperatures in
Michigan (Wehrly et al. 2006; Steen et al. 2008). The CSA allows us to estimate hierarchical
influences of landscape directly on stream temperature metrics as well as their mediating effects
on fish indirectly through stream temperature.
AMOS 21.0 (IBM Corporation 2012) was used to perform the CSA using maximum
likelihood estimation to fit sampled data to the hypothesized model. There were 30 total model
runs to predict the effects of landscape factors through two stream temperature metrics and fish
descriptors. Correlations among landscape factors were modeled based on evaluation of Pearson
correlation coefficients; R2 values less than 0.15 (and P ≥ 0.05) were considered uncorrelated in
the CSA model (Table 1.6). Fish descriptors tested in the model included metrics characterizing
assemblages (richness, diversity, and evenness) as well as taxon–specific summaries suggested
as being sensitive to temperature metrics by the ISA.
Based on the hypothesized model and sampled data, two types of outputs were used to
evaluate the CSA model. A model selection process was used to test if the sampled data (e.g.
fish taxa and community metrics, 2 stream temperature metrics) fit the a priori hypothesized
model structure through a variety of minimum fit statistics criteria for further evaluation. Model
fit was evaluated by chi–squared (χ2) and its associated probability (P > 0.05), degrees of
freedom (df), chi–squared divided by degrees of freedom (χ2/df), the root mean square error of
approximation (RMSEA ≤ 0.05), the Tucker Lewis index (TLI > 0.9), and the normed fit index
23

(NFI > 0.9) following Riseng et al. (2004), Infante et al. (2006), Arbuckle (2010), and Infante
and Allan (2010). The amount of variability in fish species explained by the models was
evaluated with squared multiple correlation coefficients (R2). Significance of the effects was
assessed with the 95% bias–corrected confidence intervals generated from Monte Carlo
bootstrap procedure and t–distribution based on the degrees of freedom for the models following
Arbuckle (2010). Variables were deemed significant at α ≤ 0.05. All variables were properly
transformed to meet assumptions of normality before analysis.
Results
Patterns in fish species diversity and distribution.—A total of 61,551 fish specimens
representing 91 species in 19 families were collected from the study sites, with 57 species
collected from UP streams, 65 species from NLP streams, and 79 species from SLP streams
(Table 1.2). Fish species ranged in percent of site occurrence from 0.4% to 66%. Forty–four
species of fish were found in all three study regions, and 25 species were found within just one
of the study regions. Eleven species of fish had a site occurrence of greater than 27% across
study sites including, White Sucker, Creek Chub, Eastern Blacknose Dace, Brook Trout, Central
Mudminnow, Mottled Sculpin, Johnny Darter, Brown Trout, Rainbow Trout, Longnose Dace,
and Common Shiner with a sharp decline in species occurrence at study sites for the eighty
additional species (Table 1.2). The most widely–distributed fish species was the White Sucker
Catostomus commersonii, found at 66.1% of the 233 study sites, with a median of 5 individuals
per 100 m and 75% and 25% quantiles equal to 11 and 1, respectively (Table 1.2).
Patterns in SGCN fish species diversity and distribution.—Of the 44 fish Species of
Greatest Conservation Need (SGCN) listed for the state of Michigan (Eagle et al. 2005), only 18

24

were collected from the 233 study sites (Table 1.2). Striped Shiner Luxilus chrysocephalus was
the most widely–distributed SGCN species across study sites, with a site occurrence of 11.6%, a
median of 3.30 fish collected at sites and with 75% and 25% quantile values equal to 6 and 1
individuals per 100 m, respectively. Slimy Sculpin Cottus cognatus, was the most locally
abundant SGCN collected (e.g., 12.7 median catch of individuals/100 m), but it was found at
only 12 study sites. Six species were found at a single site: Lake Sturgeon Acipenser fulvescens,
Lake Chubsucker Erimyzon sucetta, Bigmouth Shiner Notropis dorsalis, Spotted Sucker
Minytrema melanops, River Redhorse Moxostoma carinatum and Pugnose Minnow
Opsopoeodus emiliae. SGCN species richness across the three study regions in Michigan ranged
from 8 to 13, with the highest SGCN species richness found in the SLP and the fewest in the
NLP (Table 1.2).
Patterns in stream temperature metrics.—The PCA of stream temperature variables
resulted in four axes explaining 89% of the variation across the study sites (Table 1.5). Axis 1,
explaining 39% of the variation, was weighted positively by temperature magnitudes represented
by maximum monthly temperatures, and maximum 7, 14, 21 and 30 day moving average of
maximum temperatures. We interpreted this axis as showing high values for summer maximum
temperatures. Axis 2 explained 25% of the variation in the sites and was positively weighted by
temperature variability characterized by monthly average daily range and 7, 14, 21 and 30 day
moving average of range in daily temperatures. Axis 3 explained 13% of the variation across
study sites and was positively weighted by the timing of summer maximum range in
temperatures captured by the Julian day maximum values of 7, 14, 21, and 30 days moving
average of daily range in temperatures. Axis 4 explained 12% of the variation across study sites
and was positively weighted by the timing of the summer maximum temperatures characterized

25

by Julian day of maximum values of 7, 14, 21, and 30 days moving average of daily maximum in
temperatures. The higher amounts of variation explained by Axis 1 and 2 compared to Axis 3
and 4 suggest that summer maximum temperature and temperature variability are prominent
dimensions of the thermal regime in Michigan streams represented by our data.
Based on the above results, we selected two temperature variables that weighted strongest
on the first two PCA axes in CSA analysis. These included maximum of 30 days moving
average of daily maximum in stream temperature and maximum of 30 days moving average of
daily range in stream temperature. Hereafter they were referred to as summer maximums and
summer ranges.
Indicator species for key stream temperature metrics.—Of the twenty-seven fish species
tested in the ISA, twelve species had associations with specific temperature metrics in Michigan
with both affinities for high (i.e., upper 50%) and low (i.e., lower 50%) values of various
temperature metrics (Tables 1.7). Of the twelve species that had affinities for temperature
metrics, statewide and by regions, nine species had affinities for high summer maximums, three
for low summer maximums, six for high variability and five for low variability in stream summer
temperatures (1.7). At a statewide level White Sucker, Creek Chub Semotilus atromaculatus,
and Johnny Darter Etheostoma nigrum showed an affinity for streams with high summer
maximums, while Brook Trout showed the opposite affinity; higher abundances of Brook Trout
were found in streams with low summer maximums. Creek Chub showed an association for high
summer range in stream temperatures, indicating a high relative abundance at sites with high
variability in daily ranges of temperatures.

26

The ISA highlighted patterns in taxa associations with temperature metrics at the regional
scale versus statewide. For example, the number of fish taxa significantly associated with
temperature metrics was 5 in the UP, 9 in the NLP, and 8 in the SLP, with approximately equal
numbers of species per region associating with temperature metrics across study sites (1.2%,
1.4%, 1.0%, respectively). This suggests that stream thermal regimes may play almost an equal
role, relative to the total number of species per region, in influencing stream fishes in the three
study regions (Table 1.7). There is an increase in fish taxa richness from northern latitudes in the
UP to southern latitudes in the SLP (Table 1.2). Creek Chub was the only species associated
with temperature metrics of maximum or range in temperatures statewide and in all three regions
(Table 1.8). Associations for all other species varied regionally. Brook Trout had affinities for
low summer maximums statewide in the UP and NLP. In the SLP Brook Trout only occurred at
nine out of the 74 sites likely due to the warmer stream temperatures and suboptimal thermal
habitats in the SLP. Eastern Blacknose Dace had associations with high summer maximums and
summer ranges in the UP. This potentially indicates that stream temperatures may be cooler than
optimum stream thermal conditions for Eastern Blacknose Dace during summer months in the
UP and may be suggestive of a northern latitudinal range of thermal tolerance for Eastern
Blacknose Dace (Table 1.7). Brown Trout Salmo trutta had associations to low summer
maximums in SLP, suggesting that Brown Trout prefer colder summer stream temperatures in
southern portions of stream in Michigan. A few species only showed affinities for one or a few
temperature metrics in one region or another and this may be most indicative of the home range
of individual fish taxa across Michigan. All twelve fish species that showed a high affinity in the
ISA for any stream temperature metric at the statewide level or in any of the three study regions
was selected for further analysis, along with taxa richness, diversity, and evenness.

27

Covariance structure analysis (CSA) model fit and significance.—Fifteen of the 30
models predicting fish metrics from landscape factors and stream temperature metrics fit the
hypothesized model based on the set of fit statistics evaluated (Table 1.8). Models that failed to
fit the data will not be discussed further (Table 1.8). Of the 15 models that fit the hypothesized
model structure, seven models had a significant fit to the data and also were found to have
significant relationships between landscape factors, stream temperature metrics, and fish
descriptors summarized by standardized total effects (i.e., sum of direct and indirect effects;
Table 1.9). Eight models did not have significant relationships between fish descriptors and
stream temperature metrics.
Effects of landscape features on stream temperature.—Summer temperature maximums
increased significantly with network catchment area and decreased significantly with the
proportion of forested land cover in the network catchment (Table 1.10). The mean July air
temperature in the network catchment had a significant negative effect on summer temperature
ranges (Table 1.10). Catchment area was the strongest predictor of summer maximums of
stream temperatures, and mean July air temperature was the strongest predictor of summer range
in stream temperatures, range increased with decreasing air temperature (Table 1.10).
Effects of landscape controls and stream temperature on fish species.—In the
hypothesized models describing fish descriptors and stream temperature metrics, seven models
had significant standardized total effects (Table 1.9). Summer maximums had significant
positive standardized total effects on Creek Chub and Bluntnose Minnow, and there were
significant negative standardized total effects with summer maximums on Mottled Sculpin and
Brown Trout (Table 1.9), which are known to prefer colder water (Eaton and Scheller 1996;
Lyons et al. 1996; Wehrly et al. 2003). All specific fish taxa that had significant total effects
28

with maximum stream temperatures were influenced indirectly, either positively or negatively,
by the amount of forested land cover in the sites network catchment.
Summer temperature range had significant positive standardized total effects on Creek
Chub and Eastern Blacknose Dace. Mean July air temperature in network catchment had a
significant positive effect on Creek Chub and Eastern Blacknose Dace (Table 1.9). Of the 12
fish species tested, Creek Chub and Eastern Blacknose Dace were positively affected by the
range in summer temperatures, and part of that influence was resulted indirectly from mean July
air temperature acting through temperature range. The 12 fish species responding to temperature
metrics from the ISA analysis, specific fish species have negative and positive effects caused by
two major facets of the thermal regime, maximum stream temperatures and range in stream
temperatures during the summer months in Michigan. Mean July air temperature in network
catchment had a significant negative effect on Pielou’s evenness (Table 1.9). Of the taxonomic
summaries, summer range in stream temperatures had significant negative standardized total
effects on Pielou’s evenness. The variability in daily temperature range was found to have
negative effects on community evenness. Taxon and evenness in the fish community are also
being influenced by landscape level variables, mediated through stream temperature. All
significant relationships between summer maximum stream temperatures and range in stream
temperature with fish species and community metrics all had low R2 values.
Discussion
This research showed influences of stream temperature on fishes and influences of
landscape factors on stream temperature, which highlights that landscape factors indirectly
influence fish via stream temperature. Prominent dimensions of the thermal regime of Michigan
streams characterized in our study included summer maximums and variability in summer stream
29

temperature. Fish species that are associated with summer stream temperature have regionspecific associations to specific thermal characteristics across study sites. Of the 12 species that
showed an association to stream temperature metrics only one species (Creek Chub) showed a
consistent response to stream temperature across all study regions, while most species
responding to temperature had region–specific responses. All 12 fish species that showed a
response to summer stream temperature in the ISA analysis were further investigated during the
CSA analysis. Species–specific responses were characterized by the CSA to the two temperature
metrics (e.g. maximum 30 day moving average of daily maximums and maximum 30 day
moving average of daily range) informed from PCA, which represent the two prominent thermal
characteristics (e.g. maximums and variability in summer stream temperatures) in Michigan.
Our results also indicate that these prominent thermal metrics, maximums and variability
in summer temperatures, are themselves influenced by regional climate and landscape factors
including the amount of forest cover in catchments, mean July air temperature, and catchment
area. Thermal metrics do not operate independently but are differentially influenced by specific
landscape factors. Therefore, prominent drivers (e.g. regional climate, forest cover, July air
temperature) are influencing specific thermal regime characteristics (e.g. maximums or
variability in summer stream temperatures) will dictate the spatial extent most appropriate for
restoration, and which fish species should respond to conservation and management when
incorporating stream temperature. Also specific characteristics of the thermal regime (e.g.
maximums versus variability in summer stream temperatures) will be more strongly influenced
by land–cover change versus climate change, respectively. This study offers new insights about
the importance of summer maximum and variability in summer stream temperatures to specific
fish taxa and characterizes how these influences vary regionally. These results have valuable

30

implications when managing for specific fish species, as to differences in regional response of
fishes to stream temperature metrics and the scale of specific landscape influences directly on
fish and via stream temperature. This provides new insights into how stream fishes may respond
to specific stream temperature metrics in the face of changes in land use and climate across
Michigan (Wang et al. 2003; Allan 2004) and in the region (Lyons et al. 2010; Wenger et al.
2011; Isaak et al. 2012).
Dominant patterns in thermal regimes throughout Michigan streams.—PCA of 22
temperature metrics revealed four axes representing the major characteristics of thermal regimes
in streams in Michigan (Table 1.5). The four axes describe summer maximums and ranges in
stream temperature and the timing of the highest maximums and greatest variability in stream
temperatures during summer months in Michigan. Despite the fact that temperature data used to
calculate metrics were collected during a portion of the year (June through August), it is
noteworthy that our analysis revealed four major unique axes that explain variability inherent in
thermal regimes including magnitudes, variability and timing of events in Michigan streams.
This follows Caissie (2006) who emphasized different spatial and temporal characteristics that
affect stream maximums in temperature, the variability in stream temperatures and the timing of
these events, daily and seasonally. In the Lower Peninsula of Michigan, Wehrly et al. (2003)
found that gradients of summer stream mean temperature and temperature fluctuations were
useful in describing differences across streams in fish assemblage structure. Together, these
findings underscore the importance of considering additional thermal habitat characteristics that
would address thermal magnitudes and variability in stream temperatures along with the timing
of these events, when developing new management strategies for streams and thermal habitats
for fish in Michigan.

31

Landscape–scale influences on prominent thermal characteristics.—Our CSA results
indicated that maximum and variability in summer stream temperatures across our study sites
were correlated with landscape–level controls including catchment area, forest land cover in the
network catchment and mean July air temperature in the network catchment. Specifically,
summer maximums were positively related to catchment area and negatively related to percent
forest cover in network catchments. This result is similar to Chu et al. (2010), who used
discriminant function analysis (DFA) to show not only percent riparian forest affected stream
temperature in Ontario, Canada, but also that percent surface water and groundwater discharge
were influential. Fine and coarse surficial lithology were used in our CSA model to represent
differences in geology across Michigan and also used as a surrogate for base–flow or
groundwater influence on stream temperatures. Our final CSA model is similar to the model of
Wehrly et al. (2006) in the use of similar landscape predictors. They used catchment area, local
forest, mean July air temperature, and percent catchment water and wetland land use to describe
direct landscape controls on mean July stream temperature in the Lower Peninsula of Michigan.
In their study, catchment area had the strongest effect on mean July stream temperature,
(temperature increased with area), followed by a negative effect from local forest cover. Our
results expanded on Wehrly et al. (2006), by testing the responses of fish to an additional thermal
metric that characterizes the range or variability in summer stream temperature, which was
negatively associated with mean July air temperature. This highlights that specific facets of
stream temperature (e.g. maximums and variability) may be vulnerable to different changes in
the landscape in Michigan streams. Maximum summer stream temperatures are potentially most
heavily influenced by changes in land cover (e.g. % forest land cover), while the variability in

32

summer stream temperatures may be most vulnerable to climate change (e.g. mean July air
temperature, Table 1.9) across Michigan streams.
Prolonged high summer air temperatures may lead to less fluctuation in stream water
temperature during summer months, due in part to maximum periods of solar and long–wave
radiation exposure (Caissie 2006). Water, with its higher specific heat capacity than air, would
stay warmer longer during the summer months, losing less heat than air during longer diurnal
photoperiods (Wetzel 2001). Based on this idea, the relationships between variability in summer
stream temperatures and July air temperature will change with varying amounts of precipitation
and warmer air temperatures, such as those that will occur with predicted changes in climate
(Wenger et al. 2011; Isaak et al. 2012). Our results reveal that landscape factors affect
maximums and variability in stream summer temperature, and these factors mediate both fish
assemblage structure and taxa specific responses to the dominant two characteristics of stream
temperature across Michigan. This suggests that a more comprehensive evaluation of the direct
and indirect influences of landscape factors on fish is needed, including understanding how
landscapes control dominant thermal characteristics, and how landscape processes mediate
effects on specific fish species and community structure through characteristics of stream
temperature. This is pertinent for managing fish species effectively in the face of changing land
use and climate.
Fish response to thermal characteristics and regional differences in response.—Our ISA
analyses conducted at both state–wide and regional scales illustrated that different species of
stream fish are related to different dimensions of stream thermal regimes, including magnitudes
and daily and monthly ranges in stream temperatures and the timing of these events. Wehrly et
al. (2003) used two dimensions of stream thermal regimes, stream summer mean temperature
33

and fluctuation in stream temperatures to classify regional patterns in fish distribution in
Michigan. In our study, the ISA analysis quantified which fish species responded to specific
thermal characteristics (e.g. summer maximums and/or summer variability in stream
temperatures), and the response of specific fish species changed across study regions. For
instance, Brook Trout was associated most commonly with low maximum summer temperature
metrics (e.g., Statewide, UP, NLP) and low temperature variability in the UP and NLP. Brook
Trout only occurred at 9 out of 74 (i.e., 12%) sampled sites in the SLP, but had no association
with temperature metrics in the SLP. Brook Trout in the SLP may occur only in streams with
suboptimal–optimal thermal habitats throughout the summer, but might also need stream
connectivity to cooler stream reaches for thermal refugia and long–term persistence in the SLP
region in Michigan streams.
In contrast to brook trout, Brown Trout associated with both low maximum summer
temperatures and low variability to summer temperatures in the SLP. Brown Trout in the SLP,
where mostly warm transitional streams occur, have a strong association with thermal metrics
specifically related to cold or cool streams only, across our study sites in Michigan. Creek Chub
had associations with high maximum summer temperature metrics and high range in summer
temperatures in all study regions and statewide. Creek Chub may have a feeding advantage in
warm and warm transitional streams in Michigan because Creek Chub have been shown to have
competitive success in feeding over other cool water species like Brook Trout and Brown Trout
in Rocky Mountain streams at stream temperatures ≤22°C (Taniguchi et al. 1998). White Sucker
also had an affinity for low variability in summer temperatures in the NLP. Our ISA results
showed that different groupings of fish species and specific taxa respond to different high and

34

low values of thermal regimes, and therefore, stream temperature is mediating the assemblage
structure of fishes and membership of individual taxa via its multifaceted thermal characteristics.
In Michigan, stream temperature in different regions are influenced differently by
groundwater recharge, coarseness of landform textures, depth to bedrock, and anthropogenic
land uses (Zorn et al. 2002; Baker et al. 2003; Wang et al. 2003; Allen 2004; Wehrly et al. 2006).
Utz et al. (2010) found that fishes had region–specific sensitivity to urban and agricultural land
use across adjacent geoclimatic regions. Our ISA found different regional associations of
specific fish species to temperature metrics. It highlights that of the 12 fish species that had an
association with temperature metrics only one species (e.g., Creek Chub) responded similarly to
thermal characteristics across our three study regions, and the majority of taxa had region–
specific responses to temperature metrics (Table 1.7). This finding is consistent with the finding
of Chu and Jones (2011) in that different fish taxa had associations with specific thermal regime
classes; 16% of the explained variation in fish assemblage structure was due to spatial or region–
specific differences between these thermal classes across their study sites in the Great Lakes
Basin of Ontario, Canada. This inter–regional association of specific temperature metrics on
individual fish taxa illustrates that there are region–specific aspects of thermal regime in streams
controlling fish abundances, distributions and assemblage structure in Michigan streams.
Landscape influences on stream fishes via effects on temperature.—Our CSA provides
evidence for indirect effects of landscape–level factors controlling fluvial fish assemblage
structure and specific taxa abundances via two facets of a thermal regime, summer stream
temperature maximums and summer range in stream temperatures in our study sites across
Michigan. Specific species are differentially influenced indirectly by catchment area, percent

35

forest cover in the network catchment, mean July air temperature, through two different facets of
the stream thermal regime, summer maximums and summer range in temperatures.
Catchment area has species–specific effects on fish through two different dominant
characteristics of the thermal regime, maximum summer temperature and variability in summer
temperatures. Catchment area had significant positive effects on Bluntnose Minnow through
maximum summer stream temperatures and on Eastern Blacknose Dace through variability in
summer stream temperatures. Forest land cover had significant positive effects on Mottled
Sculpin and Brown Trout through maximum summer stream temperatures and negative effects
on Creek Chub, Common Shiner, and Bluntnose Minnow via summer maximums in stream
temperature, quantifying the role that forest land cover has on cooling of streams and association
with cold/cool water fish species. Fish species that are influenced by forest cover in the
catchment are likely to be most sensitive to changes in summer maximum stream temperatures
and affected by changes in land use (Wang et al 2003; Allan 2004).
Mean July air temperature had significant negative effects on Creek Chub and Eastern
Blacknose Dace and positive effects on community evenness through variability in summer
stream temperatures (i.e., variability decreased with air temperature). Fish that are influenced by
mean July air temperature are likely most sensitive to changes in summer variability in stream
temperatures and may be most affected by a warming environment and increased precipitation
that increases variability in stream temperatures due to effects of climate change (Wenger et al.
2011; Isaak et al. 2012). Studies have described the hierarchical influence of landscape–level
factors on stream temperature (e.g., Wehrly et al. 2006) and these landscape–level processes on
physical habitat can indirectly mediate fish species abundances and assemblage structure (Infante
et al. 2006; Infante and Allan 2010). To our knowledge this is the first study to indicate that
36

summer ranges in stream temperatures mediate taxa specific abundances and community
evenness through landscape–level factors. These results are important for setting realistic
management goals for specific species. They add new insight on the differential avenues of
vulnerability that specific fish taxa may experience from landscape–level factors like land use
change or climate change, mediating the effects on fish through specific stream thermal
characteristics.
Stream temperature is a major habitat factor that has been used to describe the
distribution and abundances of fish species (Inoue and Nakano 2001; Buission et al. 2008).
Different factors of the thermal regime like summer mean stream temperatures and fluctuations
in summer stream temperatures have been successfully used to classify existing thermal regimes
in stream segments in Michigan and describe patterns of fish species distribution (Wehrly et al.
2003; 2006). Summer stream temperatures and range in stream temperatures have also been
used to help classify valley segments in Michigan’s Lower Peninsula (Seelbach et al. 2006). The
evaluation of stream temperature (e.g. maximums and variability in stream temperatures) and
incorporation of these thermal regimes into stream management are considered a more holistic
perspective of in–stream heat budgets and flow assessments (Poole and Berman 2001; Olden and
Naiman 2010). Michigan has unique stream temperature regimes due to its low elevation, glacial
landscape, and a high degree of groundwater influence. Results from our study highlight that a
more complete evaluation of multiple facets of stream temperature in conjunction with biological
data could offer insights into differential responses of stream fish to different characteristics of
stream thermal regimes.
Management implications.—Our study has highlighted some potential research and
management opportunities. The MDNR’s current practice of characterizing stream temperatures
37

is primarily targeted on thermal characteristics important to game fishes versus the entire fish
assemblage. MDNR monitors stream temperature during periods of thermal maximums like July
and August. Little stream temperature data have historically been collected during the spring
and fall seasons. Also, sampling has historically been more targeted toward salmonid species
then stratified random sampling across the state. Our results emphasized that thermal maximums
in the summer months are important to fish assemblage structure, but the variability in monthly
and daily range in stream temperatures also influences fish assemblage structure in Michigan.
These results have important implications for thermal habitat availability and suitability for
specific game fish species and SGCN species. Under predicted land use and climate changes,
there may be substantial effects on prey fish availability for game and SGCN species related to
stream temperature changes in Michigan. Environmental changes may also cause changes in
condition of specific competition interactions between competing species or in predator–prey
relationships due in part to changes in the thermal regime in Michigan streams.
We generated one of the most comprehensive datasets of community fish assemblage
with corresponding stream temperature data. Yet the summer seasonality of our dataset was not
complete enough to explore the seasonal effects that thermal regime in streams may also have in
structuring fish assemblages during critically energy times for fish like during spawning both in
the spring and fall or cold temperatures observed during winter. The temporal variability in a
streams thermal regime and the specific importance of these changes to the timing of specific
fish species life history characteristics like migration and spawning are vital to developing
species specific thermal habitat management strategies. Our results highlighted the importance
of multiple unique characteristics of a thermal regime to fish species throughout the summer
months in Michigan streams but how these unique characteristics change seasonally and their

38

importance to specific fish taxa are essential avenues for future research. This leads to
management opportunities to improve data collection and analysis to better utilize all aspects of
the in–stream thermal regime that structures fish assemblages. To better manage for changes in
fish assemblages mediated by different facets of the in–stream thermal regimes one would need
to 1) collect complete seasonal in–stream temperature data covering all accessible months for
collection and 2) manage for a richer set of characteristics of the thermal regime that structure
fish assemblages including thermal maximums across seasons as well as monthly and daily
variability in temperature ranges across seasons.
We had one of the most comprehensive datasets for Michigan, with fish assemblage
corresponding with hourly temperature measurements for the summer months. Yet this
comprehensive dataset failed to capture enough abundance of SGCN listed fish species to
evaluate stream temperature relationships of these rare SGCN fish species. Our analyses also
pointed out another potential management opportunity. Long term monitoring programs with a
stratified random sampling design were designed to capture abundances of common fish species
to represent long term trends in a region. Rare fish species occur rarely in any randomized
sampling design because of their low distributions, low species abundances and their difficulty
with being sampled. Rare fish species are less likely to be encompassed in analyses preformed
on long term monitoring datasets. These datasets are often times the best sampling efforts and
best representation of conditions statewide. The problem with long term monitoring datasets is
that because they are the best available comparable statewide datasets they are often used to
investigate patterns in rare fishes and critical habitats for management. These monitoring
programs were not designed to capture the abundances and distribution of rare species.
Therefore management of rare fish species using long term monitor datasets for observing trends

39

in changes of biodiversity and abundance of these rare species will not provide an accurate risk
or threat assessment for habitats or fish species of greatest conservation need. Specific targeted
monitoring of SGCN species and their associated critical habitats, along with other data types
(e.g. stream temperature data), would need to be incorporated into data collection procedures to
evaluate important physical habitat relationships for rare fishes and help conserve regional
distinctiveness in fish biodiversity.
Acknowledgments
We would like to thank our funding providers Michigan Department of Natural
Resources and the U.S. Geological Survey Aquatic GAP Program. We thank Yin–Phan Tsang
for her expertise with R code and the indicator species analysis. Thanks also to Troy Zorn for
his input on identifying sampling locations for this work and for sharing his personal biological
knowledge of some of the streams and rivers in Michigan. Thanks also to Todd Wills for help in
training field crews in the Status and Trends operation procedures for collecting stream fish
assemblages developed by Michigan Department of Natural Resources. It really helped the field
crews to collect comparable and consistent data. Many individuals aided in data collection for
this project and special thanks goes out to the members of the two field crews; Molly Good and
Ryan Young from MSU and Ben Turschak, Rob Cross, Jake Riley, Amanda Chambers, Ashley
Poehls, Kelly Turek from Lake Superior State University for all of your hard work. Thanks to
Wesley Daniel for his comments which helped to improve the manuscript. Comments provided
by members of my graduate guidance committee helped to improve the manuscript.

40

APPENDICES

41

APPENDIX 1.A
TABLES
Table 1.1.—Average, maximum, and minimum values of landscape variables for 233 study
sites, for the state of Michigan and by three regions, Upper Peninsula (UP), Northern Lower
Peninsula (NLP) and Southern Lower Peninsula (SLP). Variables marked with (*) indicate those
used in Covariance Structure Analysis.

Statewide
avg
max
min
Landscape variables
Drainage area (km2)*
172.72 2206.71 3.34
Coarse lithology (%)*
84.68 100.08 0.00
Fine lithology (%)*
16.30 100.00 0.00
Stream gradient (cm/m)
0.37
2.48 0.00
Mean July air temperature (°C)* 19.93
22.75 17.36
Base-flow index (%)
65.52
87.16 29.38
Catchment land cover (%)
Urban
7.78
99.16 0.29
Agriculuture
18.80
88.01 0.00
Forest*
45.87
98.41 0.53
Wetlands
18.89
58.63 0.16
Table 1.1 (cont’d)

avg

UP
max

min

Landscape variables
Drainage area (km2)*
122.74 945.78 3.34
Coarse lithology (%)*
85.08 100.00 0.79
Fine lithology (%)*
15.76 100.00 0.00
Stream gradient (cm/m)
0.59
2.48 0.00
Mean July air temperature (°C)* 18.71
20.46 17.36
Base-flow index (%)
63.23
74.60 52.52
Catchment land cover (%)
Urban
2.97
11.90 0.29
Agriculuture
1.78
21.35 0.00
Forest*
63.35
98.41 21.02
Wetlands
26.09
58.63 0.40

42

Table 1.1.—(cont’d)

avg

NLP
max

min

Landscape variables
Drainage area (km2)*
237.06 1030.67 9.95
Coarse lithology (%)*
91.38 100.01 12.01
Fine lithology (%)*
4.35 100.00 0.00
Stream gradient (cm/m)
0.21
1.19 0.00
Mean July air temperature (°C)* 19.91
20.91 18.98
Base-flow index (%)
76.26
87.16 64.04
Catchment land cover (%)
Urban
7.00
15.69 3.25
Agriculuture
10.36
44.37 0.00
Forest*
50.17
74.74 22.00
Wetlands
16.24
30.88 5.12
Table 1.1.— (cont’d)

avg

SLP
max

min

Landscape variables
Drainage area (km2)*
184.33 2206.71 5.23
Coarse lithology (%)*
78.52 100.08 0.00
Fine lithology (%)*
27.03 100.00 0.00
Stream gradient (cm/m)
0.22
1.35 0.00
Mean July air temperature (°C)* 21.57
22.75 19.95
Base-flow index (%)
59.52
79.40 29.38
Catchment land cover (%)
Urban
14.73
99.16 3.71
Agriculuture
48.20
88.01 0.07
Forest*
19.34
63.98 0.53
Wetlands
11.66
29.97 0.16

43

Table 1.2.—Fish species collected at study sites (n=233), species codes, percent site occurrence, median and quantiles of catch,
and percent site occurrence by 3 Michigan regions, Upper Peninsula (UP), Northern Lower Peninsula (NLP) and Southern Lower
Peninsula. Species marked with (*) indicate those used in CSA. Species marked with (†) indicate Species of Greatest Conservation
Need (SGCN) in Michigan.

Fish species (n=91)
White sucker (Catostomus commersonii )*
Creek chub (Semotilus atromaculatus )*
Eastern blacknose dace (Rhinichthys atratulus )*
Brook trout (Salvelinus fontinalis )*
Central mudminnow (Umbra limi )*
Mottled sculpin (Cottus bairdii )*
Johnny darter (Etheostoma nigrum )*
Brown trout (Salmo trutta )*
Rainbow trout (Oncorhynchus mykiss )
Longnose dace (Rhinichthys cataractae )*
Common shiner (Luxilus cornutus )*
Green sunfish (Lepomis cyanellus )
Rock bass (Ambloplites rupestris )*
Largemouth bass (Micropterus salmoides )
Blackside darter (Percina maculata )
Brook stickleback (Culaea inconstans )
Hornyhead chub (Nocomis biguttatus )
Bluntnose minnow (Pimephales notatus )*
Northern hog sucker (Hypentelium nigricans )
Pumpkinseed (Lepomis gibbosus )
Yellow perch (Perca flavescens )
Rainbow darter (Etheostoma caeruleum )
Burbot (Lota lota )

Species
code
catcom
sematr
rhiatr
salfon
umblim
cotbai
ethnig
saltru
oncmyk
rhicat
luxcor
lepcya
ambrup
micsal
permac
culinc
nocbig
pimnot
hypnig
lepgib
perfla
ethcae
lotlot

Catch (individuals ∙ 100 m-1 )
% of sites
75%
25%
found
quantile Median quantile
66.10
11.20
5.30
1.30
64.40
24.00
6.70
2.30
57.50
27.40
9.50
2.10
48.50
32.20
10.90
2.60
48.10
5.30
1.60
0.70
47.60
25.50
12.00
4.10
43.30
10.50
3.30
1.30
34.80
33.50
9.00
2.60
28.80
36.40
6.60
1.30
27.90
16.40
5.30
1.80
27.00
19.70
4.40
1.30
24.50
4.50
1.70
0.80
24.00
6.10
2.80
1.10
20.20
1.80
0.80
0.40
17.20
4.10
1.80
0.70
16.30
3.80
1.60
0.70
16.30
14.80
5.10
1.30
16.30
8.90
3.10
1.10
15.90
5.30
2.00
1.00
15.90
2.30
1.30
0.60
15.50
1.90
0.70
0.30
12.90
9.10
3.30
1.10
12.90
3.20
1.10
0.50

44

% of sites found by
region
n=57 n=65 n=79
UP NLP SLP
53.60 77.40 73.00
51.50 69.40 77.00
58.80 69.40 45.90
69.10 59.70 12.20
42.30 50.00 54.10
57.70 46.80 35.10
29.90 46.80 58.10
14.40 67.70 33.80
36.10 40.30 9.50
52.60 21.00 1.40
19.60 37.10 28.40
0.00 27.40 54.10
13.40 21.00 40.50
4.10 22.60 39.20
8.20 19.40 27.00
21.60 17.70 8.10
16.50 14.50 17.60
10.30 6.50 32.40
4.10 17.70 29.70
4.10 24.20 24.30
11.30 25.80 12.20
0.00 11.30 31.10
20.60 12.90 2.70

Table 1.2.—(cont’d)
-1

Fish species
Bluegill (Lepomis macrochirus )
Striped shiner (Luxilus chrysocephalus )†
Northern redbelly dace (Phoxinus eos )
Northern pike (Esox lucius )
Smallmouth bass (Micropterus dolomieu )
Logperch (Percina caprodes )
Grass pickerel (Esox americanus )†
Coho salmon (Oncorhynchus kisutch )
Central stoneroller (Campostoma anomalum )
Stonecat (Noturus flavus )
Fantail darter (Etheostoma flabellare )†
Pearl dace (Margariscus margarita )
Slimy sculpin (Cottus cognatus )†
Fathead minnow (Pimephales promelas )
Black bullhead (Ameiurus melas )
Common carp (Cyprinus carpio )
Spotfin shiner (Cyprinella spiloptera )
Golden redhorse (Moxostoma erythrurum )†
Golden shiner (Notemigonus crysoleucas )
Yellow bullhead (Ameiurus natalis )
Blacknose shiner (Notropis heterolepis )
American brook lamprey (Lampetra appendix )
River chub (Nocomis micropogon )
Iowa darter (Etheostoma exile )
Brassy minnow (Hybognathus hankinsoni )†
Blackchin shiner (Notropis heterodon )

Catch (individuals ∙ 100 m )
Species % of sites
75%
25%
code
found
quantile Median quantile
lepmac
11.60
4.50
1.40
0.70
luxchr
11.60
6.40
3.30
1.30
phoeos
11.60
2.80
1.30
0.70
esoluc
10.30
1.30
0.70
0.30
micdol
9.90
3.40
1.80
0.80
percap
9.40
2.10
1.00
0.30
esoame
9.00
1.60
0.70
0.30
onckis
9.00
15.80
6.00
0.70
camano
8.60
7.10
3.60
1.20
notfla
6.40
2.50
1.10
0.60
ethfla
5.60
6.40
3.00
0.70
marmar
5.60
8.20
2.60
0.70
cotcog
5.20
21.20
12.70
4.80
pimpro
5.20
1.70
0.80
0.50
amemel
4.70
0.70
0.70
0.30
cypcar
4.70
2.00
1.30
0.40
cypspi
4.70
4.30
2.00
1.10
moxery
4.70
4.70
1.40
0.90
notcry
4.70
1.10
0.70
0.40
amenat
3.90
1.60
0.30
0.30
notheterol
3.90
4.40
1.60
0.70
lamapp
3.40
3.30
2.00
1.00
nocmic
3.40
12.40
7.50
4.00
ethexi
3.00
2.90
0.80
0.80
hybhan
3.00
1.60
0.70
0.50
notheterod
3.00
7.10
4.40
2.10

45

% of sites found by
region
UP
2.10
14.40
14.40
5.20
5.20
12.40
0.00
18.60
0.00
0.00
7.20
9.30
3.10
4.10
1.00
0.00
0.00
0.00
3.10
0.00
5.20
2.10
2.10
2.10
3.10
3.10

NLP
12.90
1.60
16.10
11.30
9.70
11.30
3.20
3.20
4.80
0.00
0.00
4.80
12.90
1.60
9.70
1.60
3.20
3.20
6.50
3.20
4.80
6.50
4.80
1.60
4.80
1.60

SLP
23.00
16.20
4.10
16.20
16.20
4.10
25.70
1.40
23.00
20.30
8.10
1.40
1.40
9.50
5.40
13.50
12.20
12.20
5.40
9.50
1.40
2.70
4.10
5.40
1.40
4.10

Table 1.2.—(cont’d)
-1

Fish species
Black crappie (Pomoxis nigromaculatus )
Walleye (Sander vitreus )
Round goby (Neogobius melanostomus )
Sand shiner (Notropis stramineus )
Chinook salmon (Oncorhynchus tshawytscha )
Brown bullhead (Ameiurus nebulosus )†
Greenside darter (Etheostoma blennioides )
Longear sunfish (Lepomis megalotis )
Emerald shiner (Notropis atherinoides )†
Mimic shiner (Notropis volucellus )
Finescale dace (Phoxinus neogaeus )†
Pirate perch (Aphredoderus sayanus )†
Creek chubsucker (Erimyzon oblongus )
Northern brook lamprey (Ichthyomyzon fossor )
Redfin shiner (Lythrurus umbratilis )
Silver redhorse (Moxostoma anisurum )
Spottail shiner (Notropis hudsonius )
Rosyface shiner (Notropis rubellus )
Bowfin (Amia calva )
Least darter (Etheostoma microperca )†
Channel catfish (Ictalurus punctatus )
Trout-perch (Percopsis omiscomaycus )
Lake chub (Couesius plumbeus )
Threespine stickleback (Gasterosteus aculeatus )
Greater redhorse (Moxostoma valenciennesi )†
Lake sturgeon (Acipenser fulvescens )†

Species
code
pomnig
sanvit
neomel
notstr
onctsh
ameneb
ethble
lepmeg
notath
notvol
phoneo
aphsay
eriobl
ichfos
lytumb
moxani
nothud
notrub
amical
ethmic
ictpun
peromi
couplu
gasacu
moxval
aciful

Catch (individuals ∙ 100 m )
% of sites
75%
25%
found
quantile Median quantile
3.00
0.80
0.60
0.30
3.00
0.30
0.30
0.20
2.60
8.10
3.30
2.60
2.60
1.00
0.70
0.70
2.60
1.40
0.90
0.50
2.10
0.80
0.40
0.30
2.10
5.90
3.80
3.70
2.10
7.80
2.50
1.00
2.10
1.30
1.30
1.20
2.10
9.50
2.60
1.10
2.10
6.60
1.10
0.70
1.70
2.30
1.10
0.70
1.70
33.70
19.00
3.40
1.70
2.80
1.00
0.60
1.70
2.60
1.90
1.50
1.70
0.80
0.40
0.30
1.70
1.30
1.00
0.80
1.70
8.90
4.40
2.00
1.30
1.30
1.30
0.80
1.30
16.10
2.00
1.10
1.30
2.00
1.30
0.90
1.30
1.60
1.10
0.70
0.90
2.00
1.80
1.50
0.90
1.60
1.10
0.70
0.90
1.30
1.00
0.70
0.40
0.20
0.20
0.20

46

% of sites found by
region
UP
0.00
5.20
0.00
2.10
0.00
1.00
0.00
0.00
1.00
0.00
3.10
0.00
0.00
0.00
0.00
1.00
1.00
0.00
0.00
1.00
0.00
3.10
1.00
1.00
0.00
1.00

NLP
1.60
1.60
1.60
4.80
8.10
0.00
0.00
0.00
1.60
1.60
3.20
0.00
1.60
3.20
1.60
1.60
3.20
1.60
0.00
0.00
0.00
0.00
1.60
0.00
0.00
0.00

SLP
8.10
1.40
6.80
1.40
1.40
5.40
6.80
6.80
4.10
5.40
0.00
5.40
4.10
2.70
4.10
2.70
1.40
4.10
4.10
2.70
4.10
0.00
0.00
1.40
2.70
0.00

Table 1.2.—(cont’d)
-1

Species
Fish species
code
Goldfish (Carassius auratus )
caraur
Warmouth (Chaenobryttus gulosus )
chagul
Redside dace (Clinostomus elongatus )
clielo
Gizzard shad (Dorosoma cepedianum )
dorcep
Lake chubsucker (Erimyzon sucetta )†
erisuc
Muskellunge (Esox masquinongy )
esomas
Bigmouth shiner (Hybopsis dorsalis )†
hybdor
Chestnut lamprey (Ichthyomyzon castaneus )
ichcas
Spotted sucker (Minytrema melanops )†
minmel
River redhorse (Moxostoma carinatum )†
moxcar
Black redhorse (Moxostoma duquesnii )
moxduq
Shorthead redhorse (Moxostoma macrolepidotum ) moxmac
Pugnose minnow (Opsopoeodus emiliae )†
opsemi
Sea lamprey (Petromyzon marinus )
petmar
Round whitefish (Prosopium cylindraceum )
procyl
Flathead catfish (Pylodictis olivaris )
pyloli

Catch (individuals ∙ 100 m )
% of sites
75%
25%
found
quantile Median quantile
0.40
0.30
0.30
0.30
0.40
1.60
1.60
1.60
0.40
2.20
2.20
2.20
0.40
1.60
1.60
1.60
0.40
1.30
1.30
1.30
0.40
0.20
0.20
0.20
0.40
0.60
0.60
0.60
0.40
4.20
4.20
4.20
0.40
2.00
2.00
2.00
0.40
0.60
0.60
0.60
0.40
3.00
3.00
3.00
0.40
0.40
0.40
0.40
0.40
3.30
3.30
3.30
0.40
10.70
10.70
10.70
0.40
0.60
0.60
0.60
0.40
0.70
0.70
0.70

47

% of sites found by
region
UP
0.00
0.00
1.00
0.00
0.00
1.00
1.00
0.00
0.00
0.00
0.00
0.00
1.00
1.00
0.00
0.00

NLP
0.00
0.00
0.00
0.00
0.00
0.00
0.00
1.60
0.00
1.60
0.00
0.00
0.00
0.00
1.60
0.00

SLP
1.40
1.40
0.00
1.40
1.40
0.00
0.00
0.00
1.40
0.00
1.40
1.40
0.00
0.00
0.00
1.40

Table 1.3.—Taxonomic summaries statewide and by regions.

Taxonomic summaries
Variable
Taxa richness
Shannon's diversity
Pielou's evenness

Taxa richness
Shannon's diversity
Pielou's evenness

Taxa richness
Shannon Weaver diversity
Pielou's evenness

48

Taxa richness
Shannon's diversity
Pielou's evenness

Statewide
mean max min
9.54 28.00 1.00
1.41 2.64 0.00
0.67 1.00 0.00
UP
mean max min
7.67 19.00 1.00
1.28 2.38 0.00
0.67 1.00 0.00
NLP
mean max min
10.47 28.00 2.00
1.41 2.40 0.12
0.64 1.00 0.09
SLP
mean max min
11.20 25.00 2.00
1.59 2.64 0.59
0.70 0.97 0.37

48

Table 1.4.—Instream temperature metric description, codes, mean, maximums and minimums
for variables at 233 study sites, for the state of Michigan and by three regions, Upper Peninsula
(UP), Northern Lower Peninsula (NLP) and Southern Lower Peninsula (SLP). Temperature
metrics marked with (*) indicate those used in CSA. Temperature units in (°C).

49

Instream temperature metric description
Monthly average daily maximum
June
July
August
Monthly average daily range
June
July
August
Maximum of 7 days moving average of daily
range
maximum
Maximum of 14 days moving average of daily
range
maximum
Maximum of 21 days moving average of daily
range
maximum
Maximum of 30 days moving average of daily*
range*
maximum*
Julian day of maximum of 7 days moving
average of daily
range
maximum
Julian day of maximum of 14 days moving
average of daily
range
maximum
Julian day of maximum of 21 days moving
average of daily
range
maximum
Julian day of maximum of 30 days moving
average of daily
range
maximum

Statewide
max
min

Variable code

mean

ADmax6
ADmax7
ADmax8

19.37 27.87
20.42 28.09
19.52 26.10

ADrange6
ADrange7
ADrange8

3.64
3.77
3.35

11.27
11.28
11.67

8.86
8.60
8.24

0.79
1.32
0.82

Max7MovingADRT
Max7MovingAMaxT

5.10 32.02
22.25 39.54

1.43
12.88

Max14MovingADRT
Max14MovingAMaxT

4.61 27.37
21.49 36.68

1.43
12.31

Max21MovingADRT
Max21MovingAMaxT

4.45 24.51
21.10 35.00

1.26
11.93

Max30MovingADRT
Max30MovingAMaxT

4.26 23.34
20.83 33.95

1.18
11.77

JDM7MADRT
JDM7MAMaxT

181.79 267.17 126.17
196.41 249.17 141.17

JDM14MADRT
JDM14MAMaxT

181.93 266.17 126.17
198.42 238.17 164.17

JDM21MADRT
JDM21MAMaxT

184.35 263.17 131.17
200.25 235.17 164.17

JDM30MADRT
JDM30MAMaxT

184.42 258.17 134.17
199.62 232.17 163.17

49

Table 1.4.—(cont’d)

50

Instream temperature metric description
Monthly average daily maximum
June
July
August
Monthly average daily range
June
July
August
Maximum of 7 days moving average of daily
range
maximum
Maximum of 14 days moving average of daily
range
maximum
Maximum of 21 days moving average of daily
range
maximum
Maximum of 30 days moving average of daily
range
maximum
Julian day of maximum of 7 days moving
average of daily
range
maximum
Julian day of maximum of 14 days moving
average of daily
range
maximum
Julian day of maximum of 21 days moving
average of daily
range
maximum
Julian day of maximum of 30 days moving
average of daily
range
maximum

UP
max

Variable code

mean

ADmax6
ADmax7
ADmax8

18.15 25.02 11.27
19.46 25.91 11.28
18.89 25.47 11.67

ADrange6
ADrange7
ADrange8

3.67
3.83
3.45

8.85
8.11
8.24

min

0.79
1.43
1.47

Max7MovingADRT
Max7MovingAMaxT

5.12 10.95
1.43
21.36 29.33 12.88

Max14MovingADRT
Max14MovingAMaxT

4.65 10.18
1.43
20.55 28.09 12.31

Max21MovingADRT
Max21MovingAMaxT

4.55
9.90
1.26
20.21 27.60 11.93

Max30MovingADRT
Max30MovingAMaxT

4.40
9.20
1.18
19.95 26.60 11.77

JDM7MADRT
JDM7MAMaxT

178.94 245.17 126.17
198.38 249.17 145.17

JDM14MADRT
JDM14MAMaxT

180.89 247.17 126.17
200.21 238.17 169.17

JDM21MADRT
JDM21MAMaxT

182.85 252.17 131.17
202.85 235.17 167.17

JDM30MADRT
JDM30MAMaxT

183.83 248.17 134.17
201.16 230.17 163.17

50

Table 1.4.—(cont’d)
Variable code

mean

ADmax6
ADmax7
ADmax8

19.15 24.53 12.38
20.05 26.16 13.24
18.71 25.34 12.25

ADrange6
ADrange7
ADrange8

3.98
4.03
3.50

8.86
8.60
7.35

min

1.44
1.55
1.15

Max7MovingADRT
Max7MovingAMaxT

5.92 32.02
2.05
22.12 39.54 14.42

Max14MovingADRT
Max14MovingAMaxT

5.33 27.37
1.94
21.30 36.68 14.09

Max21MovingADRT
Max21MovingAMaxT

5.05 24.51
1.90
20.83 35.00 13.63

Max30MovingADRT
Max30MovingAMaxT

4.82 23.34
1.82
20.59 33.95 13.41

51

Instream temperature metric description
Monthly average daily maximum
June
July
August
Monthly average daily range
June
July
August
Maximum of 7 days moving average of daily
range
maximum
Maximum of 14 days moving average of daily
range
maximum
Maximum of 21 days moving average of daily
range
maximum
Maximum of 30 days moving average of daily
range
maximum
Julian day of maximum of 7 days moving
average of daily
range
maximum
Julian day of maximum of 14 days moving
average of daily
range
maximum
Julian day of maximum of 21 days moving
average of daily
range
maximum
Julian day of maximum of 30 days moving
average of daily
range
maximum

NLP
max

JDM7MADRT
JDM7MAMaxT

176.54 267.17 136.17
189.95 229.17 141.17

JDM14MADRT
JDM14MAMaxT

176.16 265.17 136.17
193.21 226.17 164.17

JDM21MADRT
JDM21MAMaxT

179.75 262.17 137.17
195.27 223.17 164.17

JDM30MADRT
JDM30MAMaxT

178.28 256.17 135.17
194.89 219.17 169.17

51

Table 1.4.—(cont’d)

52

Instream temperature metric description
Monthly average daily maximum
June
July
August
Monthly average daily range
June
July
August
Maximum of 7 days moving average of daily
range
maximum
Maximum of 14 days moving average of daily
range
maximum
Maximum of 21 days moving average of daily
range
maximum
Maximum of 30 days moving average of daily
range
maximum
Julian day of maximum of 7 days moving
average of daily
range
maximum
Julian day of maximum of 14 days moving
average of daily
range
maximum
Julian day of maximum of 21 days moving
average of daily
range
maximum
Julian day of maximum of 30 days moving
average of daily
range
maximum

SLP
max

Variable code

mean

ADmax6
ADmax7
ADmax8

21.12 27.87 15.61
22.00 28.09 15.62
20.99 26.10 13.92

ADrange6
ADrange7
ADrange8

3.33
3.46
3.11

7.04
6.65
6.77

min

1.18
1.32
0.82

Max7MovingADRT
Max7MovingAMaxT

4.37
9.04
1.54
23.44 29.29 16.50

Max14MovingADRT
Max14MovingAMaxT

3.97
8.36
1.46
22.81 28.66 15.88

Max21MovingADRT
Max21MovingAMaxT

3.81
8.14
1.45
22.43 28.61 15.86

Max30MovingADRT
Max30MovingAMaxT

3.58
7.05
1.33
22.13 28.25 15.77

JDM7MADRT
JDM7MAMaxT

189.73 267.17 126.17
199.48 243.17 160.17

JDM14MADRT
JDM14MAMaxT

188.09 266.17 127.17
200.63 228.17 171.17

JDM21MADRT
JDM21MAMaxT

190.27 263.17 131.17
201.44 228.67 179.17

JDM30MADRT
JDM30MAMaxT

190.65 258.17 135.17
201.98 232.17 179.17

52

Table 1.5.—Principle component analysis (PCA) results for 22 instream temperature variables,
including interpretations of axes, loading assigned to each variable for each axes, and percentage
of variation in data explained by each axis. Instream temperature variable codes are described in
Table 1.3. Total amount of variation explained was 89.24%. Temperature metrics marked with
(*) indicate those used in CSA.

Interpertation

53

Variance explained (%)
Variable code
ADmax6
ADmax7
ADmax8
Max7MovingAMaxT
Max14MovingAMaxT
Max21MovingAMaxT
Max30MovingAMaxT*
ADrange6
ADrange7
ADrange8
Max7MovingADRT
Max14MovingADRT
Max21MovingADRT
Max30MovingADRT*
JDM7MADRT
JDM14MADRT
JDM21MADRT
JDM30MADRT
JDM7MAMaxT
JDM14MAMaxT
JDM21MAMaxT
JDM30MAMaxT

Axis 1
Axis 2
Axis 3
Maximum Range in Julian day range
values
values
in values
39.32
24.96
13.12
0.92
0.95
0.93
0.95
0.96
0.97
0.97
0.14
0.23
0.19
0.14
0.14
0.15
0.14
0.14
0.11
0.14
0.15
0.03
0.12
0.00
0.01

0.12
0.16
0.11
0.21
0.20
0.19
0.18
0.89
0.90
0.89
0.95
0.96
0.96
0.97
-0.08
-0.04
-0.06
-0.08
-0.11
-0.08
-0.07
-0.11

53

0.13
0.11
0.13
0.10
0.10
0.09
0.10
-0.13
0.04
0.11
-0.13
-0.09
-0.08
-0.07
0.89
0.94
0.93
0.92
0.06
0.04
0.03
0.02

Axis 4
Julian day
maximum
11.83
-0.12
0.06
0.16
0.00
0.03
0.06
0.05
-0.13
-0.08
0.03
-0.11
-0.10
-0.09
-0.09
0.03
0.06
0.02
0.06
0.78
0.90
0.92
0.89

Table 1.6.—Pearson product moment correlations between landscape variables for covariance
structure analysis (CSA). Mean July air temperature is at the local catchment scale, coarse and
fine lithology, forest land cover and catchment area are at the network catchment scale.

Coarse
Fine
Catchment Forest land Mean July air
lithology lithology
area
cover
temperature
Coarse lithology
1.00
Fine lithology
-0.91
1.00
Catchment area
0.16
-0.19
1.00
Forest land cover
0.19
-0.21
-0.15
1.00
Mean July air
temperature
-0.06
0.24
0.13
-0.75
1.00

54

Table 1.7.—ISA values for species associations with instream temperature variables, Statewide
(e.g., ISA value > 0.65) and 3 Michigan regions (e.g., ISA value>0.70). Statewide; n=233, Upper
Peninsula; n=97, Northern Lower Peninsula; n=62, Southern Lower Peninsula; n=74. Species
codes are described in Table 1.2 and instream temperature variable codes are described in Table
3 with .h and .l depicting high and low species associations with the variable, respectively.

Instream temperature
Statewide
variables
catcom sematr rhiatr salfon umblimethnig
ADmax6.h
0.70 0.75
0.66
ADmax6.l
0.70
ADmax7.h
0.77 0.79
0.70
ADmax7.l
0.75
ADmax8.h
0.74 0.78
0.71
ADmax8.l
0.75
Max7MovingAMaxT.h
0.78 0.79
0.71
Max7MovingAMaxT.l
0.74
Max14MovingAMaxT.h
0.76 0.77
0.70
Max14MovingAMaxT.l
0.75
Max21MovingAMaxT.h
0.77 0.78
0.69
Max21MovingAMaxT.l
0.76
Max30MovingAMaxT.h
0.77 0.78
0.70
Max30MovingAMaxT.l
0.75
ADrange6.h
0.67
ADrange7.h
0.74
ADrange8.h
0.73
Max7MovingADRT.h
0.74 0.67
Max14MovingADRT.h
0.73
Max21MovingADRT.h
0.74
Max21MovingADRT.l
0.66
Max30MovingADRT.h
0.77
JDM7MADRT.h
0.66
0.66

55

Table 1.7.—(cont’d)

Instream temperature
variables
ADmax6.h
ADmax7.h
ADmax7.l
ADmax8.h
ADmax8.l
Max7MovingAMaxT.h
Max7MovingAMaxT.l
Max14MovingAMaxT.h
Max14MovingAMaxT.l
Max21MovingAMaxT.h
Max21MovingAMaxT.l
Max30MovingAMaxT.h
Max30MovingAMaxT.l
ADrange7.h
ADrange8.h
Max7MovingADRT.h
Max7MovingADRT.l
Max14MovingADRT.h
Max14MovingADRT.l
Max21MovingADRT.h
Max21MovingADRT.l
Max30MovingADRT.h
Max30MovingADRT.l
JDM14MAMaxT.l
JDM21MAMaxT.l

Upper Peninsula
catcomsematr rhiatr salfon rhicat
0.75 0.72 0.72
0.80 0.82 0.77
0.77
0.84 0.81 0.77
0.78
0.86 0.83 0.81
0.78
0.81 0.80 0.79
0.78
0.84 0.83 0.80
0.79
0.84 0.84 0.78
0.70
0.73
0.76 0.74
0.72
0.76 0.74
0.74
0.76 0.70
0.73
0.78 0.71
0.74
0.77 0.72
0.72
0.70
0.74
0.71

56

Table 1.7.—(cont’d)
Instream temperature
Northern Lower Peninsula
variables
catcom sematr rhiatr salfon umblim ethnig saltru luxcor ambrup
ADmax6.h
0.76
0.74
ADmax7.h
0.79
0.73
0.71
ADmax7.l
0.76
ADmax8.h
0.73
0.73
0.75
0.77
ADmax8.l
0.71
0.74
Max7MovingAMaxT.h
0.74
Max21MovingAMaxT.h
0.74
0.70
Max21MovingAMaxT.l
0.73
Max30MovingAMaxT.h
0.74
0.70
Max30MovingAMaxT.l
0.73
ADrange7.l
0.71
ADrange8.h
0.73
ADrange8.l
0.75
Max7MovingADRT.l
0.71
Max14MovingADRT.l
0.80
Max30MovingADRT.l
0.80
JDM7MADRT.l
0.74
0.79
0.71
JDM14MADRT.h
0.78
JDM14MADRT.l
0.75
0.85
0.77
JDM21MADRT.l
0.76
JDM30MADRT.h
0.72
JDM30MADRT.l
0.71
JDM7MAMaxT.l
0.73
JDM30MAMaxT.l
0.74

57

Table 1.7.—(cont’d)

Instream temperature
variables
ADmax6.h
ADmax6.l
ADmax7.h
ADmax7.l
ADmax8.h
ADmax8.l
Max7MovingAMaxT.h
Max7MovingAMaxT.l
Max14MovingAMaxT.h
Max14MovingAMaxT.l
Max21MovingAMaxT.h
Max21MovingAMaxT.l
Max30MovingAMaxT.h
Max30MovingAMaxT.l
ADrange6.h
ADrange7.h
Max14MovingADRT.h
Max21MovingADRT.h
JDM21MADRT.l
JDM30MADRT.l

Southern Lower Peninsula
catcom sematr umblim saltru luxcor ambrup cotbai pimnot
0.70
0.70
0.74
0.72
0.72
0.76
0.77
0.70
0.72
0.71
0.79
0.76
0.79
0.81
0.71
0.80
0.78
0.73
0.80
0.70
0.78
0.73
0.76
0.71
0.70
0.74
0.71
0.72
0.75

58

Table 1.8.—Fit statistics for CSA models predicting fish descriptor metrics directly for
landscape factors and indirectly through metrics of a streams thermal regime. Fit equals yes if X2
prob > 0.05, RMSEA ≤ 0.05, TLI > 0.9, and NFI > 0.9.

Category

Variable

Summer maximum in stream temperature
X

Taxonomic summaries
Taxa richness
Shannon diversity
Pielou evenness
Fish individual species
White Sucker
Creek Chub
E. Blacknose Dace
Brook Trout
Central Mudminnow
Mottled Sculpin
Johnny Darter
Brown Trout
Longnose Dace
Common Shiner
Rock Bass
Bluntnose Minnow
Taxonomic summaries
Taxa richness
Shannon diversity
Pielou evenness
Fish individual species
White Sucker
Creek Chub
E. Blacknose Dace
Brook Trout
Central Mudminnow
Mottled Sculpin
Johnny Darter
Brown Trout
Longnose Dace
Common Shiner
Rock Bass
Bluntnose Minnow

2

df

2

prob RMSEA X /df

16.43 7.00 NSF
16.25 7.00 NSF
10.96 7.00 0.14
11.88
7.49
7.81
12.46
9.55
7.38
11.64
9.28
22.06
7.17
14.37
8.41

NSF
NSF
0.05

TLI

NFI

Fit

2.35 0.97 0.99
2.32 0.97 0.99
1.57 0.99 0.99

no
no
yes

7.00 0.10
NSF 1.70 0.98 0.99
7.00 0.38
0.02 1.07 1.00 0.99
7.00 0.35
0.02 1.12 1.00 0.99
7.00 NSF
NSF 1.78 0.98 0.99
7.00 0.22
0.04 1.37 0.99 0.99
7.00 0.39
0.02 1.06 1.00 0.99
7.00 0.11
NSF 1.66 0.98 0.99
7.00 0.23
0.04 1.33 0.99 0.99
7.00 NSF
NSF 3.15 0.95 0.98
7.00 0.41
0.01 1.03 1.00 0.99
7.00 NSF
NSF 2.05 0.98 0.99
7.00 0.30
0.03 1.20 1.00 0.99
Summer range in stream temperature

no
yes
yes
no
yes
yes
no
yes
no
yes
no
yes

28.13 7.00 NSF
20.51 7.00 NSF
9.25 7.00 0.24

NSF
NSF
0.04

4.02 0.93 0.98
2.93 0.95 0.98
1.32 0.99 0.99

no
no
yes

15.86
9.98
6.08
14.61
11.54
8.95
10.69
10.17
15.75
9.48
21.54
12.14

NSF
0.04
0.00
NSF
NSF
0.04
0.05
0.04
NSF
0.04
NSF
NSF

2.27
1.43
0.87
2.09
1.65
1.28
1.53
1.45
2.25
1.35
3.08
1.74

no
yes
yes
no
no
yes
yes
yes
no
yes
no
no

7.00
7.00
7.00
7.00
7.00
7.00
7.00
7.00
7.00
7.00
7.00
7.00

59

NSF
0.19
0.53
NSF
0.12
0.26
0.15
0.18
NSF
0.22
NSF
NSF

0.97
0.99
1.00
0.97
0.98
0.99
0.99
0.99
0.97
0.99
0.95
0.98

0.99
0.99
1.00
0.99
0.99
0.99
0.99
0.99
0.99
0.99
0.98
0.99

Table 1.9.—Standardized total effects of landscape variable generated through CSA directly and
indirectly on individual fish descriptors and indirectly through metrics of the thermal regime
(e.g., MaxT = the Maximum of 30 days moving average of daily maximum in stream
temperature and ADRT= Maximum of 30 days moving average of daily range is stream
temperature). Landscape variables are: Regions= three study regions in Michigan (Figure 1),
percent of coarse or fine lithology in the network catchment, network catchment area, percent of
forested land cover in the network catchment, and mean July air temperature summarized at the
network scale. Variables without values were models that had non–significant fit statistics.

Variable
Taxonomic summaries
Pielou evenness
Fish individual species
Creek Chub
E. Blacknose Dace
Central Mudminnow
Mottled Sculpin
Brown Trout
Common Shiner
Bluntnose Minnow

R2
0.02

0.04

-0.01

0.10 0.30
0.03 0.01
0.02 0.01
0.02 -0.14
0.09 -0.24
0.02 0.14
0.06 0.17

-0.04
0.00
-0.01
0.02
0.03
-0.02
-0.03

R
Taxonomic summaries
Pielou evenness
Fish individual species
Creek Chub
E. Blacknose Dace
Mottled Sculpin
Johnny Darter
Brown Trout
Common Shiner

Standardized total effects
Coarse
Fine
Catchment Forest land Mean
MaxT Region lithology lithology
area
cover
July air

2

-0.09

0.05

0.01

-0.01

0.01

0.01
0.17
0.00
-0.08
0.04
0.11
0.21
0.15
0.00
0.00
-0.01
-0.05
-0.11
-0.03
0.01
-0.01
0.00
-0.03
0.04 -0.02
0.30
0.18
0.13
0.06 -0.03
0.08
0.02
-0.01
-0.04
0.02
0.10
0.16
0.19
-0.05
0.02
Coarse
Fine
Catchment Forest land Mean
ADRT Region lithology lithology
area
cover
July air

0.05 -0.17

-0.04

-0.12

0.01

0.00

-0.02

0.07

0.08 0.21
0.07 0.19
0.00 -0.02
0.04 0.10
0.05 0.03
0.01 0.09

0.05
0.05
-0.01
0.03
0.01
0.02

0.05
0.14
-0.01
0.15
0.30
0.10

0.25
0.25
-0.02
0.27
0.16
0.06

0.03
0.16
-0.04
0.14
0.12
0.00

0.02
0.02
0.00
0.01
0.00
0.01

-0.09
-0.08
0.01
-0.05
-0.01
-0.04

60

Table 1.10.—Standardized total effects coefficients of landscape variables on stream
temperature metrics were analyzed for statistical significance (t distribution, α = 0.05) using
Monte Carlo bootstrapped SEs. Table includes squared multiple correlation coefficients (R2)
describing variance explained in stream temperature metrics. MaxT = the Maximum of 30 days
moving average of daily maximum in stream temperature and ADRT= Maximum of 30 days
moving average of daily range is stream temperature.

Variable
2

R
Standardized total effects from
Region
Coarse lithology
Fine lithology
Catchment area
Forest land cover
Mean July air temperature

MaxT

ADRT

0.23

0.08

-0.15
0.02
0.20
0.35
-0.27
0.14

0.26
0.10
0.07
0.04
0.09
-0.43

61

APPENDIX 1.B

FIGURES

Figure 1.1.—Locations of study sites in Michigan having both fish community and instream
temperature data collected from 1990 to 2011 (n= 233).

62

Figure 1.2.—Model developed for covariance structure analysis to evaluate hypothesized
relationships among landscape variables and fish assemblage descriptors, with landscape effects
modeled directly and indirectly through stream temperature variables (straight arrows), based on
theorized influences. Curved arrows among landscape variables indicate correlations supported
by Pearson pairwise correlations (Table 1.5).

63

APPENDIX 1.C

SUPPLEMENTAL TABLES

Table C1.1.—List of instream temperature metric descriptions calculated for 233 sites with matching temperature and fish
assemblage data collected from study reaches. Averages (mean), maximums (max) and minimums (min) for each temperature
variable, statewide and for each of three regions of Michigan, Upper Peninsula (UP), Northern Lower Peninsula (NLP), and
Southern Lower Peninsula (SLP).

Instream temperature metric description
Monthly average daily maximum
May
June
July
August
September
Monthly average daily mean
May
June
July
August
September
Monthly average daily minimum
May
June
July
August
September

Temperature code

Statewide
mean max
min

mean

UP
max

min

3 Michigan regions
NLP
mean max
min

mean

SLP
max
min

ADmax5
ADmax6
ADmax7
ADmax8
ADmax9

15.32
19.37
20.42
19.52
16.45

21.33
27.87
28.09
26.10
23.89

9.40
11.27
11.28
11.67
10.90

14.46
18.15
19.46
18.89
15.27

20.33
25.02
25.91
25.47
19.94

9.40
11.27
11.28
11.67
10.90

15.23
19.15
20.05
18.71
15.67

21.33
24.53
26.16
25.34
19.82

10.88
12.38
13.24
12.25
12.35

16.30
21.12
22.00
20.99
18.38

21.09
27.87
28.09
26.10
23.89

12.68
15.61
15.62
13.92
13.86

ADmean5
ADmean6
ADmean7
ADmean8
ADmean9

13.24
17.43
18.41
17.73
14.86

19.00 8.41
26.35 9.70
25.96 10.16
24.68 10.50
21.42 9.32

12.28
16.21
17.41
17.05
13.79

16.07 8.41
23.38 9.70
24.14 10.16
22.31 10.50
18.99 9.32

13.08
17.03
17.90
16.85
13.98

19.00
22.10
23.91
23.27
18.06

9.85
11.19
12.23
11.39
11.40

14.40
19.35
20.15
19.33
16.74

18.39
26.35
25.96
24.68
21.42

10.76
13.84
14.03
12.91
12.79

ADmin5
ADmin6
ADmin7
ADmin8
ADmin9

11.42
15.72
16.66
16.17
13.44

16.64
24.92
24.11
23.46
20.49

10.40
14.49
15.63
15.45
12.48

14.62
22.00
22.70
21.29
18.16

11.13
15.17
16.02
15.21
12.46

16.19 8.59
20.14 9.92
22.34 11.11
21.79 10.38
17.14 10.30

12.77
17.79
18.54
17.88
15.28

16.64
24.92
24.11
23.46
20.49

9.30
11.97
12.79
12.08
11.42

64

6.49
8.42
9.11
9.08
7.96

6.49
8.42
9.11
9.08
7.96

Table C1.1.—(cont’d)

Instream temperature metric description
Monthly average daily range
May
June
July
August
September
Monthly range of means
May
June
July
August
September
Average daily mean
Spring
Summer
Maximum daily mean
Spring
Summer
Maximum of 1 days moving average of daily
range
maximum
mean
Maximum of 3 days moving average of daily
range
maximum
mean
Maximum of 7 days moving average of daily
range
maximum
mean

Temperature code

Statewide
mean max
min

mean

UP
max

min

3 Michigan regions
NLP
mean max
min

mean

SLP
max
min

ADrange5
ADrange6
ADrange7
ADrange8
ADrange9

3.90
3.64
3.77
3.35
3.00

7.89
8.86
8.60
8.24
6.87

1.49
0.79
1.32
0.82
1.06

4.06
3.67
3.83
3.45
2.79

7.89
8.85
8.11
8.24
5.78

1.49
0.79
1.43
1.47
1.06

4.09
3.98
4.03
3.50
3.21

7.63
8.86
8.60
7.35
5.92

1.60
1.44
1.55
1.15
1.28

3.53
3.33
3.46
3.11
3.10

6.65
7.04
6.65
6.77
6.87

2.24
1.18
1.32
0.82
1.45

Rmean5
Rmean6
Rmean7
Rmean8
Rmean9

8.75
6.64
6.16
6.28
7.71

17.27
15.22
15.07
14.88
14.64

0.27
0.32
1.61
2.60
0.65

10.21
7.00
6.24
6.41
8.06

17.27
15.22
15.07
14.82
14.64

3.91
2.25
2.51
2.60
0.65

7.22
7.59
6.36
6.35
7.53

14.11
14.27
12.36
13.34
12.33

0.27
0.66
2.75
3.32
2.78

8.87
5.37
5.89
6.06
7.44

14.97
12.10
11.96
14.88
12.50

5.72
0.32
1.61
2.77
3.14

SpADmean
SuADmean

13.24 19.00
8.41
17.85 25.41 10.16

12.28 16.07 8.41
16.91 23.27 10.16

13.08 19.00 9.85
17.25 22.77 11.61

14.40 18.39 10.76
19.59 25.41 13.59

SpMDmean
SuMDmean

17.93 23.93 11.01
21.87 29.44 12.37

17.84 23.91 11.01
20.84 28.03 12.37

17.11 23.93 11.80
21.62 28.23 14.24

18.91 23.52 13.26
23.43 29.44 16.38

Max1MovingADRT
Max1MovingAMaxT
Max1MovingAMeanT

6.65 37.04
1.50
23.87 44.38 14.14
21.48 29.44 12.37

6.95 17.53 1.50
23.11 33.03 14.14
20.62 28.03 12.37

7.40 37.04 3.13
23.75 44.38 15.39
20.93 28.23 14.24

5.66 11.84 2.00
24.90 31.97 17.83
22.98 29.44 16.38

Max3MovingADRT
Max3MovingAMaxT
Max3MovingAMeanT

5.74 35.49
1.50
23.13 41.88 13.31
20.91 28.21 12.09

5.80 13.49 1.50
22.27 29.97 13.31
20.01 27.79 12.09

6.67 35.49 2.32
23.04 41.88 15.04
20.41 27.15 13.60

4.88 10.18 1.67
24.24 31.06 17.08
22.41 28.21 15.64

Max7MovingADRT
Max7MovingAMaxT
Max7MovingAMeanT

5.10 32.02
1.43
22.25 39.54 12.88
20.12 27.65 11.60

5.12 10.95 1.43
21.36 29.33 12.88
19.22 27.65 11.60

5.92 32.02 2.05
22.12 39.54 14.42
19.61 26.54 13.25

4.37 9.04 1.54
23.44 29.29 16.50
21.66 27.28 15.07

65

Table C1.1.—(cont’d)

Instream temperature metric description
Maximum of 14 days moving average of daily
range
maximum
mean
Maximum of 21 days moving average of daily
range
maximum
mean
Maximum of 30 days moving average of daily
range
maximum
mean
Julian day of maximum of 1 days moving
average of daily
range
maximum
mean
Julian day of maximum of 3 days moving
average of daily
range
maximum
mean

Temperature code

Statewide
mean max
min

mean

UP
max

min

3 Michigan regions
NLP
mean max
min

mean

SLP
max
min

Max14MovingADRT
Max14MovingAMaxT
Max14MovingAMeanT

4.61 27.37
1.43
21.49 36.68 12.31
19.48 26.78 10.96

4.65 10.18 1.43
20.55 28.09 12.31
18.53 26.35 10.96

5.33 27.37 1.94
21.30 36.68 14.09
18.94 25.61 12.98

3.97 8.36 1.46
22.81 28.66 15.88
21.09 26.78 14.40

Max21MovingADRT
Max21MovingAMaxT
Max21MovingAMeanT

4.45 24.51
1.26
21.10 35.00 11.93
19.10 26.32 10.89

4.55 9.90 1.26
20.21 27.60 11.93
18.19 25.41 10.89

5.05 24.51 1.90
20.83 35.00 13.63
18.56 25.13 12.58

3.81 8.14 1.45
22.43 28.61 15.86
20.70 26.32 14.20

Max30MovingADRT
Max30MovingAMaxT
Max30MovingAMeanT

4.26 23.34
1.18
20.83 33.95 11.77
18.85 26.14 10.71

4.40 9.20 1.18
19.95 26.60 11.77
17.93 24.84 10.71

4.82 23.34 1.82
20.59 33.95 13.41
18.32 24.87 12.40

3.58 7.05 1.33
22.13 28.25 15.77
20.45 26.14 14.20

JDM1MADRT
JDM1MAMaxT
JDM1MAMeanT

180.08 272.17 123.17
193.10 251.17 139.17
195.94 252.17 147.17

180.52 272.17 125.17
193.35 251.17 139.17
198.30 252.17 147.17

174.95 270.17 126.17
185.43 218.17 141.17
188.13 218.17 158.17

183.83 270.17 123.17
199.20 248.17 159.17
199.59 244.17 160.17

JDM3MADRT
JDM3MAMaxT
JDM3MAMeanT

180.99 269.17 124.17
193.99 251.17 127.17
198.50 251.17 146.17

180.67 250.17 126.17
196.44 251.17 143.17
201.67 251.17 146.17

177.13 269.17 134.17
186.54 219.17 127.17
190.72 226.17 159.17

184.59 265.17 124.17
197.28 244.17 158.17
201.17 244.17 159.17

66

Table C1.1.—(cont’d)

Instream temperature metric description
Julian day of maximum of 7 days moving
average of daily
range
maximum
mean
Julian day of maximum of 14 days moving
average of daily
range
maximum
mean
Julian day of maximum of 21 days moving
average of daily
range
maximum
mean
Julian day of maximum of 30 days moving
average of daily
range
maximum
mean
julian day of maximum
of maximum temperature of the summer
of average temperature of the summer
Julian day of maximum of 7 days daily
maximum
mean

3 Michigan regions
NLP
mean max
min

mean

178.94 245.17 126.17
198.38 249.17 145.17
202.25 249.17 145.17

176.54 267.17 136.17
189.95 229.17 141.17
195.36 229.17 159.17

189.73 267.17 126.17
199.48 243.17 160.17
201.51 243.17 171.17

181.93 266.17 126.17
198.42 238.17 164.17
200.51 240.17 145.17

180.89 247.17 126.17
200.21 238.17 169.17
200.75 225.17 145.17

176.16 265.17 136.17
193.21 226.17 164.17
197.59 226.17 164.17

188.09 266.17 127.17
200.63 228.17 171.17
202.68 240.17 176.17

JDM21MADRT
JDM21MAMaxT
JDM21MAMeanT

184.35 263.17 131.17
200.25 235.17 164.17
202.81 232.17 165.17

182.85 252.17 131.17
202.85 235.17 167.17
205.18 230.17 168.17

179.75 262.17 137.17
195.27 223.17 164.17
200.05 232.17 165.17

190.27 263.17 131.17
201.44 228.67 179.17
202.33 229.17 181.17

JDM30MADRT
JDM30MAMaxT
JDM30MAMeanT

184.42 258.17 134.17
199.62 232.17 163.17
201.74 231.17 163.17

183.83 248.17 134.17
201.16 230.17 163.17
202.56 230.17 163.17

178.28 256.17 135.17
194.89 219.17 169.17
198.65 219.17 167.17

190.65 258.17 135.17
201.98 232.17 179.17
203.50 231.17 185.17

JDmaxTmax
JDmaxTmean

194.10 251.17 139.17
196.57 252.17 159.17

195.73 251.17 139.17
199.30 252.17 159.17

187.73 218.17 151.17
189.85 218.17 159.17

197.29 244.17 159.17
198.62 244.17 160.17

JDMM7MaxT
JDMM7MeanT

196.83 249.17 145.17
200.24 249.17 145.17

198.40 249.17 145.17
202.22 249.17 145.17

190.97 229.17 159.17
195.89 229.17 159.17

199.89 243.17 160.17
201.49 243.17 171.17

Temperature code

Statewide
mean max
min

mean

JDM7MADRT
JDM7MAMaxT
JDM7MAMeanT

181.79 267.17 126.17
196.41 249.17 141.17
200.11 249.17 145.17

JDM14MADRT
JDM14MAMaxT
JDM14MAMeanT

67

UP
max

min

SLP
max
min

Table C1.2.—List of sites with temperature and fish assemblage data collected from study
reaches. Each site’s stream name, reach code, and region code in which it is located are listed,
along with the latitude and longitude of the point marking the upstream end of the sample reach.

Reach code
12926133
13047663
6841567
6860186
13204407
12918875
14443608
6789873
6860332
6841681
6789339
12025464
9019265
11940357
6841611
12255214
11937039
11962270
12942320
12021724
6789659
6789337
12255574
6848077
12261932
6857446
11937459
11945791
11930744
13009178

Stream name
Bird Creek
Bullock Creek
Cassidy Creek
Cedar River
Cheboyganing Creek
Cherry Creek
Cook's Run
Cranberry River
Crooked Creek
DeHaas Creek
Duck Creek
E Br Salmon Trout River
Eastman Creek
Flintsteel River
Harter Creek
Hayden Creek
Hill Creek
Hunter's Brook
Johnson Creek
Laughing Whitefish River
Little Carp River
Little Iron River
Little Portage Creek
Little River
Mill Creek
N Br Ford River
Obenhoff Creek
Onion Creek
Perch River
Perry Creek

Year Latitude Longitude Region code
2010
43.97
-82.97
SLP
2010
43.53
-84.36
SLP
2010
45.86
-87.84
UP
2010
45.78
-87.53
UP
2010
43.44
-83.76
SLP
2010
43.52
-82.64
SLP
2010
46.17
-88.88
UP
2010
46.70
-89.45
UP
2010
45.60
-87.50
UP
2010
45.60
-87.73
UP
2010
46.83
-89.47
UP
2010
46.79
-87.85
UP
2010
42.38
-86.15
SLP
2010
46.75
-89.13
UP
2010
45.71
-87.80
UP
2010
42.26
-85.83
SLP
2010
47.31
-88.43
UP
2010
45.95
-87.24
UP
2010
44.20
-83.78
NLP
2010
46.39
-87.07
UP
2010
46.73
-89.87
UP
2010
46.83
-89.59
UP
2010
42.04
-85.43
SLP
2010
45.32
-87.58
UP
2010
41.87
-85.74
SLP
2010
46.15
-87.75
UP
2010
47.11
-88.74
UP
2010
46.57
-88.95
UP
2010
46.51
-88.64
UP
2010
43.30
-83.64
SLP

68

Table C1.2.—(cont’d)

Reach code
12927361
6841565
12121180
12918323
9017617
12025490
12927171
11962262
11930120
11951379
11939211
12258102
11952109
11946929
11951377
12936622
12927975
12953314
8991910
12498902
11959836
12498830
11959562
12501809
11937883
12203110
13175771
12221810
12501827
12027614

Stream name
Pigeon River
Pine Creek
Pine River
Rock Falls Creek
S Br Macatawa River
Salmon Trout River
Shebeon Creek
Silver Creek
Silver River
Sission Lilly
Sleepy River
Tallahassee Creek
Tenderfoot Creek
Tenmile Creek
Trout Creek
White Feather Creek
Wiscoggin Drain
Au Sable River
Big S Br Pere Marquette River
Black River
Bryan Creek
Canada Creek
E Br Escanaba River
E Br Maple River
Gratiot River
Haynes Creek
Huron River
Little Indian River
Maple River
Menge Creek

Year Latitude
2010
43.74
2010
45.83
2010
44.19
2010
43.81
2010
42.74
2010
46.78
2010
43.77
2010
45.88
2010
46.76
2010
46.47
2010
46.86
2010
41.83
2010
46.34
2010
46.53
2010
46.46
2010
43.90
2010
43.59
2009
44.68
2009
43.82
2009
45.08
2009
46.19
2009
45.24
2009
46.29
2009
45.55
2009
47.34
2009
44.71
2009
42.13
2009
46.20
2009
45.53
2009
46.73

69

Longitude Region code
-83.16
SLP
-87.90
UP
-85.77
NLP
-82.70
SLP
-86.04
SLP
-87.88
UP
-83.35
SLP
-87.10
UP
-88.70
UP
-89.41
UP
-88.92
UP
-84.93
SLP
-89.48
UP
-89.43
UP
-89.03
UP
-83.97
SLP
-83.47
SLP
-84.58
NLP
-86.08
NLP
-84.46
NLP
-87.57
UP
-84.23
NLP
-87.43
UP
-84.75
NLP
-88.35
UP
-83.35
NLP
-83.36
SLP
-86.53
UP
-84.77
NLP
-88.50
UP

Table C1.2.—(cont’d)

Reach code
11951613
12953268
13046755
8990046
12503387
12503419
12027134
12021146
12953068
12952968
13030993
12017474
12017476
12017482
11930116
12953512
6859854
13027941
12214737
12015636
11959888
12961971
10850156
12202316
12952808
12503351
12014046
12944838
12021166
12188951

Stream name
Mid Br Ontonagon River
N Br Au Sable River
N Br Tobacco River
Pere Marquette River
Pigeon River
Pigeon River
Ravine River
Rock River
S Br Au Sable River
S Br Au Sable River
Shiawassee River
Sucker River
Sucker River
Two Hearted River
W Br Sturgeon River
Au Sable River
Bark River
Cedar River
Davenport Creek
Days River
Hunters Brook
Mcginn Creek
Mid Br Rouge River
Ocqueoc River
Perry Creek
Pigeon River
Rapid River
Rifle River
Silver Creek
Silver Creek

Year Latitude
2009
46.28
2009
44.83
2009
43.96
2009
43.86
2009
45.27
2009
45.15
2009
46.82
2009
46.39
2009
44.54
2009
44.62
2009
42.82
2009
46.67
2009
46.60
2009
46.61
2009
46.76
2008
44.66
2008
45.63
2008
43.43
2008
46.07
2008
45.93
2008
46.02
2008
44.84
2008
42.33
2008
45.46
2008
44.66
2008
45.40
2008
46.04
2008
44.31
2008
46.40
2008
46.40

70

Longitude Region code
-89.24
UP
-84.49
NLP
-84.71
SLP
-85.86
NLP
-84.46
NLP
-84.47
NLP
-88.19
UP
-86.92
UP
-84.55
NLP
-84.45
NLP
-83.73
SLP
-85.90
UP
-85.93
UP
-85.60
UP
-88.74
UP
-84.73
NLP
-87.26
UP
-84.11
SLP
-85.25
UP
-87.05
UP
-87.41
UP
-83.77
NLP
-83.25
SLP
-84.09
NLP
-84.08
NLP
-84.53
NLP
-87.00
UP
-84.07
NLP
-87.00
UP
-85.63
UP

Table C1.2.—(cont’d)

Reach code
11962022
12188963
12017658
12953622
12014072
12014514
12953302
12021200
12206528
12206346
12183907
11948631
12206284
12184125
13228138
12183703
11931042
12185431
12502977
9004125
12953358
6790993
6790205
13007340
13039000
11951427
11930334
12222166
11937723
12231450

Stream name
Squaw Creek
Syphon Creek
Two Hearted River
W Br Big Creek
W Br Whitefish River
W Br Whitefish River
Au Sable River
Au Train River
Bear Creek
Carp River
Clear Creek
E Br Ontonagon River
Elmhurst Creek
Grant Creek
Iron Creek
Naomikong Creek
Rock River
Roxbury Creek
Sturgeon River
Bear Creek
Beaver Creek
Black River
Black River
Cass River
Chippewa River
Cisco Br Ontonagon River
Clear Creek
Creighton River
Elm River
Fish Creek

Year Latitude
2008
45.97
2008
46.39
2008
46.58
2008
44.63
2008
46.10
2008
46.22
2007
44.78
2007
46.38
2007
46.19
2007
46.06
2007
46.33
2007
46.49
2007
46.12
2007
46.47
2007
42.09
2007
46.46
2007
46.57
2007
46.48
2007
45.28
2006
43.06
2006
44.53
2006
46.51
2006
46.48
2006
43.51
2006
43.59
2006
46.41
2006
46.71
2006
46.35
2006
47.03
2006
43.28

71

Longitude Region code
-87.22
UP
-85.75
UP
-85.65
UP
-84.27
NLP
-86.90
UP
-87.04
UP
-84.76
NLP
-86.84
UP
-84.75
UP
-84.90
UP
-84.70
UP
-88.89
UP
-84.64
UP
-84.87
UP
-84.01
SLP
-84.99
UP
-88.49
UP
-85.01
UP
-84.60
NLP
-85.47
SLP
-84.65
NLP
-90.07
UP
-90.00
UP
-83.34
SLP
-84.90
SLP
-89.34
UP
-88.65
UP
-86.28
UP
-88.86
UP
-84.97
SLP

Table C1.2.—(cont’d)

Reach code
12944334
14443726
12021288
13038808
3470497
12021058
3472663
11946997
12954072
11930214
13047167
14443816
12026122
13184584
13045491
11930690
12503375
13182894
10849642
13047477
13184576
13040936
13184448
12955128
12189217
12026264
13045425
13039110
12021738
12961811

Stream name
Gamble Creek
Iron River
Mosquito River
N Br Chippewa River
Rice Creek
Rock River
Spring Brook
Trout Creek
Van Etten River
W Br Sturgeon River
Bailey Creek
Brule River
Carp River
Clinton River
E Br Tittabawassee River
Kelsey Creek
Little Pigeon River
Mid Br Clinton River
Mid Br Rouge River
N Br Salt River
Paint Creek
Pine River
Red Run
S Br Au Sable River
Tahquamenon River
Bismark Creek
Brick Creek
Cedar Creek
Chocolay River
Cole Creek

Year Latitude
2006
44.41
2006
46.09
2006
46.49
2006
43.71
2006
42.29
2006
46.45
2006
42.37
2006
46.52
2006
44.43
2006
46.72
2005
43.86
2005
46.01
2005
46.52
2005
42.63
2005
44.18
2005
46.56
2005
45.28
2005
42.64
2005
42.37
2005
43.73
2005
42.69
2005
43.48
2005
42.54
2005
44.46
2005
46.37
2004
46.63
2004
44.22
2004
43.56
2004
46.38
2004
44.89

72

Longitude Region code
-84.03
NLP
-88.64
UP
-86.47
UP
-84.86
SLP
-84.91
SLP
-86.92
UP
-85.53
SLP
-89.01
UP
-83.34
NLP
-88.81
UP
-84.63
SLP
-88.58
UP
-87.58
UP
-83.22
SLP
-84.23
SLP
-88.58
UP
-84.50
NLP
-82.93
SLP
-83.45
SLP
-84.67
SLP
-83.14
SLP
-84.53
SLP
-83.01
SLP
-84.44
NLP
-85.42
UP
-87.51
UP
-84.25
SLP
-84.90
SLP
-87.26
UP
-83.96
NLP

Table C1.2.—(cont’d)

Reach code
13045431
12211124
13038992
11951401
13045511
13015549
13007330
13040090
12117876
9005547
9005507
12264790
11930106
13019901
12961725
13045501
3471093
13031047
12502939
12026202
12134742
13057812
8990078
13006734
3470695
12951474
13226892
13031417
13057846
12944870

Stream name
Cooks Creek
E Br Munuscong River
Johnson Creek
Marshall Creek
Mid Br Tittabawassee River
Misteguay Creek
Mud Creek
N Br Chippewa River
N Br Manistee River
Norris Creek
Norris Creek
Pokagon Creek
Sturgeon River
Thread Creek
Turtle Creek
W Br Tittabawassee River
Kalamazoo River
Shiawassee River
W Br Minnehaha Creek
Yellow Dog River
Bigelow Creek
Boardman River
Little S Br Pere Marquette River
Mid Br Cass River
Rice Creek
S Br Pine River
Saline River
Shiawassee River
Swainston Creek
W Br Rifle River

Year Latitude
2004
44.22
2004
46.11
2004
43.57
2004
46.41
2004
44.18
2004
43.16
2004
43.51
2004
43.78
2004
44.64
2004
43.13
2004
43.18
2004
41.91
2004
46.78
2004
42.87
2004
44.91
2004
44.18
2003
42.26
2003
42.80
2003
45.35
2003
46.76
2002
43.45
2002
44.67
2002
43.79
2002
43.55
2002
42.06
2002
44.57
2002
42.00
2002
42.81
2002
44.64
2002
44.25

73

Longitude Region code
-84.22
SLP
-84.24
UP
-84.84
SLP
-89.57
UP
-84.28
SLP
-83.94
SLP
-83.16
SLP
-85.13
SLP
-85.03
NLP
-86.12
SLP
-86.05
SLP
-86.20
SLP
-88.62
UP
-83.54
SLP
-83.88
NLP
-84.41
SLP
-84.96
SLP
-83.65
SLP
-84.83
NLP
-87.66
UP
-85.75
NLP
-85.40
NLP
-85.76
NLP
-83.05
SLP
-84.61
SLP
-83.51
NLP
-83.62
SLP
-83.84
SLP
-85.54
NLP
-84.15
NLP

Table C1.2.—(cont’d)

Reach code
12951286
12183695
12021302
13057824
13045737
12130638
12214753
13030899
12951194
13031255
10849532
13045751
12951060
12014040
6790865
12132510
12944354
9001717
9004341
12944400
12242480
12944882
12132402
12132432
12132378
12132498
12241372
12255996
8991986
8991416

Stream name
Wallace Creek
Ankodosh Creek
Anna River
Boardman River
Cedar River
Clam River
Crow River
Jones Creek
McDonald Creek
Ore Creek
Tarabusi Creek
W Br Ceader River
W Br Pine River
W Br Whitefish River
Black River
Hersey River
Houghton Creek
Prairie Creek
Prairie Creek
Prior Creek
Sebewa Creek
Silver Creek
Hersey River
Hersey River
Hersey River
Hersey River
Looking Glass River
Nottawa Creek
S Br Pentwater River
S Br Pentwater River

Year Latitude
2002
44.55
2001
46.48
2001
46.38
2001
44.65
2001
44.03
2001
44.15
2001
46.03
2001
42.90
2001
44.57
2001
42.67
2001
42.42
2001
44.07
2001
44.64
2001
46.10
2000
46.56
2000
43.85
2000
44.39
2000
43.11
2000
43.02
2000
44.33
2000
42.75
2000
44.18
1999
43.91
1999
43.87
1999
43.93
1999
43.86
1999
42.87
1999
42.10
1999
43.69
1999
43.71

74

Longitude Region code
-83.64
NLP
-85.00
UP
-86.71
UP
-85.45
NLP
-84.65
SLP
-85.01
NLP
-85.61
UP
-83.90
SLP
-83.48
NLP
-83.77
SLP
-83.36
SLP
-84.67
SLP
-83.57
NLP
-86.90
UP
-90.08
UP
-85.44
NLP
-84.04
NLP
-84.97
SLP
-85.02
SLP
-84.07
NLP
-84.96
SLP
-84.06
NLP
-85.52
NLP
-85.49
NLP
-85.53
NLP
-85.46
NLP
-84.90
SLP
-85.21
SLP
-86.18
NLP
-86.32
NLP

Table C1.2.—(cont’d)

Reach code
12944820
12256792
12257042
12221660
12257022
12944330
3396654
12256900
12257058
12221624
8990202
12145326
6838369
12017494
12017622
11930932
12221632
12188975
12134738
12132824
12132952
12133224
8990450

Stream name
Vaughn Creek
Blue Creek
Blue Creek
Driggs River
Mill Creek
Oyster Creek
S Br Galien River
trib Saint Joseph River
trib Saint Joseph River
Hudson creek
Martin Creek
trib Coldwater River
W BrSturgeon River
E Br Two Heart
Two Hearted River
W Br Otter River
Driggs River
Tahquamenon River
Bigelow Creek
E Br Little Muskegon River
Little Muskegon River
Little Muskegon River
S Br White River

Year Latitude
1999
44.42
1998
42.13
1998
42.14
1998
46.35
1998
42.19
1998
44.44
1998
41.81
1998
42.03
1998
42.05
1997
46.42
1997
43.59
1997
42.77
1997
46.00
1996
46.64
1996
46.66
1996
46.78
1995
46.41
1995
46.39
1994
43.51
1993
43.62
1993
43.50
1993
43.57
1993
43.55

75

Longitude Region code
-84.03
NLP
-86.37
SLP
-86.41
SLP
-86.13
UP
-86.26
SLP
-84.06
NLP
-86.69
SLP
-86.44
SLP
-86.44
SLP
-86.06
UP
-85.97
NLP
-85.37
SLP
-88.06
UP
-85.48
UP
-85.53
UP
-88.87
UP
-86.14
UP
-85.80
UP
-85.76
NLP
-85.23
NLP
-85.34
NLP
-85.28
NLP
-85.77
NLP

REFERENCES

76

REFERENCES

Allan, J. D., 2004. Landscapes and riverscapes: the influence of land use on stream ecosystems.
Annual Review of Ecology, Evolution, and Systematics 35: 257–284.
Allan, J. D., and L. B. Johnson. 1997. Catchment–scale analysis of aquatic ecosystems.
Freshwater Biology 37:107–111.
Angermeier, P. L., and I. J. Schlosser. 1989. Species area relationships for stream fishes.
Ecology 70:1450–1462.
Arbuckle, J. IBM SPSS AMOS 19 User’s Guide. 2010. Crawfordville, FL: Amos Development
Corporation.
Arismendi, I., S. L. Johnson, J. B. Dunham, and R. Haggerty. 2013. Descriptors of natural
thermal regimes in streams and their responsiveness to change in the Pacific Northwest of
North America. Freshwater Biology 58:5, 880–894.
Baker, M. E., M. J. Wiley, M. L. Carlson, and P. W. Seelbach. 2003. A GIS model of subsurface
water potential for aquatic resource inventory, assessment, and environmental
management. Environmental Management 32:706–719.
Benda, L., N. L. Proff, D. Miller, T Dunne, G. Reeves, G. Pess, and M. Pollock. 2004. The
network dynamics hypothesis: how channel networks structure riverine habitats.
Bioscience, 54(5):413–427.
Bollen, K. A. 1989. Structural equations with latent variables. Wiley, New York.
Brazner, J. C., D. K. Tanner, N. I. Detenbeck, S. L. Batterman, S. L. Stark, L. A. Jagger, and V.
M. Snarski. 2005. Regional, watershed, and sit–specific environmental influences on fish
assemblage structure and function in western Lake Superior tributaries. Canadian Journal
of Fisheries and Aquatic Sciences 62:1254–1070.
Brenden, T. O., R. D. Jr., Clark, A. R. Cooper, P. W. Seelbach, L. Wang, S. S. Aichele, and E. G.
Bissell 2006. A GIS framework for collecting, managing, and analyzing multi-scale variables
across large regions for river conservation and management. in R.M. Hughes, L. Wang, and
P.W. Seelbach, Eds., Landscape influences on stream habitats and biological assemblages.
pp. 49–74. American Fisheries Society, Symposium 48, Bethesda, Maryland.
Buisson, L., L. Blanc, and G. Grenouillet. 2008. Modelling stream fish species distribution in a
river network: the relative effects of temperature versus physical factors. Ecology of
Freshwater Fish 17:244–257.
Caissie, C. 2006. The thermal regime of rivers: a review. Freshwater Biology 51:1389–1406.
Chu, C., and N. E. Jones. 2011. Spatial variability of thermal regimes and other environmental
determinants of stream fish communities in the Great Lakes Basin Ontario, Canada. River
research and Application 27:646–662.

77

Chu, C., N. E. Jones, and L. Allin. 2010. Linking the thermal regimes of streams in the Great
Lakes Basin, Ontario, to Landscape and Climate Variables. River Research and
Application 241:221–241.
Cress, J., D. Soller, R. Sayre, P. Comer, and H. Warner. 2010. Terrestrial Ecosystems, Surficial
Lithology of the Conterminous United States: U.S. Geological Survey Scientific
Investigations Map 3126, scale 1:5,000,000, 1 sheet. Available at:
(http://pubs.usgs.gov/sim/3126).
Daly, C., M. Halbleib, J. I. Smith, W. P. Gibson, M. K. Doggett, G. H. Taylor, J. Curtis, and P.
P. Pasteris. 2008. Physiographically sensitive mapping of climatological temperature and
precipitation across the conterminous United States. International Journal of Climatology
28:2031–2064.
De Caceres, M., and P. Legendre. 2009. Associations between species and groups of sites;
indices and statistical inference. Ecology 90:3566–3574.
De Caceres, M., P. Legendre, and M. Moretti. 2010. Improving indicator species analysis by
combining groups of sites. Oikos 119:1674–1684.
Dufrene, M., and P. Legendre. 1997. Species assemblages and indicator species: the need for a
flexible asymmetrical approach. Ecological Monographs 67(3):345–366.
Eagle, A. C., E. M. Hay-Chmielewski, K. T. Cleveland, A. L. Derosier, M. E. Herbert, and R. A.
Rustem, eds. 2005. Michigan's Wildlife Action Plan. Michigan Department of Natural
Resources. Lansing, Michigan. 1592 pp. http://www.michigan.gov/dnrwildlifeactionplan
Eaton, J. G. and R. M. Scheller. 1996. Effects of climate warming on fish thermal habitat in
streams of the United States. Limnology and Oceanography 41:1109–1115.
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 landscape anthropogenic activities. Ecological Restoration 29:133–151.
Esselman, P.C., D. M. Infante, L. Wang, A. R. Cooper, D. Wieferich, Y. Tsang, D. J.
Thornbrugh, W. W. Taylor. 2013. Regional fish community indicators of landscape
disturbance to catchments of conterminous United States. Ecological Indicators 26:163–
173.
Fausch, K.D., C.E. Torgersen, C.V. Baxter and W.L. Hiram. 2002. Landscapes to riverscapes:
bridging the gap between research and conservation of stream fishes. BioScience 52(6):
483–498.
Frissell, C. A., W. J. Liss, C. E. Warren, and M. D. Hurley. 1986. A hierarchical framework for
stream habitat classification: viewing streams in a watershed context. Environmental
Management 10:199–214.
Gesch, D. B., 2007. The National Elevation Dataset. Pages 99–118 in D. Maune, editor. Digital
Elevation Model Technologies and Applications: The DEM User’s Manual, 2nd Edition:
Bethesda, Maryland, American Society for Photogrammetry and Remote Sensing.

78

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.
Hayes, D. B., H. Dodd, and J. Lessard. 2008. Effects of small dams on cold water stream fish
communities. Pages 587–602 in J. Nielsen, J. J. Dodson, K. Friedland, T. R. Hamon, J.
Musick, and E. Verspoor, editors. Reconciling fisheries with conservation: proceedings
of the 4th world fisheries congress. Volume I, American Fisheries Society, Symposium
49, Bethesda, Maryland.
Helfman, G. S. 2007. Fish conservation: A guide to understanding and restoring global aquatic
biodiversity and fishery resources. Island Press, Washington, D.C.
Hynes, H. B. N. 1975. The stream and its valley. Verhandlungen Internationale Vereinigung für
Theoretische und Angewandte Limnologie 19:1–15.
Homer, C. C. Huang, L. Yang, B. Wylie, and M. Coan. 2004. Development of a 2001 National
Landcover Database for the United States. Photogrammetric Engineering and Remote
Sensing, Vol. 70, No. 7, pp. 829–840. Available at:
(http://www.mrlc.gov/nlcd_multizone_map.php).
IBM Corporation. 2012. IBM SPSS Statistics for Windows, Amos Version 21.0, Armonk, New
York.
Infante, D. M., and J. D. Allan. 2010. Response of stream fish assemblages to local–scale habitat
as influenced by landscape: a mechanistic investigation of stream fish assemblages.
Pages 371–400 in K. B. Gido and D. A. Jackson, editors. Community Ecology of Stream
Fishes. American Fisheries Society, Symposium 73, Bethesda, Maryland.
Infante, D. M., M. J. Wiley, and P. W. Seelbach. 2006. Relationships between channel shape,
watershed characteristics, and fish assemblage structure in Lower Michigan. Pages 339–
358 in R. M. Hughes, L. Wang and P. W. Seelbach, editors. Landscape influences on
stream habitats and biological assemblages. American Fisheries Society, Symposium 48,
Bethesda, Maryland.
Inoue, M., and S. Nakano. 2001. Fish abundance and habitat relationships in forest and grassland
streams, northern Hokkaido, Japan. Ecological Research 16:233–247.
Isaak, D. J., S. Wollrab, D. Horan, and G. Chandler. 2012. Climate change effects on stream and
river temperatures across the northwest U.S. from 1980–2009 and implications for
salmonid fishes. Climatic Change 113:499–524.
Jackson, D. A., P. R. Peres–Neto, and J. D. Olden. 2001. What controls who is where in
freshwater fish communities – the roles of biotic, abiotic, and spatial factors. Canadian
Journal of Fisheries and Aquatic Science 58:157–170.
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 11(4):211-217.
Johnson, L. B., and S. H. Gage. 1997. Landscape approaches to the analysis of aquatic
ecosystems. Freshwater Biology 37:113–132.

79

Legendre, P., and L. Legendre. 2012. Numerical ecology, 3rd edition. Elsevier, Amsterdam.
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:
721–732.
Lobon–Cervia, J., Y. Dgebuadze, C. G. Utrilla, P. A. Rincon, and C. Granado–Lorencio. 1996.
The reproductive tactics of dace in central Siberia: evidence for temperature regulation of
the spatio–temporal variability of its life history. Journal of Fish Biology 48:1074–1087.
Lyons, J., J. S. Stewart, and M. Mitro. 2010. Predicted effects of climate warming on the
distribution of 50 stream fishes in Wisconsin, U.S.A. Journal of Fish Biology 77:1867–
1898.
Lyons, J., L. Wang, and T. D. Simonson. 1996. Development and validation of an index of biotic
integrity for coldwater streams in Wisconsin. North American Journal of Fisheries
Management 16:241-256.
Magnuson, J. J., L. B. Crowder, and P. A. Medvick. 1979. Temperature as an Ecological
Resource. American Zoologist 19:331–343.
Martin, R. W., and T. Petty. 2011. Local Stream Temperature and Drainage Network Topology
Interact to Influence the Distribution of Smallmouth Bass and Brook Trout in Central
Appalachian Watershed. Journal of Freshwater Ecology 24:3, 497–508.
Maruyama, G. M. 1997. Basics of structural equation modeling. Sage Publications, Thousand
Oaks, California.
Matthews, W. J. 1998. Patterns in freshwater fish ecology. Chapman and Hall, New York.
Matthews, K. R. and N. H. Berg. 1997. Rainbow trout responses to water temperature and
dissolved oxygen stress in two southern California stream pools. Journal of Fish Biology
50:50–67.
Michigan's Wildlife Action Plan. 2006. Species of greatest conservation need. Michigan
Department of Natural Resources, Lansing. Available at:
(http://www.michigan.gov/dnrwildlifeactionplan), September 2010.
Olden, J. D., and R. J. Naiman. 2010. Incorporating thermal regimes into environmental flows
assessments: modifying dam operations to restore freshwater ecosystem integrity.
Freshwater Biology 55:86–107.
Olden, J. D., and N. L. Poff. 2003. Redundancy and the choice of hydrologic indices for
characterizing streamflow regimes. River Research and Application 19:101–121.
Parsons, M., and M. C. Thoms. 2007. Hierarchical patterns of physical–biological associations in
river ecosystems. Geomorphology 80(2007):127–146.
Pielou, E. C. 1975. Ecological diversity. Wiley and Sons, New York.

80

Poff, N. 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, L. N., J. D. Allan, M. B. Bain, J. R. Karr, K. L. Prestegaard, B. D. Richter, R. E. Sparks,
and J. C. Stromberg. 1997. The natural flow regime. BioScience 47(11):769–784.
Poole, G. C., and C. H. Berman. 2001. Temperature: natural heat dynamics and mechanisms of
human–caused thermal degradation. Environmental Management 27(6):787–802.
PRISM. 2013. PRISM Climate Group. Oregon State University. Available at:
(http://www.prism.oregonstate.edu/).
Quist, M. C., W. A. Hubert, and D. J. Isaak. 2004. Fish assemblage structure and relations with
environmental conditions in a Rocky Mountain watershed. Canadian Journal of Zoology
82:1554–1565.
R Core Team. 2013. R: A language and environment for statistical computing. R Foundation for
Statistical Computing, Vienna, Austria. Available at: (http://www.R-project.org/).
Riseng, C. M., M. J. Wiley, and R. J. Stevenson. 2004. Hydrologic disturbance and nutrient
effects on benthic community structure in Midwestern U.S. streams: a covariance
structure analysis. Journal of the North American Benthological Society 23:209–326.
Schlosser, I. J. 1991. Stream fish ecology: a landscape perspective. BioScience 41(10):704–712
Seelbach, P. W., M. J. Wiley, M. E. Baker, and K. E. Wehrly. 2006. Initial classification of river
valley segment across Michigan’s Lower Peninsula. Pages 25–48 in R. M. Hughes, L.
Wang, and P. W. Seelbach, editors, Landscape influences on stream habitats and
biological communities. American Fisheries Society Symposium 48, Bethesda,
Maryland.
Shannon, C. E. 1948. A mathematical theory of communication. Bell System Technical Journal,
27:379–423.
Soller, D. R., and M. C. Reheis, compilers, 2004. Surficial materials in the conterminous United
States: U.S. Geological Survey. Open–file Report 03–275, scale 1:5,000,000. Available
at: (http://pubs.usgs.gov/of/2003/of03–275/).
Steen, P. J., T. G. Zorn, P. W. Seelbach, and J. S. Schaeffer. 2008. Classification tree models for
predicting distributions of Michigan stream fish from landscape variables. Transactions
of the American Fisheries Society 137:976–996.
Taniguchi, Y., D. C. Novinger, F. J. Rahel, and K. Gerow. 1998. Temperature mediation of
competitive interactions among three fish species, brook trout, brown trout, and creek
chub that replace each other along longitudinal stream gradients. Canadian Journal of
Fisheries and Aquatic Sciences 55:1894–1901.
Thornbrugh, D. J., and K. B. Gido. 2010. Influence of spatial positioning within stream networks
on fish assemblage structure in the Kansas River basin, USA. Canadian Journal of
Fisheries and Aquatic Sciences 67:143–156.
81

Tonn, W.M. 1990. Climate change and fish communities: a conceptual framework. Transactions
of the American Fisheries Society 119: 337–352.
U.S. Environmental Protection Agency and U.S. Geological Survey (USEPA and USGS). 2005.
National hydrography dataset plus, NHDPlus Version March/June 2011 Ecological
Restoration 29:1–2
U.S. Geological Survey (USGS). 2012. National Anthropogenic Barrier Dataset (NABD) 2012.
Available at: (http://www.fishhabitat.org/content/nfhp-data-system).
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 143:688–699.
Vannote, R. L., Minshall, G. W., Cummins, K. W., Sedell, J. R., and Cushing, C. E. 1980. The
river continuum concept. Canadian Journal of Fisheries and Aquatic Science 37:130–137.
Wang, L., D. Infante, P. Esselman, A. Cooper, D. Wu, W. Taylor, D. Beard, G. Whelan, and A.
Ostroff. 2011a. A hierarchical spatial framework and database for the national river fish
habitat condition assessment. Fisheries 36:436–449.
Wang, L., D. Infante, J. Lyons, J. Stewart, and A. Cooper. 2011b. Effects of dams in river
networks on fish assemblages in non-impoundment sections of rivers in Michigan and
Wisconsin, USA. River Research and Applications 27:473–487.
Wang, L., J. Lyons, P. Rasmussen, P. Seelbach, T. Simon, M. Wiley, P. Kanehl, E. Baker, S.
Niemela and P. M. Stewart, 2003. Watershed, reach and riparian influences on stream
fish assemblages in the Northern Lakes and Forest Ecoregion, USA. Canadian Journal of
Fisheries and Aquatic Science 60: 491–505.
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.
Ward, J. V. 1998. Riverine landscapes: biodiversity patterns, disturbance regimes, and aquatic
conservation. Biological Conservation 83(3):269–278.
Wehrly, K. E., L. Wang, and M. Mitro. 2007. Field–based estimates of thermal tolerance limits
for trout: incorporating exposure time and temperature fluctuation. Transactions of the
American Fisheries Society 136(2):365–374.
Wehrly, K. E., M. J. Wiley, and P. W. Seelbach. 2006. Influence of Landscape Features on
Summer Water Temperatures in Lower Michigan Streams. Pages 113–127 in R. M.
Hughes, L. Wang, and P. W. Seelbach, editors, Landscape influences on stream habitats
and biological communities. American Fisheries Society Symposium 48, Bethesda,
Maryland.
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.
Wehrly, K. E., M. J. Wiley, and P. W. Seelbach. 1997. Landscape–based models that predict July
thermal characteristics of Lower Michigan rivers. Michigan Department of Natural
Resources, Fisheries Division Research Report 2037, Lansing, Michigan.
82

Wenger S. J., D. J. Isaak, C. H. Luce, H. M. Neville, K. D. Fausch, J. D. Dunham, D. C.
Dauwalter, M. K. Young, M. M. Elsner, B. E. Rieman, A. F. Hamlet, and J. E. Williams.
2011. Flow regime, temperature, and biotic interactions drive differential declines of trout
species under climate change. Proceedings of the National Academy of Science.
108(34):14175–14180.
Wetzel, R. G. 2001. Limnology, 3rd edition, Academic Press, San Diego.
Wiens, J. A. 2002. Riverine landscapes: taking landscape ecology into the water. Freshwater
Biology 47:501–515.
Wills, Todd C., T. G. Zorn, A. J. Nuhfer, and D. M. Infante. 2008 Draft. Stream Status and
Trends Program sampling protocols. Chapter 26 in J. C. Schneider, editor. Manual of
fisheries survey methods II: With periodic updates. Michigan Department of Natural
Resources, Fisheries Special report 25, Ann Arbor, Michigan.
Wolock, D. M., 2003. Base–flow index grid for the conterminous United States. US Geological
Survey Open–File Report 03–263.
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.

83

CHAPTER TWO

LANDSCAPE EFFECTS ON FLUVIAL FISH ASSEMBLAGE STRUCTURE:
REGIONAL RESPONSES TO HUMAN LAND USES.

Abstract
Anthropogenic activities, including urbanization, agriculture, and dams, cause
degradation to stream ecosystems and are a dominant reason for global biodiversity declines in
fluvial fish assemblages. While such trends have been well-documented regionally, differences
occur in how fish respond to human land uses, resulting in location-specific responses in fishes
from the same degree of human land use in one region vs. another. To test for such differences,
we conducted our study within five large ecoregions in the eastern United States and evaluated
data characterizing stream fish assemblages from > 10,500 locations within the study region.
Fishes were summarized by functional and taxonomic metrics describing community structure,
trophic structure, levels of tolerance to human stressors, and life history. Natural and
anthropogenic landscape factors were evaluated across regions, and prominent human land uses
were identified to test for regionally-specific influences on fishes including percent urbanization,
percent agriculture, and density of stream-road crossings and dams in stream catchments. After
controlling for natural landscape factors (including catchment area, catchment lithology, and
elevation of study sites) we used cascade multivariate regression trees (CMRT) to quantify
variance explained in fish metrics by landscape factors. Results showed differences in dominant
influences by regions, and differences in levels of human land uses influencing fishes within and
across regions. Knowledge of such regional differences lends insights into controls on stream

84

fish assemblages and can aid in developing conservation strategies to prevent biodiversity loss
from current and future human land uses.
Introduction
For over a century, freshwater fishes have experienced population declines and range
reductions globally due to anthropogenic influences in the environment (Dudgeon et al. 2006;
Helfman 2007; Vörösmarty et al. 2010). Many freshwater fishes are critically imperiled in North
America and have remained in a similar or worse state of conservation status (i.e., listed as
endangered, threatened or vulnerable) over the last two decades (Jelks et al. 2008). Due to these
trends, freshwater fishes are considered one of the most imperiled taxa on the planet (Racciardi
and Rasmussen 1999; Cooke et al. 2012). Habitat degradation, fragmentation and loss are
dominant reasons for declines in biodiversity of freshwater fishes (Sala et al. 2000; Helfman
2007; Jelks et al. 2008; Burkhead 2012), with changes in habitat resulting in part from human
land uses in surrounding landscapes. Human land uses may affect fishes by altering stream
flows and thermal regimes as well as through inputs of nutrients, sediments, and toxins, all
known to degrade habitat for stream fish assemblages (Dudgeon et al. 2006; Helfman 2007; Jelks
et al. 2008). Modifications to aquatic habitats, induced by land cover changes are expected to
continue into the future, with persistent negative effects on freshwater fishes (Ormerod 2003;
Helfman 2007; Jelks et al. 2008).
In fluvial systems, understanding the extent to which human alterations landscapes may
affect fishes has advanced greatly in the past few decades due to intensive study of relationships
between land cover in catchments and aquatic habitat quality (e.g., Allan et al. 1997; Wiens
2002; Allan 2004). This greater understanding has been accompanied by advancements in the

85

utility and application of bioassessment methods which has improved our understanding of the
types and levels of human perturbation that are negatively affecting aquatic habitats and the
fishes they support (Barbour et al. 1999; Karr and Chu 1999; 2000 Norris and Thoms 1999). For
example, urban land use has been described as a major source of degradation to habitat and
declines in biota in stream systems (Paul and Meyer 2001; Allan 2004; Walsh et al. 2005).
Studies documenting negative effects of urban land use on habitats and fish in Wisconsin (e.g.,
Wang et al. 2000; 2001) and declines of endemic fish species with urban land use across stream
catchments exist in the southeastern U.S. (e.g., Walters et. al. 2003, Scott 2006). Similarly,
agricultural land use is known to negatively affect fish in a variety of ways (Richter et al. 1997;
Allan 2004). Percent agricultural land use in catchments was associated with negative effects on
fish assemblages in 20 major river basins throughout the conterminous United States (Meador
and Goldstein 2003). Further, regional decreases in fish abundance, species richness, and fish
biotic integrity have been described for streams in Wisconsin (e.g., Wang et al. 1997; Fitzpatrick
et al. 2001) and Michigan (e.g., Roth et al. 1996) resulting from > 30% agriculture in catchments.
Besides urban and agricultural land uses, dams and others barriers to fish passage are
increasingly acknowledged as operating at a landscape-scale due to their pervasiveness
throughout river networks and to the variety of effects they have on stream fishes and habitats
(Cooper et al. In Review). Besides reducing network connectivity, dams alter flow and thermal
regimes, sediment transport, and water physicochemistry (Olden and Naiman 2010, Cooper et al.
In Review), and recent efforts have characterized cumulative effects of multiple barriers,
including dams and passageways at stream-road crossings, on stream habitats and fishes (Wang
et al. 2011b, Perkin and Gido 2012, Januchoswki-Hartley et al. 2013). Yet despite extensive and
advancing understanding of the pervasiveness of anthropogenic landscape effects on streams,

86

degradation of stream habitats and decline in stream fishes continues throughout the U.S. (Allan
and Flecker 1993; Dudgeon et al 2006; Jelks et al. 2008; Strayer and Dudgeon 2010).
One factor that confounds the ability to protect streams from degradation resulting from
human land uses includes limited understanding of the mechanisms by which landscape factors
affect streams including delivery of sediments, nutrients, and/or toxins, and alterations to flow
and thermal regimes (Allan 2004). Such diverse and often co-occurring factors like the delivery
of materials, and thermal and flow alterations make it challenging to manage for sources of
degradation to stream habitats. Stream protection is further complicated because land use
influences on streams may vary regionally due to the influences of co-occurring natural
landscape factors such as lithology and topography. This can result in regionally-specific trends
and/or differences in the types and levels of human land uses affecting stream fishes across
regions. Only a few studies have attempted to describe such regional differences in fish
responses. For example, Utz et al. (2010) found substantial regional differences in declines of
diversity and abundances of stream fishes across physiographic regions in Maryland, attributing
these differences in part to geomorphic characteristics of the regions and their ability to buffer
effects of urbanization.
The effects of landscape-scale factors like human land use may vary by the specific
compositions of fish assemblages, and this factor requires greater understanding for stream
protection. Morgan and Cushman (2005) found differences in regional sensitivity of fish species
to urban land use in Maryland, due in part to differences in tolerance of fishes between the
Eastern Piedmont and Costal Plains physiographic regions. In a similar study, Meador et al.
(2005) showed that fish species richness declined with increasing urbanization at sites in both
Boston, MA and Birmingham, AL but attributed the observed declines to different factors in
87

Boston, fluvial specialists were negatively associated with urban land in the absence of endemic
species. In Birmingham, endemic fishes were negatively associated with urban land. The authors
attributed these results to compositional differences in fish assemblages between the regions and
to regionally different mechanisms related to urbanization like water withdrawal and flow
alterations. Together, these findings underscore the observations that fish assemblages can
respond differently and be more susceptible to human land uses depending on location and types
of human stressors (Utz et al. 2010).
Based on the importance of understanding such regional differences in land use effects on
conservation and management of fluvial fish, the goal of this study was to characterize
regionally-specific differences in fish assemblage relationships to human land uses across five
large regions in the eastern portion of the U.S. We had three objectives. First, we characterized
prominent anthropogenic landscape factors that could potentially affect fish over large spatial
extents. Second, we developed a composite variable of prominent natural landscape factors
known to affect fish that could be used to incorporate natural landscape factors into assessments
of stream fishes. Finally, we characterized if fish response to human land uses vary by
ecoregions in an attempt to describe regional influences for improved understanding of landscape
controls on stream fishes.
Methods
Study area.—This study was conducted within five large freshwater ecoregions located in
the eastern U.S. These freshwater ecoregions, referred to following as “ecoregions,” were
delineated by the World Wildlife Fund and included: the Appalachian Piedmont, Chesapeake
Bay, Laurentian Great Lakes, Middle Missouri, and the Upper Mississippi (Abell et al. 2008;

88

Figure 2.1). Freshwater ecoregions were defined as areas with similar biogeographic history in
fish faunas that encompassed major river basins and large basin boundaries; they represented
ecologically-relevant units for groupings of fishes based on evolutionary patterns across large
landscapes (Maxwell et al. 1995; Matthews 1998; Abell et al. 2000; 2008; Esselman et al. 2011).
Freshwater ecoregions were chosen as the spatial units to evaluate in this study to account for
broad influences of natural landscape factors across the entire study area on stream fishes and to
incorporate a diverse range of natural and human landscape variables to facilitate regional
comparisons among ecoregions.
The study area extends from the western Great Plains to the east coastal plains of the U.S.
(Figure 2.1). Natural landscape factors within the study area are highly diverse; study ecoregions
include previously glaciated landscapes in the Chesapeake Bay and Laurentian Great Lakes and
regions that have not been glaciated including the coastal plains and highlands of the
Appalachian Piedmont ecoregion. In the eastern and central portion of the study area, the
Middle Missouri and Upper Mississippi ecoregions included areas of intensive agriculture, while
high percentages of grasslands occur within the western Middle Missouri ecoregion. A high
percentage of forested areas are located within the Laurentian Great Lakes, and the Chesapeake
Bay ecoregion includes areas of high urban land use.
Stream layer.—The 1:100,000-scale National Hydrography Dataset Plus Version 1
(NHDPlusV1) national streams layer was used as a base layer for geographic representation of
streams reaches and their catchments (USEPA and USGS 2005). We define a stream reach as a
section of river in the NHDPlusV1 that extends 1) from the stream origin to the first downstream
confluence or junction with a lake or reservoir, 2) from an upstream confluence or lake/reservoir
outflow to the next downstream confluence or lake/reservoir junction, or 3) from an upstream
89

confluence or lake/reservoir outflow to the river mouth where it meets with a lake, reservoir, or
estuary (Brenden et al. 2006; Esselman et al. 2011; Wang et al. 2011a). Catchments
summarizing information over two spatial extents were used in this analysis based on the
NHDPlusV1. “Local catchments” included all land that drains directly into an individual stream
reach without being transported via other fluvial pathways represented in the NHDPlusV1, and
“network catchments” included all land upstream of and draining into a given reach via fluvial
pathways, including the local catchment.
Natural landscape variables.—A variety of nationally-available landscape datasets were
used in this study. All natural and human landscape factors were summarized at the local
catchment and then aggregated to the network catchment, representing two spatial extents, then
attributed to NHDPlusV1 stream reaches using a geographic information system (GIS, Table
2.1). Natural land cover classes including water, wetland, barren, forest, shrub, and grassland
were developed from the National Land Cover Dataset (NLCD, Homer et al. 2004; Table 2.1).
Catchment areas were calculated using a GIS. Channel gradient (i.e. slope) and elevation were
developed from U.S. Geological Survey (USGS) National Elevation Data (NED, Gesch 2007).
Surficial lithology metrics were developed from the USGS terrestrial ecosystems surficial
lithology of the conterminous U.S. data layer (Soller and Reheis 2004; Cress et al. 2010).
Surficial lithology classes were defined by grouping major categories of substrate size (e.g.,
coarse and fine). Coarse and fine categories were used to compare streams with a more stable
hydrology and potentially higher contribution of base-flow from groundwater inputs versus
streams that may be more prone to variable flow regimes, respectively. The base-flow index, a
model of hydraulic potential characterizing the percentage of groundwater contribution to stream
flow, was modeled by USGS (Wolock 2003; Table 2.1). A range in soil hydrologic groups and

90

soil permeability metrics were developed from the U.S. Department of Agriculture (USDA)
Natural Resources Conservation Service (NRCS) State Soil Geographic (STATSGO) Data Base
(Schwarz and Alexander 1995). Mean annual air temperature and mean annual precipitation
were summarized for the climatological period from 1981 to 2010 by the PRISM Climate Group
and attributed to catchments of stream reaches (Daly et al. 2008; PRISM 2013).
Human land uses.—Major categories of human land cover for urban and agriculture
classes from the NLCD were used including low, medium, and high intensity developed land;
open space developed land; pasture/hay and cultivated crops (Homer et al. 2004). Individual
classes for urban and agricultural land uses were summarized into composite metrics to aid in
meeting assumptions of statistical tests (described below, Table 2.1). The percentage of
impervious surface in the network catchment was developed from the NLCD (Homer et al.
2004). The population density metric was developed from a data layer created by the National
Oceanic and Atmospheric Administration (NOAA, NOAA 2010). To consider cumulative
effects of catchment-wide barriers on stream fish, variables representing barriers were
incorporated for evaluation and analysis (Table 2.1). The density of stream-road crossings, road
length and canals and ditches in catchments were developed from the U.S. Census Bureau
TIGER/line files data (U.S. Census Bureau 2002). The density of dams was developed from the
National Anthropogenic Barrier Dataset (NABD, U.S. Geological Survey 2012).
Fish data.—Data characterizing stream fish assemblages were assembled and referenced
to stream reaches in the NHDPlusV1. Assemblages were sampled by state and federal programs
using methods determined to be comparable for creeks, small rivers, and medium rivers ≤ 10,000
km2 in drainage areas, over years from 1990 to 2010 (Esselman et al. 2011; 2013 Wang et al.
2011a). Fish were identified to species, and all scientific naming was standardized to the
91

Integrated Taxonomic Information System (ITIS,ITIS 2010). Twenty two fish metrics were
developed to characterize fish assemblage structure at sites across the study region (Table 2.2),
and all metrics were described as being indicators of biological integrity in streams and
responsive to human perturbations (Lyons et al. 1996; Wang et al. 1997; Angermeier et al.
2000). Metrics included species and family richness, community diversity, and community
evenness (Table 2.2, Shannon 1948; Pielou 1975; Oksanen et al. 2013). Fish assemblages were
also characterized by functional metrics describing community structure, trophic grouping, levels
of tolerance, and life history characteristics (Table 2.2). Functional metrics were represented as
percent of individuals. Assignment of trophic metrics followed Lyons (1992), assignment of
tolerance metrics measuring sensitivity of species to perturbation followed Barbour et al. (1999),
and assignment of life history characteristics followed Frimpong and Angermeier (2009).
Data analysis
Landscape metric selection.—Fifty two variables describing a range in natural and
anthropogenic landscape factors were originally considered for analysis (Table 2.1). All
variables were transformed to better meet assumptions of linearity and homoscedasticity.
Continuous variables were natural log transformed; percentage and proportional variables were
arcsin square root transformed. We ran a Pearson correlation on all fifty two variables to reduce
the number of highly correlated metrics (Table C2.1). When a pair of variables was correlated
by an absolute value of ≥ 0.7, one of the two was eliminated from further analysis. Network
catchment summaries were selected over local catchment summaries of the same factor because
they were assumed to characterize a broader and more spatially extensive set of influences from
landscape factors on streams, and other highly correlated variables were selected based on
ecological interpretability. Principal component analysis (PCA) was then used with twenty two
92

remaining natural and human landscape variables to explore dimensionality in the dataset and to
further reduce redundancy (Legendre and Legendre 2012, Table C2.2). Landscape metrics were
chosen for additional analyses by their weightings on orthogonal axes to retain unique
dimensions of explained variation in the data and based on ecological interpretability. For
example, network catchment base-flow was chosen over soil permeability because it was
assumed to be a more interpretable measurement even though it had a lower axis weighting in
the PCA. Through this process, six natural and eight anthropogenic landscape factors were
selected for use in further analysis (Table 2.3).
Fish metric selection.—Pearson correlation was also used to reduce redundancy among
the twenty two fish metrics. This analysis showed that fish species richness, family richness, and
Shannon’s diversity index were highly correlated (Table C2.3), and fish species richness was
retained over family richness and diversity because it has been commonly used in development
of IBIs (e.g., Karr and Chu 1999). Principal component analysis (PCA) was then used with the
remaining fish metrics to characterize dimensionality in the data set (Legendre and Legendre
2012). Based on weightings on orthogonal axes, a final set of seven fish metrics (Table 2.4),
were chosen to retain unique dimensions of explained variation in the data and for further
analysis (Table C2.4).
Redundancy analysis (RDA). —RDA was used to partition variance in fish metrics into
independent (pure) and confounded (inter-correlated) components explained by groupings of
natural and human landscape variables following Borcard et al. (1992), and also to characterize
how these partitions varied by ecoregion (Borcard et al. 2011; Legendre and Legendre 2012).
Six natural landscape factors and six human land use classes were used in RDA, including an
urban composite metric comprised of low, medium, and high intensity developed land to ensure
93

a relatively consistent number of variables between groupings (i.e., human, natural, and fish,
Table 2.1). A detrended correspondence analysis (DCA) was first used to quantify the gradient
length of the dataset to determine the appropriate ordination technique, either canonical
correspondence analysis (CCA) or RDA (ter Braak 1995; Esselman and Allan 2010; Pool et al.
2010). The gradient length of species metrics along the first DCA axis indicates if species
response to environmental data is closer to linear versus more unimodal. The DCA gradient
lengths for our fish metrics was < 2, suggesting a more linear species response and that RDA
versus a CCA was more appropriate for further analysis (ter Braak 1995; Esselman and Allan
2010; Pool et al. 2010). We partitioned the total variance in fish metrics explained by natural
landscape variables, human land use variables and their corresponding interaction terms for each
of the five freshwater ecoregions using CANOCO software (ter Braak and Smilauer 2002).
Developing a composite variable to characterize natural landscape influences.—PCA
was used to develop a composite variable from six natural landscape factors across the entire
study region (Table 2.1, Figure 2.1). This composite variable was developed for use in multiple
wave cascade multivariate regression tree (CMRT) analysis to aid in controlling for natural
variability known to influence fish across the five study ecoregions in a consistent and
comprehensive manner (Whittier et al. 2007; Whittier and Van Sickle 2010; Van Sickle 2013).
Individual site scores from axis 1 of the PCA results were used to create the composite variable
(Table 2.5).
Cascade multivariate regression tree (CMRT).—CMRT was performed to quantify
regional difference in fish responses to dominant human land uses after controlling for natural
landscape influences. A CMRT is based on a tiered analysis of multivariate regression trees
(MRT) and does not require that relationships between explanatory and response variables are
94

linear or that residuals be normally distributed (De’ath 2002; Ouellette et al. 2012), making the
technique appropriate for large, ecological data sets that do not meet such statistical
requirements. The CMRT is conducted in multiple waves where the first wave of data is
theorized to have dominant explanatory power over the response data, and a second wave is
applied to explain additional variance. The first wave in this CMRT analysis was performed to
associate variability in fish metrics with natural landscape influences. Human land use metrics
including percent network urban land use, percent network agriculture, and the density of
stream-road crossings and dams in the network catchment (Table 2.3) were used in the second
wave of analysis to describe additional variation explained in the fish metrics after controlling
for the natural landscape variable.
Before analysis, fish metrics were centered and a standardized z-score was calculated for
response metrics. To run the CMRT we used the function “CascadeMRT” from the package
“mvpart” in R (De’ath 2006; Therneau et al. 2010; Oksanen et al. 2013; Ouellette 2013, R Core
Team 2013). This uses a standard recursive greedy splitting algorithm to fit all MRT models
(De’ath 2002). We use the “one standard error” rule for the v-fold cross-validation to fully grow
the MRTs (rule described by De’ath 2002). The one standard error rule suggests using the
simplest model within one standard error of the model that minimizes the cross-validated relative
error, with the aim of minimizing risk and complexity of the model. This method picks the
simplest best model that minimizes both the cross-validated relative error and the number of
splits in the MRT (De’ath 2002; Ouellette 2013). The complexity parameter value is the
minimum contribution to the R2 of a MRT tree for a split to be considered for the package
“mvpart” in R (Ouellette 2013; R Core Team 2013). In our analyses, we set complexity
parameters high enough in both the first and second waves to identify only the main factors

95

determining variation in species composition yet values were still considered conservative,
ranging between 0.01-0.05.
The CMRT analysis was used to characterize the hierarchical influences on stream
systems with the predominant influence from natural landscape factors followed by influences of
human land uses on stream fishes. It allowed us to separate the explanatory power of two data
sets on stream fishes, first controlling for natural landscape factors across the entire study area
and secondly accounting for anthropogenic landscape variables by ecoregions (Ouellette et al.
2012). Site scores from the natural composite variable were used in the first wave of the CMRT
analysis to compute the main partition of explained variance (Table 2.5). Residuals from the
first wave of the CMRT analysis were then used in the second wave with the human land use
variables to compute human partitions of additional explained variance based on the theorized
controls of riverine ecosystems, with a predominant influence on stream systems and biota from
natural variables (e.g., catchment area, lithology, etc.) and additional influences from human
perturbation in the landscape (e.g., percent catchment urban and agriculture, catchment density
of dams, Vannote et al. 1980; Frissell et al. 1986; Benda et al. 2004; Grant et al. 2007). This
provides the ability to determine the influence of human land use variables after taking into
account the influence of the natural landscape variables, recognizing that one set might change as
a function of the other (Ouellette et al. 2012).
Results
Study area.—The five study ecoregions included an area of 2,287,921 km2 (Figure 2.1).
A total of 10,522 fish sampling locations were used in analyses across the study region, with
1083 sampling sites in the Appalachian Piedmont, 807 in the Chesapeake Bay, 3773 in the

96

Laurentian Great Lakes, 886 in the Middle Missouri, and 3973 in the upper Mississippi
ecoregion. Natural landscape factors at study sites are highly diverse across ecoregions (Table
2.3). Mean network catchment area of study sites ranged from 155 km2 in the Appalachian
Piedmont ecoregion to 764 km2 in the Middle Missouri ecoregion. Fine lithology in the network
catchment was lowest in the Upper Mississippi ecoregion (mean of 9%) and highest in the
Laurentian Great Lakes ecoregion (mean of 42%). Mean local elevation was lowest at study
sites in the Appalachian Piedmont ecoregion at 157 m, and ranged from 20 m to 1166 m across
all sites in the study ecoregions, characterized by 10% and 90% exceedance values, respectively
(Table 2.3). Mean local elevation was highest at study sites in the Middle Missouri ecoregion at
591 m. Climate varied across sites in the study ecoregions as well, with lowest mean annual air
temperatures in the Upper Mississippi (8 °C) and highest mean annual air temperature and
precipitation in the Appalachian Piedmont ecoregion (15 °C, 1215 ml, respectively, Table 2.3).
The lowest mean annual precipitation was in the Middle Missouri ecoregion at 709 ml.
Study sites across the five ecoregions also exhibited a wide range in human land uses.
Mean percentage of urban land use in network catchments of study sites was highest in the
Appalachian Piedmont ecoregion at 14% and lowest at study sites in the Middle Missouri at 6%
(Table 2.3). There was high variability in mean percentage of agriculture land use in the network
catchment across study ecoregions; it was highest in the Upper Mississippi ecoregion at 60%,
with 10% and 90% exceedance values of 16% and 90%, respectively. The mean network
catchment percent agriculture was lowest in Appalachian Piedmont ecoregion with a value of
21% and 10% and 90% exceedance values of 4% and 41%, respectively (Table 2.3). Density of
stream-road crossings in network catchments of study sites was highest in the Chesapeake Bay
ecoregion at 54/100 km2 and was lowest in the network catchment for study sites in the Middle

97

Missouri ecoregion at 36/100 km2. Density of dams in the network catchment of study sites was
highest in the Appalachian Piedmont ecoregion at 3.91/100 km2 and lowest in the network
catchment of study sites in the Upper Mississippi at 0.71/100 km2 (Table 2.3)
Variance partitioning with (RDA), differences across study ecoregions.—RDA variance
partitioning indicated that natural and human landscape variables contributed to between 28% to
42% of total explained variation in fish metrics across ecoregions (Figure 2.2). The maximum
variation explained in fish occurred in the Upper Mississippi ecoregion (42%) and the minimum
variation explained occurred in the Laurentian Great Lakes ecoregion (28%, Figure 2.2). Natural
landscape variables generally explained the largest proportion of variance in fish metrics as
expected, ranging from 12% to 27%, with the exception of the Laurentian Great Lakes
freshwater ecoregion, where interactions among the human and natural variables explained the
highest percentage of variance in the fish metrics (13%, Figure 2.2). Variation explained by
human land uses alone was much less than that explained by natural landscape variables;
approximately 2% of the variation in fish was explained by human landscape variables in the
Middle Missouri and the Upper Mississippi freshwater ecoregions, ranging to a maximum of 5%
in the Chesapeake Bay freshwater ecoregion (Figure 2.2). Interactions between groupings of
natural and human landscape factors contributed a large proportion of explained variation in fish
metrics from 6% in the Chesapeake Bay and Middle Missouri freshwater ecoregion to 13% in
the Appalachian Piedmont, Laurentian Great Lakes, and Upper Mississippi ecoregions (Figure
2.2).
Developing a composite variable to characterize natural landscape influences.—Six
natural landscape factors were used in PCA to develop a generalized natural composite variable
to account for major natural landscape influences for use as the primary natural landscape
98

variable in the first wave of the CMRT analysis before further analyzing variance described by
human metrics (Table 2.5). The PCA of natural landscape variables resulted in two axes
explaining 59.70% of the variation across the study sites (Table 2.5). Axis one, explaining
38.89% of the variation in the study sites, was weighted positively by air temperature, average
precipitation, and fine lithology and was negatively weighted by mean local elevation and baseflow index (Table 2.5). Axis two, explaining 20.81% of the variation in the study sites, was
weighted positively by catchment area and fine lithology, and was negatively weighted by mean
annual precipitation in the catchment and base-flow index in the catchment (Table 2.5). The first
axis of the PCA described the prominent amount of variance representing natural variation in the
data, site scores from axis one were chosen for further analysis, and used as the natural variable
for the first wave of the CMRT analysis across all of the five study ecoregions.
Variance partitioning with cascade multivariate regression trees (CMRT).—There were
differences in the amount of variance explained in fish metrics between the five ecoregions, but
general patterns were similar. The CMRT global variance indicated that natural and human
variables combined explained between 7.21% in the Appalachian Piedmont ecoregion to 13.05%
in the Chesapeake Bay ecoregion of the total explained variation in fish metrics at sites across
the five freshwater ecoregions (Table 2.6). Variance explained from the first wave represented
by the natural landscape composite variable (referred to as main natural) ranged from a low of
5.65% in the Upper Mississippi ecoregion to 10.17% in the Appalachian Piedmont ecoregion.
The combined variance explained in fish metrics from the dominant two human land uses (i.e.
human 1 + human 2) from each drop of the second wave of the CMRT ranged from 2.56% in the
Laurentian Great Lakes to 4.91% in the Middle Missouri, across study sites of the five
ecoregions (Table 2.6).

99

Regional influences and differing levels of human land use variables for the second wave
of CMRT.— The second wave of the CMRT, constrained by the composite variable of natural
landscape factors, identified the two primary human land use variables and at what levels these
factors explained variation in fish metrics. Results showed that important metrics differed by
ecoregion. For example, the Appalachian Piedmont ecoregion had both a prominent and
secondary level of urban land use that described variance in fish assemblage metrics. The
Chesapeake Bay ecoregion had urban as the prominent land use and agriculture as a secondary
human land use influencing fish assemblage metrics, while the Laurentian Great Lakes, Middle
Missouri, and Upper Mississippi ecoregions all had catchment agriculture and density of dams as
prominent and/or secondary human land use variables describing variance in fish assemblage
metrics.
In those cases when prominent human land use metrics were the same across freshwater
ecoregions there were often substantial differences in the levels of the land use that were causing
differences in stream fish assemblage metrics across sampling sites (Table 2.6). Percent urban in
the network catchment was the primary human land use in the Appalachian Piedmont at two
levels; a high level of 38.24%, which was above the 90% exceedance value of 37% and at a level
of 10% which is just below the mean level of 14% (Table 2.3 and 2.6). Percent urban in the
network catchment was also the primary human land use in the Chesapeake Bay ecoregion at a
low level of 1.30%, almost equal to the 10% exceedance value of 1.27% for the ecoregion.
Percent agriculture in the network catchment was the primary human land use for the Upper
Mississippi and the secondary human land use in Laurentian Great Lakes, Chesapeake Bay, and
Middle Missouri ecoregions at levels of 54%, 68%, 0.40% and < 0.01% respectively (Table 2.6).
The levels of percent agriculture in the network catchment causing change in fish assemblage

100

metrics across study sites were substantially different for the Upper Mississippi and Laurentian
Great Lakes, (i.e., just below the mean and substantially above the mean, respectively) versus in
the Chesapeake Bay and Middle Missouri ecoregions (i.e. below the 10% exceedance values,
Table 2.2).
A low density of dams in the network catchment was the primary human land use
variable describing variation in fish assemblage metrics for Laurentian Great Lakes, Middle
Missouri ecoregions, and secondary in the Upper Mississippi ecoregion at levels of 0.05/100
km2, 0.02/100 km2, and 0.04/100 km2, respectively (Table 2.6). The density of road crossings in
the network catchment was not found to be one of the two most influential human land use
metrics across sites for any one of the five ecoregions we evaluated in this study.
Discussion
This study characterized regional differences in types and levels of human land uses
potentially influencing stream fish assemblages across five ecoregions in the eastern portion of
the U.S. Fish assemblages were summarized by metrics including species richness; percent
intolerant and game individuals; and percent individuals that are invertivores, omnivores,
lithophiles, and rheophiles (Lyons 1992; Barbour et al. 1999; Frimpong and Angermeier 2009).
We also identified prominent anthropogenic landscape factors within our study regions including
percent of urban and agriculture land uses and densities of stream-road crossings and dams in
stream catchments. Testing for influences of these factors with CMRT analysis indicated that
urban and agricultural land use as well as dam density had regional effects on stream fishes.
Urban land use was the prominent factor associated with differences in fish assemblages in the
Appalachian Piedmont and Chesapeake Bay ecoregions, while agricultural land use was

101

prominent in the Upper Mississippi ecoregion. Agricultural land use was also identified as a
secondary factor influencing fish in the Chesapeake Bay, Laurentian Great Lakes, and Middle
Missouri ecoregions. Dam density in catchments was the prominent influence in both the
Laurentian Great Lakes and Middle Missouri ecoregions (Table 2.6), underscoring the potential
influence of multiple dams within catchments to act as a landscape-scale influence on stream
fishes. Besides identifying regional differences in influences, we also identified differences in
levels at which fish were associated with landscape disturbances (Table 2.6). For example high
percentages of catchment agriculture (i.e., 54% and 68%, respectively) were associated with
differences in fish assemblages in the Upper Mississippi and the Laurentian Great Lakes
ecoregions, while lower levels (<0.4%) were associated with differences in fish assemblages in
the Chesapeake Bay and Middle Missouri ecoregions. Our results suggest that effects of human
land uses on stream fishes vary regionally, resulting from complex relationships between
landscape factors and mechanisms controlling stream fish assemblages, but explain a low
amount of the observed variance.
Variance in fish metrics explained by human and natural landscape factors across five
ecoregions.—By partitioning variance explained in stream fish metrics by groupings of natural
landscape factors and human land use variables, RDA identified regional differences in
landscape controls on stream fishes. Total explained variance in fish metrics ranged from 28%
in the Laurentian Great Lakes to 42% in the Upper Mississippi, and of that variance, a majority
was attributed to natural landscape factors, which ranged from 12% in the Laurentian Great
Lakes to 27% in the Upper Mississippi. Previous work conducted for streams in Louisiana and
Mississippi fell within this range; Alford (2014) describes a similar percentage of variance (i.e.,
24%) explained in fish species assemblages by catchment variables including catchment area,

102

elevation, and stream density within catchments. The RDA also quantified unique proportions of
explained variance in fish metrics by human land uses, underscoring the importance of human
land uses on stream fish across large regions and also justifying our evaluation of specific human
land uses on fishes with CMRT analysis. An additional insight from the RDA was in the
identification of the degree to which natural landscape factors and human land uses were
confounded within our study regions. In all five ecoregions, amount of variance explained by
covarying natural and anthropogenic landscape factors was greater than that explained by
anthropogenic landscape factors alone. This is due in part because natural landscape
characteristics may influence the suitability of landscapes for various human land uses (Allan
2004; Scott 2006), and it supported the need to explicitly control for natural landscape factors
consistently across the five study ecoregions with development of our natural composite variable
in CMRT analyses.
Regionally-specific effects of urban land use on stream fish.—After controlling for the
influence of natural landscape factors with our natural composite variable, percent urban land use
in the network catchment explained the most variance in fish assemblages in the Appalachian
Piedmont and Chesapeake Bay ecoregions at levels of 10% and 2%, respectively. The lower
level detected in the Chesapeake Bay ecoregion was similar to levels detected in other studies.
Yoder et al. (1999) found that declines in sensitive species in Ohio streams occurred at <5%
urban land use in stream catchments, resulting in part from degraded substrate, and Miltner et al.
(2004) found changes in fish biotic integrity in Ohio streams at levels as low as 5% urban land
use in stream catchments. The higher level of urban land use detected in the Appalachian
Piedmont followed studies in Wisconsin. Wang et al. (1997) found decreases in biotic integrity
in streams with urban land use occurring at 10%; also, Wang et al. (2000; 2001) showed declines

103

in fish species richness and diversity with levels greater than 10% imperviousness in catchments.
In our study regions, percent impervious land cover in catchments was highly correlated with
percent urban land use in catchments (R2 = 0.97, Table C2.1). This is also consistent with
studies in Ohio documenting significant declines in fish IBI scores with impervious land use >
15% (Yoder et al. 1999; Miltner et al. 2004). Differences between the Appalachian Piedmont
and Chesapeake Bay ecoregions (and differences between these regions from the other three
study regions) may be related to different mechanisms from land use and/or differences in fish
composition. Variability in fish response by region to the percent of urban in the catchment may
be due in part to relationships between natural landscape factors (e.g., lithology) and human land
uses between regions (e.g., Utz et al. 2010), or by compositional differences in fish assemblages
among regions (e.g., Meador et al 2005). However, the prominence of the urban signal in the
Appalachian Piedmont and Chesapeake Bay as opposed to other ecoregions leads us to believe
that regional differences may result from landscape characteristics of the ecoregions. For
example, percent of agriculture in catchments of study sites was low in both the Appalachian
Piedmont and the Chesapeake Bay ecoregion, in comparison to the other three regions,
potentially allowing the urban signal to prevail (Paavola et al. 2006).
Two levels of urban land use in Appalachian Piedmont.—In the Appalachian Piedmont
ecoregion, we detected a secondary influence from urban land use on fish at 38% catchment
urban land. Fish response at this level of urbanization is much higher than typical levels reported
in the literature but may be related to low population abundances and/or low fish species
diversity resulting from extreme levels of urban land use. Miltner et al. (2004) found an almost
complete loss in aquatic life when levels of impervious land cover > 27% for streams in Ohio.

104

Again, these landscape factors are known to have similar effects on streams and are highly
correlated within our study regions.
The influence of urban land use on stream fishes at two levels in the Appalachian
Piedmont (again, 10% and 38%), may reflect a series of threshold responses in the grouping of
fish metrics, indicating a low and high sensitivity of stream fishes to urban land use (Allan
2004). At a low level of urban land use, fish assemblages may begin to change from those that
would exist in the absence of urban land use in catchments, with negative effects on fishes
resulting from initial habitat degradation. At the higher level, fish assemblage metrics may stop
changing because of a loss of specialist species and some functional groups (e.g., invertivores),
then level off and become dominated by a few tolerant species at low to moderate abundances
caused from increased sediments, nutrients, toxicity and/or altered thermal regimes (Yoder et al.
1999; Paul and Meyer 2000; Miltner et al. 2004). As described previously, Yoder et al. (1999)
showed differences in fish assemblages occurring at levels of <5% urban land use in stream
catchments and also at 10% to 15%; they also described responses at levels greater than 15%
including severe fish faunal degradation from increased nutrients and toxicity from urbanizations
in the catchments.
Regionally-specific effects of agricultural land use on stream fish.—Agriculture was the
dominant human land use variable influencing fish assemblages in catchments of the Upper
Mississippi ecoregion at a level of 54%. It was also the secondary human land use in the
Laurentian Great Lakes, Chesapeake Bay, and Middle Missouri, at levels of 68%, 0.40%, and
<0.01%, respectively. These results reflect a high (> 50%) and low level (< 0.40%) agriculture
land use affecting fish assemblages across ecoregions. Other studies showed fish assemblage
change with high levels of agriculture in the catchment. Fitzpatrick et al. (2001) described a
105

decline in fish IBI scores with percentage of catchment agriculture > 30%, and Wang et al.
(1997) found a decline in fish abundance, fish species richness, and fish IBI scores at levels >
50%. Regarding the low levels influencing changes in fish assemblages in the Chesapeake Bay
and Middle Missouri ecoregions, these may be due in part to an initial decline in abundances of
fishes, but effects can also be species-specific which was documented by Utz et al. (2010). Utz
et al. (2010) found that sensitivity of fish species to percent of agriculture in catchments varied
by region, with more species having negative responses catchment agriculture in the Highlands
region than in the Coastal Plain and Piedmont physiographic region, which may be due to
differences in composition of assemblages. Also, fishes sensitive to agriculture may have been
reduced in abundances and range in the Coastal Plains and Piedmont regions from previous
logging in Maryland (i.e., legacy land use effects), but not to the same degree in the Highlands,
leading to compositional differences in fish assemblages and higher sensitivity to agriculture in
the Highland region (Harding et al. 1998). Similar factors could be at play leading to differences
seen in fish assemblage metrics at low levels of agriculture in the catchment, with potential
differences related to natural landscape factors defining the suitability for types of agriculture
(e.g., pastoral versus row crop) like the Sandhills area in the Middle Missouri ecoregion (Allan
2004).
Density of dams in the catchment.—Dam density was the prominent human land use
influencing fish assemblages in the Laurentian Great Lakes and Middle Missouri ecoregions and
secondarily in the Upper Mississippi ecoregion and levels of 0.05/100km2, 0.02/100km2, and
0.04/100km2, respectively. This supports the idea that dams are acting as a landscape influence,
and that they may be a dominant human land use causing changes in fish assemblages. In the
Laurentian Great Lakes ecoregion, density of dams was the prominent human land use that

106

described differences in groups of study sites based on fish metrics, and these results are
supported by work of Wang et al. (2011b) who characterized a unique amount of variation in two
groupings of fish indicators explained by a set of dam metrics that included density of dams in
the catchment. These dam metrics explained 16% to 19% of unique variation in fish bioticintegrity indicators and fish habitat preference indicators, respectively. Wang et al. (2011b),
after selecting for stream reaches with minimal urban and agriculture in their network catchment,
found that the majority of fish assemblage metrics tested had negative associations with dam
metrics. Our study quantified that even a small proportion of dams in the network catchment, ≤
0.05/100km2 can influence fish assemblagesin three out of the five study regions.
Stream–road crossing in the catchment.—Density of stream-road crossings in catchments
was not identified as a dominant anthropogenic landscape factor affecting fishes in any of our
five study regions. Other studies have shown negative influences on fish movement from
stream-road crossings related to types of passage ways at stream-road crossings (e.g., culverts,
slab crossings, open-box, and ford, Warren and Pardew 1998; Schaefer et al. 2003) and/or
passage way design flaws or failures like perching, slope, or crossing length (Mueller et al. 2008;
Bouska and Paukert 2010). Januchowski–Hartley et al. (2013) documented the prevalence that
road crossings pose as possible barriers to fishes in the Great Lakes region and described their
potential for catchment-wide cumulative effects with in river catchments. The additive effects
and the pervasiveness of multiple barriers, including passage ways at stream-road crossings and
small and large dams in river catchments, warrants further investigation to quantify their
catchment-wide effects on fishes (Perkin and Gido 2012; Cooper et al. In Review). The
cumulative effects of stream-road crossing may be important to fish but were just not as

107

important as urban, agriculture and dams in stream catchments on fish assemblage metrics in this
study.
Spatial extent of study and number of sampling sites.—The spatial extent of our study
region included five freshwater ecoregions encompassing an area greater than two million square
kilometers. We tested trends using data from 10,522 study sites on streams and rivers occurring
within twenty two states. To our knowledge, this is the only study to date that has conducted a
regional comparison that captures the spatial extent and potential variability in fish assemblages;
our large data set provided extensive characterizations of species compositions as well as an
equally broad ranges in environmental characteristics. The number of sites evaluated in this
study likely contributed to our ability to detect differences in groups of study sites related to fish
assemblage metrics at multiple levels of influence from agriculture in the catchment, including
our detection of fish assemblage response at such low levels of percent catchment agriculture
(i.e., <0.40%) that we quantified in the Chesapeake Bay and Middle Missouri ecoregions. The
large number of study sites evaluated might have also helped us to distinguish multiple threshold
responses in fish assemblage metrics in the Appalachian Piedmont to percent urban land use in
the catchment. The trends detected in this study offer a comprehensive baseline for additional
evaluation of regional and/or local responses of fish to human land uses.
Management implications and conclusions.—Interregional comparisons into mechanisms
controlling regional fish assemblage response to human land uses may be a fruitful area of
additional investigation. Some of the potential regional differences causing different fish
assemblage response may relate to the buffering capacity of lithology and/or topographic
landscape factors as well as natural land covers in a region. The types and proportions of natural
versus human land cover and their spatial arrangement in the catchment have been shown to
108

affect the delivery of runoff and other materials to stream ecosystems (Zimmerman et al. 2003;
Blann et al. 2009; Carpenter et al. 2011). Intensity of urban classes, specific agricultural
practices, and riparian buffers to streams are a few examples of factors that may influence runoff
and the delivery of materials to riverine ecosystems (Blann et al. 2009; Carpenter et al. 2011),
and such factors can cause heterogeneous responses in stream biota across regions (Meador and
Goldstein 2003; Morgan and Cushman 2005; Utz et al. 2010). Our results and methods can be
used to assess regional dominance of classes of human land use types that are affecting changes
in fish assemblages and can provide insights into levels of human land use that are causing
change in fish assemblage metrics.
Our study underscored regional differences in dominant human land uses affecting
differences in groups of sites based on fish metrics. If dominant human land uses and their levels
of influence on critical management factors have regionally-specific influences, these influences
should be accounted for and incorporated into the planning process to develop realistic
management priorities. Studies characterizing critical human pressures in one ecoregion may not
be directly applicable to a different ecoregion because of these differences. Well-designed
regionalized studies could help further characterize regional differences in human land uses
effects on stream biota, elucidate important mechanisms of influence on stream systems, and
have transferable results or components to help meet management priorities across large spatial
extents.
In summary, we quantified unique proportions of variance explained by natural landscape
variables versus human land uses in fish assemblage metrics and described how they vary by
ecoregion. After controlling for dominant natural landscape factors known to influence fish, we
identified that dominant human land uses (i.e., percent urban and agricultural in the catchment
109

and density of dams in the catchment) affecting change in fish assemblage metrics, vary by
ecoregions. By looking over a wide spatial extent and evaluating over 10,000 study sites we
were able to detect multiple levels of influence seen by fish assemblage response to both urban
and agriculture in the catchment across and within study regions. Results from this study
highlight that there are regional differences in dominant human land uses influencing changes in
fish assemblage structure and that there are regional disparities in the levels of urban and
agriculture in the landscape that are affecting fish. This study provides a better understanding of
these regional differences which is vital information for establishing realistic management
strategies in the hopes of preserving the biodiversity of freshwater ecosystems and to stop their
continued degradation. Considering that human populations will continue to grow and land
conversion and human development of landscapes will continue to intensity, we need responsible
planning and better understanding for future development and land conversion to protect the
diversity and complexity of the freshwater resources for future generations.

110

APPENDICES

111

APPENDIX 2.A

TABLES
Table 2.1.—List of 52 landscape variables initially evaluated for regional comparison.
Variable code
Natural
Land use (%)
Water
Wetland
Barren
Forest
Shrub
Grassland
Water_c
Wetland_c
Barren_c
Forest_c
Shrub_c
Grassland_c
Catchment area
AREASQKM
AREASQKMC
Slope
Slope
Elevation
ELEV_MEAN
ELEV_MEANC
Lithology
fineC
coarseC
Soil hydrology
Base-flow index
SOILHYGRP
SOILPERM
SOILHYGRPC
SOILPERMC
Climate
temp
AREAWTMAP

Landscape variable description

Water in local catchment
Wetland in local catchment
Barren land in local catchment
Forest in local catchment
Shrub in local catchment
Grassland in local catchment
Water in network catchment
Wetland in network catchment
Barren in network catchment
Forest in network catchment
Shrub in network catchment
Grassland in network catchment
Local catchment area (km2)
Network catchment area (km2) †*
Mean local catchment slope (degrees)
Mean local catchment elevation (m) †*
Mean network catchment elevation (m)
Fine lithology in the network catchment (%) †*
Coarse lithology in the network catchment (%)
Network catchment groundwater contribution to baseflow (%) †*
Mean local catchment soil hydrologic relative rates * 10 (infiltration rates
ranged from high = 1 to very slow = 4)
Local catchment soil permeability rates (cm/hour * 100)
Mean network catchment soil hydrologic relative rates * 10 (infiltration
rates ranged from high = 1 to very slow = 4)
Mean network catchment soil permeability rates (cm/hour * 100)

Mean annual air temperature in local catchment (°C) †*
Area weighted average mean annual precipitation in network catchment
(ml) †*
† Landscape variables used in the redundancy analysis (RDA)
* Landscape variables used in the cascade multivariate regression tree (CMRT) analysis

112

Table 2.1.—(cont’d)
Variable code
Human
Land use (%)
Pasture
Crops
Pasture_Crops
Developed_open
Developed_low
Developed_medium
Developed_high
Developed_lmh
Developed_all
Pasture_c
Crops_c
Pasture_Crops_c
Developed_open_c
Developed_low_c
Developed_medium_c
Developed_high_c
Developed_lmh
Developed_all_c
IMPERVC
Population density
POPDENSCz
Roads

Landscape variable description

Pasture/hay in local catchment
Cultivated crops in local catchment
Pasture/hay and cultivated crops in local catchment
Open space developed land in local catchment
Low intensity developed land in local catchment
Medium intensity developed land in local catchment
High intensity developed land in local catchment
Low, medium, and high intensity developed land in local catchment
All developed land in local catchment
Pasture/hay in network catchment †
Cultivated crops in network catchment †
Pasture/hay and cultivated crops in network catchment *
Open space developed land in network catchment †
Low intensity developed land in network catchment
Medium intensity developed land in network catchment
High intensity developed land in network catchment
Low, medium, and high intensity developed land in network catchment †
All developed land in network catchment *
Impervious surfaces in the network catchment
Population density in the network catchment (number/km2)
2

ROAD_CROSS_CTz Road crossing in local catchment (number/100 km )
ROAD_LENz

Road density in local catchment (km/100 km2)

ROAD_CROSSCz

Road crossing in network catchment (number/100 km2) †*

ROAD_LENCz
Barriers

Road density in network catchment (km/100 km2)

DAM_CTCz
Canal/Ditch

Dam density in network catchment (number/100 km2) †*

CANAL_DITCHCz
Canal/ditch in network catchment (m/km2)
† Landscape variables used in the redundancy analysis (RDA)
* Landscape variables used in the cascade multivariate regression tree (CMRT) analysis

113

Table 2.2.—Fish metrics calculated and evaluated for a regional landscape comparison.

Fish metric description
Community
Fish family richness
Fish species richness †
Pielou community evenness index (J ')
Shannon-Weaver community diversity index (H ')
Tolerance
Proportion of individuals that are intolerant †
Proportion of individuals that are tolerant
Threatened or endangered
Proportion of individuals that are threatened or endangered
Game species
Proportion of individuals that are game fish †
Trophic
Proportion of individuals that are generalists
Proportion of individuals that are herbivores
Proportion of individuals that are invertivores †
Proportion of individuals that are omnivores †
Proportion of individuals that are piscivores
Life history
Proportion of individuals that are hiders
Proportion of individuals that are lithophilic †
Proportion of individuals that are migrating (vagile)
Proportion of individuals that are nest gurarders
Proportion of individuals that are rheophilic †
Habitat
Proportion of individuals that prefer benthic
Proportion of individuals that prefer large river
Proportion of individuals that prefer lotic
Proportion of individuals that prefer water column
† Fish metrics used in redundance analysis (RDA) and
cascade multivariate regression tree (CMRT) analyses

114

Table 2.3.—Descriptive statistics of landscape variables for the five freshwater ecoregion with
landscape variable descriptions (i.e., mean, minimum, maximum, and the 10th and 90th
percentiles).

Landscape variable description
Natural

Mean

Network catchment area (km2)
Fine lithology in network catchment (%)
Mean local catchment elevation (m)
Base-flow index, network catchment
groundwater contribution to baseflow (%)
Mean annual air temperature in local
catchment (°C)
Area weighted average mean annual
precipitation in network catchment (ml)
Human land uses
Urban (%)
Developed, open space in network
catchment *
Developed, low, medium & high intensity
in network catchment *
Developed, open space, low, medium, &
high intensity in network catchment †
Agriculture (%)
Pasture/hay in network catchment *
Cultivated crops in network catchment *
Pasture/hay & cultivated crops in network
catchment †
Road crossings
Road crossings in network catchment

155.16

Appalachian Piedmont
Min
Max
10%
1.42 8062.60

21.76
157.30
45.95

0.00
1.39
25.23

100.00
937.66
78.30

15.64

9.53

19.33

90%

11.42 180.03
0.00 100.00
20.42 298.06
30.54
60.35
14.11

17.52

1215.13 973.82 2372.90 1125.53 1317.59

8.44

0.00

77.33

2.85

18.88

5.76

0.00

77.97

0.09

16.81

14.20

0.00

98.62

3.13

37.22

14.06
7.27
21.34

0.00
0.00
0.00

63.49
68.27
72.63

1.43
0.00
3.83

32.49
26.55
40.93

48.96

0.00

345.30

15.82

89.03

3.32

0.00

57.56

0.00

9.06

2

(number/100 km ) *†
Dams
Dam density (number/100 km2) *†
* Used in RDA
† Used in CMRT analysis

115

Table 2.3.—(cont’d)

Chesapeake Bay
Min
Max
10%

Landscape variable description
Natural

Mean

Network catchment area (km2)
Fine lithology in network catchment (%)
Mean local catchment elevation (m)
Base-flow index, network catchment
groundwater contribution to baseflow (%)
Mean annual air temperature in local
catchment (°C)
Area weighted average mean annual
precipitation in network catchment (ml)
Human land uses
Urban (%)
Developed, open space in network
catchment *
Developed, low, medium & high intensity
in network catchment *
Developed, open space, low, medium, &
high intensity in network catchment †
Agriculture (%)
Pasture/hay in network catchment *
Cultivated crops in network catchment *
Pasture/hay & cultivated crops in network
catchment †
Road crossings
Road crossings in network catchment

371.17

0.82 8483.38

3.48 560.87

13.22
288.25
46.26

0.00 100.00
2.52 1050.68
26.99
64.02

0.00
99.80
21.71 491.14
38.08
54.74

9.57

6.35

14.28

7.00

90%

13.25

1027.70 806.87 1239.06 914.37 1108.89

4.55

0.00

39.90

0.95

8.50

3.61

0.00

83.76

0.01

7.09

8.15

0.00

98.44

1.27

14.85

19.55
12.04
31.59

0.00
0.00
0.00

69.61
81.75
94.91

4.84
0.51
7.72

36.71
27.43
63.49

53.99

0.00

588.24

9.40

96.04

1.10

0.00

30.53

0.00

3.04

2

(number/100 km ) *†
Dams
Dam density (number/100 km2) *†
* Used in RDA
† Used in CMRT analysis

116

Table 2.3.—(cont’d)

Landscape variable description
Natural

Mean

Network catchment area (km2)
Fine lithology in network catchment (%)
Mean local catchment elevation (m)
Base-flow index, network catchment
groundwater contribution to baseflow (%)
Mean annual air temperature in local
catchment (°C)
Area weighted average mean annual
precipitation in network catchment (ml)
Human land uses
Urban (%)
Developed, open space in network
catchment *
Developed, low, medium & high intensity
in network catchment *
Developed, open space, low, medium, &
high intensity in network catchment †
Agriculture (%)
Pasture/hay in network catchment *
Cultivated crops in network catchment *
Pasture/hay & cultivated crops in network
catchment †
Road crossings
Road crossings in network catchment

354.81

Laurentian Great Lakes
Min
Max
10%
0.23 9533.74

41.97
277.15
45.04

0.00
74.56
18.69

7.74

2.12

90%

10.27 758.72

100.00
0.00 100.00
680.78 180.28 430.92
87.95 21.99
68.45
10.56

4.69

9.68

929.80 641.04 1566.84 764.17 1123.50

5.92

0.00

58.15

0.88

12.08

5.13

0.00

93.37

0.06

12.57

11.05

0.00

99.92

1.15

26.31

10.61
29.24
39.85

0.00
0.00
0.00

62.63
94.26
94.26

0.01
0.06
0.47

26.51
79.76
83.07

48.49

0.00

413.37

8.35

91.15

1.13

0.00

73.69

0.00

2.84

2

(number/100 km ) *†
Dams
Dam density (number/100 km2) *†
* Used in RDA
† Used in CMRT analysis

117

Table 2.3.—(cont’d)

Landscape variable description
Natural

Mean

Network catchment area (km2)
Fine lithology in network catchment (%)
Mean local catchment elevation (m)
Base-flow index, network catchment
groundwater contribution to baseflow (%)
Mean annual air temperature in local
catchment (°C)
Area weighted average mean annual
precipitation in network catchment (ml)
Human land uses
Urban (%)
Developed, open space in network
catchment *
Developed, low, medium & high intensity
in network catchment *
Developed, open space, low, medium, &
high intensity in network catchment †
Agriculture (%)
Pasture/hay in network catchment *
Cultivated crops in network catchment *
Pasture/hay & cultivated crops in network
catchment †
Road crossings
Road crossings in network catchment

764.33

Middle Missouri
Min
Max
10%
1.57 9984.84

90%

15.36 2196.10

32.01
0.00 100.00
0.00 100.00
591.36 195.32 3134.90 288.62 1165.51
38.90 12.20
84.20 19.19
69.00
10.22

0.19

12.92

8.11

12.29

708.96 344.53 1113.56 441.07 929.31

3.88

0.00

34.60

1.75

5.48

2.06

0.00

82.72

0.04

2.45

5.93

0.00

92.22

1.99

7.65

9.92
41.58
51.50

0.00
0.00
0.00

80.99
93.56
94.88

0.00
2.05
4.00

39.88
82.94
87.86

35.90

0.00

254.61

3.63

70.55

3.91

0.00

73.31

0.00

11.60

2

(number/100 km ) *†
Dams
Dam density (number/100 km2) *†
* Used in RDA
† Used in CMRT analysis

118

Table 2.3.—(cont’d)

Landscape variable description
Natural

Mean

Network catchment area (km2)
Fine lithology in network catchment (%)
Mean local catchment elevation (m)
Base-flow index, network catchment
groundwater contribution to baseflow (%)
Mean annual air temperature in local
catchment (°C)
Area weighted average mean annual
precipitation in network catchment (ml)
Human land uses
Urban (%)
Developed, open space in network
catchment *
Developed, low, medium & high intensity
in network catchment *
Developed, open space, low, medium, &
high intensity in network catchment †
Agriculture (%)
Pasture/hay in network catchment *
Cultivated crops in network catchment *
Pasture/hay & cultivated crops in network
catchment †
Road crossings
Road crossings in network catchment

356.01

Upper Mississippi
Min
Max
10%
0.23 9964.22

9.49
0.00
288.86 114.85
49.85
7.84
7.56

3.07

90%

9.65 740.40

100.00
0.00
27.43
589.09 184.87 382.03
74.87 31.21
66.29
13.60

5.48

10.61

823.55 556.81 1190.27 704.84 949.22

4.56

0.00

30.15

2.32

6.81

4.09

0.00

92.83

0.10

7.25

8.65

0.00

99.46

3.24

13.05

12.34
47.87
60.21

0.00
0.00
0.00

68.33
98.06
98.06

0.39
5.14
15.54

27.99
87.11
89.80

39.58

0.00

643.09

8.88

69.17

0.77

0.00

39.03

0.00

1.88

2

(number/100 km ) *†
Dams
Dam density (number/100 km2) *†
* Used in RDA
† Used in CMRT analysis

119

Table 2.4.—Descriptive statistics (i.e., mean, range, 10th and 90th percentiles) of fish metrics in five freshwater ecoregions with
number of sample sites per region.
Freshwater ecoregion
Fish richness Intolerant
Game fish Invertivore Omnivore
Lithophilic Rheophilic
(# of sites)
count
% individual % individual % individual % individual % individual % individual
Appalachian
Piedmont (1083)
Mean
15
1.27
28.46
65.38
23.90
49.85
20.25
Range
1—34
0—95.81 0—100.00 0—100.00
0—93.44 0—100.00
0—97.13
10%
8
0.00
3.11
40.00
3.71
4.86
0.00
90%
22
2.59
67.26
88.31
45.74
86.65
48.83
Chesapeake Bay
(807)
Mean
10
12.67
22.15
35.09
14.12
45.41
19.80
Range
1—31 0—100.00 0—100.00 0—100.00 0—100.00 0—100.00 0—100.00
10%
2
0.00
0.00
0.00
0.00
0.00
0.00
90%
19
37.15
68.91
69.67
38.54
83.00
53.44
Laurentian Great
Lakes (3773)
Mean
11
12.53
20.63
38.90
18.10
27.76
29.01
Range
1—37 0—100.00 0—100.00 0—100.00 0—100.00 0—100.00 0—100.00
10%
4
0.00
0.00
4.87
0.00
0.00
0.00
90%
19
43.80
72.48
76.54
45.97
65.08
69.95
Middle Missouri
(886)
Mean
10
2.60
9.42
35.78
37.79
10.71
42.68
Range
1—29 0—100.00 0—100.00 0—100.00 0—100.00 0—100.00 0—100.00
10%
4
0.00
0.00
4.29
2.79
0.00
0.29
90%
17
3.23
26.51
70.80
81.00
35.31
83.68
Upper Mississippi
(3973)
Mean
13
11.75
15.07
42.75
25.97
23.65
24.19
Range
1—48 0—100.00 0—100.00 0—100.00 0—100.00 0—100.00 0—100.00
10%
3
0.00
0.00
3.45
0.93
0.00
0.00
90%
24
37.05
55.25
78.70
60.33
57.96
66.34

120

Table 2.5.—Principle component analysis (PCA) results for six natural landscape variables for
all study sites across the entire study area. Variable descriptions and axes weightings are
provided. Total amount of variation explained was 59.91%.

Variance explained (%)
Natural landscape variable description
Mean annual air temperature in local catchemnt
Area weighted average mean annual percipitation in
network catchment
Fine lithology in network catchment
Network catchment area
Base-flow index, network catchment groundwater
contribution to baseflow
Mean local catchment elevation

121

Axis 1 Axis 2
39.10 20.81
0.83
0.68

-0.01
-0.54

0.53
-0.02
-0.58

0.55
0.64
-0.44

-0.77

0.21

Table 2.6.—Variable, value and percent variance explained from the main (natural) and (human)
influences on fish assemblage metrics across five freshwater ecoregions from cascade
multivariate regression tree analysis (CMRT).
Freshwater ecoregion
(# of sites)
All data (10522)

Appalachian Piedmont
(1083)

Chesapeake Bay (807)

Laurentian Great
Lakes (3773)

Middle Missouri
(886)

Upper Mississippi
(3973)

Variable
Value

Main
natural
Natural
0.12

Human 1
Agriculture
53.46%

Human 2
Agriculture
64.17%

% Variance
explained

3.66%

2.01%

1.54%

Variable
Value

Natural
1.40

Urban
9.93%

Urban
38.24%

% Variance
explained

10.17%

1.83%

1.05%

Variable
Value

Natural
0.08

Urban
1.30%

Agriculture
0.40%

% Variance
explained

5.91%

4.35%

0.55%

Variable

Natural

Dams

Agriculture

Value

0.08

2

0.05/100 km

67.67%

% Variance
explained

7.04%

1.46%

1.10%

Variable

Natural

Dams

Agriculture

2

Value
% Variance
explained

-1.55

0.02/100 km

<0.01%

7.45%

2.95%

1.96%

Variable

Natural

Agriculture

Dams

Value

-0.25

53.64%

0.04/100 km

% Variance
explained

5.65%

2.78%

0.34%

122

Total variance
explained

7.21%

13.05%

10.81%

9.60%

12.36%

2

8.77%

APPENDIX 2.B

FIGURES

Figure 2.1.—Five freshwater ecoregions of the United States that comprised the study area
(Abell et al. 2008)

123

Figure 2.2.—Results from RDA partitioned variation explained in fish community metrics by
natural and human variables across the five freshwater ecoregions in our study area.

124

APPENDIX 2.C

SUPPLEMENTAL TABLES

125

IMPERVC

POPDENSC

SOILPERM

SOILHYGRP

ELEV_MEAN

Crop_c

Developed_allc

Developed_lmhc

Developed_highc

Landscape
variables
Wetland_c
0.75
Forest_c
0.82
Shrub_c
0.81
Grassland_c
0.85
Developed_med
0.82
Developed_high
0.88
Developed_lmh
0.97 0.92 0.82
Developed_all
0.79 0.90 0.86 0.77 0.93
Pasture_Crops
0.90
Developed_lowc
0.73
0.73 0.71
0.74
Developed_medc
0.70
0.71
0.90
Developed_highc
0.83
Developed_lmhc
0.71
0.74 0.72
0.72 0.98
Developed_allc
0.74
0.87 0.94
Pasture_c
0.76
Crop_c
-0.77
0.85 0.75
Pasture_Crops_c
-0.74
0.79 0.82
ELEV_MEANC
SOILHYGRPC
SOILPERMC
POPDENSC
0.81
IMPERVC
0.73 0.72
0.75 0.96
ROAD_LENCz
0.74 0.75

Developed_medc

Developed_lowc

Developed_openc

Pasture_Crops

Crops

Pasture

Developed_all

Developed_lmh

Developed_high

Developed_med

Developed_low

Developed_open

Forest_c

Grassland

Shrub

Forest

Wetland

Table C2.1.—Results of Pearson correlation among landscape metrics; only results with r values ≥.0.70 or ≤ -0.70 are displayed.

0.95
0.96 0.90
0.91 0.86 0.96

0.92
0.96
0.79
0.83
0.78 0.73 0.81 0.79
0.97 0.93 0.99 0.97
0.71
0.74 0.78

0.81
0.76 0.76

Table C2.2.—PCA results for landscape variables including variance explained by each axis and
axis interpretation.
Variance explained
Cumulative variance explained

28.12
28.12

Axis interpretation
Landscape variable description
All developed land in network catchment (%)
Low, medium, and high intensity developed land in network
catchment (%)
Low intensity developed land in network catchment (%)
Medium intensity developed land in network catchment (%)
High intensity developed land in network catchment (%)

12.73
10.08
9.59
7.88
40.85
50.93
60.52
68.39
Groundwater
Temperature Road
Urban
index
Agriculture precipitation crossing
axis 1
axis 2
axis 3
axis 4
axis 5
0.98
0.09
-0.06
0.05
-0.08

6.99
75.39
Dams
axis 6
-0.02

0.97
0.95
0.94
0.90

0.06
0.07
0.06
0.05

-0.05
-0.02
-0.09
-0.11

0.05
0.06
0.04
0.02

-0.01
-0.02
0.05
0.09

-0.06
-0.03
-0.08
-0.09

0.82
0.81

0.02
0.11

0.13
0.04

0.23
0.10

-0.12
-0.12

0.15
0.10

Road crossing in network catchment (number/100 km )
Mean annual air temperature in local catchment (°C)
Fine lithology in the network catchment (%)

0.41
0.25
0.19

0.21
0.24
0.49

0.11
0.14
0.17

0.07
0.74
0.28

-0.70
0.02
0.02

0.03
0.25
-0.49

Dam density in network catchment (number/100 km2)
Mean network catchment soil hydrologic relative rates * 10
(infiltration rates ranged from high = 1 to very slow = 4)

0.17

0.27

-0.24

0.00

0.13

0.56

0.09

0.77

0.00

-0.03

0.05

-0.30

Canal/ditch in network catchement (m/km )
Coarse lithology in the network catchment (%)

0.09
0.03

0.12
-0.53

0.32
-0.14

-0.03
-0.15

0.46
0.35

-0.33
-0.22

Mean network catchment area (km2)
Area weighted average mean annual precipitation in network
catchment (ml)
Cultivated crops in network catchment (%)
Pasture/hay in network catchment (%)
Pasture/hay & cultivated crops in network catchment (%)
Network catchment soil permeability rates (cm/hour * 100)
Base-flow index, network catchment groundwater
contribution to baseflow (%)
Mean local catchment elevation (m)

0.03

0.01

0.08

0.00

0.87

0.06

0.02
-0.06
-0.07
-0.09
-0.09

0.17
0.19
-0.07
0.19
-0.85

-0.35
0.91
0.15
0.95
-0.17

0.72
-0.06
0.20
0.02
-0.02

-0.15
0.07
-0.15
0.01
0.15

0.16
-0.15
0.71
0.12
-0.21

-0.10
-0.12

-0.80
0.02

-0.14
-0.04

-0.19
-0.89

-0.10
-0.01

-0.02
0.13

Road density in network catchment (km/100 km2)
Open space in network catchment (%)
2

2

126

Table C2.3.—Pearson Correlation of fish metrics, only results with r values ≥.0.70 or ≤ -0.70
are displayed.

Shannon's diversity index (H' ) Fish species richness
Fish species richness
0.86
Fish family richness
0.73
0.80

127

Table C2.4.—PCA results for fish metrics.
variance explained
cumulative variance explained

16.74
16.74

15.45
32.20

10.19
42.39

10.06 9.76
52.45 62.21

7.09
69.31

fish
rheophilic
river / omnivore /
Axis interpretation
richness intolerant /water column lithophilic game invertivore
Fish metric
axis 1
axis 2
axis 3
axis 4 axis 5
axis 6
Fish species richness
0.91
-0.14
0.18
0.10 0.07
0.02
Invertivore individual (%)
0.46
-0.25
0.00
-0.27 0.07
-0.62
Lithophilic individual (%)
0.30
-0.23
-0.02
0.60 -0.11
-0.19
Lotic individual (%)
0.29
-0.34
0.34
0.57 -0.22
-0.04
Large river individual (%)
0.28
-0.33
0.20
-0.18 0.69
0.29
Omnivore individual (%)
0.23
-0.29
-0.02
-0.23 0.06
0.75
Herbivore individual (%)
0.07
-0.15
0.70
0.14 -0.20
-0.02
Rheophilic individual (%)
0.03
0.13
0.75
0.12 0.33
0.04
Tolerant individual (%)
-0.02
-0.41
-0.21
0.33 -0.37
0.51
Nest gurarder individual (%)
-0.05
-0.18
-0.21
-0.66 -0.21
-0.07
Threatened or endangered individual (%)
-0.06
0.79
0.21
-0.06 -0.07
0.11
Migrating (vagile) individual (%)
-0.09
0.81
-0.32
-0.06 0.01
-0.12
Intolerant individual (%)
-0.13
0.81
0.05
0.11 -0.06
-0.19
Hiders individual (%)
-0.21
0.46
-0.01
0.65 -0.06
0.11
Game fish individual (%)
-0.23
0.00
-0.14
0.20 0.68
-0.13
Generalist individual (%)
-0.24
-0.19
-0.04
0.52 -0.66
0.06
Carnivore individual (%)
-0.27
0.70
-0.18
0.07 0.50
-0.07
Benthic individual (%)
-0.32
0.08
0.39
0.05 -0.32
0.25
Water column individual (%)
-0.39
0.03
-0.70
0.04 0.02
0.16
Pielou's evenness index (J' )
-0.58
0.04
-0.25
0.03 0.09
-0.20

128

REFERENCES

129

REFERENCES

Abell, R. A., D. M. Olsen, E. Dinerstein, P. T. Hurley, J. T. Diggs, W. Eichbaum, S. Walters, W
Wettengel, T. Allnutt, C. J. Loucks, and P. Hedao. 2000. Freshwater ecoregions of North
America: a conservation assessment. Island Press, Washington, D.C.
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. López, R. E. Reis, J. G. Lundberg, M. H.
Sabaj Pérze, and P. Petry, 2008. Freshwater ecoregions of the world: a new map of
biogeographic units for freshwater biodiversity conservation. Bioscience 58:403–414.
Alford, J. B., 2014. Multi-scale assessment of habitats and stressors influencing stream fish
assemblages in the Lake Pontchartrain Basin, USA Hydrobiologia. 738:129–146.
Allan, J. D. 2004. Landscapes and riverscapes: the influence of land use on stream ecosystems.
Annual Review of Ecology, Evolution, and Systematics 35: 257–284.
Allan, J. D., D. L. Erickson, and J. Fay. 1997. The influence of catchment land use on stream
integrity across multiple spatial scales. Freshwater Biology 37:149–61.
Allan, J. D. and A. S. Flecker. 1993. Biodiversity Conservation in Running Waters. BioScience
43:1 32–43.
Angermeier, P. L., R. A. Smogor, and J. R. Stauffer. 2000. Regional frameworks and candidate
metrics for assessing biotic integrity in Mid-Atlantic Highland streams. Transactions of
the American Fisheries Society, 129(4):962-981.
Barbour M. T., J. Gerritsen, B. D. Snyder, and J. B. Stribling. 1999. Rapid Bioassessment
Protocols for use in streams and wadeable rivers: periphyton, benthic macroinvertebrates
and fish. Second Edition. Rep. EPA/841–B-99–002, US EPA, Off. Water, Washington,
D. C.
Benda, L., N. L. Proff, D. Miller, T. Dunne, G. Reeves, G. Pess and M. Pollock. 2004. The
network dynamics hypothesis: how channel networks structure riverine habitats.
BioScience, 54:413–427.
Blann K. L., J. L. Anderson, G. R. Sands, and B. Vondracek. 2009. Effects of agricultural
drainage on aquatic ecosystems: a review. Critical Reviews in Environmental Science
and Technology 39:909–1001.
Borcard, D., F. Gillet, and P. Legendre. 2011. Numerical Ecology with R. Springer Verlag, New
York, pp. 1–319.

130

Borcard, D., P. Legendre, and P. Drapeau. 1992. Partialling out the spatial component of
ecological variation. Ecology 73:1045–1055.
Bouska, W. W. and C. P. Paukert. 2010. Road crossing designs and their impact on fish
assemblages of Great Plains streams. Transactions of the American Fisheries Society
139:214–222.
Brenden, T. O., R. D. Clark Jr., A. R. Cooper, P. W. Seelbach, L. Wang, S. S. Aichele, and E. G.
Bissell 2006. A GIS framework for collecting, managing, and analyzing multi-scale
variables across large regions for river conservation and management. Pages 49–74 in R.
M. Hughes, L. Wang, and P. W. Seelbach, editors., Landscape influences on stream
habitats and biological assemblages. American Fisheries Society, Symposium 48,
Bethesda, Maryland.
Burkhead, N. M. 2012. Extinction rates in North American freshwater fishes 1900–2010.
BioScience 62:798–808.
Carpenter, S. R., E. H. Stanley, and M. J. V. Zanden. 2011. State of the world’s freshwater
ecosystems: physical, chemical, and biological changes. The Annual Review of
Environment and Resources 36:75–99.
Cooke, S. J., C. Paukert, and Z. Hogan. 2012. Endangered river fish: factors hindering
conservation and restoration. Endangered Species Research 17:179–191.
Cooper, A. R. et al. In Review. Fragmentation by dams: spatial measures of dam effects on
streams of the conterminous U.S. Landscape Ecology.
Cress, J., D. R. Soller, R. Sayre, P. Comer, and H. Warner. 2010. Terrestrial Ecosystems,
Surficial Lithology of the Conterminous United States: U.S. Geological Survey Scientific
Investigations Map 3126, scale 1:5,000,000, 1 sheet. Available at:
(http://pubs.usgs.gov/sim/3126).
Daly, C., M. Halbleib, J. I. Smith, W. P. Gibson, M. K. Doggett, G. H. Taylor, J. Curtis, and P.
P. Pasteris. 2008. Physiographically sensitive mapping of climatological temperature and
precipitation across the conterminous United States. International Journal of Climatology
28:2031–2064.
De’ath, G. 2006. mvpart: multivariate partitioning. R package version 1.2–4.
De’ath, G. 2002. Multivariate regression trees: a new technique for constrained classification
analysis. Ecology 83:1105–1117.
Dudgeon, D., A. H. Arthington, M. O. Gessner, Z. Kawabata, D. J. Knowler, C. Lévêque, R. J.
Naiman, A. Prieur-Richard, D. Soto, M. L. J. Stiassny, and C. A. Sullivan. 2006.

131

Freshwater biodiversity: importance, threats, status and conservation challenges.
Biological Review 81:163–182.
Esselman, P. C., and J. D. Allan. 2010. Relative influences of catchment- and reach-scale abiotic
factors on freshwater fish communities in rivers of northeastern Mesoamerica. Ecology
of Freshwater Fish. 19:439–454.
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 landscape anthropogenic activities. Ecological Restoration 29:133–151.
Esselman, P. C., D. M. Infante, L. Wang, A. R. Cooper, D. Wieferich, Y. Tsang, D. J.
Thornbrugh, and W. W. Taylor. 2013. Regional fish community indicators of landscape
disturbance to catchments of conterminous United States. Ecological Indicators 26:163–
173.
Fitzpatrick, F. A., B. C. Scudder, B. N. Lenz, and D. J. Sullivan. 2001. Effects of multi-scale
environmental characteristics on agricultural stream biota in eastern Wisconsin. Journal
of the American Water Resources Association 37:1489–1507.
Frissell, C. A., W. J. Liss, C. E. Warren and M. D. Hurley. 1986. A hierarchical framework for
stream habitat classification: Viewing streams in a watershed context. Environmental
Management. 10:199–214.
Frimpong, E. A., and P. L. Angermeier. 2009. Fish Traits: a database of ecological and lifehistory traits of freshwater fishes of the United States. Fisheries 34:487–495.
Gesch, D. B., 2007. The National Elevation Dataset, in D. Maune, ed., Digital Elevation Model
Technologies and Applications: The DEM User’s Manual, 2nd Edition: Bethesda,
Maryland, American Society for Photogrammetry and Remote Sensing p. 99–118.
Grant, E. H. C., W. H. Lowe, and W. F. Fagan. 2007. Living in the branches: population
dynamics and ecological processes in dendritic networks. Ecological Letters 10(2):165–
175.
Harding, J. S., E. F. Benfield, P. V. Bolstad, G. S. Helfman, E. B. D. Jones. 1998. Stream
biodiversity: the ghost of land use past. Proceedings of the National Academy of Sciences
95:14843–14847.
Helfman, G. S. 2007. Fish conservation: A guide to understanding and restoring global aquatic
biodiversity and fishery resources. Island Press, Washington, D.C.
Homer, C., C. Huang, L. Yang, B. Wylie, and M. Coan. 2004. Development of a 2001 National
Landcover Database for the United States. Photogrammetric Engineering and Remote
Sensing, Vol. 70, No. 7, pp. 829–840. Available at:
(http://www.mrlc.gov/nlcd_multizone_map.php).

132

ITIS. 2010. Retrieved [04, 26, 2010], from the Integrated Taxonomic Information System on-line
database, Available at: (http://www.itis.gov).
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 11(4):211-217.
Jelks, H. L., S. J. Walsh, N. M. Burkhead, S. Contreras-Balderas, E. Díaz-Pardo, D. A.
Hendrickson, J. Lyons, N. E. Mandrak, F. McCormick, J. S. Nelson, S. P. Platania, B. A.
Porter, C. B. Renaud, J. J. Schmitter-Soto, E. B. Taylor, and M. L. Warren Jr., 2008.
Conservation status of imperiled North American freshwater and diadromous fishes.
Fisheries 33 (8), 372–407.
Karr, J. R. and E. W. Chu. 2000. Sustaining living rivers. Hydrobiologia 422:423:1–14
Karr, J. R. and E. W. Chu. 1999. Restoring life in running waters: better biological monitoring.
Island Press, Washington, D.C., USA.
Legendre, P. and L. Legendre. 2012. Numerical ecology, 3rd English edition. Amsterdam, The
Netherlands: Elsevier Science BV.
Lyons J., L. Wang, and T. Simonson. 1996. Development and validation of an index of biotic
integrity for cold-water streams in Wisconsin. North American Journal of Fisheries
Management 16: 241–256.
Lyons, J. 1992. Using the index of biotic integrity (IBI) to measure environmental quality in
warmwater streams of Wisconsin. USDA Forest Service. General technical report NC–
149.
Matthews, W. J. 1998. Patterns in freshwater fish ecology. Chapman and Hall, New York.
Maxwell, J. R., C. J. Edwards, M. E. Jensen, S. J. Paustian, H. Parrott, and D. M. Hill. 1995. A
Hierarchical Framework of Aquatic Ecological Units in North America (Nearctic Zone).
St. Paul (MN): USDA Forest Service, North Central Forest Experiment Station. General
Technical Report NC-176.
Meador, M. R., J. F. Coles, and H. Zappia, 2005. Fish assemblage responses to urban intensity
gradients in contrasting metropolitan areas: Birmingham, Alabama, and Boston,
Massachusetts. Pages 409–422 in L. R. Brown, R. H. Gray, R. M. Hughes, M. R.
Meador, editors, Effects of Urbanization on Stream Ecosystems. American Fisheries
Society, Bethesda.
Meador M. R., and R. M. Goldstein. 2003. Assessing water quality at large geographic scales:
relations among land use, water physicochemistry, riparian condition, and fish
community structure. Environmental Management 31:504–517.

133

Miltner, R. J., D. White, and C. Yoder. 2004. The biotic integrity of streams in urban and
suburbanizing landscapes. Landscape and Urban Planning 69:87–100.
Morgan, R. P., and S. E. Cushman. 2005. Urbanization effects on stream fish assemblages in
Maryland, USA. Journal of the North American Benthological Society 24:643–655.
Mueller, R. P., S. S. Southard, C. W. May, W. H. Pearson, and V. I. Cullinan. 2008. Juvenile
coho salmon leaping ability and behavior in an experimental culvert test bed.
Transactions of the American Fisheries Society 137:941–950.
National Oceanographic and Atmospheric Administration (NOAA). 2010. Development sprawl
impacts on the terrestrial carbon dynamics of the United States. Available at:
(http://www.ngdc.noaa.gov/eog/dmsp/download_sprawl.html).
Norris, R. H. and M. C. Thoms. 1999. What is river health? Freshwater Biology. 41:197–209.
Oksanen, J., F. G. Blanchet, R. Kindt, P. Legendre, P. R. Minchin, R. B. O'Hara, G. L. Simpson,
P. Solymos, M. H. H. Stevens and H. Wagner. 2013. vegan: Community Ecology
Package. R package version 2.0-9. Available at: (http://CRAN.Rproject.org/package=vegan).
Olden, J. D., and R. J. Naiman. 2010. Incorporating thermal regimes into environmental flows
assessments: modifying dam operations to restore freshwater ecosystem integrity.
Freshwater Biology 55:86–107.
Ormerod, S. J. 2003. Current issues with fish and fisheries: editor’s overview and introduction.
Journal of Applied Ecology 40:204–213.
Ouellette, M. H., P. Legendre, and D. Borcard. 2012. Cascade multivariate regression tree: a
novel approach for modelling nested explanatory sets. Methods in Ecology and Evolution
3(2):234–244.
Ouellette, M. H. and with contributions from P. Legendre. 2013. MVPARTwrap:
MVPARTwrap: Additional features for package mvpart.. R package version 0.1–9.2.
Available at: (http://CRAN.R-project.org/package=MVPARTwrap).
Paavola R., T. Muotka, R. Virtanen, J. Heino, D. Jackson, and A. Maki-Petays. 2006. Spatial
scale affects community concordance among fishes, benthic macroinvertebrates, and
bryophytes in streams. Ecological Applications 16: 368–379.
Paul M. J., and J. L. Meyer. 2001. Streams in the urban landscape. Annual Review of Ecology
and Systematics 32:333–65.
Perkin, J. S. and K. B. Gido. 2012. Fragmentation alters stream fish community structure in
dendritic ecological networks. Ecological Applications 22:2176–2187.

134

Pielou, E. C. 1975. Ecological diversity. Wiley and Sons, New York.
Pool, T. K., J. D. Olden, J. B. Whittier, and C. P. Paukert. 2010. Environmental drivers of fish
functional diversity and composition in the Lower Colorado River Basin. Canadian
Journal of Fisheries and Aquatic Science 67:1791–1807.
PRISM. 2013. PRISM Climate Group. Oregon State University. Available at:
(http://www.prism.oregonstate.edu/).
Ricciardi, A. and J. B. Rasmussen. 1999. Extinction rates of North American freshwater fauna.
Conservation Biology. 13(5)1220-1222.
Richter, B. D., D. P. Braun, M. A. Mendelson, and L. L. Master, 1997. Threats to imperiled
freshwater fauna. Conservation Biology 11:1081–1093.
Roth, N. E., J. D. Allan, and D. L. Erickson. 1996. Landscape influences on stream biotic
integrity assessed at multiple spatial scales. Landscape Ecology 11:141–56.
R Core Team. 2013. R: A language and environment for statistical computing. R Foundation for
Statistical Computing, Vienna, Austria. Available at: (http://www.R-project.org/).
Sala, O. E., F. S. Chapin, J. J. Armesto, E. Berlow, J. Bloomfield, R. Dirzo, E. Huber-Sanwald,
L. F. Huenneke, R. B. Jackson, A Kinzig, R. Leemans, D. M. Lodge, H. A. Mooney, M.
Oesterheld, N. L. Poff, M. T. Sykes, B. H. Walker, M. Walker, D. H. Wall, 2000.
Biodiversity – global biodiversity scenarios for the year 2100. Science 287:1770–1774.
Schaefer, J. F., E. Marsh-Matthews, D. E. Spooner, K. B. Gido, and W. J. Matthews. 2003.
Effects of barriers and thermal refugia on local movements of threatened leopard darter,
Percina pantherina. Environmental Biology of Fishes 66:391–400.
Schwarz, G. E. and R. B. Alexander. 1995. Soils data for the conterminous United States derived
from the NRCS State Soil Geographic (STATSGO) Data Base. U.S. Geological Survey
Open-File Rep 95–449, Arc 7.0 coverage.
Scott, M. C. 2006. Winners and losers among stream fishes in relation to land use legacies and
urban development in southeastern U.S. Biological Conservation 127:301-309.
Shannon, C. E. 1948. A mathematical theory of communication. Bell System Technical Journal,
27:379–423.
Soller, D. R. and M. C. Reheis, compilers, 2004. Surficial materials in the conterminous United
States: U.S. Geological Survey. Open-file Report 03-275, scale 1:5,000,000. Available at:
(http://pubs.usgs.gov/of/2003/of03-275/).

135

Strayer, D. L. and D. Dudgeon. 2010. Freshwater biodiversity conservation: recent progress and
future challenges. Journal of North American Benthological Society 29:344–358.
ter Braak, C. J. F. 1995. Ordination. In: R. H. G. Jongman, C. J. F. ter Braak, and O. F. R. van
Tongeren, eds. Data analysis in community and landscape ecology. New York, NY:
Cambridge University Press, pp. 91–173.
ter Braak, C. J. F. and P. Smilauer. 2002. CANOCO 4.5 reference manual and CanoDraw for
Windows. User’s guide to Canoco for Windows: software for canonical community
ordination. Ithaca, New York: Microcomputer Power.
Therneau, T. M., B. Atkinson, and B. Ripley. 2010. rpart: Recursive Partitioning. R package
version 3.1-46. Ported to R by Brian Ripley. http://CRAN. R-project. org/package= rpart.
U.S. Census Bureau. 2002. Redistricting Census 2000 TIGER/ Line Files [machine-readable data
files]. Washington, DC: U.S. Census Bureau. Available at: (www.census.gov/geo/mapsdata/data/tiger.html).
U.S. Environmental Protection Agency and U.S. Geological Survey (USEPA and USGS). 2005.
National Hydrography Dataset Plus, NHDPlusV1, Available at: (www.horizonsystems.com/nhdplus/).
U.S. Geological Survey (USGS). 2012. National Anthropogenic Barrier Dataset (NABD) 2012.
Available at: (http://www.fishhabitat.org/content/nfhp-data-system).
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 143:688–699.
Van Sickle J. 2013. Estimating the risks of multiple, covarying stressors in the National Lakes
Assessment. Freshwater Science 32:204–216.
Vannote, R. L., G. W. Minshall, K. W. Cummins, J. R. Sedell, and C. E. Cushing. 1980. The
river continuum concept. Canadian Journal of Fisheries Aquatic Science, 37:130–137.
Vörösmarty, C.J., P. B. McIntyre., M. O. Gessner, D. Dudgeon, A. Prusevich, P. Green, S.
Glidden, S. E. Bunn, C. A. Sullivan, C. R. Liermann, P. M. Davies. 2010. Global threats
to human water security and river biodiversity. Nature 467:555–561.
Walsh, C. J., A. H. Roy, J. W. Feminella, P. D. Cottingham, P. M. Groffman, and R. P. I.
Morgan. 2005. The urban stream syndrome: current knowledge and the search for a cure.
Journal of the North American Benthological Society 24:706–723.
Walters, D. M., D. S. Leigh, and A. B. Bearden. 2003. Urbanization, sedimentation, and the
homogenization of fish assemblages in the Etowah River Basin, USA. Hydrobiologia
494:5–10.

136

Wang, L., D. Infante, P. Esselman, A. Cooper, D. Wu, W. Taylor, D. Beard, G. Whelan, and A.
Ostroff. 2011a. A hierarchical spatial framework and database for the national river fish
habitat condition assessment. Fisheries 36:436–449.
Wang, L., D. Infante, J. Lyons, J. Stewart, and A. Cooper. 2011b. Effects of dams in river
networks on fish assemblages in non-impoundment sections of rivers in Michigan and
Wisconsin, USA. River Research and Applications 27:473–487.
Wang, L., J. Lyons and P. Kanehl, 2001. Impacts of urbanization on stream habitat and fish
across multiple spatial scales. Environmental Management 28:255–266.
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., J. Lyons, P. Kanehl, R. Bannerman, and E. Emmons. 2000. Watershed urbanization
and changes in fish communities in southeastern Wisconsin streams. Journal of the
American Water Resources Association 36(5):1173–1189.
Warren, M. L. Jr., and M. G. Pardew. 1998. Road crossings as barriers to small-stream fish
movement. Transactions of the American Fisheries Society 127:637–644.
Whittier, T. R. and J. Van Sickle. 2010. Macroinvertebrate tolerance values and an assemblage
tolerance index (ATI) for western USA streams and rivers. Journal of the North
American Benthological Society 29:852–866.
Whittier, T. R., R. M. Hughes, G. A. Lomnicky, and D. V. Peck. 2007. Fish and amphibian
tolerance values and an assemblage tolerance index (ATI) for western USA streams and
rivers. Transactions of the American Fisheries Society 136:254–271.
Wiens, J. A. 2002. Riverine landscapes: Taking landscape ecology into the water. Freshwater
Biology 47(4): 501–515.
Wolock, D. M. 2003. Base-flow index grid for the conterminous United States: U.S. Geological
Survey Open-File Report 03–263, digital data set. Available at:
(http://water.usgs.gov/lookup/getspatial?bfi48grd).
Yoder, C. O., R. J. Miltner, and D. White. 1999. Assessing the status of aquatic life designated
uses in urban and suburban watersheds. Pages16–28 in A. Everson, editor. National
conference on retrofit opportunities for water resource protection in urban environments.
U.S. Environmental Protection Agency, Report EPA/625/R-99/002, Washington, D.C.
Zimmerman, J. K. H., B. Vondracek, and J. Westra. 2003. Agricultural land use effects on
sediment loading and fish assemblages in two Minnesota (USA) watersheds.
Environmental Management 32:93–105.

137

CHAPTER THREE

REGIONAL TRENDS OF BIODIVERSITY INDICES: NATURAL LANDSCAPE AND
HUMAN LAND USE CONTROLS ON STREAM FISH ASSEMBLAGES

Abstract
Biodiversity of fluvial freshwater fishes is critically threatened globally, and a major
factor leading to loss of biodiversity is anthropogenic disturbances in landscapes drained by
fluvial ecosystems. Declines in biodiversity of freshwater fish are often identified by loss of
species or fewer individuals comprising assemblages, but biological degradation can also occur
with increases in non-native species and/or spread of fish tolerant to human disturbances,
suggesting the importance of accounting for distinctness of assemblages along with richness and
diversity to best characterize response of stream fishes to human stressors. We summarized
stream fish assemblages by multiple biodiversity indices and then quantified their responsiveness
to landscape factors in stream catchments across five large ecoregions of the eastern U.S.
Indices included species richness, Shannon’s diversity, Pielou’s evenness, taxonomic diversity,
and taxonomic distinctness. We first tested for correlations among indices across regions and
found that while species richness and Shannon’s diversity were always highly correlated,
taxonomic distinctiveness was not highly correlated with these factors. We then used multiple
linear regressions to predict indices in each of the five ecoregions to identify significant factors
from natural landscape and human land uses. Indices were consistently predicted by catchment
area and elevation, emphasizing the importance of these factors on biodiversity. In contrast,
taxonomic distinctness was consistently predicted by the amount of urban in the catchment, but
138

the direction of the relationship varied. Agriculture in the catchment was a better predictor of
species richness and Shannon’s diversity. These results indicate complex influences of natural
and anthropogenic landscape factors on unique elements of biodiversity and offers insights from
our large regional study that may enhance conservation of stream fishes now and into the future.
Introduction
Habitat degradation and loss caused by anthropogenic disturbances is a dominant reason
for global declines in biological diversity of freshwater fishes (Sala et al. 2000; Helfman 2007;
Jelks et al. 2008; Burkhead 2012). Reduced biodiversity in freshwater systems may be related to
lost species and/or fewer individuals within a species, but researchers recognize that increases in
non-native species and/or spread of fishes tolerant to human disturbance may also be indicators
of biological degradation (Rahel 2000; Scott and Helfman 2001; Olden and Poff 2004; Scott
2006). For example, anthropogenic disturbances can encourage extirpations of endemic fishes
intolerant to human disturbance in altered habitats, while more tolerant fish species to human
stressors may thrive because of modified abiotic conditions or from suppression of predators or
competitors that do not persist in altered habitats (Olden and Poff 2004; Scott 2006; Helfman
2007). Such changes may be reflected by increased faunal similarity across regions. Because of
this, measures of species richness, diversity, and characterizations of regional distinctiveness
should be used together to more fully account for changes in biodiversity that may result from
anthropogenic disturbances, and such understanding is critical for protecting and preserving
biodiversity from current and future human stressors.
Many indices representing biological diversity in fish assemblages are based on counts of
species and abundances of individuals. Species richness measures numbers of species found,

139

and it has been shown to be responsive to anthropogenic disturbances in the landscape (Hughes
et al. 1998; Pont et al. 2006). However species richness can vary widely across regions due to
biogeographic history, stream size, and other natural landscape-scale factors that may influence
suitability of habitats for different species (Schlosser 1991; Goldstein and Meador 2004).
Species richness also does not account for differences in rare species versus common species,
and richness is recommended to be used in concert with other measures that account for
abundances of individual species (Stirling and Wilsey 2001; Magurran 2004). Measures of
evenness account for abundances, with high measures indicating more equal numbers of
individuals per species in assemblages and low measures reflecting dominance in abundances of
fewer species (Magurran 2004). Shannon’s diversity index weights all species proportionately to
their abundances; high values for Shannon’s diversity indicate more equitable distributions of
individuals present in a sample, while low values may indicate dominance of a few species
comprising an assemblage (Magurran 2004). Together, these biodiversity measures can
effectively indicate changes in diversity driven by loss of species or individuals, yet not
necessarily for non-natives or species tolerant of human disturbance (Magurran and Phillip
2001).
Taxonomic distinctness measures can indicate changes in biodiversity that do not
necessarily result in changes to species richness or diversity (Warwick and Clarke 1995).
Taxonomic diversity and taxonomic distinctness indices depend on the composition of
assemblages found throughout an entire study region; they include a hierarchical measure of
relatedness between fish species within a single sample based on their taxonomy to the pool of
species found throughout a study region (Clarke and Warwick 1998; 2001). In relatively pristine
streams, fish assemblage composition may be comprised of more distinct species and an

140

assemblage that covers a wider breadth in taxonomic diversity and distinctness (Hughes et al.
1990; Angermeier and Winston 1999). These can include endemic habitat specialists, species
with specialized feeding preferences, or species with unique life history traits. Under
anthropogenic disturbance, a loss in endemic specialist species and/or a spread of tolerant species
may result in a smaller number of taxonomic groups and a decline in fish assemblage
distinctness. The loss of fish assemblage distinctness can happen through loss of endemic
specialists, losing unique genera distinctness, or loss of whole intolerant family groups (e.g.
percidae), or expansion in similarity of genera or family groups from emigration and loss of
unique fish genera or family groups. Because of this, taxonomic distinctness measures can
indicate unique changes in assemblages resulting from anthropogenic land uses and urbanization.
A better understanding of how different biodiversity indices respond to anthropogenic
disturbances, to natural factors that influence regional species pools, and habitat factors that
constrain species distributions could help to improve understanding of biological changes in
streams. Further, such a study conducted over large regions that considers anthropogenic
landscape factors as a source of disturbance to stream habitats and evaluates response of multiple
indices could yield new insights into broad patterns of biodiversity changes occurring with largescale disturbances, like human land uses. To help meet these needs, the goal of this study is to
characterize regionally-specific landscape controls affecting multiple fish biodiversity indices in
streams across five regions of eastern United States (U.S.) We have three objectives. First, we
characterized major patterns in multiple fish biodiversity indices across study regions, including
interrelationships among the various indices. Indices evaluated included species richness,
assemblage diversity, and taxonomic distinctiveness. Second, we identified dominant natural
landscape factors and human lands uses important in predicting each of the indices. Finally, we

141

quantified how influences of natural landscape factors and human lands on biodiversity indices
vary by study ecoregion. Addressing these objectives yielded new insights into influences on
stream fish assemblages across large regions and provided opportunities to enhance conservation
efforts to protect stream fishes from current and future landscape changes.
Methods
Study area.—This study was conducted within five large freshwater ecoregions located in
the eastern U.S. These freshwater ecoregions, referred to throughout as “ecoregions,” were
delineated by the World Wildlife Fund and included: the Appalachian Piedmont, Chesapeake
Bay, Laurentian Great Lakes, Middle Missouri, and the Upper Mississippi (Abell et al. 2008;
Figure 3.1). Freshwater ecoregions are defined as areas with similar biogeographic history in
fish faunas encompassing major river basins and matching large basin boundaries; they represent
ecologically-relevant units for groupings of fishes based on evolutionary patterns across large
landscapes (Maxwell et al. 1995; Matthews 1998; Abell et al. 2000; 2008; Esselman et al. 2011).
Freshwater ecoregions were chosen as the spatial units to evaluate in this study because they
encompass a wide range in fish assemblage composition, and as such have the potential for a
wide range in richness, biodiversity, and taxonomic distinctness. This ensures that these regions
will be effective test areas to evaluate the performance of biodiversity indices in showing
responsiveness to human land uses because of the diverse range in natural and human landscape
variables across regions.
The study area extends from the western Great Plains to the east coastal plains of the U.S.
(Figure 3.1). Natural landscape factors within the study area are highly diverse; study ecoregions
include previously glaciated landscapes in the Chesapeake Bay and Laurentian Great Lakes and

142

regions that have not been glaciated including the coastal plains and highlands of the
Appalachian Piedmont ecoregion. In the eastern and central portion of the study area, the
Middle Missouri and Upper Mississippi ecoregions include areas of intensive agriculture, while
high percentages of grasslands occur within the western Middle Missouri ecoregion. High
percentage of forested areas are located within the Laurentian Great Lakes, and the Appalachian
Piedmont and the Chesapeake Bay ecoregion include areas of high urban land use.
Stream layer.—The 1:100,000-scale National Hydrography Dataset Plus Version 1
(NHDPlusV1) national streams layer was used as a base layer for geographic representation of
streams reaches and their catchments (USEPA and USGS 2005). We defined a stream reach as a
section of river in the NHDPlusV1 that extends: 1) from the stream origin to the first
downstream confluence or junction with a lake or reservoir, 2) from an upstream confluence or
lake/reservoir outflow to the next downstream confluence or lake/reservoir junction, or 3) from
an upstream confluence or lake/reservoir outflow to the river mouth where it meets with a lake or
estuary (Brenden et al. 2006; Esselman et al. 2011; Wang et al. 2011). Catchments summarizing
information over two spatial extents were used in this analysis based on the NHDPlusV1. “Local
catchments” include all land that drains directly into an individual stream reach without being
transported via other fluvial pathways represented in the NHDPlusV1, and “network catchments”
include all land upstream of and draining into a given reach via fluvial pathways and including
the local catchment.
Natural landscape variables.—Six natural landscape variables were evaluated in this
study, including network catchment area, percent of fine lithology in the network catchment,
mean local catchment elevation, base-flow index in the network catchment, mean annual air
temperature in the local catchment, and mean annual precipitation in the network catchment.
143

These factors were selected in a previous analysis (Thornbrugh this volume b) as explaining
major patterns of variation in natural landscape factors across the five study regions. A variety
of nationally-available landscape datasets were used to derive these factors. Catchment areas
were calculated using a GIS. Elevation was developed from the U.S. Geological Survey (USGS)
National Elevation Data (NED, Gesch 2007). Fine surficial lithology was developed from the
USGS terrestrial ecosystems surficial lithology of the conterminous U.S. data layer (Soller and
Reheis 2004; Cress et al. 2010). The fine surficial lithology category was defined by grouping
major classes of fine substrate size from surficial lithology for use in analysis. The fine lithology
category was used to indicate the potential of streams to have variable flow regimes as opposed
to more stable hydrology and potentially higher contribution of base-flow from groundwater
inputs. The base-flow index, a model of hydraulic potential characterizing the percentage of
groundwater contribution to stream flow, was modeled by USGS (Wolock 2003). Mean annual
air temperature and mean annual precipitation were summarized for the climatological period
from 1981 to 2010 by the PRISM Climate Group (Daly et al. 2008; PRISM 2013).
Human land uses.—Four human land use variables were evaluated in this study. They
include percent of urban and agricultural land use in network catchments and the density of
stream-road crossings and dams in network catchments. These factors describe major variability
in human land uses across the five study ecoregions (Thornbrugh this volume b). Also, percent
agricultural land use, percent urban land use, and density of dams in network catchments were
shown to be important influences on stream fish assemblage metrics including taxonomic
measures, trophic preferences, tolerance to stressors, and life history traits in the study region
(Thornbrugh this volume b). Stream road crossings are theorized to be a secondary influence on
fish (Bouska and Paukert 2010, Januchowski–Hartley et al. 2013), and we tested for the

144

cumulative effects of these potential barriers to fish passage throughout stream catchments in this
study. Major categories of human land cover for urban and agriculture were developed from the
2001 National Land Cover Dataset (2001 NLCD, Homer et al. 2004, USEPA 2008). Urban land
use included low, medium, and high intensity developed land along with open space developed
land cover classes. Agricultural land use included pasture/hay and cultivated crops land cover
classes. The density of dams was developed from the National Anthropogenic Barrier Dataset
(NABD, U.S. Geological Survey 2012), and the density of stream-road crossings was developed
from the U. S. Census Bureau TIGER/line files data (U.S. Census Bureau 2002).
Fish data.—Data characterizing stream fish assemblages were assembled and referenced
to stream reaches in the NHDPlusV1. Assemblages were sampled with electrofishing by state
and federal programs using methods determined to be comparable for creeks, small rivers and
medium rivers ≤ 10,000 km2 between years from 1990 to 2010 (Wang et al. 2011; Esselman et
al. 2011; 2013). Fish data used in analysis were identified to species, and all scientific naming
was standardized to the Integrated Taxonomic Information System (ITIS; ITIS 2010).
Biodiversity indices.—Five measures of fish assemblage biodiversity were derived from
site specific data characterizing number of species and their abundances. Three of these indices
included species richness, diversity, and evenness: (i) the number of species included in a
sample, species richness (S); (ii) species diversity based on Shannon’s diversity index (H’),
where pi is the proportional of individuals belonging to the ith species and S is the number of
species at a sample site (Shannon 1948); and (iii) Pielou’s evenness (J’) based on Shannon’s
index and on richness: where the H’ is the Shannon’s index and S is the number of species
(Pielou 1975). Evenness measures relative similarity in abundances of fish species present at a
site, relativizing the overall Shannon’s index based on number of species found at sites.
145

Shannon’s index and Pielou’s evenness indices were calculated using the function “diversity”
from the package “vegan” in R (Oksanen et al. 2013; R Core Team. 2013).
(ii)

𝐻 ′ = − ∑𝑆𝑖 𝑝𝑖 log 𝑝𝑖

(iii)

𝐽′ = 𝐻 ′ /log 𝑆
The two additional indices that describe taxonomic diversity and taxonomic distinctness

of fish assemblages were used. They were derived from abundance data and were based on
relationships in composition of stream fish assemblages at individual study sites evaluated
against taxonomic relatedness from the regional species pool across all sites in our study region
(Clarke and Warwick 1998; 2001). This is in contrast to species richness, Shannon’s diversity,
and Pielou’s evenness, which are calculated from information on assemblages from a single
study site. Taxonomic relatedness is derived in part from cladistics classification of all
organisms that account for taxonomic levels at which species are related (species, genus, family,
order, and class). A numeric weighting of branch lengths of the hierarchical cladistics
classification tree accounts for interrelationships among all species in the regional species pool
(Table C3.1). A hypothetical cladogram is shown in Figure 3.2 to highlight differences in
relatedness weighting among individuals. Sites are considered to be more taxonomically related
when most assemblage membership is comprised of species of similar genera with maximal
family, order, or class relatedness, while sites are considered to be more distinct in taxonomic
assemblage when membership is comprised of species of a variety of genera, multiple families,
orders, and classes.
These indices were derived as follows: (iv) taxonomic diversity (Δ): where xi denotes the
abundance of ith species (i =, ..., S), n is the total number of individuals in a sample, and the wij is
146

the “distinctness weight,” a taxonomic distance measure that links species i and j in a taxonomy
(Clarke and Warwick 2001): and (v) taxonomic distinctness (Δ*) is defined as (Δ) divided by the
value it takes when the taxonomic hierarchical tree has the simplest possible structure, that of all
species belonging to the same genus (Warwick and Clarke 1995; Clarke and Warwick 1998;
2001).

(iv)

∆ = [∑ ∑𝑖<𝑗 𝜔𝑖𝑗 𝑥𝑖 𝑥𝑗 ]/[𝑛(𝑛 − 1)/2]

(v)

∆∗ = [∑ ∑𝑖<𝑗 𝜔𝑖𝑗 𝑥𝑖 𝑥𝑗 ]/[∑ ∑𝑖<𝑗 𝑥𝑖 𝑥𝑗 ]

Five taxonomic levels were used, species, genus, family, order, and class, to calculate the
distinctness weight (wij) from the master fish species list for the study region (Table C3.1). The
distinctness weight was calculated using function “taxa2dist” and taxonomic diversity and
taxonomic distinctness indices were calculated using the function “taxondive” from the package
“vegan” in R (Oksanen et al. 2013; R Core Team 2013).
Data analysis
Pearson correlation among fish biodiversity indices.—A Pearson correlation was run to
compare co-variability among fish biodiversity indices within regions and to characterize how
these interrelationships varied across the five study ecoregions. All variables were transformed
to better meet assumptions of linearity and homoscedasticity. Species richness was square-root
transformed; all other indices were natural log transformed. Correlations were judged to be high
when R2 values were ≥ 0.65; moderate at 0.35 to 0.65 and low when values were < 0.35.
Regression analysis.—Multiple linear regression models were developed to identify the
most important natural landscape variables and human land uses predicting each of the five fish
147

biodiversity indices and to evaluate the variability in best models across the five study
ecoregions. All variables were transformed before regression analysis. The best linear
regression models were developed utilizing an exhaustive search with a branch-and-bound
algorithm for the best subset of five or fewer natural landscape variables and/or human land use
variables (from a total of 10 possible predictors, Table 3.1) predicting each of five fish
biodiversity indices in all five study ecoregions (Little et al. 1963; Miller 2002; Lumley 2009).
Best models were selected using Bayesian Information Criterion (BIC). A standardized
regression coefficient (β) value was calculated to evaluate the importance of variables selected
for inclusion in the models predicting biodiversity indices. To identify collinearity among
explanatory variables, a variance inflation factor (VIF) was calculated. Multiple linear
regression models were calculated using the function “regsubsets” from the package “leaps” in R
(Lumley 2009; R Core Team 2013). The standardized regression coefficient (β) was calculated
using the function “lm.beta” from the package “QuantPsyc” in R (Fletcher 2012; R Core Team
2013).
Results
Study area.—The five study ecoregions included an area greater than 2,000,000 km2
(Figure 3.1), and data from more than 10,500 fish sampling locations were used in analyses.
Natural landscape factors across the study regions are highly diverse (Table 3.1). Mean network
catchment area of study sites ranged from 155 km2 in the Appalachian Piedmont ecoregion to
764 km2 in the Middle Missouri ecoregion. Fine lithology in the network catchment was lowest
in the Upper Mississippi ecoregion (mean of 9%) and highest in the Laurentian Great Lakes
ecoregion (mean of 42%). Mean local elevation was lowest at study sites in the Appalachian
Piedmont ecoregion at 157 m, and ranged from 20 m to 1166 m across all sites in the study
148

ecoregions, characterized by 10% and 90% exceedance values, respectively (Table 3.1). Mean
local elevation was highest at study sites in the Middle Missouri ecoregion at 591 m. Climate
varied across sites in the study ecoregions, with lowest mean annual air temperatures in the
Upper Mississippi (8 °C) and highest mean annual air temperature and precipitation in the
Appalachian Piedmont ecoregion (15 °C, 1215 ml, respectively Table 3.1). The lowest mean
annual precipitation was in the Middle Missouri ecoregion at 709 ml.
There was also wide variability in study sites across ecoregions in the four human land
uses. Mean percentage of urban land use in network catchments of study sites was highest in the
Chesapeake Bay ecoregion at 32% and lowest at study sites in the Middle Missouri at 6% (Table
3.1). There was a high variability in levels of mean percentage of agriculture land use in the
network catchment across study ecoregions; it was highest in the Upper Mississippi ecoregion at
60%, with 10% and 90% exceedance values of 16% and 90%, respectively. The mean network
catchment percent agriculture was lowest in Appalachian Piedmont ecoregion with a value of
21%, and with 10% and 90% exceedance values of 4% and 41%, respectively. Density of
stream-road crossings in network catchments of study sites was highest in the Chesapeake Bay
ecoregion at 54/100 km2 and was lowest in the network catchment for study sites in Middle
Missouri ecoregion at 36/100 km2. Density of dams in the network catchment of study sites was
highest in the Appalachian Piedmont ecoregion at 3.91/100 km2 and lowest in the network
catchment of study sites in the Upper Mississippi at 0.71/100 km2.
Regional patterns in biodiversity indices.— Study regions supported 312 species of fish,
comprising 98 genera, 34 families, 21 orders, and two classes (Table C3.1). Patterns in
biodiversity indices varied across study ecoregions. Species richness across all sites ranged from
one in every ecoregion to a maximum of 48 in the Upper Mississippi. While mean species
149

richness was highest in the Appalachian (15 species), it ranged to a low of 10 in both the
Chesapeake and Middle Missouri ecoregions (Table 3.2). Differences across study regions were
also shown when considering the 10% exceedance values for species richness. In the
Chesapeake Bay ecoregion, the 10% exceedance value for species richness was two, while in the
Appalachian Piedmont, it was eight, indicating a substantial number of sites with low species
richness in the Chesapeake (Table 3.2).
Shannon’s diversity and Pielou’s evenness values were also variable. Similar to species
richness, mean Shannon’s diversity was highest in the Appalachian (1.89, Table 3.2), while
evenness was highest in the Chesapeake Bay (0.61, Table 3.2). This likely reflects the fact that
while sites in the Chesapeake support fewer species on average compared to other regions, the
species comprising assemblages are more even in numbers. Mean diversity and mean evenness
across sites were lowest in the Middle Missouri. Finally, the highest value for Shannon’s
diversity was found in the Appalachian Piedmont at 2.77 (Table 3.2).
Patterns in taxonomic diversity varied from patterns expressed by species richness and
Shannon’s diversity across study ecoregions. For example, mean taxonomic diversity and
distinctness were highest in the Appalachian Piedmont (54.05 and 69.68, respectively, Table
3.2), following highest values for species richness and Shannon’s diversity in the Appalachian
Piedmont ecoregion. In contrast, the second highest mean taxonomic diversity and distinctness
value were in the Middle Missouri ecoregion (41.89 and 66.40, respectively). While mean
species richness and Shannon’s diversity were low here, this result emphasizes that even in
regions with lower species richness or Shannon’s diversity; sites may still support assemblages
with a high degree of taxonomic diversity and taxonomic distinctness.

150

Regional patterns in correlation of biodiversity indices.—Patterns in correlation among
biological indices were similar across study regions (Table 3.3). Species richness and Shannon’s
diversity highly correlated in each of the five study regions and ranged in strength from a high of
R2 = 0.86 in the Chesapeake Bay ecoregion to a low of R2 = 0.69 in the Middle Missouri
ecoregion. Shannon’s diversity also had a high correlation with taxonomic diversity across all
regions; the highest value was R2 = 0.87 in the Upper Mississippi ecoregion and the low was R2 =
0.77 in the Laurentian Great Lakes (Table 3.3). Taxonomic diversity was highly correlated with
taxonomic distinctness across study regions (high of R2 = 0.77 to a low of R2 = 0.59 in both the
Chesapeake Bay and Laurentian Great Lakes ecoregions). Interestingly, while taxonomic
diversity was moderately correlated with species richness and highly correlated Shannon’s
diversity, taxonomic distinctness was never highly correlated with these factors (highest R2 =
0.46 in the Upper Mississippi), emphasizing that this metric captures a different biodiversity
component (Table 3.3). Evenness was not strongly correlated with any other metrics with the
exception of species richness; it was moderately correlated with richness in all regions (Table
3.3).
Predicting biodiversity indices from landscape factors with linear regression models.—
All biodiversity indices were identified as being significantly predicted by at least two landscape
factors. Adjusted R2 ranged from 0.52 for species richness in the Upper Mississippi ecoregion to
0.04 for Pielou’s evenness in the Appalachian. All natural landscape factors and human land use
variables selected for inclusion into models were significant (p ≤ 0.01). Variance inflation
factors (VIF) were low (i.e., maximum VIF = 5.2) for all natural landscape factors and human
land uses that were selected as predictor variables in the best models across all five study
ecoregions. The low VIF values below 5.2 for all selected predictor variables indicated that

151

there was low collinearity between variables selected in the best models predicting biodiversity
indices (Legendre and Legendre 2012).
Trends in predicting biodiversity indices from landscape factors.— Biodiversity indices
were variably predicted across regions, but the best predicted index in all ecoregions was species
richness with a mean adjusted R2 = 0.37 (Figure 3.3). The second best predicted index across
ecoregions was taxonomic distinctness (mean adjusted R2 = 0.25), while the worst predicted
biodiversity index was Pielou’s evenness across study ecoregions with a mean adjusted R2 = 0.13
(Figure 3.3). The most important predictor of biodiversity indices was catchment area.
Catchment area was the strongest or second strongest predictor in 24 of the 25 best models
predicting all biodiversity indices across the five study ecoregions (Table 3.4). Local catchment
elevation was also important at predicting taxonomic diversity and taxonomic distinctness
indices in four of the five ecoregions (Table 3.4). Catchment agriculture was an important
human land use predicting species richness and Shannon’s diversity in three of the five
ecoregions, and it also significantly predicted species richness in Middle Missouri and Shannon’s
diversity in the Laurentian Great Lakes ecoregions. Agriculture, however, was never a predictor
for taxonomic distinctness (Table 3.4). Catchment urban land use was a consistent predictor of
taxonomic distinctness in four of the five ecoregions, but the direction of that influence varied by
region (Table 3.4). Both local air temperature and catchment precipitation were important
predictors of indices across all ecoregions, yet no consistent patterns could be discerned in their
ability to predict specific indices or in regional trends.
Regional differences in predicting biodiversity indices from landscape factors.—
Biodiversity indices were strongly influenced by catchment fine lithology in the Laurentian
Greats Lakes (all indices except distinctiveness) and Middle Missouri ecoregions (richness,
152

Shannon’s diversity, and distinctiveness), and metrics in both regions had no influence from
catchment base-flow index. While fine lithology and base-flow index occasionally predicted
indices in other regions, there was never an occasion where both catchment fine lithology and
catchment base-flow index occurred together as predictors in a model. Regression models
showed regional differences in the influences of catchment urban and agricultural land use on
biodiversity indices. Catchment urban was a good predictor for taxonomic distinctness in all
ecoregions but the direction of the influence varied across ecoregions. Catchment urban had a
positive influence on taxonomic distinctness in the Appalachian Piedmont and the Chesapeake
Bay ecoregions and a negative influence on taxonomic distinctness in the Laurentian Great
Lakes, Middle Missouri, and Upper Mississippi ecoregions. Although they had comparatively
less predictive capacity than other metrics, density of stream road crossing and dams in the
catchment were significant predictors of species richness and Shannon’s diversity in the
Chesapeake Bay ecoregion. Density of stream-road crossing in catchments was a significant
predictor of species richness in the Middle Missouri ecoregions and the density of dams in the
catchment was also and significant predictor of Shannon’s diversity in the Laurentian Great
Lakes ecoregion (Table 3.4).
Discussion
This study characterized trends in stream fish assemblages across five ecoregions in the
eastern portion of the U.S. using biodiversity indices including species richness, Shannon’s
diversity, Pielou’s evenness, taxonomic diversity, and taxonomic distinctness (Shannon 1948;
Pielou 1975; Clarke and Warwick 1998; 2001). General patterns in correlations among
individual indices were variable, but relationships among indices were similar across ecoregions.
For example, species richness and Shannon’s diversity were always highly correlated in each of
153

the five study ecoregions, yet taxonomic distinctiveness was never highly correlated with these
factors, emphasizing the independence of distinctness from richness and Shannon’s diversity.
Results of multiple linear regression models suggested differences in landscape influences on
groups of metrics. Natural landscape factors including stream size and elevation were important
and consistent predictors of biodiversity indices, while climatological factors and factors
indicating groundwater delivery to streams had more variable influences. Percent of agriculture
and urban in the catchment were important predictors of biodiversity across all regions.
Catchment agriculture had a fairly consistent response on richness and diversity, while
catchment urban land use most consistently predicted taxonomic distinctness, yet that response
varied by region. High amounts of catchment urban land use could negatively affect taxonomic
distinctness causing shifting fish assemblage composition from endemic faunas to more tolerant
fish fauna promoting fish assemblages with more regional similarity (McKinney and Lockwood
1999; Rahel 2002). Together, these results suggest that multiple measures of fish biodiversity
should be used to characterize response of stream fish assemblages to landscape influences, not
only because they capture unique attributes of the fish assemblage, but also because indices are
being influenced by land uses differently.
Correlation among biodiversity indices.— Measures of biodiversity were variably
intercorrelated, but showed similar trends in correlations across ecoregions. Species richness and
Shannon’s diversity were always highly correlated, Shannon’s diversity and taxonomic diversity
were always highly correlated, and taxonomic diversity and taxonomic distinctness were
moderately to highly correlated in each of the five study ecoregions. In contrast, taxonomic
distinctness was never strongly correlated with species richness, Shannon’s diversity, or Pielou’s
evenness. Similar relationships between biodiversity indices were seen in macroinvertebrate

154

communities from stream sites in western and central Finland (Heino et al. 2007); they found
that taxonomic distinctness of macroinvertebrate assemblages was not correlated with Shannon’s
diversity or Pielou’s evenness (Heino et al. 2007). In a similar study, Heino et al. (2005) found
that fish species richness and taxonomic distinctness were moderately to weakly correlated for
lake fish assemblages in northeastern Finland. Results of these studies, along with ours,
emphasize that these biodiversity indices represent unique attributes of stream fish assemblages,
and that commonly used metrics such as richness, Shannon’s diversity, and Pielou’s evenness
may not be fully effective at capturing differences in regional distinctiveness of freshwater
assemblages, including differences resulting from anthropogenic landscape factors.
Natural landscape factors predicting biodiversity indices.—Of the major natural and
anthropogenic gradients analyzed, biodiversity indices were most strongly influenced by
catchment area, it was a significant predictor of every biodiversity index in all ecoregions. This
result follows from species-area relationships in fish faunas (Eadie et al. 1986; Angermeier and
Schlosser 1989; Thornbrugh and Gido 2010); with increasing stream size, there is an increase in
stream productivity, habitat heterogeneity, and habitat area, promoting fish assemblage diversity
(Sheldon 1968; Vannote et al. 1980; Marsh-Matthews and Matthews 2000). Local elevation was
also an important natural landscape predictor of taxonomic diversity and taxonomic distinctness
indices with negative effects in four of the five ecoregions (Table 3.4). This may be due to an
increase in assemblage taxonomic complexity at lower elevations where increased stability in
flow and temperature regimes and spatial connectivity in fluvial systems contribute to increased
taxonomic diversity with additions of unique genus and family groups (Angermeier and Winston
1998; Jackson et al. 2011). Together, these findings underscore the role of natural landscape

155

factors on biodiversity indices and the sensitivity of indices to catchment area as a control of
habitat heterogeneity and influence on biodiversity in fish assemblages of fluvial ecosystems.
Catchment fine lithology was also found to be influential in biodiversity indices for the
Laurentian Great Lakes ecoregion, positively influencing species richness, Shannon’s diversity
and taxonomic diversity in the ecoregion. Percent of fine lithology in the catchment has been
shown to have a landscape-scale influence directly controlling fish assemblage structure by
distinguishing streams with different thermal and hydrologic conditions, and it also acts as a
surrogate measure for groundwater potential to base-flow (Zorn et al. 2002; Wehrly et al. 2003).
Fine catchment lithology has also been shown to mediate fish assemblages by controlling
specific aspects of stream thermal regimes like maximums and variability in stream summer
temperatures in Michigan (Thornbrugh this volume a).
Both mean annual air temperature and precipitation were important predictors of
biodiversity indices across all ecoregions, yet no consistent patterns could be interpreted in their
ability to predict specific indices or in regional trends. For example, species richness and
Shannon’s diversity were positively associated with air temperature in the Chesapeake Bay and
Laurentian Great Lakes ecoregions, but were negatively associated with temperature in the
Appalachian Piedmont. Also, catchment precipitation had a positive influence on species
richness, Shannon’s diversity and both taxonomic indices in the Upper Mississippi, yet
catchment precipitation was not a good predictor for any biodiversity indices in the Appalachian
Piedmont ecoregion. Despite the lack of consistent trends across indices or regions, the variable
associations between air temperature and precipitation on indices does suggest their importance
in controlling stream fish assemblages. Further, it underscores the fact that fish biodiversity has
the potential to be influenced by future climate changes (Dudgeon et al. 2006; Buisson et al.
156

2008; Olden et al. 2010), and that a better understanding of regional influences of climate on
stream fish assemblages is warranted.
Human land uses predicting biodiversity indices.— Percent of catchment agriculture and
urban land use were important predictors of biodiversity across all regions; however associations
of these land uses with indices were variable. Catchment agriculture was shown to generally
have a positive effect on fish species richness and Shannon’s diversity across ecoregions, except
for a negative influence on species richness in the Middle Missouri ecoregion. Wang et al.
(2001) also found that species richness and diversity of stream fishes from 47 sites in
southeastern Wisconsin, U.S., were positively associated with catchment agriculture. Their
study sites also had a wide range in human land use in their catchments, including some with
high levels of urban and agricultural lands. They surmised that, if impervious surfaces
associated with urban land uses were low in catchments, sites with high levels of agriculture still
had the potential for high species richness and diversity. However, Wang et al. (1997) showed
that with high levels of agriculture (>50% in catchments), fish IBI scores from streams decreased
across study sites that had moderate levels of agriculture in their catchments. This suggests that
agriculture may have negative effects on stream fish assemblages that are not necessarily
reflected in changes in species richness and diversity. Such changes could include losses of
intolerant fishes and co-occurring increases in tolerant fish or nonnatives, and it emphasizes the
importance of using multiple metrics to best characterize changes in biodiversity that may result
in streams from human land uses. Our findings, along with those of the studies referenced
above, could also result from differences in types of agriculture occurring on the landscape
and/or specific mechanisms by which agriculture could influence streams, and investigations that
accounted for specific changes in stream habitat that occurred with agriculture (e.g., increased

157

nutrient loadings, and/or changes in sediment, thermal or flow regimes), would likely yield more
specific insights into how fish respond to agriculture.
Catchment urban land use was a consistent predictor of taxonomic distinctness in four of
the five ecoregions, although its influence on taxonomic distinctness was positive in the
Appalachian Piedmont and negative in the Laurentian Great Lakes, Middle Missouri, and Upper
Mississippi ecoregions. These varying influences on taxonomic distinctness may be due in part
to regional influences related to the specific proportion or intensity of urban land use in the
ecoregion or the interaction between natural landscape factor like lithology and topography and
their control on the delivery of runoff and other materials to stream systems (Utz et al. 2010,
Thornbrugh this volume b). More significantly, these results suggest that high amounts of
catchment urban land use negatively affecting taxonomic distinctness, that may reflect the loss of
endemic or sensitive fish taxa and increases in more tolerant taxa, leading to homogenization of
assemblages across study regions (Scott 2006; Helfman 2007). These results emphasize how
taxonomic distinctness measures can indicate unique changes occurring in stream fish
assemblages resulting from anthropogenic land uses, and that taxonomic distinctness may also be
an effective metric for bioassessment of streams along with richness and diversity (Stoddard et
al. 2008).
The density of road crossings and dams in the catchment were also significant predictors
of species richness and Shannon’s diversity. However their influence was always positive on
these indices in the three ecoregions where they were significant predictors. Studies have
emphasized how these landscape factors may have the potential to influence stream fishes (e.g.,
Januchowski–Hartley et al. 2013; Perkin and Gido 2012), and our results support this. However,
they also suggest that greater numbers of dams and/or stream road crossings may potentially
158

make stream habitats more favorable for more species than could be negatively affected by
barriers. In a study of dam influences on stream fishes in Michigan and Wisconsin, Wang et al.
(2011) showed increases in IBI scores with more upstream dams. They attributed their finding to
potential modifications in habitat that benefited species including reduced sediment loads and
potentially more stable stream flow regimes in streams downstream of dams.
Spatial extent of study and scope of natural and human environmental gradients
captured.— The spatial extent over which this study was conducted offers important insights
into the response of stream fish assemblages to natural landscape factors and human land uses.
Our study area included five freshwater ecoregions encompassing an area greater than two
million square km, and we tested associations between stream fishes and landscape factors using
data from over 10,500 study sites on streams and rivers occurring within twenty two states. This
has provided a unique opportunity for conducting regional comparisons based on extensive
variability in fish assemblages and characterizations of species compositions that were vital in
summarizing patterns in biodiversity. Also, our large data set allowed us the ability to capture
equally broad ranges in environmental characteristics to associate with biodiversity indices.
While the amount of variance predicted in the biodiversity indices were moderately low, given
the spatial extent of the study, species richness and taxonomic distinctness indices were
relatively well predicted across ecoregions (i.e., mean adjusted R2 = 0.37 and 0.25, respectively,
Figure 3.3). For Shannon’s diversity, Pielou’s evenness, and taxonomic diversity we were likely
unable to capture regionally-specific mechanisms that may be most strongly influencing these
indices.
Management implications.—This study emphasized that, across large regions,
biodiversity indices for stream fish assemblages are significantly influenced by both natural
159

landscape factors and human land uses. Catchment area was a strong predictor of all biodiversity
indices across study regions, and should be controlled for when utilizing biodiversity indices as
response metrics in assessments of stream integrity. Taxonomic distinctiveness, found to
represent a unique attribute of stream fish assemblages compared to diversity and richness, was
also predicted by local elevation, which should similarly be accounted for when using this index
for assessment.
This study also showed the importance of human land uses in predicting biodiversity
indices, underscoring their value in identifying broad changes in stream fish assemblages
resulting from anthropogenic stressors. Taxonomic distinctness was consistently predicted by
catchment urban land use, while catchment agriculture was a better predictor of species richness
and Shannon’s diversity across ecoregions. These findings reflect how unique attributes of fish
assemblages are influenced by human land uses differently, and that those changes are likely due
in part to shifting species membership, and not always due to the loss of species or individuals
(Scott and Helfman 2001; Scott 2006), Based on our results, we recommend using multiple
biodiversity measures to capture unique attributes of stream fish assemblages when testing for
effects of human stressors. This has important implications for development of multimetric
indices of biotic integrity to assess streams, and it highlights the importance of identifying
species-specific targets when implementing actions to conserve or protect biodiversity in stream
systems.

160

APPENDICES

161

APPENDIX 3.A

TABLES
Table 3.1.—Descriptive statistics of landscape variables for the five freshwater ecoregion with
landscape variable descriptions (i.e., mean, minimum, maximum, and the 10th and 90th
percentiles).
Landscape variable description
Natural

Mean

Appalachian Piedmont
Min
Max
10%

155.16
1.42 8062.60
11.42
Network catchment area (km2)
Fine lithology in network catchment (%)
21.76
0.00 100.00
0.00
Mean local catchment elevation (m)
157.30
1.39 937.66
20.42
Base-flow index, network catchment
45.95 25.23
78.30 30.54
groundwater contribution to baseflow (%)
Mean annual air temperature in local catchment
15.64
9.53
19.33 14.11
(°C)
Area weighted average mean annual
1215.13 973.82 2372.90 1125.53
precipitation in network catchment (ml)
Human land uses
Urban (%)
Developed, open space, low, medium, & high
14.20
0.00
98.62
3.13
intensity in network catchment
Agriculture (%)
Pasture/hay & cultivated crops in network
21.34
0.00
72.63
3.83
catchment
Road crossings
Density of road crossings in network catchment
48.96
0.00 345.30
15.82

90%
180.03
100.00
298.06
60.35
17.52
1317.59

37.22

40.93

89.03

2

(number/100 km )
Dams
2

3.32

Dam density (number/100 km )

162

0.00

57.56

0.00

9.06

Table 3.1.—(cont’d)
Landscape variable description
Natural

Mean

Chesapeake Bay
Min
Max
10%

90%

371.17
0.82 8483.38
3.48 560.87
Network catchment area (km2)
Fine lithology in network catchment (%)
13.22
0.00 100.00
0.00 99.80
Mean local catchment elevation (m)
288.25
2.52 1050.68
21.71 491.14
Base-flow index, network catchment
46.26 26.99
64.02 38.08 54.74
groundwater contribution to baseflow (%)
Mean annual air temperature in local catchment
9.57
6.35
14.28
7.00 13.25
(°C)
Area weighted average mean annual
1027.70 806.87 1239.06 914.37 1108.89
precipitation in network catchment (ml)
Human land uses
Urban (%)
Developed, open space, low, medium, & high
8.15
0.00
98.44
1.27 14.85
intensity in network catchment
Agriculture (%)
Pasture/hay & cultivated crops in network
31.59
0.00
94.91
7.72 63.49
catchment
Road crossings
Density of road crossings in network catchment
53.99
0.00 588.24
9.40 96.04
2

(number/100 km )
Dams
2

1.10

Dam density (number/100 km )

163

0.00

30.53

0.00

3.04

Table 3.1.—(cont’d)
Landscape variable description
Natural

Mean

Network catchment area (km2)
Fine lithology in network catchment (%)
Mean local catchment elevation (m)
Base-flow index, network catchment
groundwater contribution to baseflow (%)
Mean annual air temperature in local catchment
(°C)
Area weighted average mean annual
precipitation in network catchment (ml)
Human land uses
Urban (%)
Developed, open space, low, medium, & high
intensity in network catchment
Agriculture (%)
Pasture/hay & cultivated crops in network
catchment
Road crossings
Density of road crossings in network catchment

354.81

Laurentian Great Lakes
Min
Max
10%
0.23 9533.74

41.97
277.15
45.04

0.00
74.56
18.69

7.74

2.12

90%

10.27 758.72

100.00
0.00 100.00
680.78 180.28 430.92
87.95 21.99 68.45
10.56

4.69

9.68

929.80 641.04 1566.84 764.17 1123.50

11.05

0.00

99.92

1.15

26.31

39.85

0.00

94.26

0.47

83.07

48.49

0.00

413.37

8.35

91.15

1.13

0.00

73.69

0.00

2.84

2

(number/100 km )
Dams
2

Dam density (number/100 km )

164

Table 3.1.—(cont’d)
Landscape variable description
Natural

Mean

Network catchment area (km2)
Fine lithology in network catchment (%)
Mean local catchment elevation (m)
Base-flow index, network catchment
groundwater contribution to baseflow (%)
Mean annual air temperature in local catchment
(°C)
Area weighted average mean annual
precipitation in network catchment (ml)
Human land uses
Urban (%)
Developed, open space, low, medium, & high
intensity in network catchment
Agriculture (%)
Pasture/hay & cultivated crops in network
catchment
Road crossings
Density of road crossings in network catchment

764.33

Middle Missouri
Min
Max
10%
1.57 9984.84

90%

15.36 2196.10

32.01
0.00 100.00
0.00 100.00
591.36 195.32 3134.90 288.62 1165.51
38.90 12.20
84.20 19.19 69.00
10.22

0.19

12.92

8.11

12.29

708.96 344.53 1113.56 441.07 929.31

5.93

0.00

92.22

1.99

7.65

51.50

0.00

94.88

4.00

87.86

35.90

0.00

254.61

3.63

70.55

3.91

0.00

73.31

0.00

11.60

2

(number/100 km )
Dams
Dam density (number/100 km2)

165

Table 3.1.—(cont’d)
Landscape variable description
Natural

Mean

Network catchment area (km2)
Fine lithology in network catchment (%)
Mean local catchment elevation (m)
Base-flow index, network catchment
groundwater contribution to baseflow (%)
Mean annual air temperature in local catchment
(°C)
Area weighted average mean annual
precipitation in network catchment (ml)
Human land uses
Urban (%)
Developed, open space, low, medium, & high
intensity in network catchment
Agriculture (%)
Pasture/hay & cultivated crops in network
catchment
Road crossings
Density of road crossings in network catchment

356.01

Upper Mississippi
Min
Max
10%
0.23 9964.22

9.49
0.00
288.86 114.85
49.85
7.84
7.56

3.07

90%

9.65 740.40

100.00
0.00 27.43
589.09 184.87 382.03
74.87 31.21 66.29
13.60

5.48

10.61

823.55 556.81 1190.27 704.84 949.22

8.65

0.00

99.46

3.24

13.05

60.21

0.00

98.06

15.54

89.80

39.58

0.00

643.09

8.88

69.17

0.77

0.00

39.03

0.00

1.88

2

(number/100 km )
Dams
Dam density (number/100 km2)

166

Table 3.2.—Descriptive statistics (i.e., mean, range, 10th and 90th percentiles) of fish biodiversity indices in five freshwater
ecoregions.
Species
Shannon's
Pielou's
Taxonomic Taxonomic
richness (S ) diversity (H' ) evenness (J' ) diversity (Δ) distinctness (Δ*)

Freshwater ecoregion
Appalachian Piedmont
Mean
Range
10%
90%

15
1–34
8
22

1.89
0–2.77
1.28
2.37

0.51
0.11–1.00
0.33
0.69

54.05
0–76.62
38.79
66.62

69.98
21.94–96.17
58.57
79.73

Mean
Range
10%
90%

10
1–31
2
19

1.46
0–2.94
0.36
2.29

0.61
0.14–1.00
0.37
0.95

38.51
0–84.01
6.55
62.48

58.35
21.94–100.00
38.80
80.15

Mean
Range
10%
90%

11
1–37
4
19

1.56
0–2.87
0.74
2.23

0.55
0.10–1.00
0.33
0.82

37.31
0–84.01
18.62
56.98

54.08
21.94–95.65
38.26
74.39

Mean
Range
10%
90%

10
1–29
4
17

1.38
0–2.69
0.63
2.03

0.50
0.12–1.00
0.25
0.76

41.89
0–70.06
16.05
59.91

66.40
21.94–84.01
42.83
82.05

Mean
Range
10%
90%

13
1–48
3
24

1.61
0–3.11
0.62
2.38

0.54
0.07–1.00
0.33
0.78

38.35
0–84.01
15.25
60.74

54.95
21.94–100.00
39.10
74.21

Chesapeake Bay

Laurentian Great Lakes

Middle Missouri

Upper Mississippi

167

Table 3.3.—Pearson’s correlations between the five biodiversity indices for the five study
freshwater ecoregions.

Freshwater ecoregion
Variable
Appalachian Piedmont
Species richness (S )
Shannon's diversity (H' )
Pielou's evenness (J' )
Taxonomic diversity (Δ)
*

Taxonomic distinctness (Δ )
Chesapeake Bay
Species richness (S )
Shannon's diversity (H' )
Pielou's evenness (J' )
Taxonomic diversity (Δ)
*

168

Taxonomic distinctness (Δ )
Laurentian Great Lakes
Species richness (S )
Shannon's diversity (H' )
Pielou's evenness (J' )
Taxonomic diversity (Δ)
*

Taxonomic distinctness (Δ )
Middle Missouri
Species richness (S )
Shannon's diversity (H' )
Pielou's evenness (J' )
Taxonomic diversity (Δ)
*

Taxonomic distinctness (Δ )
Upper Mississippi
Species richness (S )
Shannon's diversity (H' )
Pielou's evenness (J' )
Taxonomic diversity (Δ)
*

Taxonomic distinctness (Δ )
*P <0.05; **P <0.01.

S

H'

J'

.
0.73**
-0.38
0.50**

.
0.25**
0.80**

.
0.33**

0.24**

0.25**

-0.01

.
0.84**
-0.44
0.57**

.
-0.06
0.74**

.
0.09*

0.26**

0.28**

-0.01

.
0.77**
-0.49
0.47**

.
0.02
0.71**

.
0.22**

0.15**

0.13**

0.05**

.
0.66**
-0.52
0.53**

.
0.17**
0.80**

.
0.10**

0.33**

0.16**

-0.31

.
0.81**
-0.47
0.69**

.
-0.05
0.83**

.
0.01

0.46**

0.38**

-0.16

168

Δ

Δ

*

.
0.72**

.

.
0.79**

.

.
0.70**

.

.
0.62**

.

.
0.76**

.

Table 3.4.—Results of multiple linear regression models for relationships of the biodiversity indices to natural landscape variables
and human land use variables, with degrees of freedom, adjusted R2, and standardized regression coefficient (β) describing the
variable strength in the model and direction of influence on biodiversity indices. All variables selected for inclusion in best models
as predictors for biodiversity indices were significant at p-value ≤ 0.01.
Freshwater ecoregion
Variable
Appalachian Piedmont
Species richness (S )
Shannon's diversity (H' )
Pielou's evenness (J' )
Taxonomic diversity (Δ)

d. f. Adj R 2 Area_c Elev Fine_c Baseflow_c Temp_air Precip_c Urban_c Ag_c Rd_dc Dam_dc
1077
1077
1080
1078

0.25
0.16
0.04
0.09

+0.42
–
–
+0.32
–
–
-0.18
–
+0.14
+0.25 -0.12
–

-0.14
-0.14
–
–

-0.14
-0.11
–
–

–
–
–
–

-0.10
-0.11
–
–

+0.10
+0.09
–
–

–
–
–
–

–
–
–
–

Taxonomic distinctness (Δ*)
Chesapeake Bay
Species richness (S )
Shannon's diversity (H' )
Pielou's evenness (J' )
Taxonomic diversity (Δ)

1073

0.14

+0.22 -0.26

–

–

–

–

+0.12

–

–

–

801
801
803
795

0.33
0.24
0.11
0.23

+0.59
–
+0.50
–
-0.31
–
+0.42 -0.19

–
–
–
–

–
–
–
–

+0.19
+0.16
–
–

–
–
-0.09
–

Taxonomic distinctness (Δ*)
Laurentian Great Lakes
Species richness (S )
Shannon's diversity (H' )
Pielou's evenness (J' )
Taxonomic diversity (Δ)

737

0.33

+0.42 -0.45

–

+0.16

–

–

–

–

–

–

3767
3767
3767
3753

0.35
0.17
0.16
0.15

+0.42 +0.16 +0.29
+0.31
–
+0.16
-0.17 -0.08 -0.25
+0.30 -0.13 +0.13

–
–
–
–

+0.31
+0.07
-0.18
–

-0.13
–
+0.16
–

–
–
–
-0.06

–
+0.09
–
–

–
–
–
–

–
+0.05
–
–

Taxonomic distinctness (Δ*)
Middle Missouri
Species richness (S )
Shannon's diversity (H' )
Pielou's evenness (J' )
Taxonomic diversity (Δ)

3652

0.22

+0.39 -0.13

–

+0.11

–

-0.15

-0.12

–

–

880
882
881
879

0.40
0.21
0.22
0.23

+0.47 -0.65 -0.20
+0.24 -0.44 -0.15
-0.39
–
–
+0.28 -0.45
–

–
–
–
–

–
–
-0.22
–

–
–
-0.31
–

–
–
–
–

Taxonomic distinctness (Δ*)
Upper Mississippi
Species richness (S )
Shannon's diversity (H' )
Pielou's evenness (J' )
Taxonomic diversity (Δ)

857

0.24

+0.26 -0.28 +0.12

–

–

+0.19

-0.16

–

–

–

3967
3967
3970
3947

0.52
0.34
0.14
0.25

+0.63
–
+0.50 +0.22
-0.36
–
+0.42 +0.14

–
–
–
–

-0.20
-0.23
–
-0.17

-0.18
–
–
–

+0.22
+0.22
–
+0.20

–
–
–
–

+0.15
+0.09
-0.06
+0.08

–
–
–
–

–
–
–
–

Taxonomic distinctness (Δ*) 3843

0.34

+0.43

+0.07

–

+0.15

+0.19

-0.07

–

–

–

–

–

169

–
+0.16 +0.13
–
+0.14 +0.13
–
-0.15
–
+0.13 +0.16
–

-0.20 +0.10
–
–
+0.13
–
–
–

+0.12
+0.11
–
–

–
–
–
–

APPENDIX 3.B

FIGURES

Figure 3.1.—Five freshwater ecoregions of the United States that comprised the study area
(Abell et al. 2008).

170

Figure 3.2.—Part of a taxonomic cladogram, showing examples of path length weights {wij}
used to define taxonomic diversity and taxonomic distinctness (example from Clark and
Warwick 1998)

171

0.6

0.5

Adjusted R

2

0.4

0.3

0.2

0.1

0.0
s
s
s
y
y
nes
nes
nes
rsit
rsit
t
e
e
h
n
c
e
c
v
v
n
i
i
v
ri
cd
isti
se
's d
es
'
i
i
d
n
u
c
m
o
c
e
i
lo
no
Sp
ann
om
Pie
xo
n
a
Sh
o
T
x
Ta

Biodiversity indices
Figure 3.3.—Range in Adjusted R2 values in biodiversity indices across five study ecoregions
from the multiple linear regressions. The dots are mean values with the whiskers maximum and
minimum values.

172

APPENDIX 3.C

SUPPLEMENTAL TABLES
Table C3.1.—List of species use in biodiversity calculations. (n=312)
Common name
Least brook lamprey
American brook Lamprey
Sea lamprey
Chestnut lamprey
Northern brook lamprey
Southern brook lamprey
Silver lamprey
Lake sturgeon
Shovelnose sturgeon
Paddlefish
Longnose gar
Spotted gar
Shortnose gar
Florida gar
Bowfin
American eel
American Shad
Blueback herring
Alewife
Skipjack herring
Gizzard shad
Threadfin shad
Goldeye
Mooneye
Pink salmon
Coho salmon
Chinook salmon
Cutthroat trout
Rainbow trout
Atlantic salmon
Brown trout
Brook trout
Round whitefish
Rainbow smelt
Northern pike
Redfin pickerel
Chain pickerel

Scientific name
Lampetra aepyptera
Lethenteron appendix
Petromyzon marinus
Ichthyomyzon castaneus
Ichthyomyzon fossor
Ichthyomyzon gagei
Ichthyomyzon unicuspis
Acipenser fulvescens
Scaphirhynchus platorynchus
Polyodon spathula
Lepisosteus osseus
Lepisosteus oculatus
Lepisosteus platostomus
Lepisosteus platyrhincus
Amia calva
Anguilla rostrata
Alosa sapidissima
Alosa aestivalis
Alosa pseudoharengus
Alosa chrysochloris
Dorosoma cepedianum
Dorosoma petenense
Hiodon alosoides
Hiodon tergisus
Oncorhynchus gorbuscha
Oncorhynchus kisutch
Oncorhynchus tshawytscha
Oncorhynchus clarkii
Oncorhynchus mykiss
Salmo salar
Salmo trutta
Salvelinus fontinalis
Prosopium cylindraceum
Osmerus mordax
Esox lucius
Esox americanus
Esox niger

Family
Petromyzontidae
Petromyzontidae
Petromyzontidae
Petromyzontidae
Petromyzontidae
Petromyzontidae
Petromyzontidae
Acipenseridae
Acipenseridae
Polyodontidae
Lepisosteidae
Lepisosteidae
Lepisosteidae
Lepisosteidae
Amiidae
Anguillidae
Clupeidae
Clupeidae
Clupeidae
Clupeidae
Clupeidae
Clupeidae
Hiodontidae
Hiodontidae
Salmonidae
Salmonidae
Salmonidae
Salmonidae
Salmonidae
Salmonidae
Salmonidae
Salmonidae
Salmonidae
Osmeridae
Esocidae
Esocidae
Esocidae

173

Order
Petromyzontiformes
Petromyzontiformes
Petromyzontiformes
Petromyzontiformes
Petromyzontiformes
Petromyzontiformes
Petromyzontiformes
Acipenseriformes
Acipenseriformes
Acipenseriformes
Semionotiformes
Semionotiformes
Semionotiformes
Semionotiformes
Amiiformes
Anguilliformes
Clupeiformes
Clupeiformes
Clupeiformes
Clupeiformes
Clupeiformes
Clupeiformes
Osteoglossiformes
Osteoglossiformes
Salmoniformes
Salmoniformes
Salmoniformes
Salmoniformes
Salmoniformes
Salmoniformes
Salmoniformes
Salmoniformes
Salmoniformes
Osmeriformes
Esociformes
Esociformes
Esociformes

Class
Cephalaspidomorphi
Cephalaspidomorphi
Cephalaspidomorphi
Cephalaspidomorphi
Cephalaspidomorphi
Cephalaspidomorphi
Cephalaspidomorphi
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii

Table C3.1.—(cont’d)
Common name
Muskellunge
Eastern mudminnow
Central mudminnow
Common carp
Goldfish
Cutlips minnow
Tonguetied minnow
Eastern silvery minnow
Mississippi silvery minnow
Plains minnow
Western silvery minnow
Brassy minnow
Golden shiner
Rosyside dace
Redside dace
Fallfish
Creek chub
Sandhills chub
Eastern blacknose dace
Longnose dace
River chub
Bluehead chub
Hornyhead chub
Bull chub
Comely shiner
Bridle shiner
Ironcolor shiner
Spottail shiner
Swallowtail shiner
Rosyface shiner
Emerald shiner
Ghost shiner
Sand shiner
Weed shiner
Mimic shiner
Whitemouth shiner
Highfin shiner
Pugnose shiner
River shiner
Bigeye shiner
Redlip shiner
Greenhead shiner
Dusky shiner
Wedgespot shiner
Blackchin shiner
Blacknose shiner

Scientific name
Esox masquinongy
Umbra pygmaea
Umbra limi
Cyprinus carpio
Carassius auratus
Exoglossum maxillingua
Exoglossum laurae
Hybognathus regius
Hybognathus nuchalis
Hybognathus placitus
Hybognathus argyritis
Hybognathus hankinsoni
Notemigonus crysoleucas
Clinostomus funduloides
Clinostomus elongatus
Semotilus corporalis
Semotilus atromaculatus
Semotilus lumbee
Rhinichthys atratulus
Rhinichthys cataractae
Nocomis micropogon
Nocomis leptocephalus
Nocomis biguttatus
Nocomis raneyi
Notropis amoenus
Notropis bifrenatus
Notropis chalybaeus
Notropis hudsonius
Notropis procne
Notropis rubellus
Notropis atherinoides
Notropis buchanani
Notropis stramineus
Notropis texanus
Notropis volucellus
Notropis alborus
Notropis altipinnis
Notropis anogenus
Notropis blennius
Notropis boops
Notropis chiliticus
Notropis chlorocephalus
Notropis cummingsae
Notropis greenei
Notropis heterodon
Notropis heterolepis

Family
Esocidae
Umbridae
Umbridae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae

174

Order
Esociformes
Esociformes
Esociformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes

Class
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii

Table C3.1.—(cont’d)
Common name
Longnose shiner
Yellowfin shiner
Taillight shiner
Cape Fear shiner
Ozark minnow
Coastal shiner
Silver shiner
Sandbar shiner
Mirror shiner
Telescope shiner
Topeka shiner
Bigeye chub
Silverjaw minnow
Highback chub
Rosyface chub
Channel shiner
Suckermouth minnow
Central stoneroller
Largescale stoneroller
Bluefin stoneroller
Bluntnose minnow
Fathead minnow
Bullhead minnow
Lake chub
Grass carp
Northern redbelly dace
Southern redbelly dace
Finescale dace
Mountain redbelly dace
Rudd
Silver carp
Bighead carp
Satinfin shiner
Ocmulgee shiner
Greenfin shiner
Whitetail shiner
Thicklip chub
Bannerfin shiner
Red shiner
Whitefin shiner
Fieryblack shiner
Spotfin shiner
Blacktail shiner
Steelcolor shiner
Altamaha shiner
Santee chub
Gravel chub

Scientific name
Notropis longirostris
Notropis lutipinnis
Notropis maculatus
Notropis mekistocholas
Notropis nubilus
Notropis petersoni
Notropis photogenis
Notropis scepticus
Notropis spectrunculus
Notropis telescopus
Notropis topeka
Hybopsis amblops
Notropis buccatus
Hybopsis hypsinotus
Hybopsis rubrifrons
notropis wickliffi
Phenacobius mirabilis
Campostoma anomalum
Campostoma oligolepis
Campostoma pauciradii
Pimephales notatus
Pimephales promelas
Pimephales vigilax
Couesius plumbeus
Ctenopharyngodon idella
Phoxinus eos
Phoxinus erythrogaster
Phoxinus neogaeus
Phoxinus oreas
Scardinius erythrophthalmus
Hypophthalmichthys molitrix
Hypophthalmichthys nobilis
Cyprinella analostana
Cyprinella callisema
Cyprinella chloristia
Cyprinella galactura
Cyprinella labrosa
Cyprinella leedsi
Cyprinella lutrensis
Cyprinella nivea
Cyprinella pyrrhomelas
Cyprinella spiloptera
Cyprinella venusta
Cyprinella whipplei
Cyprinella xaenura
Hybopsis zanema
Erimystax x-punctatus

Family
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae

175

Order
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes

Class
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii

Table C3.1.—(cont’d)
Common name
White shiner
Crescent shiner
Striped shiner
Warpaint shiner
Common shiner
Bleeding shiner
Bandfin shiner
Rosefin shiner
Ribbon shiner
Redfin shiner
Speckled chub
Silver chub
Pearl dace
Pugnose minnow
Flathead chub
Longnose sucker
Mountain sucker
Quillback
River carpsucker
Highfin carpsucker
Lake chubsucker
Creek chubsucker
Shorthead redhorse
Silver redhorse
Bigeye jumprock
River redhorse
Blacktip jumprock
Golden redhorse
V-lip redhorse
Striped jumprock
Greater redhorse
Northern hog sucker
Roanoke Hogsucker
Blue sucker
Smallmouth buffalo
Bigmouth buffalo
Black buffalo
Spotted sucker
Oriental weatherfish
Blue catfish
Channel catfish
Tadpole madtom
Margined madtom
Freckled madtom
Slender madtom
Stonecat
Carolina madtom

Scientific name
Luxilus albeolus
Luxilus cerasinus
Luxilus chrysocephalus
Luxilus coccogenis
Luxilus cornutus
Luxilus zonatus
Luxilus zonistius
Lythrurus ardens
Lythrurus fumeus
Lythrurus umbratilis
Macrhybopsis aestivalis
Macrhybopsis storeriana
Margariscus margarita
Opsopoeodus emiliae
Platygobio gracilis
Catostomus catostomus
Catostomus platyrhynchus
Carpiodes cyprinus
Carpiodes carpio
Carpiodes velifer
Erimyzon sucetta
Erimyzon oblongus
Moxostoma macrolepidotum
Moxostoma anisurum
Moxostoma ariommum
Moxostoma carinatum
Moxostoma cervinum
Moxostoma erythrurum
Moxostoma pappillosum
Moxostoma rupiscartes
Moxostoma valenciennesi
Hypentelium nigricans
Hypentelium roanokense
Cycleptus elongatus
Ictiobus bubalus
Ictiobus cyprinellus
Ictiobus niger
Minytrema melanops
Misgurnus anguillicaudatus
Ictalurus furcatus
Ictalurus punctatus
Noturus gyrinus
Noturus insignis
Noturus nocturnus
Noturus exilis
Noturus flavus
Noturus furiosus

Family
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Catostomidae
Cobitidae
Ictaluridae
Ictaluridae
Ictaluridae
Ictaluridae
Ictaluridae
Ictaluridae
Ictaluridae
Ictaluridae

176

Order
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Siluriformes
Siluriformes
Siluriformes
Siluriformes
Siluriformes
Siluriformes
Siluriformes
Siluriformes

Class
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii

Table C3.1.—(cont’d)
Common name
Speckled madtom
Brindled madtom
Northern madtom
Flathead catfish
Snail bullhead
White catfish
Black bullhead
Yellow bullhead
Brown bullhead
Flat bullhead
Spotted bullhead
Swampfish
Pirate perch
Trout-perch
Burbot
Banded killifish
Mummichog
Golden topminnow
Northern plains killifish
Blackspotted topminnow
Plains killifish
Northern studfish
Lined topminnow
Blackstripe topminnow
Speckled killifish
Plains topminnow
Starhead topminnow
Bluefin killifish
Mosquitofish
Least killifish
Inland silverside
Brook silverside
Threespine stickleback
Ninespine stickleback
Fourspine stickleback
Brook stickleback
Slimy sculpin
Mottled sculpin
Banded sculpin
Potomac sculpin
White perch
Striped bass
White bass
Yellow bass
Mud sunfish
Rock bass
Roanoke bass

Scientific name
Noturus leptacanthus
Noturus miurus
Noturus stigmosus
Pylodictis olivaris
Ameiurus brunneus
Ameiurus catus
Ameiurus melas
Ameiurus natalis
Ameiurus nebulosus
Ameiurus platycephalus
Ameiurus serracanthus
Chologaster cornuta
Aphredoderus sayanus
Percopsis omiscomaycus
Lota lota
Fundulus diaphanus
Fundulus heteroclitus
Fundulus chrysotus
Fundulus kansae
Fundulus olivaceus
Fundulus zebrinus
Fundulus catenatus
Fundulus lineolatus
Fundulus notatus
Fundulus rathbuni
Fundulus sciadicus
Fundulus dispar
Lucania goodei
Gambusia affinis
Heterandria formosa
Menidia beryllina
Labidesthes sicculus
Gasterosteus aculeatus
Pungitius pungitius
Apeltes quadracus
Culaea inconstans
Cottus cognatus
Cottus bairdii
Cottus carolinae
Cottus girardi
Morone americana
Morone saxatilis
Morone chrysops
Morone mississippiensis
Acantharchus pomotis
Ambloplites rupestris
Ambloplites cavifrons

Family
Ictaluridae
Ictaluridae
Ictaluridae
Ictaluridae
Ictaluridae
Ictaluridae
Ictaluridae
Ictaluridae
Ictaluridae
Ictaluridae
Ictaluridae
Amblyopsidae
Aphredoderidae
Percopsidae
Gadidae
Fundulidae
Fundulidae
Fundulidae
Fundulidae
Fundulidae
Fundulidae
Fundulidae
Fundulidae
Fundulidae
Fundulidae
Fundulidae
Fundulidae
Fundulidae
Poeciliidae
Poeciliidae
Atherinopsidae
Atherinopsidae
Gasterosteidae
Gasterosteidae
Gasterosteidae
Gasterosteidae
Cottidae
Cottidae
Cottidae
Cottidae
Moronidae
Moronidae
Moronidae
Moronidae
Centrarchidae
Centrarchidae
Centrarchidae

177

Order
Siluriformes
Siluriformes
Siluriformes
Siluriformes
Siluriformes
Siluriformes
Siluriformes
Siluriformes
Siluriformes
Siluriformes
Siluriformes
Percopsiformes
Percopsiformes
Percopsiformes
Gadiformes
Cyprinodontiformes
Cyprinodontiformes
Cyprinodontiformes
Cyprinodontiformes
Cyprinodontiformes
Cyprinodontiformes
Cyprinodontiformes
Cyprinodontiformes
Cyprinodontiformes
Cyprinodontiformes
Cyprinodontiformes
Cyprinodontiformes
Cyprinodontiformes
Cyprinodontiformes
Cyprinodontiformes
Atheriniformes
Atheriniformes
Gasterosteiformes
Gasterosteiformes
Gasterosteiformes
Gasterosteiformes
Scorpaeniformes
Scorpaeniformes
Scorpaeniformes
Scorpaeniformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes

Class
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii

Table C3.1.—(cont’d)
Common name
Shadow bass
Flier
Blackbanded sunfish
Bluespotted sunfish
Banded sunfish
Redbreast sunfish
Green sunfish
Warmouth
Bluegill
Pumpkinseed
Orangespotted sunfish
Dollar sunfish
Longear sunfish
Redear sunfish
Spotted sunfish
Largemouth bass
Spotted bass
Redeye bass
White crappie
Black crappie
Everglades pygmy sunfish
Banded pygmy sunfish
Swamp darter
Tessellated darter
Sawcheek darter
Glassy darter
Bluntnose darter
Slough darter
Orangethroat darter
Johnny darter
Mud darter
Greenside darter
Rainbow darter
Carolina darter
Saluda darter
Iowa darter
Fantail darter
Savannah darter
Christmas darter
Turquoise darter
Longfin darter
Spotted darter
Pinewoods darter
Least darter
Goldstripe darter
Riverweed darter
Seagreen darter

Scientific name
Ambloplites ariommus
Centrarchus macropterus
Enneacanthus chaetodon
Enneacanthus gloriosus
Enneacanthus obesus
Lepomis auritus
Lepomis cyanellus
Chaenobryttus gulosus
Lepomis macrochirus
Lepomis gibbosus
Lepomis humilis
Lepomis marginatus
Lepomis megalotis
Lepomis microlophus
Lepomis punctatus
Micropterus salmoides
Micropterus punctulatus
Micropterus coosae
Pomoxis annularis
Pomoxis nigromaculatus
Elassoma evergladei
Elassoma zonatum
Etheostoma fusiforme
Etheostoma olmstedi
Etheostoma serrifer
Etheostoma vitreum
Etheostoma chlorosoma
Etheostoma gracile
Etheostoma spectabile
Etheostoma nigrum
Etheostoma asprigene
Etheostoma blennioides
Etheostoma caeruleum
Etheostoma collis
Etheostoma saludae
Etheostoma exile
Etheostoma flabellare
Etheostoma fricksium
Etheostoma hopkinsi
Etheostoma inscriptum
Etheostoma longimanum
Etheostoma maculatum
Etheostoma mariae
Etheostoma microperca
Etheostoma parvipinne
Etheostoma podostemone
Etheostoma thalassinum

Family
Centrarchidae
Centrarchidae
Centrarchidae
Centrarchidae
Centrarchidae
Centrarchidae
Centrarchidae
Centrarchidae
Centrarchidae
Centrarchidae
Centrarchidae
Centrarchidae
Centrarchidae
Centrarchidae
Centrarchidae
Centrarchidae
Centrarchidae
Centrarchidae
Centrarchidae
Centrarchidae
Elassomatidae
Elassomatidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae

178

Order
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes

Class
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii

Table C3.1.—(cont’d)
Common name
Banded darter
Yellow perch
Logperch
Stripeback darter
Shield darter
Dusky darter
Channel darter
Piedmont darter
Gilt darter
Blackside darter
Blackbanded darter
Slenderhead darter
Roanoke darter
River darter
Western sand darter
Eastern sand darter
Freshwater drum
Striped mullet
Round goby
Southern flounder
Hogchoker
Pallid shiner
Clear chub
Pinewoods shiner
Scarlet shiner
Sailfin shiner
Notchlip redhorse
Crystal darter
Holiday darter
Smallmouth bass
White sucker
Black redhorse
Torrent sucker
Shoal chub
Blue Ridge sculpin
Bayou topminnow
Sauger
Walleye
Bigmouth shiner
Carmine shiner
Western blacknose dace

Scientific name
Etheostoma zonale
Perca flavescens
Percina caprodes
Percina notogramma
Percina peltata
Percina sciera
Percina copelandi
Percina crassa
Percina evides
Percina maculata
Percina nigrofasciata
Percina phoxocephala
Percina roanoka
Percina shumardi
Ammocrypta clara
Ammocrypta pellucida
Aplodinotus grunniens
Mugil cephalus
Neogobius melanostomus
Paralichthys lethostigma
Trinectes maculatus
Hybopsis amnis
Hybopsis winchelli
Lythrurus matutinus
Lythrurus fasciolaris
Pteronotropis hypselopterus
Moxostoma collapsum
Crystallaria asprella
Etheostoma brevirostrum
Micropterus dolomieu
Catostomus commersonii
Moxostoma duquesnii
Thoburnia rhothoeca
Macrhybopsis hyostoma
Cottus caeruleomentum
Fundulus notti
Sander canadensis
Sander vitreus
Hybopsis dorsalis
Notropis percobromus
rhinichthys obtusus

Family
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Percidae
Sciaenidae
Mugilidae
Gobiidae
Paralichthyidae
Achiridae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Cyprinidae
Catostomidae
Percidae
Percidae
Centrarchidae
Catostomidae
Catostomidae
Catostomidae
Cyprinidae
Cottidae
Fundulidae
Percidae
Percidae
Cyprinidae
Cyprinidae
Cyprinidae

179

Order
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Perciformes
Mugiliformes
Perciformes
Pleuronectiformes
Pleuronectiformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Perciformes
Perciformes
Perciformes
Cypriniformes
Cypriniformes
Cypriniformes
Cypriniformes
Scorpaeniformes
Cyprinodontiformes
Perciformes
Perciformes
Cypriniformes
Cypriniformes
Cypriniformes

Class
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii
Actinopterygii

REFERENCES

180

REFERENCES

Abell, R. Á., D. M. Olsen, E. Dinerstein, P. T. Hurley, J. T. Diggs, W. Eichbaum, S. Walters, W
Wettengel, T. Allnutt, C. J. Loucks, and P. Hedao. 2000. Freshwater ecoregions of North
America: a conservation assessment. Island Press, Washington, D.C.
Abell, R. M., 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. López, R. E. Reis, J. G. Lundberg, M. H.
Sabaj Pérze, and P. Petry, 2008. Freshwater ecoregions of the world: a new map of
biogeographic units for freshwater biodiversity conservation. Bioscience 58:403–414.
Angermeier, P. L. and I. J. Schlosser. 1989. Species-area relationship for stream fishes. Ecology
70(5):1450–1462.
Angermeier, P. L. and M. R. Winston. 1998. Local vs. regional influences on local diversity in
stream fish communities of Virgina. Ecology 79(3):911–927.
Angermeier, P. L. and M. R. Winston. 1999. Characterizing fish community diversity across
Virginia landscapes: prerequisite for conservation. Ecological Application 9:335–349.
Brenden, T. O., R. D. Clark Jr., A. R. Cooper, P. W. Seelbach, L. Wang, S. S. Aichele, and E. G.
Bissell 2006. A GIS framework for collecting, managing, and analyzing multi-scale
variables across large regions for river conservation and management. in R. M. Hughes,
L. Wang, and P. W. Seelbach, editors, Landscape influences on stream habitats and
biological assemblages. pp. 49–74. American Fisheries Society, Symposium 48,
Bethesda, Maryland.
Bouska, W. W. and C. P. Paukert. 2010. Road crossing designs and their impact on fish
assemblages of Great Plains streams. Transactions of the American Fisheries Society
139:214–222.
Buisson, L., W. Thuiller, S. Lek, P. Limp. and G. Grenouillet. 2008. Climate change hastens the
turnover of stream fish assemblages. Global Change Biology 14:2232–2248.
Burkhead, N. M. 2012. Extinction rates in North American freshwater fishes 1900–2010.
BioScience 62:798–808.
Cress, J., D. Soller, R. Sayre, P. Comer, and H. Warner, 2010. Terrestrial Ecosystems, Surficial
Lithology of the Conterminous United States: U.S. Geological Survey Scientific
Investigations Map 3126, scale 1:5,000,000, 1 sheet. Available at:
(http://pubs.usgs.gov/sim/3126).
181

Clarke K. R., and R. M. Warwick. 1998. A taxonomic distinctness index and its statistical
properties. Journal of Applied Ecology 35:523–531.
Clarke, K. R., and R. M. Warwick. 2001. A further biodiversity index applicable to species lists:
variation in taxonomic distinctness. Marine Ecology Progress Series 216:265–278.
Daly, C., M. Halbleib, J. I. Smith, W. P. Gibson, M. K. Doggett, G. H. Taylor, J. Curtis, and P.
P. Pasteris. 2008. Physiographically sensitive mapping of climatological temperature and
precipitation across the conterminous United States. International Journal of Climatology
28:2031–2064.
Dudgeon, D., A. H. Arthington, M. O. Gessner, Z. Kawabata, D. J. Knowler, C. Lévêque, R. J.
Naiman, A. Prieur-Richard, D. Soto, M. L. J. Stiassny, and C. A. Sullivan. 2006.
Freshwater biodiversity: importance, threats, status and conservation challenges.
Biological Review 81:163–182.
Eadie, J. McA., T. A. Hurly, R. D. Montgomerie, and K. L. Teather. 1986. Lakes and rivers as
islands: species–area relationships in the fish faunas of Ontario. Environmental Biology
of Fishes 15:81–89.
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 landscape anthropogenic activities. Ecological Restoration 29:133–151.
Esselman, P.C., D. M. Infante, L. Wang, A. R. Cooper, D. Wieferich, Y. Tsang, D. J.
Thornbrugh, W. W. Taylor. 2013. Regional fish community indicators of landscape
disturbance to catchments of conterminous United States. Ecological Indicators 26:163–
173.
Fletcher, T. D. 2012. QuantPsyc: Quantitative Psychology Tools. R package version 1.5.
Available at: (http://CRAN.R-project.org/package=QuantPsyc).
Gesch, D. B., 2007. The National Elevation Dataset, in D. Maune, ed., Digital Elevation Model
Technologies and Applications: The DEM User’s Manual, 2nd Edition: Bethesda,
Maryland, American Society for Photogrammetry and Remote Sensing p. 99–118.
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.
Heino, J., H. Mykrä, H. Hämäläinen, J. Aroviita, and T. Muotka. 2007. Responses of taxonomic
distinctness and species diversity indices to anthropogenic impacts and natural
environmental gradients in stream macroinvertebates. Freshwater Biology 52:1846–1861.

182

Heino, J., J. Soininen, J. Lappalainen, and R. Virtanan. 2005. The relationship between species
richness and taxonomic distinctness in freshwater organisms. Limnology and
Oceanography 50:978–986.
Helfman, G. S. 2007. Fish conservation: A guide to understanding and restoring global aquatic
biodiversity and fishery resources. Island Press, Washington, D.C.
Homer, C. C. Huang, L. Yang, B. Wylie and M. Coan. 2004. Development of a 2001 National
Landcover Database for the United States. Photogrammetric Engineering and Remote
Sensing 70(7):829–840.
Hughes, R. M., P. R. Kaufmann, A. T. Herlihy, T. M. Kincaid, L. Reynolds, and D. P. Larsen.
1998. A process for developing and evaluating indices of fish assemblage integrity.
Canadian Journal of Fisheries and Aquatic Sciences 55:1618–1631.
Hughes, R. M., C. M. R. Whittier, and D. P. Larsen. 1990. A regional framework for
establishing recovery criteria. Environmental Management 14:673–683.
ITIS. 2010. Retrieved [04, 26, 2010], from the Integrated Taxonomic Information System.
Available at: (http://www.itis.gov).
Jackson, D. A., P. R. Peres-Neto, and J. D. Olden. 2011. What controls who is where in
freshwater fish communities – the roles of biotic, abiotic, and spatial factors. Canadian
Journal of Fisheries and Aquatic Science 58:157–170.
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 11(4):211-217.
Jelks, H. L., S. J. Walsh, N. M. Burkhead, S. Contreras-Balderas, E. Díaz-Pardo, D. A.
Hendrickson, J. Lyons, N. E. Mandrak, F. McCormick, J. S. Nelson, S. P. Platania, B. A.
Porter, C. B. Renaud, J. J. Schmitter-Soto, E. B. Taylor, M. L. Warren Jr., 2008.
Conservation status of imperiled North American freshwater and diadromous fishes.
Fisheries 33 (8), 372–407.
Little, J. D. C., K. G. Murty, D. W. Sweeney, C. Karel. 1963. An algorithm for the traveling
salesman problem. Operations Research 11(6):972–989.
Legendre, P. and L. Legendre. 2012. Numerical ecology, 3rd English edition. Amsterdam, The
Netherlands: Elsevier Science BV.
Lumley, T., using Fortran code by Alan Miller (2009). Leaps: regression subset selection. R
package version 2.9. Available at: (http://CRAN.R-project.org/package=leaps).
Magurran, A. E. 2004. Measuring Biological Diversity, 2nd edition. Blackwell, Oxford.

183

Magurran, A. E. and D. A. T. Phillip. 2001. Implications of species loss in freshwater fish
assemblages. Ecography 24:645-650.
Marsh-Matthews, E. and W. J. Matthews. 2000. Geographic, terrestrial and aquatic factors:
which most influence the structure of stream fish assemblages in the Midwestern United
States? Ecology of Freshwater Fish 9:9–21.
Matthews, W. J. 1998. Patterns in freshwater fish ecology. Chapman and Hall, New York.
Maxwell, J. R., C. J. Edwards, M. E. Jensen, S. J. Paustian, H. Parrott, D. M. Hill. 1995. A
Hierarchical Framework of Aquatic Ecological Units in North America (Nearctic Zone).
St. Paul (MN): USDA Forest Service, North Central Forest Experiment Station. General
Technical Report NC-176.
McKinney, M. L., and J. L. Lockwood. 1999. Biotic homogenization: A few winners replacing
many losers in the next mass extinction. Trends in Ecology and Evolution 14:450-453.
Miller, A. J. 2002. Subset Selection in Regression. 2nd edition. Chapman and Hall, New York.
Olden, J. D., M. J. Kennard, F. Leprieur, P. A. Tedesco, K. O. Winemiller and E. GarcíaBerthou. 2010. Conservation biogeography of freshwater fishes: recent progress and
future challenges. Diversity and Distributions 16:496–513.
Olden, J. D., and N. L. Poff. 2004. Ecological processes driving biotic homogenization: Testing
a mechanistic model using fish fauna. Ecology 85:1867-1875.
Oksanen, J., F. G. Blanchet, R. Kindt, P. Legendre, P. R. Minchin, R. B. O'Hara, G. L. Simpson,
P. Solymos, M. H. H. Stevens and H. Wagner. 2013. vegan: Community Ecology
Package. R package version 2.0-9. Available at: (http://CRAN.Rproject.org/package=vegan).
Perkin, J. S. and K. B. Gido. 2012. Fragmentation alters stream fish community structure in
dendritic ecological networks. Ecological Applications 22:2176–2187.
Pielou, E. C. 1975. Ecological diversity. Wiley and Sons, New York.
Pont, D., B. Hugueny, U. Beier, D. Goffaux, A. Melcher, R. Noble, C. Rogers, N. Roset and S.
Schmutz. 2006. Assessing river biotic condition at the continental scale: a European
approach using functional metrics and fish assemblages. Journal of Applied Ecology, 43,
70–80.
PRISM. 2013. PRISM Climate Group. Oregon State University. Available at
(http://www.prism.oregonstate.edu/).
R Core Team. 2013. R: A language and environment for statistical computing. R Foundation for
Statistical Computing, Vienna, Austria. Available at: (http://www.R-project.org/).
184

Rahel, F. J. 2002. Homogenization of freshwater faunas. Annual Review of Ecology and
Systematics 33:291-315.
Rahel, F. J. 2000. Homogenization of fish faunas across the United States. Science 288:854-856.
Sala, O. E., F. S. Chapin, J. J. Armesto, E. Berlow, J. Bloomfield, R. Dirzo, E. Huber-Sanwald,
L. F. Huenneke, R. B. Jackson, A Kinzig, R. Leemans, D. M. Lodge, H. A. Mooney, M.
Oesterheld, N. L. Poff, M. T. Sykes, B. H. Walker, M. Walker and D. H. Wall, 2000.
Biodiversity – global biodiversity scenarios for the year 2100. Science 287:1770–1774.
Schlosser, I. J. 1991. Stream fish ecology: a landscape perspective. BioScience 41(10):704–712.
Scott, M. C. 2006. Winners and losers among stream fishes in relation to land use legacies and
urban development in the southeastern US. Biological Conservation 127:301–309.
Scott, M. C. and G. S. Helfman, 2001. Native invasions, homogenization, and the mismeasure of
integrity of fish assemblages. Fisheries 26:6-15.
Shannon, C. E. 1948. A mathematical theory of communication. Bell System Technical Journal
27:379–423.
Sheldon, A. L. 1968. Species diversity and longitudinal succession in stream fishes. Ecology
49:193–198.
Soller, D. R., and M. C. Reheis, compilers, 2004. Surficial materials in the conterminous United
States: U.S. Geological Survey. Open–file Report 03–275, scale 1:5,000,000. Available
at: (http://pubs.usgs.gov/of/2003/of03–275/).
Stirling, G. and B. Wilsey. 2001. Empirical relationships between species richness, evenness,
and proportional diversity. The American Naturalist 158:286-99.
Stoddard, J.L., A.T. Herlihy, D.V. Peck, R.M. Hughes, T.R. Whittier and E. Tarquinio. 2008. A
process for creating multimetric indices for large-scale aquatic surveys. The North
American Benthological Society 27(4):878-891.
Thornbrugh, D. J. This volume a. Multiple facets of streams thermal regime: effects of stream
temperature on Michigan’s fluvial fish assemblages. in Regional influence of landscape
features and processes on fluvial fish assemblages, Doctoral Dissertation, Michigan State
University, East Lansing, Michigan.
Thornbrugh, D. J. This volume b. Landscape effects on fluvial fish assemblage structure:
regional responses to human land uses. in Regional influence of landscape features and
processes on fluvial fish assemblages, Doctoral Dissertation, Michigan State University,
East Lansing, Michigan.

185

Thornbrugh, D. J., and K. B. Gido. 2010. Influence of spatial positioning within stream networks
on fish assemblage structure in the Kansas River basin, USA. Canadian Journal of
Fisheries and Aquatic Sciences 67:143–156.
U.S. Census Bureau. 2002. Redistricting Census 2000 TIGER/ Line Files [machine-readable data
files]. Washington, DC: U.S. Census Bureau. Available at: (www.census.gov/geo/mapsdata/data/tiger.html).
U.S. Environmental Protection Agency (USEPA). 2008. United States Environmental Protection
Agency Multi-Resolution Land Characteristics Consortium. Available at:
(www.epa.gov/mrlc/).
U.S. Environmental Protection Agency and U.S. Geological Survey (USEPA and USGS). 2005.
National Hydrography Dataset Plus, NHDPlusV1, Available at: (www.horizonsystems.com/nhdplus/).
U.S. Geological Survey (USGS). 2012. National Anthropogenic Barrier Dataset (NABD) 2012.
Available at: (http://www.fishhabitat.org/content/nfhp-data-system).
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 143:688–699.
Vannote, R. L., G. W. Minshall, K. W. Cummins, J. R. Sedell and C. E. Cushing. 1980. The river
continuum concept. Canadian Journal of Fisheries and Aquatic Science 37:130–137.
Wang, L., D. Infante, P. Esselman, A. Cooper, D. Wu, W. Taylor, D. Beard, G. Whelan, and A.
Ostroff. 2011. A hierarchical spatial framework and database for the national river fish
habitat condition assessment. Fisheries 36:436–449.
Wang, L., J. Lyons and P. Kanehl, 2001. Impacts of urbanization on stream habitat and fish
across multiple spatial scales. Environmental Management 28:255–266.
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.
Warwick, R. M., and K. R. Clarke. 1995. New ‘biodiversity measures reveal a decrease in
taxonomic distinctness with increasing stress. Marine Ecology Progress Series 129:301305.
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.
Wolock, D. M., 2003. Base-flow index grid for the conterminous United States: U.S. Geological
Survey Open-File Report 03–263, digital data set. Available at:
(http://water.usgs.gov/lookup/getspatial?bfi48grd).
186

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.

187

CHAPTER FOUR

CONCLUSIONS: REGIONAL INFLUENCES ON FISH ASSEMBLAGES – SUMMARY
OF FINDINGS AND IMPLICATIONS FOR MANAGEMENT

In the concluding chapter, I provide a brief overview of the principle findings of my
dissertation, Further I will highlight some implications for management opportunities that
presented themselves through results and research findings of these studies.
Principle Findings
Chapter one: multiple facets of a streams thermal regime: effects of stream temperature
on Michigan’s fluvial fish assemblages.—In this study, we summarized stream water
temperature data collected from 233 stream reaches located throughout three ecoregions within
the state of Michigan into multiple thermal metrics and identified major patterns in thermal
regimes of streams representative of temperature magnitudes, variability of ranges in
temperatures, and the timing of both summer maximums in temperature and variability of
temperatures. We then showed how various thermal metrics were influential to species of fishes
common to the state of Michigan. Thermal preferences of fishes were shown to vary across
ecoregions, but dominant temperature metrics influencing fishes in Michigan were those
describing maximums in stream summer temperatures and variability in stream summer
temperature. This chapter provides additional insight into which fluvial fish species in Michigan
are being affected by different thermal characteristic and highlights some of the regional
variability in influences across the state.

188

Results from this chapter also showed that abundances of some fish species in Michigan
are being indirectly influenced by landscape-level factors through their mediating effects on
stream temperature. This highlights that specific facets of stream temperature (including
summer thermal maximums and variability in summer stream temperatures) may be vulnerable
to different changes in landscapes drained by Michigan streams. Maximum summer stream
temperatures are potentially most heavily influenced by changes in land cover (e.g. % forest land
cover), while the variability in summer stream temperatures may be most vulnerable to climate
change (e.g. mean July air temperature, Table 1.9) across Michigan streams. This adds critical
new insight into the mediating effects that landscape-scale influences like forest land cover and
watershed July air temperature may have on controlling abundances and distribution of stream
fishes in Michigan, and points to two different sources of environmental risk (i.e., land use
change and climate change) that may be future sources influencing changes if fish assemblages
in Michigan streams.
Chapter two: landscape effects on fluvial fish assemblage structure: regional responses
to human land uses.—In this study, we evaluated influences of human land uses on stream fish
assemblages from more than 10,500 locations across five large regions of the eastern United
States. Natural landscape factors (i.e., catchment area, percent catchment fine lithology, mean
local elevation, catchment base-flow index, local mean annual air temperature, and catchment
mean annual precipitation) and human land uses (i.e., percent urban and agricultural in the
catchment and density of dams in the catchment) were identified as describing major patterns of
variation in landscape factors of the study regions. Fish assemblages were summarized by
metrics including species richness; percent intolerant and game individuals; and percent
individuals that are invertivores, omnivores, lithophiles, and rheophiles (Lyons 1992; Barbour et

189

al. 1999; Frimpong and Angermeier 2009). These metrics have been used in many investigations
for assessing stream integrity, and they describe variability in assemblages across the study
regions. After controlling for dominant natural landscape factors, we identified that dominant
human land uses (i.e., percent urban and agricultural in the catchment and density of dams in the
catchment) were associated with change in fish assemblage metrics occurring across study sites,
and that these influences varied by study regions.
Results highlighted that there are regional differences in dominant human land uses
influencing changes in fish assemblage metrics and that levels of catchment urban and
agriculture land use affecting stream fishes varied by study regions. Findings from this study
add new depth to our understanding of regional influences on streams fishes and highlights that
location can play a big part in dictating which and at what levels human stressors are influencing
stream fish assemblages.
Chapter three: regional trends of biodiversity indices: natural landscape and human land
use controls on stream fish assemblages.—In this chapter, characterized trends in stream fish
assemblages from more than 10,500 locations across five ecoregions in the eastern portion of the
United States using biodiversity indices including species richness, Shannon’s diversity, Pielou’s
evenness, taxonomic diversity, and taxonomic distinctness (Shannon 1948; Pielou 1975; Clarke
and Warwick 1998; Clarke and Warwick 2001). General patterns in correlations among
individual indices were variable, but relationships among indices were similar across ecoregions.
For example, species richness and Shannon’s diversity were always highly correlated in each of
the five study ecoregions, yet taxonomic distinctiveness was never highly correlated with these
factors, emphasizing the independence of distinctness from richness and Shannon’s diversity.
Results of multiple linear regression models suggest differences in landscape influences on
190

groups of metrics. Natural landscape factors including stream size and elevation were important
and consistent predictors of biodiversity indices, while climatological factors and factors
indicating groundwater delivery to streams had more variable influences. Percent of agriculture
and urban lands in catchments were important predictors of biodiversity across all regions.
Catchment agriculture had a fairly consistent response on richness and diversity, while
catchment urban land use most consistently predicted taxonomic distinctness, yet that response
varied by region. High amounts of catchment urban land use could negatively affect taxonomic
distinctness causing shifting fish assemblage composition from endemic faunas to more tolerant
fish fauna promoting fish assemblages with more regional similarity (McKinney and Lockwood
1999; Rahel 2000). Together, these results suggest that multiple measures of fish biodiversity
should be used to characterize response of stream fish assemblages to landscape influences, not
only because they capture unique attributes of the fish assemblage, but also because indices are
being influenced by land uses differently.
Management Implication
Chapter one: management implications.— Our results emphasized that thermal
maximums in the summer months are important to fish assemblage structure, but the variability
in monthly and daily range in stream temperatures also influences fish assemblage structure in
Michigan. These results have important implications for thermal habitat availability and their
suitability for specific game fish species and species of greatest conservation need (SGCN).
Under predicted land use change and climate changes, there may be substantial effects on prey
fish availability for game species and SGCN species related to stream temperature changes in
Michigan. Environmental changes may also cause changes in interactions among competing

191

species or in predator–prey relationships due in part to changes in the thermal regime in
Michigan streams (Taniguchi and Nakano 2000).
While this study was conducted with comprehensive datasets of community fish
assemblages with corresponding stream temperature data from throughout Michigan, the
seasonality of our dataset was not complete enough to fully explore all of the seasonal effects
that streams thermal regimes may also have in structuring fish assemblages. Other seasons are
known to be energy expensive times for fishes, including spring and fall due to spawning and/or
migration as well as winter, during which fish may be limited by cold temperatures and/or
feeding opportunities (Carter 2005). Understanding the full annual variability in a stream’s
thermal regime is vital for developing management strategies that maximize the suitability of
conditions for stream fishes of interest. This requires additional research and underscores the
importance of sampling programs that capture stream thermal characteristics throughout the year.
Also, to the degree that our results showed the importance of multiple unique characteristics of
thermal regimes in Michigan, management efforts targeting the collection of continuous
temperature data (vs., for example, maximum and minimum temperatures only) would be
beneficial for understanding the complex set of responses of stream fishes to thermal regimes,
including fish response to variability, timing, and frequency of key thermal events. Together,
such information would ensure better understanding or relationships between stream fishes and
thermal regimes of streams, and would encourage development of more effective management
strategies in the face of current and future changes to stream temperatures.
Chapter two: management implications.— Our results and the approaches taken can be
used to assess regional influences of human land uses that are affecting changes in fish
assemblages and can provide insights into what levels of human land use by regions will be
192

expected to cause change in assemblages. This study underscored regional differences in
dominant human land uses like catchment urban and agriculture and density of dams in the
catchment and quantified that different levels of human land uses like percent of urban and
agriculture in the catchment are effecting change in fish assemblage metrics, but that these levels
can be regionally specific. If dominant human land uses and their levels of influence on critical
management factors have regionally-specific influences, these influences should be accounted
for and incorporated into the planning process to develop realistic management priorities.
Studies characterizing critical human pressures in one ecoregion may not be directly applicable
to a different ecoregion because of these differences. Well-designed regionalized studies could
help further characterize regional differences in human land uses effects on stream biota,
elucidate important mechanisms of influence on stream systems, and have transferable results or
components to help meet management priorities across large spatial extents.
Results elucidate that interregional comparison into mechanisms controlling regional fish
assemblage response to human land uses may be a fruitful area of additional investigation. Some
of the potential regional differences causing different fish assemblage response may relate to the
buffering capacity of lithology and/or topographic landscape factors as well as natural land
covers in a region. The types and proportions of natural versus human land cover and their
spatial arrangement in the catchment have been shown to affect the delivery of runoff and other
materials to stream systems (Zimmerman et al. 2003; Blann et al. 2009; Carpenter et al. 2011).
Intensity of urban classes, specific agricultural practices, and riparian buffers to streams are a
few examples of factors that may influence runoff and the delivery of materials to riverine
ecosystems (Blann et al. 2009; Carpenter et al. 2011), and such factors can cause heterogeneous
responses in stream biota across regions (Meador and Goldstein 2003; Morgan and Cushman

193

2005; Utz et al. 2010). Specific methods from in this study can be used to assess regional
dominance of classes of human land use types and at what levels these of human land uses might
be expected to cause changes in fish assemblages.
Chapter three: management implications.—This study showed that, across large regions,
biodiversity indices for stream fish assemblages are strongly influenced by both natural
landscape factors. Catchment area was a strong predictor of all biodiversity indices across study
regions, and should be controlled for when utilizing biodiversity indices as response metrics in
assessments of stream integrity. Taxonomic distinctiveness, found to represent a unique attribute
of stream fish assemblages compared to diversity and richness, was also predicted by local
elevation, which should similarly be accounted for when using this index for assessment.
This study also showed the importance of human land uses in predicting indices,
underscoring their value in identifying broad changes in stream fish assemblages resulting from
anthropogenic stressors. Taxonomic distinctness was consistently predicted by catchment urban
land use, while catchment agriculture was a better predictor of species richness and Shannon’s
diversity across ecoregions. These findings reflect how unique attributes of fish assemblages are
influenced by human land uses differently, and that those changes are likely due in part to
shifting species membership, and not always loss of species or individuals (Scott and Helfman
2001; Scott 2006). Based on our results, we recommend using multiple biodiversity measures to
capture unique attributes of stream fish assemblages when testing for effects of human stressors.
This has important implications for development of multimetic indices of biotic integrity to
assess streams, and it highlights the importance of identifying species-specific targets when
implementing actions to conserve or protect biodiversity in stream systems.

194

REFERENCES

195

REFERENCES

Barbour M. T., J. Gerritsen, B. D. Snyder, and J. B. Stribling. 1999. Rapid Bioassessment
Protocols for use in streams and wadeable rivers: periphyton, benthic macroinvertebrates
and fish. Second Edition. Rep. EPA/841–B-99–002, US EPA, Off. Water, Washington,
D. C.
Blann K. L., J. L. Anderson, G. R. Sands, and B. Vondracek. 2009. Effects of agricultural
drainage on aquatic ecosystems: a review. Critical Reviews in Environmental Science
and Technology 39:909–1001.
Carpenter, S. R., E. H. Stanley, and M. J. V. Zanden. 2011. State of the world’s freshwater
ecosystems: physical, chemical, and biological changes. The Annual Review of
Environment and Resources 36:75–99.
Carter, K. 2005. The effects of temperature on steelhead trout, coho salmon, and Chinook
salmon biology and function by life stage. Implications for the Klamath River total
maximum daily loads. California Regional Water Quality Control Board. North Coast
Region, Santa Rosa, California.
Clarke K. R., and R. M. Warwick. 1998. A taxonomic distinctness index and its statistical
properties. Journal of Applied Ecology 35:523–531.
Clarke, K. R., and R. M. Warwick. 2001. A further biodiversity index applicable to species lists:
variation in taxonomic distinctness. Marine Ecology Progress Series 216:265–278.
Frimpong, E. A., and P. L. Angermeier. 2009. Fish Traits: a database of ecological and lifehistory traits of freshwater fishes of the United States. Fisheries 34:487–495.
Lyons, J. 1992. Using the index of biotic integrity (IBI) to measure environmental quality in
warmwater streams of Wisconsin. USDA Forest Service. General technical report NC–
149.
McKinney, M. L., and J. L. Lockwood. 1999. Biotic homogenization: A few winners replacing
many losers in the next mass extinction. Trends in Ecology and Evolution 14:450-453.
Meador M. R., and R. M. Goldstein. 2003. Assessing water quality at large geographic scales:
relations among land use, water physicochemistry, riparian condition, and fish
community structure. Environmental Management 31:504–517.
Morgan, R. P., and S. E. Cushman. 2005. Urbanization effects on stream fish assemblages in
Maryland, USA. Journal of the North American Benthological Society 24:643–655.
Pielou, E. C. 1975. Ecological diversity. Wiley and Sons, New York.

196

Rahel, F. J. 2000. Homogenization of fish faunas across the United States. Science 288:854-856.
Scott, M. C. 2006. Winners and losers among stream fishes in relation to land use legacies and
urban development in the southeastern US. Biological Conservation 127:301–309.
Scott, M. C. and G. S. Helfman, 2001. Native invasions, homogenization, and the mismeasure of
integrity of fish assemblages. Fisheries 26:6-15.
Shannon, C. E. 1948. A mathematical theory of communication. Bell System Technical Journal,
27:379–423.
Taniguchi, Y., and S. Nakano. 2000. Condition–specific competition: implications for the
altitudinal distribution of stream fishes. Ecology 81:2027–2039.
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 143:688–699.
Zimmerman, J. K. H., B. Vondracek, and J. Westra. 2003. Agricultural land use effects on
sediment loading and fish assemblages in two Minnesota (USA) watersheds.
Environmental Management 32:93–105.

197