w .
was: .r

.3

 

 

L
V..-

1 g

 

G"\ ‘ ..
{T D )- MICHIGAII‘g'ATr-z {SIS
NIVERSITY
[/21 6 Cl 6 CH EAST LANSING. MICH 48824-1048

This is to certify that the
dissertation entitled

BIOLOGICAL EVALUATION OF
NON-WADEABLE RIVERS
IN MICHIGAN

presented by

KELLY JAMES WESSELL

has been accepted towards fulfillment
of the requirements for the

Ph. degree In ENTOMOLOGY

 

 

@fl/séi”

Major Profé’ssor’ s Signature

16 DECEMBER, 2004
Date

 

MSU is an Affirmative Action/Equal Opportunity Institution

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/ClRC/DateDue.p65-p.15

BIOLOGICAL EVALUATION OF NON-WADEABLE RIVERS IN MICHIGAN

Kelly James Wessell

A DISSERTATION
Submitted to
Michigan State University
in partial fitlfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Entomology

2004

ABSTRACT
BIOLOGICAL EVALUATION OF NON-WADEABLE RIVERS IN MICHIGAN
By
Kelly James Wessell

Compared to smaller, wadeable streams, non-wadeable rivers are relatively understudied.
Currently, protocols exist in most states, including Michigan, to evaluate the ecological
condition of wadeable streams, but none such protocols exist for larger, non-wadeable
rivers. The goal of this research was to establish sampling protocols and develop a
multimetric index of biological integrity for Michigan’s non-wadeable rivers. I sampled
28 unique non-wadeable river reaches in Michigan that encompassed a wide range of
human impacts and ecological conditions. In each reach, I took physical, chemical, and
macro invertebrate samples. I found that sample reaches had unique physical, chemical,
and biological characteristics that allowed the evaluation of ecological health at the reach
scale. Using several techniques to eliminate redundancy among metrics and identify
those biological attributes that accounted for the most among-reach variation in
macroinvertebrate communities, I developed a useful protocol that will allow the rapid
bioassessment of non-wadeable rivers in Michigan. When used together with the Habitat
Index, the NW—IBI will allow the objective evaluation of non-wadeable rivers that may

be applicable to other regions.

This dissertation is dedicated to my grandparents, Helen and James Hines,

Tyrus Wessell, Sr., and Alice Bennett. If only everyone could be so lucky. ...

iii

ACKNOWLEDGEMENTS

I would first like to thank Dr. Richard Merritt, my major professor and
mentor throughout graduate school. Without his support, I couldn’t have enjoyed
this project and my time at Michigan State University nearly as much. Thanks also
go to Drs. Ken Cummins, Jan Stevenson, Mike Kaufman, and Ned Walker, who
served on my doctoral committee and provided invaluable encouragement and
support throughout this study. I am also grateful to Dr. Dave Allan, with whom I
worked very closely on this project, for the time and effort he put into making it
something I can be proud of.

I was also lucky enough to have a great field and lab crew. Todd White, I
who is one of the best naturalists I have ever met, deserves special recognition for
his expertise in macroinvertebrate identification and his tireless efforts in the field.
Rachael Harris and Jeremy Moore also worked very hard for me, especially during
my summer 2002 field season. Jo Wilhelm, the non-wadeable habitat guru, was
great to work with and supplied me with much needed GIS and other habitat data.

My family and friends, especially my wife, Ngoc Kieu, my mom, Linda
Hines Wessell, and my dad, Tyrus Wessell, should be commended for putting up
with me when I was working late hours, and consumed with the stress that goes
along with being a graduate student. They supported and encouraged me
throughout my time in grad school, and I love them very much. Osvaldo
Hernandez, my office mate and great fi'iend, helped me keep my sanity during our

time together at MSU by talking science over beers at the Peanut Barrel.

iv

TABLE OF CONTENTS

LIST OF TABLES ......................................................................................................... vi

LIST OF FIGURES ........................................................................................................ xi

CHAPTER 1: INTRODUCTION AND LITERATURE REVIEW ................................ 1
Introduction ........................................................................................................ 2
Literature Cited ................................................................................................... 14

CHPATER 2: SPATIAL AND TEMPORAL VARIATION IN THE PHYSTICAL

AND CHEMICAL ASPECTS OF NON-WADEABLE RIVERS IN MICHIGAN .......... 18
Introduction ........................................................................................................ 19
Methods and Materials ........................................................................................ 24
Results ................................................................................................................ 26
Discussion .......................................................................................................... 29
Literature Cited ................................................................................................... 36
Tables ................................................................................................................. 39
Flgures .......... 47

CHAPTER 3: BIOLOGICAL EVALUATION OF MICHIGAN ’S NON-

WADEABLE RIVERS USING MACROINVERTEBRATES ....................................... 66
Introduction ........................................................................................................ 67
Methods and Materials ........................................................................................ 70
Results ................................................................................................................ 77
Discussion .......................................................................................................... 86
Conclustion ......................................................................................................... 97
Literature Cited ................................................................................................... 100
Tables ................................................................................................................. 105
Figures ................................................................................................................ 129

APPENDIX 1: COMPOSITE ASSESSMENT METRICS AND STRESSOR-

RESPONSE RELATIONSHIPS ..................................................................................... 149

APPENDIX 2: LARGE WOODY DEBRIS (LWD) ASSESSMENT METRICS

AND STRESSOR-REPSONSE RELATIONSHIPS ....................................................... 152

APPENDIX 3: RECORD OF DEPOSITION OF VOUCHER SPECIMENS .................. 163

LIST OF TABLES

CHAPTER 2

Table 2.1.

Table 2.2.

Table 2.3.

Table 2.4.

Table 2.5.

Table 2.6.

Table 2.7.

Table 2.8.

Table 2.9.

Mean values for physical/chemical parameters by site. Parameters were
recorded with a YSI 6600 multiparameter data sonde. TOT N=Total

Nitrogen (ppm); TOT P=Total Phosporous (ppb); TEMP=Temperature (C);
COND=Conductivity (mS/cm); PH=pH; DO=Dissolved Oxygen (mg/L);
TURB=Turbidity (NTU); CHL=Suspended Chlorophyll (jg/L). See Table

3.1 for site codes. ......................................................................................... 40

Mean values for landuse/habitat values by site. All values fi'om
Wilhelm (2002) except LWD. Landuse values are percentages of urban
(Ur), agricultural (Ag), and natural (Nat) and are either for the entire
watershed (WS) or in a 100m riparian buffer zone at each site (RIP).
LWD=the number of transects with large woody debris to sample. See

Table 3.1 for site codes. ............................................................................... 41
Basic statistics of sonde parameters from all sites. TEMP: Temperature

(C); COND: Conductivity (mS/cm); DO: Dissolved oxygen (mg/L);

TURB: Turbidity (NTU); CHL: Suspended chlorophyll (ug/L). ................... 42
ANOVA results for physical parameters at all sites. .................................... 42
AN OVA results for nutrient concentrations for all 2001 sites. ...................... 42
ANOVA results for chlorophyll concentrations for all 2001 sites. ................ 43
Basin effects for sonde parameters. 2001 sites only. This table shows the
results of AN OVA tests for differences among sites fi'om the same river
catchment................ ..................................................................................... 43
Basin effects for nutrient concentrations. 2001 sites only. This table

shows the results of AN OVA tests for differences among sites from the

same river catchment .................................................................................... 44
Basin effects for sonde parameters. 2002 sites only. This table shows the

results of AN OVA tests for differences among sites from the same river
catchment ..................................................................................................... 45

Table 2.10. Mean differences and associated p-values from individual T-tests on

physical parameters between 2001 and 2002 field seasons. A negative
difference indicates values decreased. A positive difference indicates
values increased. N/A: Data not available. See Table 3.1 for site codes. 46

vi

CHAPTER 3

Table 3.1.

Table 3.2.

Table 3.3.

Table 3.4.

Table 3.5.

Table 3.6.

Table 3.7.

Table 3.8.

Table 3.9.

List of sites sampled for macroinvertebrates ............................................... 106

Catchment disturbance gradient (CDG) scores for each sample site. The
seven individual measures were scored 0-4 with zero indicative of a natural
system and 4 suggestive of a highly disturbed system. The scores for each
metric were summed to give a total CDG score. See Table 3.1 for site

codes. (Modified fi'om Wilhelm 2002). ...................................................... 107

Total riparian disturbance gradient (RDG) score for each sample site based

on the number of gaps in the riparian area and the mean riparian width.

The two metrics were scored on a scale fiom 0-4 and were summed to

yield a total score. A low number indicates a natural site, while a high
number indicates a highly disturbed site. See Table 3.1 for site codes
(Modified from Wilhelm 2002). ................................................................. 108

List of macroinvertebrates found in Michigan non-wadeable rivers. Note

that fimctional group assignments are at the family level and may vary
within each family. ..................................................................................... 109

Potential biological attributes to be used as metrics in the final protocol
Potential metrics are categorized based on whether they are population,
community, or functional attributes. Taxonomic resolution is to the family
level. Asterisks indicate metrics retained for the composite assessment but
not for the LWD assessment (EPT_RICH and P_RICH). Italics indicate
metrics retained for the LWD assessment but not for the composite (SCR). .....
.................................................................................................................. 1 12

Pearson correlation matrix of population level attributes with family-level
data from composite samples. Bold numbers indicate highly correlated
metrics (>0.70). Codes are listed in Table 3.5. ......................................... 113

Pearson correlation matrix of population level attributes from LWD

samples. Bold numbers indicate highly correlated metrics (>0.70).
Taxonomic resolution is to the family level. Codes are listed in Table 3.5.
.................................................................................................................. 113

Pearson correlation matrix of community level attributes with family-level
data from composite samples. Bold numbers indicate highly correlated
metrics (>0.70). Codes are listed in Table 3.5. ........................................... 114

Pearson correlation matrix of community level attributes from LWD

samples. Bold numbers indicate highly correlated metrics (>O.70).
Taxonomic resolution is to the family level. Codes are listed in Table 3.5.
.................................................................................................................. 115

vii

Table 3.10. Pearson correlation matrix for functional attributes with family—level data
fi'om composite samples. Bold numbers indicate highly correlated metrics
(>0.70). Codes are listed in Table 3.5. ....................................................... 116

Table 3.11. Pearson correlation matrix for functional attributes from LWD samples.
Bold numbers indicate highly correlated metrics (>0.70). Taxonomic
resolution is to the family level. Codes are listed in Table 3.5 .................... 117

Table 3.12. Results of principal component analysis for both composite and large
woody debris (LWD) samples. Bold values indicate axes that were firrther
examined for metric selection. .................................................................... 118

Table 3.13. Principal component loadings for composite samples. In general, those
metrics with the highest loadings per axis were retained for the final
protocol. When two or more metrics had similarly high loadings, two were
chosen for the final protocol. Bold numbers indicate which metrics were
retained from each axis. Some metrics with high loadings were not
retained due to inaccuracy (e.g., PER_OLIG). See Table 3.5 for metric
codes .......................................................................................................... 119

Table 3.14. Principal component loadings for LWD samples. In general, those
metrics with the highest loadings per axis were retained for the final
protocol. When two or more metrics had similarly high loadings, 2-3 were
chosen for the final protocol. Bold numbers indicate which metrics were
retained fi'om each axis. Some metrics with high loadings were not
retained due to ambiguous response to hunmn influences (e.g., LNSH).
See Table 3.5 for metric codes. .................................................................. 120

Table 3.15. Metrics retained and their corresponding weights in the final protocol
(composite samples only). Weights were determined by summing
eigenvalues from retained axes and calculating the total variance explained
by each axis. Scores were then approximated based on a lOO-point scale.
When 2 or more metrics were retained from each axis, scores were divided
equally based on the variance calculated for the entire axis. See Table 3.5
for metric codes. ........................................................................................ 121

Table 3.16. Metrics retained and their corresponding weights in the final protocol
(LWD samples only). Weights were determined by summing eigenvalues
from retained axes and calculating the total variance explained by each
axis. Scores were then approximated based on a 100 point scale. When 2
or more metrics were retained from each axis, scores were divided equally
based on the variance calculated for the entire axis. See Table 3.5 for
metric codes. .............................................................................................. 121

viii

Table 3.17. Scoring criteria for biological monitoring of non-wadeable rivers using
composite samples with all habitats. ........................................................... 122

Table 3.18. Scoring criteria for biological monitoring of non—wadeable rivers using
large woody-debris samples only. .............................................................. 123

Table 3.19. Results fiom multiple linear regression of composite metrics with
environmental variables. Forward stepwise regression (p—value to
enter=0.15; tolerance=0.001) and backward stepwise regression (p—value to
remove=0. 15; tolerance 0.001) sometimes selected different environmental
variables to which metrics respond. Only the first 4 variables to enter
(forward) or be removed (backward) are shown here. 1General response to
riparian landuse. 2General response to watershed landuse. *Denotes
variables with p<0.05. "Denotes variables with p<0.001. In all cases,
p<0. 10. See Table 3.5 for metric codes and Tables 19-20 for
environmental variable codes. .................................................................... 124

Table 3.20. Results fi'om multiple linear regression of LWD metrics with
environmental variables. Forward stepwise regression (p-value to .
enter=0.15; tolerance=0.001) and backward stepwise regression (p-value to
remove=0. 1 5; tolerance 0.001) sometimes selected different environmental
variables to which metrics respond. Only the first 4 variables to enter
(forward) or be removed (backward) are shown here. 1General response to
riparian landuse. *Denotes variables with p<0.05. ”Denotes variables
with p<0.001. In all cases, p<0. 10. See Table 3.5 for metric codes and
Tables 19-20 for environmental variable codes ........................................... 125

Table 3.21. Results from the discriminant fimction analysis. Two analyses were
done: One in which approximately one-half of the sites fi'om each
classification type were used to generate the model, while the other half
were used to test the model. The other analysis used jackknifing to
evaluate the model. .................................................................................... 126

Table 3.22. Site rankings based on CDG, RDG, and an overall ranking based on CDG
and RDG rankings combined with the number of transects with LWD
(Mean Rank). Hl=Non-wadeable Habitat Index (Wilhelm 2002). Site
scores for composite and LWD assessments are also listed. These data
were used to evaluate the NW-IBI sensitivity (composite vs. LWD) to
differing scales of human impacts. ............................................................. 127

Table 3.23. Regression data for composite assessments. In all cases, the composite
NW-IBI was the dependent variable. . Independent variables reflect
differing scales of human influence. CDG=Catchment Disturbance
Gradient; RDG=Riparian Disturbance Gradient; Mean Rank is based on
catchment-wide, riparian, and in-stream habitat quality. HI=Non-wadeable
Habitat Index (Wilhelm 2002). ................................................................... 128

Table 3.24. Regression data for LWD assessments. In all cases, the LWD NW-IBI
was the dependent variable. Independent variables reflect differing scales
of human influence. CDG=Catchment Disturbance Gradient;
RDG=Riparian Disturbance Gradient; Mean Rank is based on catchment-
wide, riparian, and in-stream habitat quality. HI=Non-wadeable Habitat
Index (Wilhelm 2002) ................................................................................ 128

LIST OF FIGURES

CHAPTER 2
Figure 2.1. Actual values from chl a extraction vs. sonde values for (a)
phytoplankton and 0)) periphyton samples. ............................................... 48
Figure 2.2. Mean (a) conductivity and (b) pH for all 2001 sites. Error bars indicate
standard error. See Table 3.1 for site codes ............................................... 49
Figure 2.3. Mean (a) dissolved oxygen (DO) and (b) turbidity for all 2001 sites.
Error bars indicate standard error. See Table 3.1 for site codes. ................ 50
Figure 2.4. Mean chlorophyll values for all 2001 sites. Error bars indicate standard
error. See Table 3.1 for site codes ............................................................. 51
Figure 2.5. Mean (a) total N and (b) nitrate for all 2001 sites. Error bars indicate
standard error. See Table 3.1 for site codes ...................................... . ......... 52
Figure 2.6. Mean ammonia for all 2001 sites. Error bars indicate standard error. See
Table 3.1 for site codes .............................................................................. 53
Figure 2.7. Mean (a) total P and (b) soluble reactive P (SRP) for all 2001 sites. Error
bars indicate standard error. See Table 3.1 for site codes. ......................... 54
Figure 2.8. Mean (3) phytoplankton and (b) periphyton for all 2001 sites. Data are

fi'om sonde measurements. Error bars indicate standard error. See Table
3.1 for site codes. ...................................................................................... 55

Figure 2.9. Mean (a) conductivity and (b) pH for all 2002 sites. Error bars indicate

Figure 2.10.

Figure 2.11.

Figure 2.12.

standard error. See Table 3.1 for site codes ............................................... 56

Mean (a) dissolved oxygen and (b) turbidity for all 2002 sites. Error bars
indicate standard error. See Table 3.1 for site codes .................................. 57

Mean chlorophyll values for all 2002 sites. Error bars indicate standard
error. No chlorophyll data exist for the AuSable or the Manistee sites.
See Table 3.1 for site codes. ...................................................................... 58

Comparison of 2001 and 2002 physical parameters from the Grand

River @ Grand Rapids (gd_gr). All parameters were measured with a

YSI 6600 multiparameter data sonde. Y-axis shows mean values +/- SE. .....
................................................................................................................. 59

Figure 2.13.

Figure 2.14.

Figure 2.15.

Figure 2.16.

Figure 2.17.

Figure 2.18.

Comparison of 2001 and 2002 physical parameters from the Grand
River @ Ionia (gd_ion). All parameters were measured with a YSI 6600
multiparameter data sonde. Y-axis shows mean values +/— SE. ................. 60

Comparison of 2001 and 2002 physical parameters from the Grand River
@ Johnson Pk (gd_jon). All parameters were measured with a YSI 6600
multiparameter data sonde. Y-axis shows mean values +/- SE. ................. 61

Comparison of 2001 and 2002 physical parameters fi'om the Grand

River @ Comstock Riverside (gd_cmr). All parameters were measured
with a YSI 6600 multiparamter data sonde. Y-axis shows mean values

+/- SE ........................................................................................................ 62

Comparison of 2001 and 2002 physical parameters from the AuSable
River @ Whirlpool (as_whp). All parameters were measured with a YSI
6600 multiparameter data sonde. Y-axis shows mean values +/- SE. ........ 63

Comparison of 2001 and 2002 physical parameters from the Manistee @
High Bridge (ma_hbr). All parameters were measured with a YSI 6600
multiparameter data sonde. Y-axis shows mean values +/- SE. ................. 64

Box and whisker plot of diel oxygen change for selected 2001 sites.
Data are from a YSI 600 multiparameter data logging sonde. .................. 65

CHAPTER 3

Figure 3.1.

Figure 3.2.

Figure 3.3.

Figure 3.4.

Diagram of a non—wadeable study reach I chose a standard 2000m
reach, and sampled macroinvertebrates at transects placed every 200m. .. 130

Mean taxa richness and Shannon diversity (H’) from selected 2001 study
sites. LWD (a and c) and F POM (b and d) samples only. Note that

despite coming from the same habitat, richness and diversity still varied
considerably in LWD samples, while variability was less pronounced in
FPOM samples. Numbers on x-axis indicate different sites. ................... 131

Scree plot of eigenvalues (1») for each PCA axis. Axes 1-5 were further
examined for composite metric selection. Axes 1-4 were further
examined for LWD metric selection. Criteria: 74>]. ................................ 132

PCA site scores (axes 1 vs. 2) for the composite data. Site scores are

plotted with rivers identified (see Table 3.1 for river codes). This plot
shows that sites did not cluster by river or catchment. ............................. 133

xii

Figure 3.5.

Figure 3.6.

Figure 3.7.

Figure 3.8.

Figure 3.9.

Figure 3.10.

Figure 3.11.

Figure 3.12.

Figure 3.13.

Figure 3.14.

Figure 3.15.

Figure 3.16.

Figure 3.17.

Figure 3.18.

PCA site scores identified by ecoregion. (a) Axis I vs. Axis 2; (b) Axis

2 vs. Axis 3. Sites clustered based on ecoregion along axis 1, but not

along the other axes. SLP: Southem Lower Peninsula; NLP: Northern
Lower Peninsula; UP: Upper Peninsula. ................................................. 134

(a) Mean RDG and (b) mean CDG by ecoregion (:t SE) (See Wilhelm
2002 for information regarding the development of the RDG and CDG). ......
............................................................................................................... 135

Theoretical example of how a 25-point and a 5-point metric were scored
based on inter-quartile ranges .................................................................. 136

NW—IBI scores for all sites comparing composite and LWD assessments.
LWD assessments appeared to be more sensitive than composite
assessments. ............................................................................................ 136

Composite NW-IBI scores for each non-wadeable study site. Individual

metric scores are shown in each bar. See Table 3.1 for site codes. .......... 137
LWD NW—IBI scores for each non-wadeable study site. Individual '
metric scores are shown in each bar. See Table 3.1 for site codes. .......... 138
Composite vs. LWD NW-IBI scores for non-wadeable river study sites. .......
............................................................................................................... 1 39
Mean Site Ranking vs. NW—IBI for composite assessments ..................... 140
Mean Site Ranking vs. NW-IBI for LWD assessments ............................ 141
Saginaw River @ Zilwaukee (sg_zi101) riparian view. This site scored
lowest in the composite NW—IBI, and was classified as “poor” by both
types of assessment. ................................................................................ 142
Grand River @ Ionia (gr_ion01, gr_ion02). This site was sampled in
both 2001 and 2002, receiving a score of “fair” each time (composite
assessment). ............................................................................................ 143
Manistee River @ High Bridge (ma_hbr). This site was scored as
“good” in both 2001 and 2002 (composite assessment). Inset: Note clear
water and slightly embedded coarse substrate. ......................................... 144
Manistee River @ Coates Rd (ma_cts02). This site scored “excellent” in
both assessment types .............................................................................. 145
Mean number of new taxa collected in successive LWD samples. Note

that after 8 samples, there were never new taxa collected, indicating tlmt

xiii

8 samples should be sufficient for LWD assessments. These data were
tabulated fi'om 6 randomly chosen sites. .................................................. 146

Figure 3.19. Flowchart illustrating the steps involved in the biological assessment of
non-wadeable rivers in Michigan. ............................................................ 147

xiv

CHAPTER 1

INTRODUCTION AND LITERATURE REVIEW

CHAPTER 1
INTRODUCTION AND LITERATURE REVIEW

INTRODUCTION

Large river ecosystems have long been associated with the development of human
civilization. They provide water for irrigation of agricultural land, help in replenishing
nutrients on their floodplains, and carry away waste generated by inhabitants of their
watersheds. Humans have historically settled in areas within close proximity to rivers for
the above reasons, and have invariably modified and impacted these systems on which
they rely so heavily. As the ultimate sink for all upstream and watershed-wide processes,
large rivers have been subjected to many different types of impacts fi'om their human
inhabitants including: 1) nutrient enrichment; 2) other non-point source pollution like
pesticide inputs and sedimentation fi'om intensive agriculture; 3) channel modification for
navigation and drainage purposes; and 4) industrial pollution from point-source discharge
of waste. Since the Clean Water Act, state and federal agencies in the United States have
established various procedures to protect and evaluate lotic ecosystem integrity. With the
application of scientific theory to set expectations for river health and monitoring of
current conditions and trends, our riverine systems are slowly improving, but this

progress has been extremely biased toward smaller, wadeable streams.

LARGE RIVER ECOLOGY
Much of the science of stream ecology has been developed on small streams.
This is due to several factors: Smaller streams are easier environments to sample,

because there is no need for a boat or access sites. Also, many stream ecologists focus on

working in pristine streams, free of anthropogenic influences. One is hard-pressed to find
a large river that is not affected by humans, whether these effects are in terms of
watershed land use, irnpoundments, or channel modification. Despite large rivers being
relatively understudied compared to small streams, there have been several important
theories developed in the last 50 years that have advanced our understanding of how large

rivers are structured and how they function.

River Continuum Concept

In their classic paper, Vannote et al. (1980) synthesized what was known about
the structure and firnction of lotic ecosystems by describing an orderly shift in energy
sources and consumer groups along a predictable gradient of physical factors from
headwaters to mouth. Specifically, the river continuum concept (RCC) assumes that
energy for biological production comes fi'om three sources: local inputs of organic matter
form riparian sources (allochthonous inputs), primary production from within the stream
(autochthonous inputs), and transport of organic matter fi'om upstream. The relative
importance of each energy source varies along the river continuum, and is predicable
based on a dynamic equilibrium with the physical environment (Johnson et a1.
l995a;Vannote et al. 1980).

Small, low order streams have a relatively constant environment due to
continuous groundwater inputs and small watershed area. Within a forest, the canopy
reduces the amount of light the stream receives, thus limiting photosynthesis. Because of
this, the RCC predicts that the dominant energy source for low order (orders 1-3) streams

will be coarse particulate organic matter (CPOM) derived from terrestrial leaf litter that

falls into the stream. As a result, the primary macroinvertebrate groups in these small
streams are shredders who feed on the CPOM and collectors who filter out fine particles
from the water column.

The physical processes of mid-sized rivers (order 4-6) are more variable,
exhibiting the largest range of temperatures and hydraulic conditions. The forest canopy
Opens, allowing light available for autochthonous production, but reducing leaf litter
inputs into the stream. This shift in energy source results in a shift in the dominant
invertebrate firnctional feeding groups present. Here, scrapers and collectors will
dominate, and because of the diversity of physical forces (temperature especially), mid-
sized rivers are predicted to have the highest biological diversity.

In larger rivers (orders 6 and above), temperature and hydraulic changes are
buffered by the large volume of water. Leaf litter inputs are minor due to large channel
width, but primary production is limited by turbidity. As a result, the main energetic
inputs to these rivers are from upstream processing of CPOM (fragments and feces), and
collectors are the dominant macroinvertebrate fimctional feeding group.

A corollary to the RCC is the serial discontinuity concept (Ward and Stanford
1983), which addresses the effects of dams on rivers. According to this concept, dams
cause a longitudinal discontinuity of physical and biological features. This should shift
the continuum either up or down the stream order axis depending on the dam’s location.
For example, a dam placed on a mid-sized river should stabilize temperatures and flows
downstream, reducing the biological diversity that was originally maintained by the

physical diversity. A dam on a large river should reduce turbidity downstream Item the

darn, resulting in more autochthonous production, and causing the river to function more
like a mid-order river (Ward and Stanford 1983).

The river continuum concept was developed for North American forested stream
ecosystems. Because it does not always accurately describe other types of lotic systems,
the RCC has come under criticism since its inception. New Zealand workers first pointed
out that in areas where there are no significant riparian inputs, such as desert, deep
canyon, or prairie streams, the RCC does not adequately describe the structure and
function of these systems. Winterbourn et a1. (Winterbourn et a1. 1981) showed that, in
contrast with the RCC, stream invertebrates in New Zealand systems show little
longitudinal shift in fimctional group dominance, and attribute this to the
geomorphological and landscape-level differences in New Zealand streams compared to
North American streams. The lack of retention of CPOM in the typically high gradient
streams of New Zealand, along with the climatically unpredictable nature of these
streams has resulted in a lack of shredders along with a macroinvertebrate fauna that is
functionally flexible with asynchronous life histories (Winterbourn et a1. 1981).
However, in a later paper, Cummins (1988) noted that this was an exception, which
reflected various degrees of alteration from the aboriginal condition. Minshall et al
(1983) found that watershed climate and geology, riparian conditions, tributaries, and
other location-specific factors can also cause a river to deviate fi'om its predicted position
along the river continuum based on what might be initially expected based on stream
order alone. In many situations, the RCC holds, and in situations where conditions
deviate fi'om what is predicted, the RCC still serves as a useful paradigm for

understanding lotic ecosystems (Cummins 1988; Minshall et al. 1985).

Perhaps the most serious questioning of the RCC has come fiom work on large
floodplain rivers. The RCC predicts that much of the energy fueling large river
production will come from upstream processing of CPOM. Some authors maintain that
the river continuum concept underestimates the importance of the floodplain in providing
energy to the system (Sedell et a1. 1989). The RCC was tested on several large rivers:
The Moisie and Salmon Rivers are ninth-order rivers with constrained channels and no
floodplain. Both rivers exhibited carbon flow characteristics that closely adhered to the
RCC (Sedell et a1. 1989). However, the Amazon and Parana-Plata Rivers are large,
tropical rivers with extensive floodplains. These rivers showed differences in carbon
processing depending on whether or not areas had close associations with the floodplain.
River areas with the most interactions with their floodplains were the most productive,
indicating a substantial energetic input from floodplain areas (Sedell et al. 1989). These
data show that the RCC holds for large rivers confined to their banks, but that river-

floodplain interactions can seriously disrupt predictions of the river continuum concept.

Flood Pulse Concept

The flood pulse concept introduces a lateral dimension to the theory of lotic
ecosystems. The flood pulse concept applies to large, floodplain rivers in temperate and
tropical areas, and states that the most important hydrologic feature of large rivers is the
predicable flooding of the river over its banks (Junk et a1. 1989). Floodplains are highly
productive, typically contain a wide variety of aquatic habitats, and are periodically and

predictably inundated.

During a flood, aquatic organisms migrate onto the floodplain to use the available
recourses. As floodwaters recede, nutrients and organic matter are funneled back into the
channel, and this replenishes the resources depleted from the system since the last flood.
Because the flooding is often predictable, biological communities show adaptations to
using the floodplain resources (Johnson et a1. l995a;Junk et al. 1989). For example, in
rivers that regularly flood, fish species, via environmental cues such as temperature or
day length, anticipate the floods and spawn before or during the rise of water levels
(Bayley 1995). As a result of this close tie between flooding and biological production,
the flood pulse concept also predicts that hydrologic alteration, such as impoundments
and channelization, along with watershed land use changes, can deleteriously affect
biological communities (Bayley 1995;Johnson et a1. l995b). This is well illustrated by
the channelization and irnpoundment of the Kissimmee River in Florida, along with the
subsequent drainage and conversion of floodplain habitat for intensive agriculture (Toth
1990).

While both longitudinal (RCC) and lateral (flood pulse) attributes of lotic
ecosystems are important in determining their structure and function, it is really a
hierarchical combination of large scale and small-scale processes that define the structure
and fitnction of large river systems. Large-scale processes such as plate tectonics
(influencing underlying geological features) and climate (affecting rainfall and flooding)
tend to influence river morphology and species pools. Within these higher levels of
organization, smaller-scale processes such as species interactions (competition and

predation) and flow characteristics (influencing substrate particle size) operate to give

each river its characteristic population-level, community-level, and ecosystem-level

structure (Johnson et al. 1995a).

ANTHROPOGENIC IMPACTS ON LARGE RIVER ECOSYSTEMS

There are few pristine large river ecosystems remaining in the world.
Agricultural and urban development along river corridors have had serious impacts to
large river systems related to clearing the riparian area, and the subsequent loss of
retention of nutrients, sediments, and toxins, along with a reduction in the overall
terrestrial inputs of CPOM and large woody debris (LWD). Hydrologic alterations, in the
form of impoundments and channelization can reduce interaction between the river and
its floodplain (Stanford 1996). The following is a brief discussion of the impacts humans

have had on large river systems.

Watershed Land Use

The conversion of land for intensive agriculture or urban development has had
both direct and indirect effects on lotic ecosystem function (Allan and F lecker 1993).
Forested watersheds generally act as filters for the stream channels, buffering sediment
and nutrient loads, and capturing toxins before they enter the river (Karr 1991). Riparian
cover also buffers temperature changes, resulting in lower maximal values in the summer
and higher minimal values in the winter (Allan and F lecker 1993). When a forested
watershed is converted for agriculture, this buffering capacity is lost. Nonpoint source
pollutants such as suspended sediments (Wolrnan 1971) and nutrients (Smith et al. 1987)

generally increase, and the addition of pesticides and herbicides to the watershed often

have adverse effects on stream metabolism (Young and Huryn 1999). Urban
development, specifically the addition of roads and other non-porous substrates to the
watershed, increase the overland flow of water into the river, also resulting in increased
sedimentation (Kart 1991) and fossil firel runoff. Another way in which urban
development has affected large river systems is through atmospheric deposition of toxic

substances such as PCBs and metals (Smith et al. 1987).

Hydrologic Alteration

Hydro logic alteration of rivers has also had serious impacts on large river
systems. As discussed above, the serial discontinuity hypothesis (Ward and Stanford
1983) addresses the effects of dams on the river continuum. Dams can also affect the
flood pulse (Johnson et al. 1995a). By controlling the release of impounded water, water
level variation below the dam can be substantially reduced. This has overall impacts on
water temperature and dissolved oxygen (depending on whether the dam is “surface
release” or “deep release”), as well as sediment and nutrient retention, which can affect
most of the river’s important ecological processes (Ligon et al. 1995).

Channel alteration, including complete channelization has been shown to have
deleterious effects on river structure and function. In the late 1960’s, the US. Army
Corps of Engineers began channelization of the Kissimmee River in Florida for the
purpose of flood control and floodplain development. Since channelization, there has
been a 90% decrease in wading bird and waterfowl populations (Weller 1995), along with
a decline in the once outstanding largemouth bass fishery (Merritt et al. 1996). The loss

of littoral fringe habitat in the main channel reduced the available habitat for

macroinvertebrates and fishes, and the reduction of flow through the remnant channel
resulted in a shift in the invertebrate community to one more characteristic of lentic

systems (Merritt et a1. 1996; Merritt et al. 1999).

ASSESSING IMPACTS ON LARGE RIVER ECOSYSTEMS

In order to address the threats to lotic ecosystem integrity discussed above, there
must be an objective means of monitoring changes in river health and quantifying effects
of anthropogenic effects. Using living organisms to evaluate water quality has its roots in
the concept of the saprobian system (Cairns and Pratt 1993). This idea builds on the
concept that certain organisms, because of their differing tolerances toward organic
enrichment, could be used as indicators of ecosystem stress. This concept of indicator
organisms is still used today (Hilsenhoff 1987; Hilsenhoff 1988). However, as more was
learned about stream ecology and aquatic insects, problems with the saprobian indicator
concept became apparent. Most aquatic insects have restricted distributions or seasonal
fluctuations, which preclude their use as indicators except in specified areas where they
occur and during the season in which they are most likely to be captured as part of any
sampling protocol. These problems are compounded with the diversity of hydro logic and
habitat characteristics of any given stream or river. Additionally, the lack of solid
autecological information on most species makes their use as indicators tenuous at best,
and dangerous at worst (Cairns and Pratt 1993).

Because of these problems, community measures of ecosystem integrity were
introduced within the field of bioassessment. These provide information on community

structure that goes beyond simple indicator species. Commonly, these types of indices

10

are manifested in the form of the diversity index. The Shannon Diversity Index is
perhaps the most widely used of these. Generally, a diversity index incorporates some
measure of total taxa richness combined with the relative representation of each taxon.
James Karr and others (Fore et al. 1996; Karr 1987; Karr 1999; Karr and Chu
1999; Reynoldson et al. 1997; Thorne and Williams 1997) developed the idea of the
multimetric index, which incorporates population measures (such as indicator groups)
along with community measures (such as diversity). Each metric is scored based on
benthic invertebrate samples from the area to be assessed, and individual metric scores
are combined to give an overall assessment of the stream, river, or lake that is being
evaluated. Currently, such a scoring system is used by the Michigan Department of
Environmental Quality (MDEQ) for wadeable stream biomonitoring (MDNR 1991).
Multivariate bioassessment protocols, which evaluate ecosystem health based on
an expected biota are gaining wider acceptance in the field of biomonitoring (Hawkins et
al. 2000;Moss et al. 1999;Wright 1995). The observed community (from the samples) is
then compared to the expected community (derived fiom reference areas), and
assessment is based on the difference. The main problem with multivariate methods is
the taxonomic resolution required for such studies (Cao et al. 1996). While much of the
state and federal programs in the US. require only family-level identification of
macroinvertebrates, multivariate methods often require species-level identification. As
mentioned above, multivariate assessments require the use of reference sites to set initial
expectations for benthic communities. This is difficult with non-wadeable rivers because

of the size of the watershed and the historical usage of these rivers for irrigation, logging,

11

and transportation—there simply are no large rivers in Michigan that have not at one time
been impacted by human use.

Another approach to monitoring lotic systems has been proposed by Merritt et al.
(Merritt et a1. 1996;Merritt et al. 1999), which builds on the concepts of
macroinvertebrate fimctional groups (Cummins and Klug 1979) and the RCC (V annote et
al. 1980). This approach uses macroinvertebrate fimctional group ratios to assess
ecosystem function. The premise is that aquatic invertebrate communities predictably
respond to changes in their physical environment, and these changes are reflected in the
community-wide functional group representation. This type of assessment involves
constructing macroinvertebrate firnctional group ratios that serve as analogues to .
ecosystem parameters. For example, the ratio of autochthonous to allochthonous inputs,
or the ratio of photosynthetic production to respiration (P/R) can be approximated by the
ratio of live vascular plant shredders + scrapers as a proportion of CPOM detritivorous
shredders + total collectors (Merritt et al. 1996). Habit and voltinism groups can also be
used as ecosystem analogues. For instance, to obtain a measure of habitat stability, one
can use the ratio of clingers + climbers as a proportion of burrowers + sprawlers +
swimmers (habit groups) or species with > 1 generation per year as a proportion of
species with 51 generations per year (voltinism groups) (Merritt et al. 1996). The use of
functional group analogues to evaluate ecosystem function and stability has been
successful in the Kissimmee (Merritt et a1. 1996;Merritt et al. 1999) and Caloosahatchee
(Merritt et al. 2002) Rivers in Florida.

In general, biological assessment protocols commonly use a variety of

assemblages to infer ecological condition. Fish (Jennings et al. 1995) and

12

macroinvertebrates (citation) are generally the most common assemblages used, but
benthic algae communities have recently received attention in the literature and are
especially useful due to the extensive autecological knowledge of algae. Because each
has its advantages and disadvantages relating to temporal and spatial responses to
anthropogenic stressors, many protocols use multiple assemblages in ecological
assessments in addition to physical and chemical parameters important to the system of
interest.

Currently, there are many state and federal biomonitoring programs that use
macroinvertebrates to evaluate the integrity of streams. Most of these programs have
been developed solely for smaller, wadeable streams. Because of the inherent differences
between wadeable and non-wadeable river structure and fimction, these methods are not

suitable for non-wadeable river assessment in Michigan.

13

LITERATURE CITED

Allan,J.D. and A.S.Flecker, 1993. Biodiversity conservation in running waters.
Bioscience 43: 32-43.

Bayley,P.B., 1995. Understanding large river floodplain ecosystems. Bioscience 45: 153-
158.

Cairns,J.J. and J.RPratt, 1993. A history of biological monitoring using benthic
macroinvertebrates. In: Rosenberg,D.M. and V.H.Resh (eds), Freshwater
Biomonitoring and Benthic Macro invertebrates, Chapman and Hall, New York,
pp. 10-27.

Cao,Y., A.W.Bark, and W.P.Williams, 1996. Measuring the responses of
macroinvertebrate communities to water pollution: A comparison of multivariate
approaches, biotic and diversity indices. Hydrobiologia 341: 1-19.

Cummins, K.W., 1988. The Study of Stream Ecosystems: A Functional View. In:
Pomeroy,L.R and J .J .Alberts (eds), Concepts of Ecosystem Ecology: A
Comparative View, Springer-Verlag, New York, pp. 247-261.

Cummins,K. W. and M.J.Klug, 1979. Feeding ecology of stream invertebrates. Ann. Rev.
Ecol. Syst. 10: 147-172.

Fore,L.S., J.R.Karr, and R.W.Wisseman, 1996. Assessing invertebrate responses to
human activities: Evaluating alternative approaches. Journal of the North
American Benthological Society 15: 212-231.

Hawkins,C.P., R.H.Norris, J.N.Hogue, and J.W.Feminella, 2000. Development and
evaluation of predictive models for measuring the biological integrity of streams.
Ecological Applications 10: 1456-1477.

Hilsenhoff,W.L., 1987. An improved biotic index of organic stream pollution. Great
Lakes Entomologist 20: 31-40.

HilsenhoflLW.L., 1988. Rapid field assessment of organic pollution with a family-level
biotic index. Journal of the North American Benthological Society 7: 65-68.

14

Jennings, M. J., Fore, L. S., and Karr, J. R. Biological monitoring of fish assemblages in
Tennessee Valley reservoirs. 1995. La Crosse, WI (USA). Conference on
Sustaining the Ecological Integrity of Large Floodplain Rivers: Application of
Ecological Knowledge to River Management. 1994.

Johnson,B.L., W.B.Richardson, and T.J.Naimo, 1995a. Past, Present, and Future
Concepts in Large River Ecology. BioScience 45:134-141.

Johnson,B. L, W. B. Richardson, and T. J. Naimo, 1995b. Past, Present, and Future
Concepts 1n Large River Ecology: How rivers fitnction and how human activities
influence river processes. Bioscience 45:134-141.

Junk, W. J., Bayley, P. B., and Sparks, R. E. The flood pulse concept in river-floodplain
ecosystems. Dodge, D. P. 106, 110-127. 1989. Proceedings of the International
Large River Symposium.

Karr,J.R, 1987. Biological monitoring and environmental assessment: A conceptual
fiamework. Environmental Management 11:249-256.

Karr,J.R., 1991. Biological integrity: A long-neglected aspect of water resource
management. Ecological Applications 1:66-84.

Karr,J.R., 1999. Defining and measuring river health. Freshwater Biology 41: 221-234.

Karr,J.R. and E.W.Chu, 1999. Restoring Life in Running Waters, Island Press,
Washington, DC.

Ligon,F.K., W.B.Dietrich, and W.J.Trush, 1995. Downstream ecological effects of dams.
Bioscience 45: 183-192.

[MDNR] Michigan Department of Natural Resources. Qualitative Biological and Habitat
Survey Protocols for Wadable Streams and Rivers: Great Lakes and
Environmental Assessment Section (GLEAS) Procedure 51. 1991. Lansing, MI,
Michigan Department of Natural Resources.

MerritLR.W., M.J.Higgins, K.W.Cummins, and B.Vandeneeden, 1999. Seasonal
Differences in Invertebrate Functional Feeding Group Relationships of the
Kissirnmee River-floodplain Ecosystem, Florida. In: D.Batzer, R.B.Rader, and

15

S.AWissinger (eds), Invertebrates in Freshwater Wetlands of North America,
John Wiley & Sons, Inc., pp. 55-79.

Merritt, RW; KW Cummins; MB Berg; JA Novak; MJ Higgins; KJ Wessell; and JL
Lessard. 2002. Development and application of a macro invertebrate firnctional-

group approach in the bioassessment of remnant river oxbows in southwest
Florida. Journal of the North American Benthological Society 21(2):290-310.

Merritt,R.W., J.R.Wallace, M.J.Higgins, M.KAlexander, M.B.Berg, W.T.Morgan,
KW.Cummins, and B.Vandeneeden, 1996. Procedures for the firnctional analysis

of invertebrate communities of the Kissimmee River-floodplain ecosystem.
Florida Scientist 59: 216-274.

MinshalLG.W., K.W.Cummins, R.C.Petersen, C.E.Cushing, D.A.Bruns, J.RSedell, and
R.L.Vannote, 1985. Developments in stream ecosystem theory. Canadian Journal
of Fisheries and Aquatic Sciences 42: 1045-1055.

MinshalLG.W., RC.Petersen, KW.Cummins, T.L.Bott, J.R.Sedell, C.E.Cushing, and
RL.Vannote, 1983. Interbiome comparison of stream ecosystem dynamics.
Ecological Monographs 53: 1-25.

Moss,D., J.F.Wright, M.T.Furse, and R.T.Clarke, 1999. A comparison of alternative
techniques for prediction of the fauna of running-water sites in Great Britain.
Freshwater biology. Oxford 41 :167-181.

Reynoldson,T.B., R.H.Norris, V.H.Resh, K.E.Day, and D.M.Rosenberg, 1997. The
reference condition: A comparison of multimetric and multivariate approaches to
assess water-quality impairment using benthic macroinvertebrates. Journal of the
North American Benthological Society 16:833-852.

Sedell, J. R., Richey, J. E., and Swanson, F. J. The river continuum concept: A basis for
the expected ecosystem behavior of very large rivers? 49-55. 1989. Proceedings
of the lntemational Large River Symposium (LARS)

Smith,R.A., R.B.Alexander, and M.G.Wolman, 1987. Water quality trends in the nation's
rivers. Science 235: 1607-1615.

Stanford,J.A., 1996. Landscapes and catchment basins. In: Hauer,F.R and G.A.Lamberti
(eds), Methods in Stream Ecology, Acedemic Press, San Diego, CA, pp. 3-22.

16

Thorne,RS. and W.P.Williams, 1997. The response of benthic macroinvertebrates to
pollution in developing countries: A multimetric system of bioassessment.
Freshwater Biology 37: 671-686.

Toth,L.A., 1990. Impacts of channelization on the Kissimmee River ecosystem.
Proceedings of the Kissimmee River Restoration Symposium 47-56.

Vannote,R.L., G.W.Minshall, K.W.Cummins, J.RSedell, and C.E.Cushing, 1980. The
river continuum concept. Can. J. Fish. Aquat. Sci. 37: 130-137.

Ward,J.V. and J .A.Stanford, 1983. The Serial Discontinuity Concept of Lotic
Ecosystems. In: Fontaine,T.D. and S.M.Bartell (eds), Dynamics of Lotic
Ecosystems, pp. 29-42.

Weller,M.W., 1995. Use of Two Waterbird Guilds as Evaluation Tools for the
Kissimmee River Restoration. Restoration Ecology 3: 211-224.

Winterbourn,M.J., J.S.Rounick, and B.Cowie, 1981. Are New Zealand Stream
Ecosystems Really Different? New Zealand Journal of Marine and Freshwater
Research 15: 321-328.

Wolman,M.G., 1971. The nation's rivers. Science 174: 905-918.

Wright,J.F., 1995. Development and use of a system for predicting the macroinvertebrate
fauna in flowing waters. Australian journal of ecology. Oxford 20: 181-197.

Young,R.G. and A.D.Huryn, 1999. Effects of land use on stream metabolism and organic
matter turnover. Ecological Applications 9: 1359-1 376.

17

CHAPTER 2
SPATIAL AND TEMPORAL VARIATION IN THE PHYSICAL AND

CHEMICAL ASPECTS OF NON-WADEABLE RIVERS IN MICHIGAN

18

CHAPTER 2
SPATIAL AND TEMPORAL VARIATION IN THE PHYSICAL AND

CHEMICAL ASPECTS OF NON-WADEABLE RIVERS IN MICHIGAN

INTRODUCTION

Water quality and habitat quality are intimately linked with macroinvertebrate
diversity, life history, and growth. Some of these parameters change predictably along
the river continuum (e.g., turbidity) in a natural manner (V annote et a1. 1980), and others
are influenced by seasonal factors like flooding (Junk et al. 1989). Invariably,
anthropogenic influences shape the physical and chemical traits of lotic ecosystems, and
these influences manifest themselves on multiple spatial and temporal scales. While
much work has been done on the relationships between physical aspects of lotic
environments and the associated biota, these relationships are poorly understood in non-
wadeable rivers because of their large watersheds and corresponding multitude of
processes that determine a non-wadeable river’s physico-chernical signature. When
undertaking a reach scale bioassessment project such as the one in the following chapter,
it is important to discern the spatial and temporal variability of these parameters at the
reach scale, especially when determining stressor-response relationships.

Water quality is ultimately reflected in the macroinvertebrate communities, and
this presents several advantages compared to simply measuring these parameters directly.
Macroinvertebrates are usefirl indicators of water quality because they are present in
ahnost every aquatic environment. They are also indicative of environmental quality at

the local scale, as compared to fish, which are much less sedentary in nature. Unlike

19

organisms with shorter life cycles (e.g., algae), they also allow the effects of regular or
intermittent perturbation to be detected (Resh et al. 1996). However, the synergistic
effects of various stressors any given macroinvertebrate population or community
encounters throughout its life cycle can make it more difficult to isolate the effects of any
one stressor by studying them in their natural environments. Macroinvertebrates do,
however, lend themselves nicely to experimental studies. This allows information fi‘om
monitoring studies such as this one to be tested and causal mechanisms to be analyzed
(Resh et al. 1996). The following is a short review of some of these parameters’
effects—both direct and indirect—on large river macroinvertebrates.

Temperature is one of the most important variables for river biota, affecting
macroinvertebrates in many different ways. A direct effect of temperature is on
metabolic rates. Like all exothermic organisms, macroinvertebrate growth rates are
determined largely by temperature. Temperature also affects solubility of gases in water
such as oxygen. Typically, the greatest source of heat in large rivers is direct solar
radiation, since most their surfaces are directly exposed to sunlight.

Conductivity is the measure of electrical conductance of water, and an
approximate indicator of dissolved ions. Variation in conductivity depends on the
relative influence of underlying geological features and precipitation in providing the
system with dissolved ions, with overall discharge and evaporation as secondary
influences (Allan 1995). Studies have shown that stream organisms require water of
some minimal ionic concentration (Willoughby and Mappin 1988), and periphyton
productivity has been shown to be higher in waters with more dissolved ions (Hill and

Webster 1982). In general, water of very low ionic concentrations appears to support a

20

reduced fauna, and the number of species commonly increases with ionic concentrations
(Allan 1995).

Another physical aspect of non-wadeable rivers of importance to
macroinvertebrates is pH. The acidity of water has been shown to increase
macroinvertebrate drift and survival, with mayflies showing the most significant
sensitivity (Courtney and Clements 1998). This factor has also been shown to influence
leaf pack processing rates, with neutral streams showing the fastest rates, alkaline streams
with intermediate rates, and acidic streams with the slowest rates (Griffith and Perry
1993). This effect can be direct (inhibition of macroinvertebrates themselves) or indirect
due to the inhibition of microbial activity on the leaf surface (Griffith and Perry 1993;
Tuchman 1993). While the range of pH in most Michigan rivers is not in itself wide
enough to cause direct damage to macroinvertebrate communities (Schell and Kerekes
1989), there are other important indirect relationships between pH and possible efl'ects on
macroinvertebrates. For example, in areas with heavy metal pollution, metal dissolution
in the water column has been shown to be heavily influenced by pH, which may lmve
direct effects on macroinvertebrates or cascading effects caused by algal sensitivity to
metals (Aliotta et al. 1983; Kozitskaya and Komarenko 1995; Peterson et al. 1984).

Dissolved oxygen G)O) also greatly affects aquatic life. The amount of dissolved
oxygen in water bodies is influenced by abiotic aspects of the system such as temperature
and barometric pressure, as well as biotic processes like respiration (decomposition) and
photosynthesis. Low D0 is often a result of nutrient enrichment stemming fi'om
agricultural runoff in streams, and this is commonly what is targeted in multimetric

indices. In addition to nutrient enrichment, DO levels may decline to dangerous levels in

21

areas with high quantities of organic matter. Macroinvertebrates have differing
tolerances to low DO, and the relative abundance of tolerant groups is often used as a
means of evaluating organic pollution in small streams (Hilsenhoff 1987;1-Iilsenhoff
1988)

Fine sediments in the water column can also influence macroinvertebrate
communities. Turbidity increases with riparian clearing, increases in agricultural
landuse, as well as an increase in impervious substrates such as roads due to urban
development and construction projects in the watershed (Wood and Armitage 1997).
When this sediment settles to the bottom, it causes embeddedness of coarser substrates,
which is detrimental to some fish species (Tumpenny and Williams 1980). Suspended as
well as benthic fine particulate organic matter (FPOM) can also result in shifts in
functional feeding groups of macroinvertebrates. For example, high suspended sediment
loads may inhibit filtering collectors by clogging nets (e.g., Hydropsychidae) (Lemly
1982) or scrapers clingers (Kaller 2002) by covering hard substrates such as boulders,
rocks, or large woody debris (LWD). Increased suspended sediments has also been
documented to reduce macroinvertebrate density, biomass, as well as EPT taxa richness
(Angradi 1999).

As mentioned above, nutrient enrichment is often the focus of environmental
monitoring studies because of the effects it has on dissolved oxygen levels. In a nutrient
limited system like small streams, enrichment of nitrogen or phosphorous causes algae to
bloom out of control. These algae eventually die and settle to the bottom where
decomposers begin to break them down. Because these decomposers are heterotrophic

(e.g., bacteria), oxygen levels begin to decline rapidly as a result of increased biological

22

oxygen demand. In many cases, DO declines to levels that are lethal to the more
sensitive groups of macroinvertebrates and can also result in fish kills. Nutrient
enrichment is primarily caused by intensive agriculture in the watershed and is facilitated
by the clearing of riparian vegetation that would normally buffer nutrient inputs.

All of the parameters discussed above act to influence the algal community—both
suspended and benthic. Algae need light to photosynthesize, so they are influenced by
suspended solids. Indeed, most large rivers are primarily limited by light because they
are deeper and have much higher suspended sediment loads than small streams (V annote
et al. 1980), and this phenomenon is often seasonal in nature (Knowlton and Jones 1996;
Koch et al. 2004). Aside from the effects of suspended and benthic algae have on. DO
levels, these communities are also important food sources for filtering collectors and
scrapers.

The factors discussed above work in concert with habitat parameters such as
availability of large woody debris (LWD), substrate size, riparian vegetation, and
hydrologic variation to shape lotic macroinvertebrate communities, and a large-scale
study of Michigan’s non-wadeable river physical and chemical properties has never been
done that describes these properties. The objectives of this study were to 1) describe the
range of physical and chemical parameters in Michigan’s non-wadeable rivers; 2)
determine which parameters are common to entire catchments and which are more
descriptive of reach scale properties; and 3) determine the year to year variability in these

parameters from reaches visited multiple times.

23

METHODS AND MATERIALS

I used a YSI 6600 multiparameter data sonde to record temperature, pH,
conductivity, dissolved oxygen, turbidity, and suspended chlorophyll at each study reach
(Table 3.1). Readings were taken at each transect in the vicinity of macroinvertebrate
sampling. The sonde was calibrated daily for dissolved oxygen, turbidity, and
chlorophyll. I used barometric pressure to calibrate the oxygen probe to 100% saturation.
In the field, the turbidity and chlorophyll probes were calibrated daily using distilled
water. Turbidity was also calibrated weekly in the laboratory using a turbidity standard
(100 NTU). I calibrated pH on a weekly basis using the 3-point method (pH=4,7,and 10).
Conductivity was also calibrated weekly (conductivity=0 mS/cm and 100 mS/cm).

Another sonde was deployed at most sites for a period of 24b and set to log
environmental data every 15 minutes. This sonde was used to look at daily changes in
dissolved oxygen. This was done only in the 2001 sampling year, and the DO probe
malfunctioned for many of the sites, and as a result, diel oxygen data are only available
for 10 sites: sw_sg, sg_zil, ra__dun, mk_trk, mk_thp, kz_ver, kz_cus, gr_ion, gr _gr, and

gr_cmr (see Table 3.1 for site codes).

Nutrient Samples

Water samples were taken at transects A, G, and K (Figure 3.1) at each site by
placing 250 mL Nalgene bottles below the surface of the water, allowing water to flow
into the bottle. Each time, I rinsed the bottles with river water 3 times prior to capping
the bottle. Bottles were acid washed by soaking them in 70% H2804 for 48 hours prior to

use. In the field, bottles were placed on ice in a large cooler. At the end of each day, all

24

bottles were returned to the lab and samples were frozen until they were sent to Michael
Grant (analytical chemist, UMBS). Each sample was analyzed for nitrate, ammonia, total

N, SRP, and total P.

Chlorophyll Samples

Periphyton was sampled by scraping 2 10cm2 subsamples from large rocks (at
sites with cobble) or large woody debris. Periphyton was scraped with a toothbrush and
rinsed into the calibration cup of the YSI 6600 multiparameter sonde with distilled water
so the total volume of the sample was 200mL. Chlorophyll was measured with the sonde
in the total sample before filtering a 25 mL subsample through a Whatman GFC filter
paper in the field to be used for actual chl a analysis. Samples were taken at transects A,
G, and K.

I sampled phytoplankton with a standard plankton net by taking five sweeps of
the net through the water column (in order to make sure algal concentrations would be
high enough for chl a analysis). The sample was rinsed into the collection jar at the
bottom of the net and poured into the sonde calibration cup. At this point, I recorded the
sonde chlorophyll readings, and then filtered a 50 mL subsample through a Whatman
GFC filter paper to be used for actual chl a analysis. Samples were taken at transects A,
G, and K.

In both cases, filter papers were placed in covered plastic petri dishes and
wrapped in aluminum foil in the field before they were placed on ice. At the end of each

day, all chlorophyll samples were frozen for chlorophyll a extraction analysis. The actual

25

values (from chl a analysis) were then compared to the sonde values in order to evaluate
the accuracy of the chlorophyll probe with linear regression.

I analyzed the sonde data by performing an AN OVA on all sites using individual
transect data as replicates. This allowed the coarse evaluation of overall differences
among sites for each parameter. To examine differences among sites on the same river I
performed separate AN OVAs on sonde measurements using only values from sites
within the same catchment (e.g., Grand River) and within the same sampling year (e.g.,
2001 or 2002) in order to isolate catchment vs. year to year variation in values. Year to
year variation in physical parameters was analyzed using ANOVA on sites that were
repeated each year. See Table 3.1 for a list of sites and corresponding sampling dates. I
also used AN OVA to examine differences in nutrients, phytoplankton, and periphyton
among sites from 2001 only. My 2002 samples were subjected to thawing before they
were sent for nutrient or chlorophyll a analysis, and were therefore inaccurate. Year-to-
year differences in sonde parameters were examined by performing individual T-tests on
reaches that were sampled in both years. See Tables 2.1 and 2.2 for average values for

parameters measured at each site.

RESULTS
Overall Range of Physical and Chemical Parameters

All of the sonde parameters were highly variable among sites (Table 2.3), and
significant differences (p<0.05) were detected in all cases (Table 2.4). In addition,
nutrient concentrations as well as phytoplankton and periphyton all showed overall

differences at the reach scale (Tables 2.5-2.6). Temperature ranged between 19C and

26

30C, while conductivity ranged between 0001-0999 mS/cm. Michigan’s non-wadeable
rivers also ranged in pH, although none of the sites sampled were acidic in nature.
Dissolved oxygen also showed a wide range in overall values (Table 2.3). Turbidity and
suspended chlorophyll, however, showed the widest range in values across sites (Table
2.3). The results of the regression analysis showed significant relationships between
sonde values and measured values of chlorophyll a for both periphyton and
phytoplankton (Figure 2.1). Since temperature is confounded with time of day as well as
time of year, I believe that any real differences in temperature are not necessarily
biologically relevant or the result of anthropogenic impacts. For this reason, differences

in average temperature will not be discussed in this chapter.

Difl'erences Among Sites from the Same Catchment

OveralL most sonde parameters varied significantly at the reach scale in 2001
(Table 2.4). In the Grand River basin, all parameters showed differences among sites
except for turbidity, which was highly variable. The Kalamazoo River sites varied at the
reach scale in suspended chlorophyll, DO, and turbidity, but there were no significant
differences between these two sites in neither conductivity nor pH. The two sites on the
Manistee River sampled in 2001 showed no differences in suspended chlorophyll, but all
other parameters were significantly different. All parameters were significantly different
between the two Muskegon River sites except pH. The two sites on the River Raisin
showed significant differences among all physical variables (Table 2.7). Overall
differences in phytoplankton and periphyton densities were detected (Table 2.6), but

because of the high correlation between actual chlorophyll a from extraction methods and

27

the sonde values (Figure 2.1), only sonde values will be discussed for the remainder of
this chapter. See Figures 2.2-2.4 for actual differences in sonde parameters among 2001
sites.

Overall differences in nutrient concentrations were detected among sites (Table
2.5). However, these differences in dissolved nutrients among sites in the same
watershed were not so found to be statistically significant, and this is likely due at least in
part to the smaller number of nutrient samples taken at each site (n=3) compared to the
sonde measurements (n=11). Sites within the same basin almost always had nutrient
concentrations that were approximately the same. The exception to this was the Grand
River. Except for ammonia and SRP concentrations, nutrients levels were significantly
different among the 4 sites. The only other difference in nutrient levels between sites was
in the River Raisin, which showed differences in SRP between the two sites sampled in
2001 (Table 2.8). See Figures 2.5-2.7 for actual differences in nutrient concentrations
among 2001 sites.

Despite the lack of significant differences in nutrient concentrations, sites fi'om
the 2001 sampling season showed differences in both phytoplankton and periphyton as
measured by the chlorophyll probe on the sonde (Table 2.6; Figure 2.8).

When basin effects were considered for 2002 sites, similar results were found.
While most variables were significantly different among sites on the same river, there are
exceptions. Interestingly, Grand River sites showed differences in turbidity in 2002, but
showed no differences in conductivity. Similarly, the sites on the Manistee River showed

no differences in turbidity in 2002. The Tahquarnenon River sites were most similar

28

fiom a water quality standpoint—the only significant difference between the two sites

was conductivity (Table 2.9). Nutrient data were not available for the 2002 samples.

Year-to- Year Variation in Water Quality Parameters

When differences between 2001 and 2002 values of water quality parameters
were tested with individual T-tests, I found that most sites showed different water quality
signatures between years. All sites that were repeated in the two field seasons showed
significant differences in conductivity, with the largest difference occurring at the Grand
River @ Grand Rapids (Table 2.9). All sites but the Grand River @ Johnson Park
showed significant differences in pH. Dissolved oxygen levels were also different at all
sites but the Grand River @ Johnson Park and the Manistee River @ High Bridge.
Turbidity and suspended chlorophyll were the most similar water quality parameters
between the two years. Only two sites on the Grand River (Ionia and Comstock
Riverside) showed differences in turbidity between years, and one site (Grand River @
Comstock Riverside) showed significant difference in suspended chlorophyll (Table 2.9).
Interestingly, all differences in conductivity and suspended chlorophyll were the results
of decreases in values, but all other parameters showed some sites that increased and
some sites that decreased in values from 2001 to 2002 (Table 2.9). See Figures 2.12-2.17

for actual measurements of year-to-year variation in physical and chemical values.

DISCUSSION
In general, the physical/chemical signature of non-wadeable river reaches is

unique even among sites from the same catchment, suggesting that despite upstream and

29

catchment-wide influences, reaches maintain a unique physical environment. Most
parameters were highly variable even among sites fi'om the same catchment, and this is
likely due to the many factors that influence water quality parameters in a large river
ecosystem. These factors include underlying geological features, magnitude and timing
of precipitation (influencing discharge), and the composition of this precipitation (Allan
1995; Castillo et al. 2000). In addition to the natural fluctuations in the physical and
chemical properties of rivers, these parameters are also heavily influenced by
anthropogenic influences like landuse, habitat modification, impoundment, and industrial
or municipal effluents.

Conductivity appears to be highly related to geographical (geological) differences
among sites, with the lowest values in conductivity occurring at the northern sites
(AuSable, Muskegon, Manistee, Tahquamenon, and Menominee Rivers) in both years
(Figure 2.23 and 2.9a). All sites were slightly alkaline in both years (Figures 2.2b and
2.9b), and the upper peninsula sites were close to neutral (Figure 2.9), suggesting
geological differences may confer a greater buffering capacity in these sites.

Dissolved oxygen is relatively high in all sites despite some sites’ heavily
agricultural watersheds. This suggests the non-wadeable rivers I sampled were not
nutrient limited—it is more likely they were light-limited (Knowlton and Jones 1996;
Koch et a1. 2004). While those sites with intensive agriculture and their associated
nutrient enrichment do have higher ranges in D0 (e.g., gr_ion), the minimum DO values
at even these sites are not dangerously low (Figure 2.18). The sites with the lowest range

in diel D0 are those that are most shaded (e.g., ra_dun, kz_cus) (Figure 2.18). Notably,

30

the River Raisin @ Dundee (ra_dun) is also one of the most turbid sites visited in 2001
(Figure 2.3).

Both turbidity and suspended chlorophyll varied most among sites in both years
(Figures 2.3b, 2.4, 2.1% and 2.11). Both parameters are highly dependant on discharge,
and so the variability within each site is expected. Variability in turbidity among sites
was most likely due to landuse differences (along with underlying geology) (Young and
Huryn 1999). Variability in suspended chlorophyll among sites could be due to many
factors including conductivity, turbidity, riparian shading, and nutrient inputs (Allan
1995)

Nutrient concentrations also varied significantly among sites (Table 2.5), and this
was again likely due to differences in landuse—both catchment wide and riparian.
Interestingly, some of the sites with the highest nutrient concentrations had relatively low
suspended chlorophyll (phytoplankton) as well as periphyton (as measured by the
chlorophyll probe). Many of these same sites also had the highest turbidity. For
example, in 2001, the River Raisin @ Dundee (ra_dunOl) had the highest nitrate levels
(Figure 2.5b) and highest SRP levels of all sites (Figure 2.7b). Yet this site had lower
suspended chlorophyll (Figure 2.4) and phytoplankton (Figure 2.8a) levels than most
sites. Presumably, this is due to the fact that this site also had the highest turbidity of all
sites visited in 2001 (Figure 2.3b). The site with the lowest turbidity in 2001 (ma_hbrOl)
(Figure 2.3b) also had the lowest suspended chlorophyll (Figure 2.4) as well as low
values of all nutrients measured (Figure 2.5-2.7). This site also had low phytoplankton,
but relatively high periphyton concentrations (with low standard error) (Figure 2.8). This

suggests that nutrient concentrations are adequate for maintaining consistently high

31

periphyton levels in the absence of high suspended solids, which act to shade benthic

algae.

Difi’erences Among Sites fi'om the Same Catchment

When assessing non-wadeable river environmental health at the reach scale, it is
important that biological communities do not simply represent catchment-specific
signatures. It has been shown that habitat quality varies at the reach scale (Wilhelm
2005), and these data suggest that water quality is also unique at the reach scale. Almost
all physical parameters from sonde data were significantly different among sites fi'om the
Grand River and between sites fi'om the Kalamazoo, Manistee, Muskegon, and Raisin
rivers in 2001 (Table 2.7). Those parameters that were not significantly different among
sites item the same basin were those that were the most variable. For example, turbidity
levels were the same for all Grand River sites (Table 2.7), and this is likely due to the fact
that turbidity levels within each site on the Grand River were highly variable compared to
other parameters (Figure 2.3b). The Manistee River sites showed a similar pattern in
suspended chlorophyll (Table 2.7; Figure 2.4). Some of these patterns may be related to
proximity of sites to each other. For example, suspended chlorophyll levels in the Grand
River increased fi'om upstream to downstream (the order of these sites along the
continuum, from upstream to downstream is: gr_ion, gr_cmr, gr _gr, gr Jon) (Figure 2.4).
A similar pattern in chlorophyll levels is seen in the River Raisin fiom the upstream site
(ra_dun) to the downstream site (ra_mon) (Figure 2.4). This is a common pattern related
to canopy cover and discharge predicted by stream ecosystem theory (Minshall et al.

1985;Vannote et al. 1980). Similar patterns were observed in 2002, with the exception of

32

the two sites on the Tahquamenon River. The only significant difference in sonde
parameters on this river was in conductivity (Table 2.9). The upstream site (tq_nwb02)
had higher conductivity than the downstream site (tq_pdsOZ) (Figure 2.9a).

This reach-specific distinctiveness in sonde parameters was not seen in nutrient
concentrations. Almost none of the sites tested for differences in nutrient concentrations
showed significant differences when compared with sites from the same catchment
(Table 2.8). As mentioned above, this is probably at least partially due to the low number
of nutrient samples taken at each reach (n=3). The exception to this is the Grand River,
where total N, nitrate, and total P all showed differences among the 4 sites sampled
(Table 2.8). One possible reason for this is the proximity of the Grand River sites to
minor impoundments. Directly downstream fi'om the Comstock Riverside site (gr_cmr),
there are two small dams. This could explain why nutrient levels drop so low in this site
as well as the site directly downstream from the dams (gr _gr) (Figures 2.5 and 2.7)
(Castillo et al. 2000). This could also help explain the abundance of periphyton directly
downstream from the dam (gr _gr). This site had the lowest turbidity of all the Grand
River sites, and this is likely because the suspended solids settled out directly upstream
from the dam, increasing water clarity (Ward and Stanford 1983). While most of the
non-wadeable rivers in Michigan are impounded in some way, none of the other sites I

visited were in such close proximity (both upstream and downstream) fi'om a dam.

Year-to- Year Variation in Water Quality Parameters

Most physical parameters varied significantly fi'om 2001 to 2002. In the 6 sites

that were sampled in both years, conductivity was significantly different in all sites

33

(Table 2.10), and this was always due to in increase in this parameter (Figures 212-217).
An explanation for this is the increased discharge during 2002 due to increased
precipitation. The 2001 sampling season was one of the driest summers in recent years,
which would have increased the concentration of dissolved ions in river waters
throughout the state of Michigan. In 2002, however, there was much more rain, which
diluted these ions and lowered the conductivity (personal observation). A similar pattern
was found in pH fi'om year to year. This is not surprising since pH and conductivity are
highly correlated (Figures 2.12-2.17) (Allan 1995).

Patterns in D0 and turbidity were not so clear. While most sites showed
significant differences in D0 from year to year, these values increased at some sites and
decreased at others (Table 2.10). For example, mean DO levels decreased by 2.875 mg/L
at the Grand River @ Comstock Riverside (gd_cmr) in 2002 as compared to 2001 (Figure
2.15). This could be due to an overall decrease in suspended chlorophyll at this site
(Figure 2.15). However, at the Grand River @ Grand Rapids (gr _gr), the opposite trend
in D0 was observed (Table 2.9; Figure 2.12). There were also increases and decreases in
turbidity levels, though none of the increases were significantly different from year to
year (Table 2.9). As mentioned above, turbidity was highly variable even within each
study reach.

The only site in which suspended chlorophyll levels were significantly different
fiom year to year was the Grand River @ Comstock Riverside (gr_cmr) (Table 2.9,
Figure 2.15). The chlorophyll probe malfirnctioned for the Manistee and AuSable sites in
the year 2002, so no data are available on year-to-year variation. See Figures 2.12-2.17

for graphical representation of year-to-year variation in sonde parameters.

34

Implications for the Bioassessment of Non-wadeable Rivers in Michigan

Constructing an assessment protocol of any type for non-wadeable rivers is challenging
because of the large watersheds and the associated anthropogenic irnpacts at differing
spatial scales that shape the physical and chemical environments in each reach. It is
important, when undertaking such a project as developing a biomonitoring protocol for
these systems, that the biological community of interest be unique at the scale at which
assessment is to take place. Because biological communities in rivers are shaped by
water quality and habitat quality, it is important that these factors differ at the reach scale
and equally important to understand the source of this variability. This study showed that
water quality parameters such as conductivity, pH, DO, turbidity, chlorophyll, and
nutrients are highly variable at the reach scale, and this presumably is what causes

macro invertebrate communities to be unique at the same scale (see Chapter 3).

However, the robustness of any bioassessment protocol is of equal importance.
The data reported in this study suggest that year-to-year variation in a river’s physical and
chemical environment is an important consideration. Indeed, this will likely be the
reason for the differences in IBI scores between the two main sampling seasons since
habitat at any given site changes much more slowly.

To be of most use, a dataset like this, combined with macroinvertebrate data from
the same sites, could be used to formulate hypotheses that lend themselves to
experimental work. This would allow actual study of mechanistic relationships between
water quality (and the human behaviors that modify it) and macroinvertebrate population

and community responses.

35

LITERATURE CITED

Aliotta,G., G.Pinto, and APollio, 1983. Observations on tolerance to heavy metals of
four green algae in relation to pH. Giomale Botanico Italiano 117: 247-251.

Allan,J.D., 1995. Stream Ecology: Structure and function of running waters. Chapman
and Hall, London.

Angradi,T.R, 1999. Fine sediment and macroinvertebrate assemblages in Appalachian
streams: A field experiment with biomonitoring applications. Journal of the North
American Benthological Society 18: 49-66.

Castillo,M.M., J.D.Allan, and S.Brunzell, 2000. Nutrient concentrations and discharges
in a midwestern agricultural catchment. Journal of Environmental Quality 29:
1142-1151. .

Courtney,L.A and W.H.Clements, 1998. Effects of acidic pH on benthic
macroinvertebrate communities in stream microcosms. Hydrobiologia 379: 135-
145.

Griffith,M.B. and S.A.Perry, 1993. Colonization and processing of leaf litter by
macro invertebrate shredders in streams of contrasting pH. Freshwater Biology.
30: 93-103.

HilsenhoflLW.L., 1987. An improved biotic index of organic stream pollution. Great
Lakes Entomologist 20: 31-40.

Hilsenhoff,W.L., 1988. Rapid field assessment of organic pollution with a family-level
biotic index. Journal of the North American Benthological Society 7: 65-68.

Junk, W. J., Bayley, P. B., and Sparks, R. E. The flood pulse concept in river-floodplain
ecosystems. Dodge, D. P. 106, 110-127. 1989. Proceedings of the International
Large River Symposium.

Kaller,M.D., 2002. Effects of sediment upon benthic macroinvertebrates in forested
northern Appalachian streams.

36

Knowlton,M.F. and J .R.Jones, 1996. Experimental evidence of light and nutrient
limitation of algal grth in a turbid midwest reservoir. Archiv fur Hydrobiologie
135: 321-335.

Koch,R. W., D.L.Guelda, and P.A.Bukaveckas, 2004. Phytoplankton growth in the Ohio,
Cumberland and Tennessee Rivers, USA: inter-site differences in light and
nutrient limitation. Aquatic Ecology 38:17-26.

Kozitskaya,V.N. and Y.Komarenko, 1995. Effect of environmental pH on growth of
algae. Hydrobiol. J. 31: 35-45.

Lemly,A.D., 1982. Modification of Benthic Insect Communities in Polluted Streams:
Combined Effects of Sedimentation and Nutrient Enrichment. Hydrobiologia 87:
229-245.

MinshalLG.W., K.W.Cummins R.C.Petersen, C.E.Cushing, D.A.Bruns, J.R.Sedell, and
R.L.Vannote, 1985. Developments in stream ecosystem theory. Canadian Journal
of Fisheries and Aquatic Sciences 42: 1045-1055.

Peterson,H.G., F .P.Healey, and R.Wagemann, 1984. Metal toxicity to algae: A highly pH
dependent phenomenon. Canadian Journal of Fisheries and Aquatic Sciences 41:
974-979.

Resh,V.H., M.J.Myers, and M.J.Hannaford, 1996. Macroinvertebrates as Biotic
Indicators of Environmental Quality. In: Hauer,F.R. and G.A.Lamberti (eds),
Methods in Stream Ecology, Academic Press, San Diego, pp. 647-668.

Schell, V. A. and Kerekes, J. J. 1988. Distribution, abundance and biomass of benthic
macro invertebrates relative to pH and nutrients in eight lakes of Nova Scotia,
Canada. 1989. Nova Scotia, Canada, Wolfville, NS. (Canada). Symp. on the
Acidification of Organic Waters in Kejirnkujik National Park.

Tuchman,N.C., 1993. Relative importance of microbes versus macroinvertebrate
shredders in the process of leaf decay in lakes of differing pH. Canadian Journal
of Fisheries and Aquatic Sciences 50: 2907-2712.

Turnpenny,A.W.H. and R.Williams, 1980. Effects of sedimentation on the gravels of an
industrial river system. J. Fish Biol. -693.

37

Vannote,R.L., G.W.Minshall, KW.Cummins, J.R.Sedell, and C.E.Cushing, 1980. The
river continuum concept. Can J. Fish. Aquat. Sci. 37: 130-137.

Ward,J.V. and J .A Stanford, 1983. The Serial Discontinuity Concept of Lotic
Ecosystems. In: Fontaine,T.D. and S.M.Bartell (eds), Dynamics of Lotic
Ecosystems, pp. 29-42.

Wood,P.J. and P.D.Armitage, 1997. Biological eflects of fine sediment in the lotic
environment. Environmental Management 21: 203-217.

Young,R.G. and A.D.Huryn, 1999. Effects of land use on stream metabolism and organic
matter turnover. Ecological Applications 9: 1359-1376.

38

TABLES

39

Table 2.1. Mean values for physical/chemical parameters by site. Parameters were
recorded with a YSI 6600 multiparameter data sonde. TOT N=Total Nitrogen (ppm);
TOT P=Total Phosporous (ppb); TEMP=Temperature (C); COND=Conductivity

(mS/cm); PH=pH; DO=Dissolved Oxygen (mg/L); TURB=Turbidity (NTU);

CHL=Suspended Chlorophyll ( 1g/L). See Table 3.1 for site codes.

 

 

Site Code TOT N TOT P TEMP COND PH no TURB CHL
as__whp01 0.28 7.20 25.49 0.30 8.58 9.62 0.04 1.17
gd_cmr01 0.84 30.27 24.95 0.65 8.41 11.30 20.02 46.76
gd_gr01 0.77 29.00 25.98 0.72 8.47 10.59 27.78 40.76
gd_ion01 2.63 80.33 24.67 0.68 8.34 8.98 21.34 33.00
gd ~161101 1.90 53.00 24.94 0.68 8.50 11.51 34.62 46.88
kz_cusOl 1.08 38.17 23.17 0.60 7.99 7.22 13.69 7.02
ma_hbr01 0.27 6.60 22.54 0.34 8.26 8.87 0.55 0.73
ma_rbw01 0.27 3.53 23.07 0.35 8.20 9.37 12.63 2.42
mk_thpOl 0.52 19.13 22.71 0.33 8.23 9.44 0.88 5.79
mk_trkOl 0.59 27.50 24.77 0.36 8.16 8.81 3.05 3.54
ra_dun01 1.52 52.35 22.40 0.64 7.97 6.36 75.41 8.46
ra_monOl 2.34 59.57 23.52 0.53 8.44 9.00 31.50 11.95
sg_zilOl 1.40 32.40 27.79 0.87 8.18 5.83 38.07 15.00
sh_sg01 1.19 33.97 25.91 0.68 8.33 7.99 80.06 17.74
tb_sag01 0.74 36.10 29.52 0.88 8.80 11.51 46.18 23.48
as_mth02 0.21 9.00 27.57 0.32 8.23 8.50 N/A N/A
as_whp02 0.13 4.50 26.01 0.30 8.07 7.46 N/A N/A
gd_cmr02 1.48 19.07 22.40 0.57 8.17 8.42 1 1.12 21.46
gd_gh02 1.26 27.70 23.56 0.61 8.51 15.17 24.96 62.90
gd_gld02 1.93 37.37 N/A N/A N/A N/A N/A N/A
gd_grO2 1.47 20.80 21.12 0.59 8.39 12.12 17.18 35.66
gd_ion02 1.42 16.53 24.37 0.56 8.15 7.74 53.05 30.87
gd_jon02 1.23 17.67 22.10 0.62 8.50 11.88 131.51 46.44
ma_ctsO2 0.27 4.33 22.78 0.30 8.08 9.81 N/A N/A
ma_hbr02 0.17 5.97 24.01 0.34 8.05 9.24 N/A N/A
ma_mn302 0.23 6.27 22.47 0.33 7.90 7.02 N/A N/A
me_kss02 0.28 16.53 25.45 0.23 7.39 6.15 39.59 11.65
me_stb02 0.35 15.80 23.34 0.23 7.46 7.51 5.75 7.27
mk_br02 N/A N/A 25.45 0.35 7.96 7.41 10.76 9.07
mk_nwg02 0.43 3.83 19.88 0.29 7.70 8.49 2.96 4.36
sj_mv102 0.73 6.77 28.90 0.57 7.72 7.06 4.06 9.45
sj_rvw02 1.08 16.63 27.00 0.60 8.21 9.61 10.56 110.14
tq_nwb02 0.46 11.20 23.26 0.16 7.12 6.84 30.83 11.28
tq_pd302 0.51 6.70 22.32 0.15 7.13 6.65 9.61 11.51

 

40

Table 2.2. Mean values for landuse/habitat values by site. All values from Wilhelm
(2002) except LWD. Landuse values are percentages of urban (Ur), agricultural (Ag),
and natural (Nat) and are either for the entire watershed (WS) or in a 100m riparian
buffer zone at each site (RIP). LWD=the number of transects with large woody debris to
sample. See Table 3.1 for site codes.

 

 

Site Code Ur ws Ag ws Nat ws Ur RIP AgRIP NatRIP LWD
as_whp01 3.67 2.34 86.29 0.00 0.00 100.00 8
gd_cmr01 7.70 59.50 23.85 93.62 0.00 6.38 4
gd_gr01 8.20 59.05 23.67 66.67 0.00 23.81 2
gd_ion01 8.66 63.56 20.08 0.00 20.00 80.00 9
gd_jon01 8.37 58.84 23.67 5.13 0.00 89.74 10
kz_cusOl 8.78 55.29 28.27 0.00 0.00 95.65 10
ma_hbr01 1.81 13.31 75.83 0.00 0.00 93.18 8
ma_rbw01 1.87 13.35 74.58 6.98 0.00 93.02 9
mk_thpOl 3.22 23.31 62.97 0.00 0.00 100.00 8
mk_trkOl 3.32 24.95 61.52 13.16 71.05 13.16 8
ra_dunOl 6.55 72.38 15.14 51.11 17.78 31.11 | 10
ra_monOl 6.36 74.67 13.25 54.17 0.00 31.25 1
sg_zilOl 8.06 48.68 31.43 38.78 0.00 14.29 3
sh_ngl 9.56 57.13 21.49 0.00 0.00 100.00 0
tb_sag01 4.93 37.19 45.97 27.66 4.26 42.55 5
as__mth02 3.74 3.89 84.65 63.64 0.00 27.27 7
as_whp02 3.67 2.34 86.29 0.00 0.00 100.00 7
gd_cmr02 7.70 59.50 23.85 93.62 0.00 6.38 5
gd_gh02 8.62 59.09 23.31 2.22 0.00 97.78 8
gd_gld02 14.46 52.19 24.93 60.87 0.00 32.61 8
gd_gr02 8.20 59.05 23.67 66.67 0.00 23.81 3
gd_ion02 8.66 63.56 20.08 0.00 20.00 80.00 10
gd_jon02 8.37 58.84 23.67 5.13 0.00 89.74 10
ma_ct502 1.81 13.22 76.92 0.00 0.00 100.00 7
ma_hbr02 1.81 13.31 75.83 0.00 0.00 93.18 10
ma_mns02 2.05 12.09 76.06 86.36 0.00 0.00 6
me_kss02 1.94 5.74 88.35 31.82 0.00 68.18 8
me_stb02 1.90 3.16 91.29 0.00 0.00 100.00 7
mk_br02 3.37 20.56 66.07 10.64 2.13 57.45 10
mk_nwg02 3.28 23.33 63.01 8.11 16.22 70.27 10
sj_mv102 4.25 68.24 22.99 31.48 1.85 57.41 10
sj_rvw02 5.25 66.09 23.41 61.11 2.78 27.78 7
tq_nwb02 1.28 1.95 88.95 0.00 0.00 100.00 0
tq_pd502 0.66 0.79 94.52 2.17 0.00 97.83 10

 

41

Table 2.3. Basic statistics of sonde parameters from all sites. TEMP: Temperature (C);
COND: Conductivity (mS/cm); DO: Dissolved oxygen (mg/L); TURB: Turbidity (NTU);

CHL: Suspended chlorophyll (ug/L)

 

 

 

 

 

 

 

TEMP COND PH DO TURB CHL

N ofcases 374 374 374 374 372 319

Minimum 19.120 0.001 7.000 5.350 -1 . 100 -50.000

Maximum 30.400 0.999 8.880 16.230 317.300 387.500

Range 1 1.280 0.998 1.880 10.880 318.4 437.500

Mean 24.301 0.481 8.107 8.879 20.471 22.442

Standard Dev 2.226 0.201 0.393 2.065 34.041 36.398

Table 2.4. AN OVA results for physical parameters at all sites.

Dependant .

Variable df F-ratlo P-value
Temperature 33 149.017 <0.001
Conductivity 33 209.744 <0.001

pH 33 295.815 <0.001

Dissolved Oxygen 33 127.708 <0.001

Turbidity 33 5.401 <0.001

Chlorophyll 28 7. 1 72 <0.001

Table 2.5. AN OVA results for nutrient concentrations for all 2001 sites.

Dependant .

Variable df' F-ratro P-value

Total N 15 7.748 <0.001

Nitrate 1 5 8.792 <0.001

Ammonia 15 2.968 0.005

Total P 15 7.527 <0.001

SRP 15 9.153 <0.001

 

42

Table 2.6. ANOVA results for chlorophyll concentrations for all 2001 sites.

 

Dependant

 

Variable df F-ratro P-value
Phytoplankton CHL 15 40.059 <0.001
Periphyton CHL 15 2.741 0.009

 

Table 2.7. Basin efl‘ects for sonde parameters. 2001 sites only. This table shows the
results of AN OVA tests for differences among sites from the same river catchment.

 

 

River Parameter df' F-ratio P-value
Grand Chlorophyll 3 10.273 <0.001
Grand Conductivity 3 30.381 <0.001
Grand DO 3 19.185 <0.001
Grand pH 3 13.330 <0.001
Grand Turbidity 3 1.619 0.200
Kalamazoo Chlorophyll 1 42.680 <0.001
Kalamazoo Conductivity 1 0.070 0.794
Kalamazoo DO 1 19.876 <0.001
Kalamazoo pH 1 2.583 0.124
Kalamazoo Turbidity 1 16.620 0.001
Manistee Chlorophyll l l .525 0.23 1
Manistee Conductivity 1 34.946 <0.001
Manistee DO 1 5.531 0.029
Manistee pH 1 12.744 0.002
Manistee Turbidity 1 7.101 0.015
Muskegon Chlorophyll 1 5.453 0.030
Muskegon Conductivity 1 133.008 <0.001
Muskegon DO 1 18.864 <0.001
Muskegon pH 1 3.161 0.091
Muskegon Turbidity 1 9.553 0.006
Raisin Chlorophyll I 4.635 0.044
Raisin Conductivity 1 4.893 0.039
Raisin DO 1 266.914 <0.001
Raisin pH 1 363.795 <0.001
Raisin Turbidity 1 17.455 <0.001

 

43

Table 2.8. Basin effects for nutrient concentrations. 2001 sites only. This table shows
the results of AN OVA tests for differences among sites from the same river catchment.

 

 

River Parameter df F-ratio P-value
Grand Total N 3 ' 51.452 <0.001
Grand Nitrate 3 35.864 <0.001
Grand Ammonia 3 0.940 0.465
Grand Total P 3 19.883 <0.001
Grand SRP 3 3.485 0.090
Kalamazoo Total N 1 0.384 0.569
Kalamazoo Nitrate 1 0.343 0.590
Kalamazoo Ammonia 1 0.169 0.702
Kalamazoo Total P 1 0.186 0.688
Kalamazoo SRP 1 1.569 0.279
Manistee Total N 1 0.001 0.980
Manistee Nitrate 1 0.249 0.644
Manistee Ammonia 1 8.508 0.043
Manistee Total P 1 3.398 0.139
Manistee SRP 1 1.690 0.263,
Muskegon Total N 1 0.286 0.621
Muskegon Nitrate 1 0.049 0.836
Muskegon Ammonia 1 4.21 1 0.109
Muskegon Total P 1 2.097 0.221
Muskegon SRP 1 0.664 0.461
Raisin Total N 1 0.550 0.512
Raisin Nitrate 1 0.329 0.607
Raisin Ammonia 1 0.353 0.594
Raisin Total P 1 0.104 0.769
Raisin SRP 1 312.006 <0.001

 

Table 2.9. Basin effects for sonde parameters. 2002 sites only. This table shows the
results of ANOVA tests for differences among sites from the same river catchment.

 

 

River Parameter df P-value
AuSable Chlorophyll 1 n/a n/a
AuSable Conductivity 1 19.212 <0.001
AuSable DO 1 19.93 5 <0.001
AuSable pH 1 16.502 0.001
AuSable Turbidity 1 0.889 0.357
Grand Chlorophyll 4 63 .881 <0.001
Grand Conductivity 4 0.976 0.429
Grand DO 4 223.180 <0.001
Grand pH 4 181.809 <0.001
Grand Turbidity 4 2.727 0.040
Manistee Chlorophyll 2 n/a n/a
Manistee Conductivity 2 391 .937 <0.001
Manistee DO 2 168.892 <0.001
Manistee pH 2 41 .673 <0.001
Manistee Turbidity 2 0.749 0.482.
Menominee Chlorophyll 1 6.518 0.019
Menominee Conductivity 1 0.442 0.514
Menominee DO 1 41 . 1 83 <0.001
Menominee pH 1 1.358 0.251
Menominee Turbidity 1 1 .571 0.224
Muskegon Chlorophyll 1 1 8.586 <0.001
Muskegon Conductivity I 291 .267 <0.001
Muskegon DO 1 95 .095 <0.001
Muskegon pH 1 43 .773 <0.001
Muskegon Turbidity 1 63 .320 <0.001
St. Joseph Chlorophyll 1 4.598 0.044
St. Joseph Conductivity l 401 . 1 l7 <0.001
St. Joseph DO 1 435.887 <0.001
St. Joseph pH 1 316.283 <0.001
St. Joseph Turbidity 1 45.765 <0.001
Tahquamenon Chlorophyll 1 0.014 0.907
Tahquamenon Conductivity 1 202.661 <0.001
Tahquamenon DO 1 0.987 0.332
Tahquamenon pH 1 0.342 0.565
Tahquamenon Turbidity 1 3.735 0.068

 

45

 

 

$2 $2 :66 9.3+ :3 23+ 83v :3- ~85 33. sales
<2 <2 2 3 MR? so? 4m _ .N- 83v :3- See 82.. Beta
8% 93. we; 83? Sod 22. 83v 83- e2; 23- some»
83 32- :2 82:- 83 an: 83 sod- so? :8. elem
53v Sen- so? 83- 53v mate: 53v 32. 83v 6%.? Belem
:3 2.2. 83v ”was- 82 83+ 82 83: 82v e3? 830m
2:57.“ a 92?»... G 2:9»-.— 9 039»..- G 2:57: 9

a»: =Eaeezo 3.2 case: a»... on E. seas £62.25 ea

 

.880 8?. com _.m 033. com .0328;
8: Sam ”32 .3382: mo£e> $329: ooeocotfiu Period < .uommoeoou mes—g 8:335 ooeocomav gnawed < .meOmmom Emu
Sam 98 Saw 5953 £30838 Bambi :0 meme“? 2323?: Sea mos—eta eocfioomme use av mooeouotmu 502 .3 .m 2an

46

FIGURES

47

 

 

 

 

 

 

 

 

 

 

(a)
<4
c»
3
715‘
3
i
>
01
'0
C
o
L”,
2
_1
_1 I V I l l l
-7 -6 -5 -4 -3 -2 -1
LN(Actual Value) uglL
500 1 1 1" (b)
R2=0.660
_’ 4004. p<0.001 - —
3
3 300 —-
to
>
8
c 200 —‘
o
(I)
100 —
0 1 1 1
0 1 2 3 4

Actual Value (ug/cm2)

Figure 2.1. Actual values fi'om chl a extraction vs. sonde values for (a) phytoplankton
and (b) periphyton samples.

48

 

tb_sagO1
sh_ng1
sg_zil01
ra_mon01
ra_dun01
mk_trk01
mk_thp01
ma_rbw01
ma_hbr01
kz_vb01
kz_cus01
gd_jon01
gd_ion01
gd_gr01
gd_cmr01
as_whp01

I I l
0.00 0.25 0.50 0.75 1.00
CONDUCTIVITY (mS/cm)

(8)

SITE CODE

IIIMIIJIIIIIIIII

 

 

 

tb_sagO1 (b)
sh_ng1
sg_zil01

ra_mon01

ra_dun01
mk_trk01
mk_thp01
ma_rbw01
ma_hbr01
kz_vb01
kz_cus01
gd_jon01
gd_ion01
gd_gr01
gd_cmr01
as_whp01

I l l I
5.0 5.8 6.6 7.4 8.2 9.0
PH

Figure 2.2. Mean (a) conductivity and (b) pH for all 2001 sites. Error bars indicate
standard error. See Table 3.1 for site codes.

SITE CODE

llLlllllJllllIll

 

 

49

 

l

tb_sagO1
sh_ng1
sg_zil01
ra_mon01
ra_dun01
mk_trk01
mk_thp01
ma_rbw01
ma_hbr01
kz_vb01
kz_cus01
gd_jon01
gd_ion01
gd_gr01
gd_cmr01
as_whp01 1

l
0 5 10
DO (mg/L)

(8)

l

IllIIlII

SITECODE

IILIIJ

 

.1.
0'1

 

l

tb_sagO1 (b)
sh_ng1
sg_zil01
ra_mon01
ra_dun01
mk_trk01
mk_thp01
ma_rbw01
ma_hbr01
kz_vb01
kz_cusO1
gd_jon01
gd__ion01
gd_gr01
gd_cmr01

as_whp01 1 1
0 20 40
TURBIDITY (NTU)

I I

I

IILIIII

SITECODE

IIIII

 

 

05
0

Figure 2.3. Mean (a) dissolved oxygen (DO) and (b) turbidity for all 2001 sites.
Error bars indicate standard error. See Table 3.1 for site codes.

50

 

tb_sagO1
sh_ng1
sg_zil01
ra_mon01
ra_dun01
mk_trk01
mk_thp01
ma_rbw01
ma_hbr01
kz_vb01
kz_cusO1
gd_jon01
gd_ion01
gd_gr01
gd_cmr01
as_whp01 1 1
0 20 40
CHLOROPHYLL (ug/L)

SITECODE
IIILIIIIIIIIIIH

I

 

 

O)
C

Figure 2.4. Mean chlorophyll values for all 2001 sites. Error bars indicate
standard error. See Table 3.1 for site codes.

51

SITE CODE

SITE CODE

tb_ng1
sw_sgo1
sg_zil01
ra_mon01
ra_dun01
mk_trk01
mk_thp01
ma_rbw01
ma_hrb01
kz_ver01
kz_cus01
gr _jon01
gr_ion01
gr_gr01
gr_cer1
as_whp01

tb_ng1
sw_ng1
sg_zil01
ra_mon01
ra_dun01
mk_trk01
mk_thp01
ma_rbw01
ma_hrb01
kz_ver01
kz_cusO1
gr_jon01
gr_ion01
gr _gr01
gr_cmr01
as_whp01

 

 

 

 

 

 

 

I I
d (a)
i.
I I _
2 3 4
TOTAL N (ppm)
1 1 _ (b)
I I ‘—
1 2 3
NITRATE (ppm)

Figure 2.5. Mean (a) total N and (b) nitrate for all 2001 sites. Error bars
indicate standard error. See Table 3.1 for site codes.

52

 

tb_ng1
sw_sgo1
sg_zil01
ra_mon01
ra_dun01
mk_trk01
mk_thp01
ma_rbw01
ma_hrb01
kz_ver01
kz_cus01
gr _jon01
gr_ion01
gr _gr01
gr_cmr01
as‘Whpm 1 1 1 1 —

0.00 0.04 0.08 0.12 0.16 0.20
AMMONIA (ppm)

SITE CODE

IIIIIIIIIIIIII

 

Figure 2.6. Mean ammonia for all 2001 sites. Error bars indicate
standard error. See Table 3.1 for site codes.

53

 

tb_ng1
sw_ngt
sg_zilOt
ra_mon01
ra_dun01
mk_tr‘k01
mk_thp01
ma_rbw01
ma_hrb01
kz_ve101
kz_cu301
gr _jon01
gr_ion01
gr _gr01
gr_cmr01
as_whp01

SITE CODE

I I L I l I I

I

I

(a)

llllllllllllllli

 

 

0 10 20 3O 40 50 60 7O 80 90100

TOTAL P (ppb)

 

tb_ng1
sw_ng1
sg_zil01
ra_mon01
ra_dun01
mk_t11<01
mk_thp01
ma_rbw01
ma_hrb01
kz_ver01
kz_cus01
gr _jon01
gr_ion01
gr _gr01
gr_cmr01
as_whp01

SITE CODE

I l
0 10 20
SRP (99b)

0))

lllllillllllll

 

OJ
C

Figure 2.7 . Mean (a) total P and (b) soluble reactive P (SRP) for all 2001
sites. Error bars indicate standard error. See Table 3.1 for site codes.

54

 

tb_ng1 (a)
sw_ng1
sg_zi101
ra_mon01
ra_dun01
mk_trk01
mk_thp01
ma_rbw01
ma_hrb01
kz_ver01
kz_cusO1
gr _jon01
gr_ion01
gr _gr01
gr_cm101
as_whp01

SITE CODE

IIIIIIIIIIIIILII

 

l I I I I I l
0 10 20 30 40 50 60 70
PHYTOPLANKTON (uglL)

1 1 1 1 (b)

8

 

 

SITE CODE
;
'5
§

3

I”

:r

E
IIIILLIIIIIIIII

 

1 1 1 m
0 100 200 300 400 500
PERIPHYT ON (ug/L)

 

Figure 2.8. Mean (a) phytoplankton and (b) periphyton for all 2001 sites.
Data are fi'om sonde measurements. Error bars indicate standard error. See
Table 3.1 for site codes.

55

 

 

 

 

tq_pds02 (a)
tq_nwb02
sj_n'v02
sj_mv|02
mk_nwg02
mk_br02
me_stb02
me_kss02
ma_man02
ma_hbr02
ma_ctsOZ
gd__jon02
gd_ion02
gd_gr02
gd_gh02
gd_cmr02
as_whp02

as_mth02 1

I I
0.0 0.2 0.4 0.6 0.8
CONDUCTIVITY (mS/cm)

SITE CODE

IJLIIIIIIIIIIIIIII

 

 

tq_pdsOZ
tq_nwb02
sj_n'v02
sj_mv|02
mk_nwg02
mk_br02
me_stb02
me_kssOZ
ma_man02
ma_hbr02
ma_cts02
gd_jon02
gd_ion02
gd_gr02
gd_gh02
gd_cmr02
as_whp02
as_mth02

6 7 8
PH

Figure 2.9. Mean (a) conductivity and (b) pH for all 2002 sites. Error bars indicate
standard error. See Table 3.1 for site codes.

0’)

SITE CODE

IIIIIIIIIIIIIIIIII

 

(O

56

 

tq_pds02
tq_nwb02
sj_riv02
sj_mv|02
mk_nwg02
mk_br02
me_stb02
me_kssOZ
ma_man02
ma_hbr02
ma_ctsOZ
gd_ion02
gd_ion02
gd_gr02
gd_gh02
gd_cm102
as_whp02
as_mth02

I
0 6 12
DO (mg/L)

A
3
v

SITE CODE

IIIIIIILIIIIIIIIII

 

A
on

 

tq_pdsOZ
tq_nwb02
sj_riv02
sj_mv|02
mk_nwg02
mk_br02
me_stb02
me_kssOZ
ma_man02
ma_hbr02
ma_ctsOZ
gd_ion02
gd_ion02
gd_gr02
gd_gh02
gd_cmr02
as_whp02
as_mth02 1 1
0 10 20
TURBIDITY (NTU)

Figure 2.10. Mean (a) dissolved oxygen and (b) turbidity for all 2002 sites.
Error bars indicate standard error. See Table 3.1 for site codes.

(b)

SITE CODE

IIIIIIIIIIIIIIIIII

 

 

(a)
O

57

 

tq_pds02 —
tq_nwb02 e
sj_riv02 —
sj_mv|02 —
mk_nwg02 —
mk_br02 —
me_stb02 .,
me_kss02 e
gd_jon02 —
gd_ion02 —
gd_gr02 —
gd_gh02 —
gd_cmr02 —
I I l I l I
0 10 20 30 40 50 6 70
CHLOROPHYLL (ug/L)

SITE CODE

 

Figure 2.11. Mean chlorophyll values for all 2002 sites. Error bars indicate
standard error. No chlorophyll data exist for the AuSable or the Manistee sites. See
Table 3.1 for site codes.

58

{I

 

 

.mm ..\+ mos—m.» 508 950% minur .038 83 c0~0§3mm==8 OOOO 5% 0 "EB 3.5808 0.83 £20883
=< 31283862 85 © 85 Eco 2: see aaoasa Reese 88 as, some as_aeeoo .2 .N 2:5

 

 

 

 

 

 

 

 

 
 
 

 

 

 

 

 

 

 

 

 

30> 30>
NOON POON NOON .OON O
I.
I 1 S n
0 W.
H nlv.
1| 17
n T 1 Dow M
t 1 8 .5” I.
n
r . on L . on?
30> 30> 30.»
NOON «OON NOON SON NO NOON FOON n6
1 1 Or m
t t S G t t n a t t 0 o N
O O
I I NF w I I. W C d I- ; N.° w
5 H
/ S
1 1 9 _I W
t t e. o t L o o w
r 1 I.
p n “F n P 0.0 p h m 0

59

8000833.“ =< 831.03 383 ® 32% 835 05 80b 8000:33q 306an N

30>

NOON

 

.mm ..\+ 0033, 308 @508 83> .088 03c 338338038 OOOO ~m> a as? 3.8308 0003
a 8806 208888 .23 one:

.. 88

 

OF

 

 

«F

115111 co

30>

NOON

I

U

FOON

 

 

 

30>

NOON

 

«OON

 

 

 

fa

NOW

mam

 

 

Yd

“1160 ‘IHO

IICI

30>

NOON

POON

 

OO.

OON

 

 

 

30>

«cow .. room

 

Oon

 

 

 

uI‘D/SW CINOO

n.I.N Ailmqmr

60

.mm ..\+ 002.3 308 36% 33> .288 8% 3808338038 OOOO ~m> 3 53» 00.8308 033 80808338
=< .221 3 Va 88:2 © 3,3 “.55 2: 520 525% .8333 88 Ba 88 18 aafigaoo .2 .N 2:5

 

 

 

 

 

 

 

 

 

Ina/3w CINOO

 

30 > 30 >
38
on «Sm 88
I.
O n
.l l 0' H m-
... m
n H...
mm 1 I. S A
T 1 CD -I N
I 1 cm I.
1 1 8 n
. . 8 . . 8
30 > 30 > 30 >
88 38 «SN 88 . «SN 58 .
n a B o
3o
0 1 1 to 8...
0 :3
d 8...
w T 1 0.0
6 H 8.0
I
.l 8o .
r 1 2 1 1 o a 86
86
b _ z + _ 3 r . 93

 

 

 

 

 

 

 

 

 

61

.Mm 1\+ m0=_3> 8308 $68 83> .0338 333 33833888 OOOO _m> 3 fit: 308.308 03>» 8808338
=< 880433 02803M 0608800 © 32m 83.5 05 80h 8030835“ 323% NOON 83 SON mo 383800 .m_.N 03me

30 >
Noo~

 

FOON

 

N—

 

 

nw

'l/5UJ OCI

 

NOON

30 >
.83

,OF

ON

on

Ov

 

 

 

NOON

30>

POON

on

 

 

GO

'IIBn ‘IHO

 

NOON

30 >

 

FOON

 

Ow

ON

 

 

 

NOON

30>

rOON

 

on

end
umd
omd
mod
OOO

POO 1

NOd

away.

vwd
mwd

 

 

mod

n.LN Autumn;

“113/8w CINOO

62

NOON

 

.Mm 4+ 323> 308 36% 33> .0338 333 33338338 coco Hm> 3 535 303308 0303 8030833a
=< 380.33 03330.5“ xoo88o0 ® 33m 33.5 08 88.3 820833q 303.38 NOON 33 SON mo 38m3q800 «fin 03wE

30>

wOON

 

 

 

N?

9.

'II5UJ OG

NOON

 

30 >
.83

 

 

 

NOON

 

30 >
.83

 

 

0.3

'I/Bn 1H0

30 >
383 53

 

 

Ow

ON

 

 

30 >
383 38

 

 

on

omO
hmd
and
and
8O
56
NOO

8O .

voO
8O

 

 

8O

niN KHPMJHJ.

“JO/8w CINOO

62

COND mS/cm

DO mg/L

0.31

0.30

0.29

0.28

11

 

 

2001

 

 

 

 

2001

Turbidity NTU

 

2002
Year

8.7

20

10

 

 

 

2001

 

2002

 

 

 

2001

 

2002

Figure 2.16. Comparison of 2001 and 2002 physical parameters from the
AuSable River @ Whirlpool (as_whp). All parameters were measured with a
YSI 6600 multiparameter data sonde. Y-axis shows mean values +/- SE.

63

COND mS/cm

DO mg/L

 

0.345

 

0.340 -

0.335 -

 

 

 

 

0.330

 

 

 

 

9.5 - -
10 ~ -

Turbidity NTU

8.5 1—

 

 

 

 

     

-10
2001 2002 2001 2002

Year Year

Figure 2.17. Comparison of 2001 and 2002 physical parameters from the Manistee
@ High Bridge (ma_hbr). All parameters were measured with a YSI 6600
multiparamter data sonde. Y-axis shows mean values +/- SE.

8.0

 

 

sw_sg f w. _
.._.1 i °H:i:H° ~

gr_gr i [EH -
9.....- H134- ~
mk_thp - m -
mk_trk - 1]} 1
ra_dun - fl]. -
gr_ion - Cir—r I ]——IO 1
kz_ver - {IS—1 0 «
kz_cus - W 4

l l l l l

0 2 4 6 8 10 12 14 16

Site Code

 

 

 

 

-
_

 

Dissolved Oxygen (mg/L)

Figure 2.18. Box and whisker plot of diel oxygen change for selected 2001 sites. Data
are from a YSI 600 multiparameter data logging sonde.

65

CHAPTER 3:

BIOLOGICAL EVALUATION OF MICHIGAN’S NON-WADEABLE RIVERS

USING MACROINVERTEBRATES

66

CHAPTER 3:
BIOLOGICAL EVALUATION OF MICHIGAN’S NON-WADEABLE RIVERS

USING MACROINVERTEBRATES

INTRODUCTION

Large river ecosystems have long been subjected to problems caused by human
settlement. Since the beginning of civilization, these systems have been relied-upon
heavily for irrigation, navigation, waste removal, and drinking water, and have
subsequently paid a heavy price for their utility to human inhabitants. Because of their
large watersheds, large rivers are subjected to a much more complicated suite of _
problems than small streams. These include upstream and watershed influences from
agriculture, logging, and urban development, introduction of exotic species from
international shipping, as well as damming and dredging for navigational purposes. This
has resulted in many of our large, non-wadeable rivers being severely degraded (Paton
1979; Sparks et al. 1990; Gore and Shields 1995; Sparks 1995).

Biological assessment has become an accepted way of evaluating the synergistic
effects humans have on lotic environments (Cairns 1990; Cairns and Pratt 1993; Karr
1993; Karr 1987; Karr and Chu 1999; Kerans and Karr 1994). These protocols are
preferable to chemical monitoring primarily because they integrate effects of short-term
environmental variation and combined effects of water and habitat quality, as opposed to
mere snapshot of water quality at the time when chemical samples are taken. Biological
assemblages commonly used in such protocols include fish, macroinvertebrates, and

algae, all with different levels of spatial and temporal resolution. Macroinvertebrates are

67

particularly useful because they are abundant in most streams, provide indication of
localized (reach scale) conditions, are easy to collect and identify, and serve as a food
base for higher trophic levels (Cairns and Pratt 1993).

The development of effective indicators to assess the ecological condition of large
river ecosystems is becoming a priority for state and federal agencies (U SEPA).
However, most bioassessment protocols for macroinvertebrates have been developed
almost exclusively for small, wadeable streams as a means to evaluate nutrient
enrichment and oxygen depletion (e.g., MDNR 1991). Because of the fundamental
differences between small stream and large river structure and fimction (e. g., Vannote et
al. 1980), some metrics included in these protocols are not necessarily applicableto large
rivers.

Our understanding of large river ecology has lagged compared to that of smaller,
wadeable streams. Sampling difficulties relating to their depth, discharge, and structural
complexity is a main reason for this. Additionally, the complicated factors which shape
non-wadeable biological communities (watershed landuse, in-stream habitat, and water
quality issues), make causal relationships between stressors and biota difficult to define
using what is known from research on wadeable streams. However, the application of
fundamental stream theories such as the River Continuum Concept (V annote et aL 1980),
the Serial Discontinuity Concept (Ward and Stanford 1983), and the Flood Pulse Concept
(Junk et al. 1989) to large rivers have received attention (Bayley 1995). These
fundamental theories have facilitated our understanding of large river ecosystems

(Johnson et al. 1995), in that they have served as templates to set expectations of large

68

river structure and function, and have helped formulate ideas regarding the effects
humans have on these systems (Johnson et al. 1995).

The need for ecosystem management of large river systems is extremely
important (Sparks 1995). Currently, some governmental agencies are in the process of
developing habitat and biological sampling protocols for non-wadeable river assessment
(Lazorchak et al. 2000), and these protocols include methods for assessing
macroinvertebrate communities (Klemm et al. 1999).

My primary objective was to develop a non-wadeable index of biological integrity
(N W-IBI) using macroinvertebrate attributes that best describe variability in ecological
condition of Michigan’s non-wadeable rivers. I developed the protocol through a
systematic approach of reducing variable redundancy, and determining which
macro invertebrate attributes were most responsible for among-site differences. Metrics
selected for the final protocol describe population, community, and functional differences
among sites. The goal was to develop two separate IBIS—one using composite samples
from all habitats present in each study reach, and one using samples from targeted
habitats.

When used in conjunction with the non-wadeable habitat index (NWT-II; Wilhelm
et al. 2005; Wilhelm 2002), the NW-IBI will provide an objective means of evaluating
anthropogenic impact and targeting specific rivers and segments of rivers for

conservation or restoration.

69

METHODS
Defining Non-wadeable Rivers

The first step in undertaking such a project is to define what ‘non-wadeable’
means. Intuitively, a non-wadeable river is either too deep or discharge is too high to
permit one to safely wade into in order to acquire samples needed for biological (or
habitat) assessment. However, this necessitates actual field visits, which is not
necessarily cost-effective. For this reason, it is desirable to established basic guidelines
to help us identify non-wadeable segments of rivers before going out in the field.

Large rivers have been defined in the literature in many ways, including those
whose drainage basins exceed 1600 km2 (Ohio EPA 1989), have an average depth of
greater than one meter (Stalnaker et al. 1989), or a stream order of 6th or higher (Sedell et
al. 1989; Johnson et al. 1995; Vannote et al. 1980). For a more detailed discussion, see
Wilhelm (2002).

I used USGS gauge data and various GIS layers to define and identify non-
wadeable rivers of Michigan as those of order 5 and above, with drainage areas greater
than 1600 kmz, mainstem lengths exceeding 100 km, and mean annual discharge greater
than 15 m3/s. These criteria usually translated to those rivers and river segments which
are either too deep or discharge is too high to safely acquire samples without the use of a

boat, and identify 22 rivers in Michigan with non-wadeable sections (Table 3.1).
Study Sites

Rivers visited ranged fi'om 5"1 to 7'“ order and were subject to a range of human

influences and natural variability. Through the course of this study, I sampled 33 non-

70

wadeable river reaches in 11 rivers in Michigan (Table 3.1). Several of these sites were
repeated. I sampled sites in eleven major watersheds, ranging in size fi'om 16,856 km2
(Saginaw River) to 2,124 km2 (Taquamenon River) (Table 3.1; Wilhelm 2002). Overall,
sites were selected fiom the three major ecoregions in Michigan. Six watersheds were in
the Southern Lower Peninsula (SLP), three in the Northern Lower Peninsula (NLP), and
two were in the Upper Peninsula (UP) (Table 3.1). These ecoregions encompass a
considerable range in climate, vegetation, geology, and human landuse. A discernible
gradient in current and historical landuse exists from north to south in Michigan’s
watersheds. Natural areas dominate the UP (90% forested or covered by wetlands).
Though logged in the late 19th century, most of the NLP today is dominated by mixed
conifers and deciduous trees (75% natural; <4% urban, <11% agricultural). Our NLP
sites were the most heavily influenced by humans—less than 25% remained as natural

land, with 57% agricultural and over 8% urban (Wilhelm 2002).

Site Selection

When undertaking a project such as this, especially when no true reference
condition exists, it is important to sample the full range of impact levels to obtain an
accurate picture of what these impacts have on macroinvertebrate communities. In
choosing sites, I wanted to sample the entire gamut of ecological conditions, fi'om those
most highly impacted by urban development and/or agricultural development to those
that are relatively undisturbed.

Wilhelm (2002) rated sites based on percent agriculture and urban landuse in the

basin and in a 100m riparian buffer, dam density, NPDES permit density, and road

71

density. These scores were summed and incorporated into an overall watershed-level
Catchment Disturbance Gradient (CDG) (Table 3.2). Sites were also rated based on
number of gaps in the riparian and riparian width using aerial photographs, and these
scores were incorporated into a Riparian Disturbance Gradient (RDG) (Table 3.3).

Sites visited ranged widely in both scores (Wilhelm 2002). Of the sites visited by
the biological crew, the Tahquamenon River at Paradise scored best for watershed-level
disturbance (CDG=0), while the River Raison at Dundee scored the worst (CDG=13)
(Table 3.2). The AuSable River at Whirlpool scored the best for Riparian disturbance
(RDG=0), while the St. Joseph River at Riverview scored the worst (RDG=8) (Table
3.3). It should be noted that a reach with a low score for either of these measures of
anthropogenic influence should not necessarily be considered a reference reach, rather, it
should be considered to be the least impacted condition. Because of the size of non-
wadeable river watersheds, virtually all large rivers in Michigan are impacted by human

settlement or other landuse issues in some way.

Field Methods

Reaches were defined as a 2000m section of river in which 11 evenly spaced
transects (every 200m) were sampled for habitat quality, water quality, and
macroinvertebrates (Figure 3.1). US. EPA studies have found that, for Midwestern
streams, a reach length of 40 channel widths is sufficient to characterize the variability in
fish communities (Lazorchak et al. 2000). Despite the fact that the average wetted width
of our sites was 89m (range=32-183m), I believed this to be a sufficient length in

capturing most of the natural variability in macroinvertebrate communities (as well as

72

habitat) in each reach, while keeping the length of the study reach feasible for rapid
protocols.

At each study reach, I used procedures modified from US EPA protocols for
sampling non-wadeable rivers macroinvertebrates. I sampled all available habitats at
each transect. Habitats were categorized into 6 different types: fine particulate organic
matter (FPOM), sand, coarse substrates (gravel and small stones), cobble, large woody
debris (LWD), and macrophytes. Each habitat was sampled with a D-fi'ame aquatic dip
net with 500 rm mesh in 15 second timed sweeps to standardize the sampling effort.
Samples were then rinsed in a sieve (500 um) and preserved in the field with 70% EtOH.

Each reach was sampled twice. The first set of samples was kept separate by
habitat to look at habitat-specific differences in macroinvertebrate community structure.
The second set of samples was combined into one large composite sample for each site.
This overall sampling scheme is a slight modification of the US. EPA protocol for
sampling non-wadeable rivers (Lazorchak et al. 2000).

Samples were sorted and identified to family level in the laboratory. Habitat-
specific samples were completely sorted, while the large composite samples were
subsampled into quarters before being processed.

Many state and federal agencies target specific habitats to help control for
differences in macroinvertebrate populations and communities due to differences in
habitat. For example, cobble samples from two different rivers might be more similar
(e.g., in terms of diversity or FF G composition) than a cobble and a sand sample from the
same river. Another benefit of habitat-specific sampling is that it reduces the amount of

fieldwork for assessment crews and can help reduce the amount of detritus in samples.

73

At each transect, I recorded the different habitat types present and kept one set of
samples separate to determine whether or not habitat-specific sampling was possible for
the biological assessment of non-wadeable rivers. This process was mainly one of
observation. Which habitats were common to all of our study reaches? Is
macro invertebrate structure in any given habitat variable as a result of water quality or

other human influences? How does habitat type affect metric robustness?

Data Reduction and Metric Selection

For each site, an array of different summary attributes, or potential metrics, were
calculated. These potential metrics described population, community, and functional
levels of organization. In total, 26 metrics were included in the initial analysis (Table
3.5). However, many of these metrics were highly correlated with each other, and some
did not vary among sites. Because of this, we used a stepwise process in selecting
metrics to include in the final protocol.

The first step was to eliminate redundancy among metrics. A Pearson correlation
analysis was done on all potential metrics to see which ones were highly correlated. If
two or more individual metrics were highly correlated (correlation coefficient > 0.70),
metrics were removed from further analyses based on several criteria, precision of metric
response to individual stressors, biological meaningfiilness, ease of measurement, and
distributional characteristics. Correlation analysis was done on population, community,
and filnctional metrics separately to ensure that each type of metric was represented in the

final protocol.

74

The second step in metric selection was to perform principal components analysis
(PCA) on metrics retained fiom the correlation analysis. This provided information on
which biological attributes were most responsible for among-site variation Again,
criteria were set for the retention of metrics fiom the PCA. Only axes with eigenvalues
greater than 1 were further examined, and retained metrics were those with the highest
loadings on each retained axis. For this part of the analysis, all metrics were standardized
by mean and standard deviation because of vast differences in scale of individual metrics.

When necessary, metrics were also transformed to approximate normal distributions.

Metric Response to Stressors

Examining metric response to human influences is a crucial part of the metric
selection process. In the first two steps of the data reduction process, metrics were
identified that provide unique information and explain significant among-site variation.
However, if metric response to human influences is not log-linear or exponential, an
evaluation of sites based on tint metric will be ambiguous. For example, if a metric
value is plotted for each site along a gradient of human influence and the response is
quadratic, highly impacted sites and unirnpacted sites may receive the same score
(Stevenson and Smol 2001). The metrics examined (Table 3.5) are generally agreed to
have predictable responses to human influences, with the exception of some of the F F G
attributes (Ohio EPA 1989; Barbour et al. 1992).

To determine the specific stressors to which individual metrics respond, we used
multiple linear regression (MLR) with individual metrics as the dependent variables and a

suite of physical, chemical, habitat, and landuse variables as the independent variables.

75

Both forward and backward stepwise regression (tolerance=0.001, p to
enter/remove=0. 15) helped us compare environmental variables to which metrics
respond. Significant metric/stressor relationships as well as overall explanatory power
(R2) helped us evaluate each regression model.

Environmental variables were recorded at each transect at each site, and mean
values per site were used in the regression analysis. Physical and chemical variables
(DO, temperature, pH, conductivity, turbidity, and suspended chlorophyll) were
measured with a YSI® 6600 multiparameter sonde. Water samples were taken at 3
transects (A, G, K; Figure 3.1) per site for measuring nutrient (Total P, Total N) levels at
each site. Landuse variables (% urban, agricultural, and natural) at the watershed scale
and in a 100m riparian buffer were calculated using GIS data by Wilhelm (2002) and
Wilhelm et al 2005.

These analyses were conducted separately for both the composite samples and the

habitat-specific samples.

Evaluation of the NW-IBI

The NW-IBI was evaluated using several techniques. The first technique was to
randomly choose a subset of sites within each assessment category (“poor”, “fair”,
“good”, “excellent”) and designate them as model sites. I used discriminant function
analysis (DFA) on the model sites with all metrics retained for the final protocol as
descriptors. We then used these functions to assign the remaining test sites to assessment

categories. We also performed DFA on all sites with jackknifmg, where one site is

removed from the list of sites in an iterative fashion to construct the overall model. In

76

both cases, DF A determines the percent of correct classifications, and this can be used as
a means of evaluating the overall model’s efficiency at classifying sites (Legendre and
Legendre 1998).

This method of evaluation, while valuable, could be considered circular because I
selected metrics using all of our study sites. We also evaluated the NW-IBI by plotting
scores from each sites against mean site rankings. These rankings were based on the
CDG, the RDG (Wilhelm 2002), as well as number of transects with large woody debris.
Site rankings were calculated based on these factors individually, and then mean site
ranking was determined (Table 3.24). By comparing these rankings to the NW-IBI
scores for each site, the protocol is evaluated with independent expectations for each site.

In addition to evaluating the NW-IBI scoring system with independent
expectations of site-specific ecological integrity, we also evaluated the NW—IBI’s
sensitivity to anthropogenic influences at different spatial scales (e. g., watershed vs.
riparian vs. overall habitat). We used regression analysis to look at the relationships
between the NW-IBI and the RDG, CDG, and the overall Habitat Index (HI) developed
by Wilhelm (2002). This also helped use evaluate the composite and the habitat-specific
assessment types’ sensitivity to the collection of influences that may impact biological

integrity.

RESULTS
Non-wadeable Macroinvertebrates
We found a wide variety of macroinvertebrates in our study reaches. Overall,

there were a total of 17 non-insects, with gastropods accounting for the most non-insect

77

richness and crustaceans accounting for the highest non-insect abundance. There were a
total of 76 insect families in all of our samples with Diptera, Ephemeroptera, and

Trichoptera making up the majority of insects collected (Table 3.4).

Habitat-Specific Assessment

From our observations in the field, we noticed that the only habitat common to
ahnost all of our study reaches was fine particulate organic matter (FPOM), and there
was very little difference between macroinvertebrate diversity or taxa richness in FPOM
samples taken fi'om a highly impacted river and FPOM samples fi'om rivers that were
fairly unimpacted (Figure 3.2). Cobble habitats were extremely rare in most sites, and
macrophyte habitats varied seasonally. Sand and coarse substrates varied more by
watershed, and would presumably be more influenced by high discharge events (personal
observations). Because large woody debris (LWD) habitats were usually fairly abundant
in most of our reaches (2 sites contained no LWD: sw_sg and tq_nwb) (Table 3.1), we
decided to use LWD as our target habitat for the habitat-specific assessment. Large
woody debris has been shown to be an extremely important habitat for benthic
macroinvertebrates (Hilderbrand et al. 1997; Abbe and Montgomery 1996), and has been
shown to harbor the highest insect diversity in large rivers (Merritt et al. 1996). Samples
taken from LWD indicated that taxa richness and diversity showed relatively little
variance within reaches, yet varied significantly among rivers (Figure 3.2).

This enabled me to construct two types of biological assessments: one using
composite data for all habitats present, and the other using only large woody debris

(LWD) sample data. While a composite assessment is more time consuming due to the

78

increased amount of detritus compared to LWD samples, it may be necessary in rivers
with insufficient amounts of LWD. The advantage to habitat-specific sampling is that, by
focusing on a single habitat, water quality issues are less likely to be masked by the
variation inherent in samples fi'om different habitats (Parsons and Norris 1996).
However, composite assessments may be the only option in some rivers with insufficient

LWD habitat.

Data Reduction: Correlation Analysis

The correlation analysis resulted in a total of 9 metrics being discarded due to
redundancy, and these results hold true for both the composite data and the LWD data
unless otherwise noted. In the group of population-level metrics, E_RICH and T_RlCH
were both discarded because of the high correlation with EPT_RICH (Tables 3.6-3.7).
Of the community-level metrics, PER_E and EPT_DIP were discarded due to their high
correlation with PER_EPT, and DIV was discarded because of its high correlation with
both RICH and PER_DOM. (Tables 3.8-3.9). PER_OLIG was retained for the composite
sample analysis, but discarded for the LWD analysis due to very low numbers of
Oligochaeta in the LWD samples (Table 3.5). The correlation analysis resulted in the
elimination of P_R, C_FPOM, and T_BFPOM due to high correlations with SCR, SHD,
and CF, respectively (Tables 3.10-3.11). The decision to retain or discard certain highly
correlated metrics was based on ease of measurement and best professional judgment.
For example, RICH and DIV were highly correlated with each other, but RICH is much
easier to measure, so it was retained. See Table 3.5 for metric codes and a summary of

metrics retained during the correlation analysis.

79

Data Reduction: Principal Components Analysis

Based on the criteria outlined above, the first 5 PCA axes were further examined
for the composite samples (Figure 3.3). All other axes had eigenvalues less than 1
(Figure 3.3; Table 3.12). Because some metrics had similar loadings on individual axes,
sometimes more than one metric was retained for the final protocol fiom any given axis.
Overall, the PCA analysis for the composite metrics resulted in a total of 8 metrics
retained (Table 3.13). From axis 1, two functional metrics were retained (the Habitat
Stability F F G surrogate and F FG Diversity). Only Percent Trichoptera was retained from
axis 2. The population level metric, EPT Richness, and the community level metric,
Taxa Richness, were both retained from axis 3. Plecoptera Richness and Diptera
Richness were both retained fi'om axis 4, while only Percent Dominance was retained
from axis 5 (Table 3.13). See Table 3.5 for a summary of metrics retained in the PCA
analysis.

Because all metrics did not describe an equal amount of among-site variation,
each was weighted based on the axis fi'om which it was retained. Both the habitat
stability F FG surrogate and FF G diversity were retained from the first PC axis. This axis
described 50% of the overall variation among sites, and so each metric is based on a 25-
point scale. Likewise, % Trichoptera abundance is based on a 20 point scale, EPT taxa
richness an 8 point scale, total taxa richness a 7 point scale, and the remainder are based
on a 5 point scale. The total possible score for the composite NW-IBI is 100 points

(Table 3.15).

80

Principal components analysis resulted in the further examination of axes 1-4
(Figure 3.3; Table 3.12), and a total of 6 metrics were retained for the final protocol for
the LWD assessment (Table 3.14). Percent Dominance, Habitat Stability FFG surrogate,
and FF G Diversity were all retained fi'om axis 1. Percent Trichoptera, Diptera Richness,
and percent Scrapers were retained from axes 2, 3, and 4, respectively, based on their
principal component loadings (Tables 3.14-3. 17).

As in the composite assessments, individual metrics should have weighted
contributions to the overall LWD NW—IBI based on the axis from which it was retained
and that axis’ corresponding contribution to the overall variation among sites. Axis I
described approximately 60% of the overall variation, and since 3 metrics (HAB_STAB,
FFG_DIV, and PER_DOM) were retained from this axis, each are scored on a 20 point
scale. Percent Trichoptera and Diptera taxa richness are both scored on a 15 point scale,
and % scraper abundance is scored on a 10 point scale. Like the composite assessment,
the total possible score for the LWD NW-IBI is 100 points (Tables 3.16—3.l8).

Overall, sites showed no real tendency to cluster based on river basin (e. g., Figure
3.4); however, they showed a slight tendency to cluster based on ecoregion on axis 1
(Figure 3.5a). This is not surprising, due to the condition of sights in the northern
ecoregions (NLP and UP). ANOVA showed an overall difference in RDG (F=5 .560,
df=2, p=0.009) and CDG (F=45.332, df=2, p<0.001) scores by ecoregion Fisher’s LSD
pairwise test showed significantly higher RDG as well as CDG scores for SLP compared
to UP sites (Figure 3.6). This overall pattern of sites clustering by ecoregion was not

observed on subsequent axes (e.g., Figure 3.5b).

81

Metrics Included in the Final Protocol and Stressor-Response Relationships

Defining metric response to stressors is key to effectively managing aquatic
resources. Ideally, individual metrics are responsive to a particular stressor, which is
caused by a particular human behavior. Macroinvertebrates in large rivers are subjected
to a suite of stressors, ranging fiom water chemistry issues (e. g., pH, nutrients), instream
habitat, riparian landuse, and watershed landuse. MLR helped identify specific stressors-
response relationships, and overall, these relationships were relatively weaker for
composite metric-stressor relationships (mean R2=0.47) than LWD metric-stressor

relationships (mean R2=0.64).

Composite Assessment Metrics

Overall, metrics included in the final composite assessment protocol responded to
a wide variety of human influences on several different spatial scales, from watershed
landuse to instream habitat and water chemistry and instream habitat. Several metrics
showed significant negative correlation with percent urban (e.g., HAB_STAB, RICH,
EPT_RICH) and agricultural (e.g., HAB_STAB) landuse in the watershed. Other metrics
showed significant correlations with percent natural landuse in the watershed (e.g.,
HAB_STAB) and in the riparian zone (e.g., PER_DOM). Percent Trichoptera showed
significant correlations with agricultural landuse in the riparian zone. Both EPT_RICH
and P_RICH had significant correlations with instream large woody debris.

Some metrics also correlated with various water quality parameters such as
suspended chlorophyll (FFG_DIV, DIP_RICH), turbidity (F FG_DIV , DIP_RICH),

conductivity (HAB_STAB, FFG_DIV, PER_DOM), and pH (DIP_RICH). Metrics also

82

responded significantly to total phosphorous (FFG_DIV, PER_DOM) as well as total
nitrogen (DIP_RICH). For a detailed of composite metrics’ response to specific

stressors, see Table 3.19 and Appendix 1.

L WD Assessment Metrics

LWD metrics also showed a wide range of factors to which they respond, and it
should be noted that the same metric sometimes showed different responses to stressors
in the LWD assessment compared to the composite style assessment (Table 3.20). Some
were significantly correlated with watershed landuse (e.g., HAB_STAB, PER_DOM,
SCR) and riparian landuse (e.g., PER_DOM, PER_T).

Several LWD metrics were significantly related to water quality parameters such
as pH (FFG_DIV, SCR), dissolved oxygen (PER_DOM), and conductivity (SCR). Like
the composite metrics, some LWD metrics were also significantly correlated with total
nitrogen (HAB_STAB, FF G_DIV, DIP_RICH) and total phosphorous (FFG_DIV, SCR).
For a more details regarding LWD metrics’ response to specific stressors, see Table 3.20

and Appendix 2.

Integration of Biological Metrics: The Non-wadeable Biotic Index (NW-IBI)

I set the scoring criteria to correspond to quartiles of each metric in a 4 bin
scoring system by combining all our sites and calculating the 25‘”, 50'“, and 75th
percentiles for each metric. For example, for a metric scored on a 25 point scale (e.g.,
HAB_STAB in the composite assessment), sites scoring below the 25th percentile

received a score of 0 for that metric. Sites scoring between the 25m and 50“I percentiles

83

received a score of 8. Sites scoring between the 50th and 75'” percentile received a score
of 16, and sites scoring above the 75th percentile received the full 25 points for that
individual metric (Figure 3.7). This was done for all metrics for both the composite and
the LWD assessment types (Tables 3.17-3.18).

The Non-wadeable Index of Biological Integrity (NW-IBI) has a total possible
score of 100 for both composite and LWD assessments. A total score between 0-25
results in a classification of ‘poor’ for that site; a site with a total score ranging between
26-50 is classified as ‘fair’; a site with a total score between 51-75 is classified as ‘good’;
and a site with a total score between 76-100 is classified as ‘excellent’ (Tables 3.14-
3.15).

When the NW-IBI was calculated for all sites, results differed slightly for
composite vs. LWD type assessments. Using the composite method, there were 3 sites
classified as excellent, 11 sites classified as good, 12 sites classified as fair, and 8 sties
classified as poor. Using the LWD samples only, 5 sites were classified as excellent, 10
sites were classified as good, 8 were classified as fair, and 9 as poor (Figure 3.8).
Overall, LWD assessments had more sites classified as poor and excellent compared to
composite methods. Interestingly, however, both types of assessments usually classified

sites similarly in terms of overall ranking (Figures 3.9-3.11).

Evaluation of the NW -IBI
Discriminant function analysis (DFA) was used to evaluate the robustness and
sensitivity of both forms of the NW-IBI. When the overall DFA model was calculated

based on a subset of model sites fiom each classification type (e. g., “poor”, “fair”,

84

“good”, “excellent”) based on the composite assessment methods, 75% of the test sites
were correctly classified as poor, 33% were correctly classified as fair, 67% were
correctly classified as good, and 100% were correctly classified as excellent (Table 3.23).
When jackknifmg was used, 88% of the sites were correctly classified as poor, 50% were
correctly classified as fair, 55% were correctly classified as good, and 67% were
correctly classified as excellent.

When DF A was performed based on the LWD assessments using a subset of
(model) sites to compute the model, 100% of the sites were correctly classified as poor,
25% were correctly classified as fair, 100% were correctly classified as good, and 33% of
test sites were correctly classified as excellent. When jackknif'mg was used, 78% of sites
were correctly classified as poor, 75% as fair, 60% as good, and 50% as excellent (Table
3.23).

When evaluated using independent expectations for site condition (e.g., Mean
Rank), composite assessment scores (Figure 3.12) showed a slightly weaker relationship
to site ranking compared to the LWD assessment scores (Figure 3.13). Both assessment
methods showed significant correlations with the CDG and the Habitat Index (HI),
although the correlation was lower with the composite assessments (Table 3.25) than the
LWD assessment (Table 3.26). In both cases, there was a higher correlation between the

biological community scores and the H] (Tables 3.25-3.26).

85

DISCUSSION
General Discussion of the NW-IBI

Throughout the development of this protocol, we sampled a wide variety of non-
wadeable sites throughout Michigan. These sites were subjected to a variety of
anthropogenic influences, and ranged from reaches that were considered to be in
excellent condition to those in dire need of mitigation. This is extremely important, since
no real reference condition existed for rivers of this size in Michigan due to their large
watershed size and concomitant human influences. These influences invariably affect a
river’s biological constituents (Karr 1997).

Both the composite and the LWD assessment protocols are designed to be
independent of river basin. PCA shows us that site scores do not seem to cluster by river
basin. In other words, metrics do not simply identify watersheds. Each study reach has a
unique signature, and that signature depends on local processes (e. g., anthropogenic
impacts associated with that particular reach) (Figure 3.4). The NW-IBI was also
designed to apply widely throughout the state, independent of ecoregion. While overall
site scores are higher along PC axis 1 (defining functional differences) (Figure 3.5a), this
difference is not seen in subsequent axes (Figure 3.5b). As mentioned earlier, CDG and
RDG scores are significantly lower in the SLP ecoregion compared to the NLP and UP
regions (Figure 3.6). This suggests that NW-IBI scores should naturally be higher
Northern Michigan. Because many of the stressors identified by MLR as having
significant correlations with the two functional metrics (FFG_DIV and HAB_STAB) are
directly correlated with the RDG (e.g., riparian landuse) and the CDG (e. g., watershed

landuse), I believe that the differences among sites were due to differences in actual

86

ecological condition that are results of, at least in part, anthropogenic influences that go
beyond ecoregional differences.

It was my goal to construct a protocol that helped discern human impacts on non-
wadeable rivers in a hierarchical manner, incorporating metrics that describe ecosystem
or fimctional attributes, community composition, and population-level changes. These
different levels of resolution are subject to different spatial and temporal scales of
impacts (Noss 1990). In both forms of the NW-IBI, fiinctional and community level
metrics were weighted higher than population level metrics (Tables 3.17-3.18). Although
this is a result of the principal component analysis, it makes sense fi'om an ecological
standpoint. It is generally agreed upon that higher levels of organization (i.e., functional
or community level) are more reliable than lower levels (i.e., population level) because of
the high degree of background variation to which these lower levels are subjected
(Cottingham and Carpenter 1998).

The Habitat Stability FF G surrogate (Merritt et al. 2002) and F FG Diversity both
provide ways to evaluate the functional, or ecosystem-level, differences among non-
wadeable rivers. While some of this difference is expected to occur naturally (V annote et
al. 1980; Cummins 1988), human landuse practices and their associate stressors have
been shown to affect functional composition of riverine systems (Merritt et al. 2002;
Cummins 1993; Cummins et al. 1989). The fact that these functional metrics retained
from the first principal component axes and were included in both types of assessments
suggests that the overriding differences among sites (and so the overall effect of

stressors) results in primary functional degradation.

87

Metrics included in the final protocol also describe differences among non-
wadeable macroinvertebrates at the community level. Total Taxa Richness, Percent
Dominance and Percent Trichoptera have been shown to vary predictably with human
influence (Ohio EPA 1989; Barbour et al. 1992). High macroinvertebrate taxa richness
generally indicates functional redundancy (e. g., many different scrapers, shredders, etc.) a
diverse food base for fish and other vertebrate predators. Percent dominance is highly
correlated with diversity (Tables 3.8-3.9). Often, if one particular species or group is
particularly dominant in an ecological community, it is because that group is especially
tolerant to the stressor to which the system is subjected (e.g., low dissolved oxygen or
high turbidity). This idea of tolerance is particular to certain groups of
macroinvertebrates. Percent Trichoptera individuals reflect the overall composition of
the community comprised by caddisflies. Caddisflies are generally considered a group
that is intolerant to low DO levels (Hilsenhoff 1988;.lacob and Walther 1981). Many of
the families that dominate non-wadeable rivers in Michigan are also dependant upon firm
substrates like cobble or LWD for attachment (e.g., Brachycentridae and
Hydopsychidae), low turbidity (so as not to clog nets), and are reliant upon healthy
upstream processing of coarse organic matter as a food base (V annote et al. 1980).

The number of Ephemeroptera, Plecoptera, and Trichoptera (EPT) families
describes population level differences among sites, and is a metric in many IBIs (Karr
1987; Kerans and Karr 1994; Jacob and Walther 1981; Hayashi 1989; Cairns 1990;
Barbour et al. 1992; Marshall 1993; 1993; Pinel-Alloul et al. 1996; [MDNR] Michigan
Department of Natural Resources 1991). The number of EPT families has been shown to

decrease due to nutrient enrichment from agricultural landuse (resulting in low DO), and

88

turbidity. They are also considered desirable because of their prevalence in the drift,
providing certain fish species with an adequate food supply. Many EPT families,
especially Plecoptera families, require clean, firm substrate free of sedimentation because
of trophic (e.g., scraper mayflies) or respiratory constraints. Diptera Richness is another
population level metric, which is included in both protocols, has a negative relationship
with anthropogenic impacts. In highly-impacted streams, often the only Diptera group
present is Chironomidae, which are known overall to be highly tolerant to low DO levels
(Hilsenhoff 1988), which is often related to intensive agricultural landuse and
sedimentation of fine organic matter. In systems without these influences, Diptera
Richness tends to be higher, and often several functional groups are representedby the
Diptera community (Table 3.4).

It is clear that individually, these metrics describe ecologically meaningful
differences among sites based on functional, community, and population level attributes.
When taken together as a single index, the NW-IBI, they will help describe the ecological
condition of non-wadeable rivers at the reach scale. For example, the lowest scoring site
scored with composite metrics, Saginaw River @ Zilwaukee (sg_zilOl, NW-IBI
composite score=0) (Figure 3.9), is subjected to a variety of human influences, and
scored zero for all metrics. This reach of the Saginaw River is used extensively for
shipping, is dredged, and its riparian vegetation is almost completely altered (Figure
3.14). This site also scored “poor” in the LWD assessment (Figure 3.10).

An example of a site scoring “fair” in the composite assessment is the Grand
River @ Ionia (Figure 3.9). At first glance, this site looks like it’s in good shape, but

upon further examination, one sees that the riparian buffer, is very narrow (Figure 3.15),

89

with intensive agriculture dominating the overall landuse for the watershed (AgWS=63%;
Table 3.20). This is an example of a site with “fair” scores in almost every individual
metric, suggesting a moderate amount of human impact at all spatial scales.

Interestingly, the LWD assessment classified this site as “poor” in both years. This is
likely due to the heavy siltation on much of the LWD habitat, which would cause the
LWD NW-IBI score to be much lower than the composite score, which masks some of
the water quality issues because it is dependent on overall habitat quality.

The Manistee River @ High Bridge scored “good” in both 2001 and 2002 when
scored with the composite method (Figure 3.9). This site is characterized by an intact
riparian zone, clear water, and a variety of macroinvertebrate habitat, including a hrge
amount of LWD (Figure 3.16). This particular site scored low on the habitat stability
F FG metric, and this is likely due to embeddedness of its normally coarse and sandy
substrate (Figure 3.16). In 2001, this site also scored “good” in the LWD assessment, but
in 2002, it scored “excellent” (Figure 3.10). This suggests that LWD habitat improved in
some way in the second year this site was evaluated. In the summer of 2002, flows were
relatively higher because of much more precipitation (personal observation), which could
have acted to scour the LWD habitat of its fine layer of sediment, subsequently making
this habitat more favorable to a variety of macroinvertebrate groups. This is reflected
mainly in a higher score for the habitat stability FF G surrogate (Figure 3.10).
Alternatively, the LWD assessment is simply more sensitive to water quality parameters,
and these parameters vary to a greater degree temporally than landuse or habitat

parameters.

90

There were only three sites classified as “excellent” by the composite assessment.
The site that scored the highest was the Manistee River @ Coates Rd (Figure 3.9). This
study site scored the highest possible for almost every individual metric in the composite
assessment. In general, the Manistee River is a fairly unimpacted river, but this particular
site had exceptional habitat quality (macrophytes, LWD, and coarse substrates with very
little FPOM). Overall, this site had much hydrologic variability, with slow-moving areas
dominated by macrophyte beds, as well as deeper, faster areas with cobble and clean
LWD (Figure 3.17). This site also scored “excellent” in the LWD assessment (Figure
3.10).

In general, the failure of the NW—IBI to classify one site similarly fi'om year to
year or by composite versus LWD can be helpful when diagnosing ecological condition
or a particular study reach. For example, incongruities between composite assessments
from year to year may help diagnose effects of drought (low flows). Inconsistencies
between composite and LWD assessments in the same year (especially when combined
with the HI) may help diagnose problems with specific habitats (e.g., embeddedness).
However, the variation in scores may simply be a result of the great degree of natural
variation in these systems, which is a primary reason an IBI for non-wadeable rivers is
such a challenge. Alternatively, these incongruities associated with variation in NW-IBI
scores from one year to another could simply be a result of real differences in water

quality parameters fi'om year-to-year (see Chapter 2).

91

Metric Response to Stressors

The technique I used to discern relationships between metrics and the stressors to
which they respond needs fiirther work. While the technique of establishing metric-
stressor relationships after metric selection is often used, this could presumably result in
losing some precision in the overall NW-IBI as well as with individual metrics. This is
because metrics that were not selected for the overall index may have had higher overall
correlations with individual stressors. For example, in a correlation analysis, percent
Chironomidae had a the highest correlation of any of the original 26 biological attributes
with urban landuse in the watershed. Metrics with similarly high correlations with this
landuse measure include FFG_DIV, but it could be argued that the methods I used
resulted in the loss of some precision. However, the results of PCA did not show
significant variation among-site variation in metrics that had high correlations with any
single stressor or human behavior, and this could be due to the complex interaction
among multiple stressors and overall river quality. This is an issue that needs further
examination.

Using regression techniques to quantify metric-stressor relationships, while
encouraged (U .S.EPA 2000), should be done with caution. Aside fi'om defining
correlations instead of actual causes, there are other pitfalls in the way that we examined
relationships with metrics and environmental parameters. First of all, the suite of
possible stressors is far fi'om complete (I ables 2.1-2.2). For example, we made no
measure of sediment toxicity (metals, PCBs, pesticides/herbicides), and this is commonly
a problem in large rivers, especially those draining urban watersheds (Smith et al. 1987;

Young and Huryn 1999). Also, it is important to note the difference between stressors

92

and the human activities that influence the stressor. For example, a common stressor in
lotic ecosystems is low DO. In order to manage this stressor, the human activities
causing it must be identified, and in the case of DO, these activities may include clearing
riparian buffer strips in agricultural areas, thus allowing nutrient enrichment of the
waterway. While we included both landuse practices as well as the stressors caused by
those practices in the MLR models for both types of metrics, technically, this is not
appropriate.

Despite this, including the activities (e.g., landuse) that cause stressors to increase
(e. g., water quality parameters) allows us to evaluate metric sensitivity to scale. It
provides us insight into the many factors influencing macroinvertebrates in large rivers.
Very few of the metrics in either protocol have only the actual stressor included in its
MLR model—they also contain some sort of measure of landuse (Tables 319-3.20).
This is likely because of the many ways a particular landuse affects different stressor
levels. For example, a high percentage of urban landuse in the riparian zone would cause
a decrease in riparian filtering capacity, which would cause elevated turbidity fi'om road
run-off (also causing deposition of other pollutants), decreased LWD habitat, increased
embeddedness, and may even increase suspended chlorophyll by reducing shading
(although this would be less of a factor in large rivers). Intensive agricultural landuse,
whether in the riparian zone or the entire watershed, could cause increased nutrient
levels, turbidity, and pesticide run-off, resulting in reduced levels of DO, among other
things.

In summary, non-wadeable rivers are subjected to a wide range of stressors.

Some of these originate upstream, some are a product of watershed-scale activities, some

93

are caused by riparian landuse practices, and some are a result of channel modification
(e.g., dredging and impoundment). To compound this, human activities at different
scales may result in the same type of stressors, and macroinvertebrates integrate the
effects of these stressors regardless of scale. For these reasons, it is extremely difficult to
establish extremely strong relationships between any one stressor and metric values in
non-wadeable rivers. Inferring stressor-response relationships in large rivers should be
undertaken case-by-case, using the guidelines in Appendix 1-2, but relying heavily on

professional judgment.

Evaluation of the NW -IBI

When dealing with variable systems such as non-wadeable rivers, it is important
to be aware of where this variation arises, what it means, and how each type of
assessment can be affected by it. The discriminant function analysis (DF A) can tell us a
little about each assessment type’s ability to classify sites consistently. For the test sites
only, DFA does a better job of classifying extremely poor sites and good (LWD) to
excellent (composite) sites (Table 3.23). In the jackknifed DF A, % correct classification
is increased overall, again suggesting the metrics chosen do a reasonably good job at
detecting among-site differences (Table 3.23). However, the cut-offs in scores for each
classification type are somewhat arbitrary—a site scoring 49 is classified as “fair”, while
a site scoring 51 is classified as “good”. Some of the misclassification in both DF A
models is likely due to this factor. These categories, while useful as narrative

descriptions of overall ecological integrity, should nonetheless be used with caution.

94

Regression analysis with NW-IBI scores and Mean Rank also show a general
trend that is expected—as a site’s overall ranking decreases, so does its NW-IBI score
(Figures 3.11-3.12). The low correlation coefficients for both assessment types, while
initially troubling, should be expected because the factors used to compute Mean Rank
(e. g., landuse and large woody debris), while useful for setting independent expectations
for sites, summarize a relatively small amount of expected variability among sites. For
example, no water quality parameters or substrate composition data were included in the
calculation of Mean Rank.

As far as the NW-IBI’s sensitivity to scale-specific factors, both types of
assessment showed significant correlations with the CDG, Mean Rank, as well as the
Habitat Index (HI) (Wilhelm 2002). The correlation with the HI was the highest (Tables
325-326). This was not surprising, because the HI incorporates differences among sites
at many different spatial scales, from the entire watershed to the littoral fringe habitats
sampled for macro invertebrates. It is widely acknowledged that one of the main
advantages to biomonitoring is that biological communities integrate synergistic effects
of the multiple scales of influence that humans have on ecosystems e.g., (Fausch et al.
1984; Karr 1997; Rosenberg and Resh 1993). Aquatic organisms are affected by landuse
within the entire watershed, local (riparian) landuse, upstream processes, water quality,

and instream habitat quality.

Comparing Composite and LWD Methods of Assessment

The evaluation of both the composite and LWD protocols must additionally be

addressed in terms of differences between each type of assessment: composite vs. LWD.

95

Upon immediate examination, it may seem preferable to simply conduct the LWD
assessment. It is quicker due to shorter sample processing times, it seems to be more
sensitive, and it is more independent of overall habitat quality. However, LWD
assessments may not always be the way to proceed, and this depends largely on the
overriding goal of the assessment. One must consider both the robustness and the
sensitivity of each assessment method.

For example, the composite assessment was relatively more robust than the LWD
assessment at correctly classifying sites into four broad categories (Table 3.23).
However, when one compares this with the regression of NW-IBI scores and Mean Rank,
we see that LWD assessments are more sensitive to differences among sites (Figures
3.1 1-3. 12). The composite assessment seems to be more effective at incorporating
overall variation among sites and consistently evaluating them, while the LWD
assessment is more sensitive to variation, presumably because LWD harbors a more
stable macro invertebrate community in general. Put another way, the composite
assessment is more robust to small changes in the environment, and the LWD assessment
is more adept at detecting these changes.

This has wide ranging management implications. For instance, if the goal of the
assessment is trend monitoring, the composite method may be preferable. Composite
assessment tends to be more robust to overall change, and would probably not detect
subtle changes in water quality that may occur on a relatively short temporal scale.
However, if the goal is to compare two sites within a short time fi'ame, LWD assessments
may be more favorable because this type of assessment would be more sensitive to subtle

differences between sites.

96

It should be noted that LWD assessments are not always possible because some
large rivers in Michigan do not have sufficient LWD habitat to conduct such an
assessment. Figure 3.18 shows the number of new taxa (families) acquired from each
transect in a 2000m reach from a subset of LWD samples. After 8 transects were
sampled, on average, no new taxa were collected (Figure 3.18). While this may not
substantially affect some metrics (e.g., those involving relative abundance), there are
clear implications for richness metrics (e.g., EPT Richness, Total Richness). Therefore, it
is our recommendation that unless there is LWD habitat at no less than 8 of the 11
transects in a study reach (Figure 3.1), only composite assessments can be relied upon
(Figure 3.19). Unfortunately, this means that only sites with at least 8 transects
containing large woody debris can reliably assessed using the habitat specific methods,
biasing the LWD assessment toward only relatively high-quality sites. General
procedures for both types of assessment, along with materials needed to conduct the

assessment, are summarized in Appendix 3.

CONCLUSION

While the approach outlined above provides general methods for the development
of ecological assessments that may be applied to other systems and assemblages, it is
important to note the limitations of this approach. The problem with developing a
protocol using the methods outlined above lies with the identification of metric-stressor
relationships. While there are many different approaches used to develop biological
evaluations including the examination of rare species based on predictive models

(Hawkins, Norris et al. 2000) and the use of functional feeding group surrogates for

97

ecosystem processes (Merritt et al. 2002), recent research points to the importance of
developing metrics that describe valued ecological attributes (Ohio EPA 1989) and
specific risk management strategies (Stevenson 1998). This invariably requires precise
metric-stressor relationships to be defined. However, due to the relatively low number of
non-wadeable rivers in Michigan, the complex nature of non-wadeable rivers, and the
ways in which multiple stressors interact at different spatial scales (see Chapter 2), it is
difficult to define single stressor-response relationships with the methods outlined in this
chapter. This is the main weakness of my approach, and will require further examination.

The metrics selected for both NW-IBI assessment types have well-documented
relationships to anthropogenic influences (Ohio EPA 1989; Karr and Chu 1999). While
assessment of non-wadeable rivers will require more field as well as laboratory work than
many of the currently used wadeable protocols, I believe that the NW-IBI, when used in
conjunction with the Habitat Index (Wilhelm 2002; Wilhelm et al. 2005), provides an
objective means of assessing the biological integrity of non-wadeable rivers in Michigan
at the site scale.

The overall procedures outlined for assessing non-wadeable rivers in Michigan
(Appendix 3) can be completed by a field team of 2 individuals in approximately 1 day,
and depending on which type of assessment is chosen, meroinvertebrate processing may
be conducted on site. The decision as to which type of assessment to use depends on the
overall goal of the assessment and the number of transects with LWD.

The NW-IBI will undoubtedly require periodic fine-tuning and adjustment as
additional data and experience arise. However, it appears to be a robust and sensitive

means of evaluating the biotic integrity of Michigan’s non-wadeable rivers. Future

98

research should be directed at refining metric selection criteria so as to enable the precise
evaluation of stressors based on biological attributes, incorporating a fish procedure into
the overall non-wadeable river assessment protocol. This will add a unique spatio-
temporal dimension to the protocol and help MDEQ better-evaluate the ecological health

of Michigan’s non-wadeable river systems.

99

LITERATURE CITED

Abbe,T.B. and D.R.Montgomery, 1996. Large woody debris jams, channel hydraulics
and habitat formation in large rivers. Regulated Rivers-Research & Management
12: 201 -221.

Barbour, M. T., Plafltin, J. L., Bradley, B. P., Graves, C. G., and Wisseman, R. W. 1989.
Evaluation of EPA's rapid bioassessment benthic metrics: Metric redundancy and
variability among reference stream sites. 1992. Toronto, Ont. (Canada).
Symposium on Community Metrics to Detect Ecosystem Effects.

Bayley,P.B., 1995. Understanding large river floodplain ecosystems. Bioscience 45: 153-
158.

Cairns,J.J., 1990. The genesis of biomonitoring in aquatic ecosystems. Environ. Profess.
12: 169-176.

Cairns,J.J. and J.R.Pratt, 1993. A history of biological monitoring using benthic
macroinvertebrates. In: Rosenberg,D.M. and V.H.Resh (eds), Freshwater
Biomonitoring and Benthic Macroinvertebrates, Chapman and Hall, New York,
pp. 10-27.

Cottingham,K.L. and S.R.Carpenter, 1998. Population, Community, and Ecosystem
Variates as Ecological Indicators: Phytoplankton Responses to Whole-Lake
Enrichment. Ecological Applications 8: 508-530.

Cummins, K.W., 1988. The Study of Stream Ecosystems: A Functional View. In:
Pomeroy,L.R. and J.J.Alberts (eds), Concepts of Ecosystem Ecology: A
Comparative View, Springer-Verlag, New York, pp. 247-261.

Cummins,K.W., 1993. Bioassessment and analysis of functional organization of running
water ecosystems, Lewis Publishers, Boca Raton, FL (USA).

Cummins,K.W., M.A.Wilzbach, D.M.Gates, J.B.Perry, and W.B.Taliaferro, 1989.
Shredders and riparian vegetation. Bioscience 39: 24-30.

100

Fausch,KD., J.R.Karr, and P.RYant, 1984. Regional application of an index of biotic
integrity based on stream fish communities. Transactions of the American
Fisheries Society 113: 39-55.

Gore,J.A. and F.D.Shields, 1995. Can Large Rivers Be Restored? Bioscience 45:142-152.

Hawkins, C. P., R H. Norris, et al. (2000). Development and evaluation of predictive
, models for measuring the biological integrity of streams. Ecological Applications
10(5): 1456-1477.

Hayashi,F., 1989. Respiratory responses of aquatic insects to low oxygen concentration
Teisansonodo ni taisuru suisei konchu no kokyu hanno. Jap. J. Lirnnol.
/Rikusuizatsu. 50: 255-268

Hilderbrand,R.H., A.D.Lemly, C.A.Dolloff, and K.L.Harpster, 1997. Effects of large
woody debris placement on stream channels and benthic macroinvertebrates.
Canadian Jourrml of Fisheries and Aquatic Sciences 54: 931-939.

Hilsenhoff,W.L., 1988. Rapid field assessment of organic pollution with a family-level
biotic index. Journal of the North American Benthological Society 7: 65-68.

Jacob,U. and H.Walther, 1981. Aquatic Insect Larvae as Indicators of Limiting Minimal
Contents of Dissolved Oxygen. Aquatic Insects 3: 219-224.

Johnson,B.L., W.B.Richardson, and T.J.Naimo, 1995. Past, Present, and Future Concepts
in Large River Ecology: How rivers function and how human activities influence
river processes. Bioscience 45: 134-141.

Junk, W. J., Bayley, P. B., and Sparks, R. E. The flood pulse concept in river-floodplain
ecosystems. Dodge, D. P. 106, 110-127. 1989. Proceedings of the International
Large River Symposium.

Karr,J.R., 1987. Biological monitoring and environmental assessment: A conceptual
fiamework. Environmental Management 11: 249-256.

Karr,J.R., 1993. Biological monitoring: Challenges for the firture. Lewis Publishers,
Boca Raton, FL (USA).

101

Karr,J.R., 1997. The future is now: Biological monitoring to ensure healthy waters.
Northwest Science [Northwest Sci. ] 71: 254—257.

Karr,J.R. and E.W.Chu, 1999. Restoring Life in Running Waters, Island Press,
Washington, DC.

Kerans,B.L. and J .RKarr, 1994. A benthic index of biotic integrity (B-IBI) for rivers of
the Tennessee Valley. Ecological Applications [Ecol Appl. ] 4: 768-785.

Klemm,D.J., J.M.Lazorchak, and D.V.Peck, 1999. Benthic Macroinvertebrates. In:
Lazorchak,J.M., B.H.Hill, D.KAverill, D.V.Peck, and D.J.Klemm (eds), Field
operations and methods for measuring the ecological condition of non-wadeable
rivers and streams, United States Environmental Protection Agency, Washington,
DC, pp. 133-150.

Lazorchak,J.M., B.H.Hill, D.K.Averill, D.V.Peck, and D.J.Klemm, 2000. Environmental
Monitoring and Assessment Program -Surface Waters: Field Operations and
Methods for Measuring the Ecological Condition of Non-wadeable Rivers and
Streams, US. Environmental Protection Agency, Cincinnati, OH.

Legendre,P. and L.Legendre, 1998. Numerical Ecology, Elsevier, Amsterdam.

[MDNR] Michigan Department of Natural Resources. Qualitative Biological and Habitat
Survey Protocols for Wadable Streams and Rivers: Great Lakes and
Environemental Assessment Section (GLEAS) Procedure 51. 1991. Lansing, MI,
Michgian Department of Natural Resources.

MarshalLKE., 1993. The literature of biomonitoring, Chapman and Hall, New York
(USA).

Merritt, RW; Wallace, JR; Higgins, MJ; Alexander, MK; Berg, MB; Morgan, WT;
Cummins, KW; Vandeneeden, B. 1996. Procedures for the functional analysis of
invertebrate communities in the Kissimmee River-floodplain ecosystem. Florida
Scientist 59(4):216—274.

Merritt,R.W., K.W.Cummins, M.B.Berg, J.A.Novak, M.J.Higgins, K.J.Wessell, and
J .L.Lessard, 2002. Development and application of a macroinvertebrate
functioml-group approach in the bioassessment of remnant river oxbows in

102

southwest Florida. Journal of the North American Benthological Society 21: 290-
310.

Noss,RF., 1990. Indicators for Monitoring Biodiversity: A Hierarchical Approach.
Conservation Biology 4: 355-364.

Ohio EPA. Biological criteria for the protection of aquatic life. Vol. III. Standardized
field sampling and laboratory methods for assessing fish and macroinvertebrate
communities. 1989. Columbus, OH, Ohio EPA, Division of Water Quality
Monitoring and Assessment.

Parsons,M. and R.H.Norris, 1996. The effect of habitat-specific sampling on biological
assessment of water quality using a predictive model. Freshwater Biology 36:
419-434.

Paton,A., 1979. The harnessing of large rivers. pg1-13 In: Tidal power and estuary
management., Scientechnica; Bristol (UK) Colston Papers.

Pinel-Alloul,B., G.Methot, L.Lapierre, and A.Willsie, 1996. Macroinvertebrate
community as a biological indicator of ecological and toxicological factors in
Lake Saint-Francois (Quebec). Environmental Pollution 91: 65-87.

Resh, V.H. and J .K. Jackson. 1993. Freshwater biomonitoring and benthic
macroinvertebrates, Chapman and Hall, New York (USA).

Rosenberg,D.M. and V.H.Resh, 1993. Introduction to fieshwater biomonitoring and
benthic macroinvertebrates. In: Rosenberg,D.M. and V.H.Resh (eds), Freshwater
Biomonitoring and Benthic Macro invertebrates, Chapman and Hall, New York,

pp. 1-9.

Sedell, J. R., Richey, J. E., and Swanson, F. J. The river continuum concept: A basis for
the expected ecosystem behavior of very large rivers? 1989. Proceedings of the
International Large River Symposium

Smith,RA., RB.Alexander, and M.G.Wolman, 1987. Water quality trends in the nation's
rivers. Science 235: 1607-1615.

Sparks,RE., 1995. Need for Ecosystem Management of Large Rivers and Their
Floodplains. Bioscience 45: 168-182.

103

Sparks,R.E., P.B.Bayley, S.L.Kohler, and L.L.Osborne, 1990. Disturbance and recovery
of large floodplain rivers. Environmental Management 14: 699-709.

Stalnaker,C.B., RT.Milhous, and KD.Bovee, 1989. Hydrology and hydraulics applied to
fishery management in large rivers.

Stevenson, R. J. 1998. Diatom indicators of stream and wetland stressors in a risk
management framework. Environmental Monitoring and Assessment. 51: 1-2.

Stevenson,RJ. and J.Smol, 2001. Use of Algae in Environmental Assessments. In:
Wehr,J.D. and R.G.Sheath (eds), Freshwater Algae in North America:

Classification and Ecology, Academic Press, San Diego.

U.S.EPA. Stressor Identification Guidance Document. EPA-822-B-00-025. 2000.
Washington, DC.

Vannote,R.L., G.W.Minshall, K.W.Cummins, J.RSedell, and C.E.Cushing, 1980. The
river continuum concept. Can. J. Fish. Aquat. Sci. 37: 130-137.

Ward,J.V. and J.AStanford, 1983. The Serial Discontinuity Concept of Lotic
Ecosystems. In: Fontaine,T.D. and S.M.Bartell (eds), Dynamics of Lotic
Ecosystems, pp. 29-42.

Wilhelm, J .G.O. 2002. A Habitat Rating System for Non-wadeable Rivers of Michigan.
Master’s Thesis. University of Michigan, Ann Arbor MI

Wilhelm, J.G.O., J.D. Allan, K.J. Wessell, R.W. Merritt, and K.W. Cummins. 2005 (in
press). Habitat evaluation of non-wadeable rivers. Ecological Applications

Young,R.G. and A.D.Huryn, 1999. Effects of land use on stream metabolism and organic
matter turnover. Ecological Applications 9: 1359-1376.

104

TABLES

105

Table 3.1. List of sites sampled for macroinvertebrates.

 

 

 

Stream and site name Code Name Date Stream and site name Code Name Date
Sampled Sampled
Au Sable @ Whirlpool as_whpOI 08/ 14/01 Au Sable @ Mouth as_mth02 07/30/02
Grand @Comstock gd_cerI 07/18/01 Au Sable @ Whirlpool as_whp02 07/31/02
0111511565333 Rapids gd _gr'01 07/19/01 Grand @ Comstock gd_cmr02 06/13/02
Grand @ Ionia gd_ion01 06/27/01 Grlzlnxdegcgrand Haven gd _gi102 06/19/02
Grand @ Johnson gd_jon01 06/28/01 prand @ Grand Ledge gd_gld02 06/12/02
Kalamazoo @ Custer kz_cus01 06/19/01 Grand @ Grand Rapids gd_gr02 06/18/02
Kalamazoo @ Verberg kz_verOl 06/21/01 Grand @ Ionia gd_ion02 06/11/02
Manistee @ High Bridge ma_hbr01 08/02/01 Grand @ Johnson gd_jon02 06/18/02
Manistee @ Rainbow ma_rbw01 08/01/01 arristee @ Coates ma_cts02 07/23/02
Muskegon @ Thomapple mk_thp01 07/11/01 anistee @ Higr Bridge ma_hbr02 07/24/02
Muskegon @ Truckey mk_trkOl 07/10/01 anistee @ Manistee rm__mns02 07/25/02
Raisin @ Dundee ra_dunOI 07/04/01 enominee @ Koss me_kss02 07/09/02
Raisin @ Monroe ra_mon01 07/03/01 enominee @ Sturgeon me_stb02 07/10/02
Saginaw @ Zilwaukee sg_zilOl 07/31/01 Egon @ Big Rapids mk_br02 06/25/02
Shiawassee @ Saginaw sw_ngl 07/26/01 uskegon @ Newaygo mk_nwg02 06/26/02
Tittabawassee @ Saginaw tb_ngl 08/07/01 St. Joseph @ Mottville sj_mv102 07/02/02
St. Joseph @ Riverview sj_rvw02 07/18/02
Tahquamenon @ tq_nwb02 07/12/02
Newberry
Tahquamenon @ Paradise tq_pds02 07/13/02

 

 

106

Table 3.2. Catchment disturbance gradient (CDG) scores for each sample site. The
seven individual measures were scored 04 with zero indicative of a natural system and 4
suggestive of a highly disturbed system. The scores for each metric were summed to give
a total CDG score. See Table 3.1 for site codes. (Modified from Wilhelm 2002)

 

Total
Site % Urban, % Ag, % Urban, % Ag, Dam NPDES Road CDG
Code Buffer Buffer Basin Basin Density Density Density Score

 

tq_pds 0 0 0 0 0 0 0 0
ma_cts 0 0 0 0 l 0 0 1
ma_hbr 0 0 0 0 1 0 0 I
ma_rbw 1 0 0 0 1 0 0 2
me_stb 0 0 0 0 3 1 0 3
mk_thp 0 0 1 l l l l 3
tq_nwb 0 0 0 0 3 0 l 3
as_whp 0 0 1 0 3 0 0 4
me_kss 2 0 0 0 3 l 0 5
mk_br 1 1 1 1 1 1 1 5
ma_mns 4 0 1 O 1 0 1 6
mk_nwg l 2 l l 1 l 2 ‘ 6
as_mth 3 O l 0 3 0 0 7
kz_cus O 0 3 3 1 3 3 7
mk_trk l 4 1 1 1 1 1 8
gd_gh l 0 3 3 2 3 3 9
gd_jon 1 0 3 3 2 3 3 9
kz_alg 0 0 3 2 4 3 3 9
sw_sg 0 0 3 3 3 3 3 9
tb_sg 2 1 2 l 3 2 1 9
gd_ion 0 3 3 3 l 3 2 10
sg_zil 2 0 3 2 3 2 2 10
gd_gld 3 0 4 2 2 4 4 1 I
gd_gr 3 0 3 3 2 3 3 1 1
kz_con 4 0 3 3 1 3 3 1 1
ra_mon 3 O 2 4 2 3 3 1 I
sj_mvl 2 1 2 4 2 2 2 I I
gd_cmr 4 0 3 3 2 3 3 12
gd__low 1 4 3 3 1 2 3 12
sj_rvw 3 1 2 4 2 2 3 12
kz_ver 4 O 4 2 3 3 4 I3
ra_dun 3 2 2 4 2 3 3 l3

 

107

Table 3.3. Total riparian disturbance gradient (RDG) score for each sample site based on
the number of gaps in the riparian area and the mean riparian width. The two metrics
were scored on a scale from 0-4 and were summed to yield a total score. A low number
indicates a natural site, while a high number indicates a highly disturbed site. See Table
3.1 for site codes (Modified from Wilhelm 2002).

 

Site Code Number of Gaps Riparian Width Total RDG Score
as_whp O O
ma_cts
me_stb
sw_sg
tq_nwb
tq_pds
ma_hbr
mk_nwg
gd_low
kz_cus
me_kss
gd_gh
gd__ion
kz_alg
mk_br
mk_trk
gd_jon
ma_rbw
mk_thp
sj_mvl
tb_sg
gd_gld
sg_zil
ra_dun
ra_mon
as_mth
gd_cmr
gd_gr
kz_ver
kz_con
ma_mns
sj_rvw

 

AA-‘kDJUJWADJNNNNI—NN—‘fi-‘ON—‘H—‘OOOOOOOOO
Ab-h-hAAWMAWWNWNNWNw—‘NN—NNHHOOOOO
OOOOOONQNQONONMMA-h-h-hhwwwMWNNNH—‘OOOOOO

 

108

Table 3.4. List of macroinvertebrates found in Michigan non-wadeable rivers. Note that
functional group assignments are at the family level and may vary within each family.

 

 

Functional
Order (or other higer taxon) Family Feeding Group
Non-Insects
Bivalvia Sphaeriidae CF
Bivalvia Corbiculidae CF
Bivalvia Dreissenidae CF
Bivalvia Unionidae CF
Gastropoda Ancylidae Sc
Gastropoda Hydrobiidae Sc
Gastropoda Lymnaeidae Sc
Gastropoda Physidae Sc
Gastropoda Planorbidae Sc
Gastropoda Pleuroceridae Sc
Gastropoda Pomtiopsidae Sc
Gastropoda Valvatidae Sc
Gastropoda Viviparidae Sc
Crustacea Amphipoda Sh
Crustacea Argulidae P
Crustacea Decapoda CG
Crustacea Isopoda Sh
Insecta
Coleoptera Carabidae P
Coleoptera Chrysomelidae Sh
Coleoptera Curculionidae Sh
Coleoptera Dytiscidae P
Coleoptera Elrnidae CG
Coleoptera Gyrinidae P
Coleoptera Haliplidae Pi
Coleoptera Hydrophilidae P
Coleoptera Noteridae P
Coleoptera Psephenidae Sc
Coleoptera Scirtidae Sc
Diptera Athericidae P
Diptera Ceratopogonidae P
Diptera Chaoboridae P
Diptera Chironomidae CG
Diptera Culicidae CF
Diptera Dolichopodidae P
Diptera Empididae P
Diptera Ephydridae Sh
Diptera Muscidae P

 

109

Table 3.4 (continued)

 

 

Functional
Order (or other higher taxon) Family Feeding Group
Diptera Simuliidae CF
Diptera Stratiomyidae CG
Diptera Tabanidae P
Diptera Tipulidae CG
Ephemeroptera Baetidae CG
Ephemeroptera Baetiscidae CG
Ephemeroptera Caenidae CG
Ephemeroptera Ephemerellidae Sc
Ephemeroptera Ephemeridae CG
Ephemeroptera Heptageniidae Sc
Ephemeroptera Isonychiidae CF
Ephemeroptera Leptophlebiidae CG
Ephemeroptera Polymitarcyidae CG
Ephemeroptera Potomanthidae CF
Ephemeroptera Trichorythidae CG
Hemiptera Belostomatidae P
Hemiptera Corixidae
Hemiptera Gerridae P
Hemiptera Hebridae P
Hemiptera Mesoveliidae P
Hemiptera Nepidae P
Hemiptera Notonectidae P
Hemiptera Pleidae P
Hemiptera Saldidae P
Hemiptera Veliidae P
Hirudinea Hirudinea P
Hymenoptera Mymaridae P
Lepidoptera Nepticulidae Sh
Lepidoptera Pyralidae Sh
Megaloptera Corydalidae P
Megaloptera Sialidae P
Megaloptera Sisyridae P
Odonata Aeshnidae P
Odonata Calopterygidae P
Odonata Coenagrionidae P
Odonata Corduliidae P
Odonata Gomphidae P
Odonata Lestidae P

 

110

Table 3.4 (continued).

 

 

Functional
Order (or other higher taxon) Family Feeding Group
Odonata Libellulidae P
Oligocheata Oligochaeta CG
Plecoptera Perlidae P
Plecoptera Pteronarcyidae Sh
Trichoptera Brachycentridae CF
Trichoptera Dipseudopsidae CF
Trichoptera Glossosomatidae Sc
Trichoptera Helicopsychidae Sc
Trichoptera Hydropsychidae CF
Trichoptera Hydroptilidae Sc
Trichoptera Lepidostomatidae Sh
Trichoptera Leptoceridae Sh
Trichoptera Limnephilidae Sh
Trichoptera Philopotamidae CF
Trichoptera Phryganeidae Sh
Trichoptera Psychomyiidae CG
Trichoptera Uenoidae Sc

Trichoptera Polycentropodidae P

111

Table 3.5. Potential biological attributes to be used as metrics in the final protocol.
Potential metrics are categorized based on whether they are population, community, or
functional attributes. Taxonomic resolution is to the family level. Asterisks indicate
metrics retained for the composite assessment but not for the LWD assessment
(EPT_RICH and P_RICH). Italics indicate metrics retained for the LWD assessment but

 

 

 

 

 

not for the composite (SCR).
A’I'I‘RIBUTE Code Expected Status
Response Corr PCA
Population Level
Ephemeroptera Richness E_RICH - Discar' ded Discarded
Plecoptera Richness P_RICH - Retained Retained“
Trichoptera Richness T_RICH - Discarded -
EPT Richness EPT_RICH - Retained Retained“
Diptera Richness DIP_RICH - Retained Retained
Community Level
% Ephemeroptera PER_E - Discarded -
% Plecoptera PER_P - Retained Discarded
% Trichoptera PER_T - Retained Retained
% EPT PER_EPT - Retained Discarded
% Diptera PER_DIP + Retained Discarded
% Chironomidae PER_CI-IIR + Discarded -
% Oligochaeta PER_OLIG + Retained Discarded
Taxa Richness RICH - Retained Retained
Shannon Diversity DIV - Discarded -
% Dominance PER_DOM + Retained Retained
EPT/EPT+DIP EPT_DIP - Discarded -
Functional Group Metrics
or Surrogates
% Shredders SHD O Retained Discarded
% Scrapers SCR 0 Retained Retained
% Collector F ilterers CF 0 Retained Discarded
% Collector Gatherers CG 0 Retained Discarded
% Predators PRED O Retained Discarded
F FG Diversity FF G_DIV - Retained Retained
Habitat Stability F FG* HAB_STAB - Retained Retained
P/R F FG P_R 0 Discarded -
CPOM:F POM FFG C_F POM Discarded -
Transport:Benthic F POM T_BFPOM Discarded -

 

112

 

 

83 82 82 Rod- 82 5:95
83 8»... 82 as... moaIEm
83 :3 84.8 52%
83 $3 moan”
I I I I 83 :03 m
moi an :02 .Em 52 .8 mom. a moEIm

 

.m.m 033. E 38: 2m 8600 .36. fig 05 9 mm coca—82 280:0me .AondAv 852:
383.380 raw“: 2865 238:: 29m .3388 95A Bob moasnttm :52 mousing mo x53 5:32.80 cowaom Rd 033.

 

 

113

83 82 8:. Ed 22 5:95
83 5... 82 £3. moEIEm
83 82. 3.8 ERIE
83 82. moEIm
I I 83 I :02 m
521.35 muEIEm moi e 52 a :02 m

 

.3 2,5 5 BE as $80 .625 8:38 8032.8 same 3885
8038:: 20m .3388 83388 Bob 3% 357383 53> 3:553 _o>o~ :28?qu go 358: 83.38.80 cofimom dd 2an

 

coo; mwmd 036

$5.:

omvd-

whmd-

3N6-

Smd

mam...

wad-

N26

H

 

imIimi
83 Soc 83.- 82- 2:8- 83- 888 its 83- 83 iii
83 8.2 83- 82- :88- 83 8% 488- 83.- iIimi
83 82- 22- 83- 83 8:. 82- 8:8 mIimi
. . . . - . - . . - o I
83 83 83 82 83 22 Rm 8 50 ii
. . . - . - . . - i I
83 an 8 n8 8 88 8 m8 8 3. 8 io ii
83 $2- $8- 8.8 23. moIimi
83 8% 8.2- 8:. >55
83 .53- 82 moIim
. . - 2 I
88 3 8m 8 05 ii
83 moi
H I I I 30 ~55 I I 205
imIimi i. ii i ii 5 ii Iii Iii .5 ii >5 me im Iii moi

 

.m.m 2an 5 moi: oi mouoo .AondAv 3508 moi—0:8 meE 88%.:
338:: Eom .8338 ommomEoo 80¢ 88 B>2$=mim 5m? 8:553 _o>o_ bmSEEOo mo x58: motfloboo memiom .w.m 033.

114

 

 

83 £88 888 ~88 ~88- E8- ~88 $88 «88- 888 imIi-ni
83 828 888- 3:8- 888- :88 888 888- 898 iii
83 8~8 2:8- 388- -§ :88 888- $~8 iIimi
83 828- E8- 888 888 «88- 288 mIimi
83 R~8 88. ~88- $~8 888- iEoIimi
83 8.88- 88- ~88 888- maIimi
83 E8 288. 58 >5
83 ~88- 888 moIim
83 «$8- EooIimi
83 moi
imIimi inimi iIimi mIimi mfiw moIimi >5 mmIim Min-Wm moi

 

.m.m 033. 8 688: 28 8800 ._o>o_ 388m 05 8 mm mots—88 ogomoxa-w .SndAv 8308
moi—oboe bme 8868 8388 20m .3388 925 89a 888585 _o>o_ 383888 mo 5.88 828388 gem-ion dd 3an

115

 

 

coo; movd 3 md m Rd mmvd cmod 3a.:- onvd nwmd nmvd >5l0mlr-H
coo._ Cod- wmwé momd Sod- mood.- maé ham-o cued 209mm H
coo; owed- owe-o mm fl .0. omvd- mvod- mood 23.: zonal U
coo; awed wmod- «N56. v3.0 god NmodI mum-rm g
coo; 3 fic- mamd- mood mead @86- M m
coo; S _.o @86- w: .o- m ~ Nd.- Oman
coo; on m .0. $66. meio- OD
ooo.— coed mvcd m0
ooo.— mmod MUm
o2: Qmw
>~Q 20mm I mama-w I
Iona mlb 20mm 0 lg m m Oman 00 m0 «Um Qmm

 

88 038. 8 BE 28 88o .8ch8 8:88 380:8 :82 3888

8388 35 82588 880583 80¢ 83 E>£$=8fl 53> 8:555 _8ouo8d Sm 5.588 8350.88 888$ .26 £an

116

 

 

117

mcvd onwd cemd owed- 593 a $6 mmvd >MQIOMIM
3a.: 2 Md- mmod- Elm .o- Q65 2: .o- Smd EOMMW M.
Sod 3nd mhmd ahmd- vmod- 306.. 03.0 EOMMI U
084 ommd $55 $06. an»... mmmd 53.0 mme mé
o2: coed mute- m3 .o- «3.: wood M M
coo; 52.? v» M .o- Mmmd wmod QmMM
coo; momd- mwmd- cemd- OD
coo. _ awed- vmmd MU
coo; omod- MUm
0004 93m
>5 20mm I m<Hm I
IOMM MIH EOMM U Im<m M M QMMM 00 Mo MUm him

 

.m.m 033‘ 5 you: 0.8 mocou 426— »:Efl 2.: 8 mm 3:282 280:0me .AondAv
8:qu Baa—oboe meB 8365 8038:: Rem .moEEwm 925 Bob 8:553 3:283 SM x59: :23?on acmaom .M _.m 2an

Table 3.12. Results of principal component analysis for both composite and large woody
debris (LWD) samples. Bold values indicate axes that were filrther examined for metric
selection.

 

 

PCA Axis Eigenvalues

Composite L WD

1 6.51766 6.28627
2 2.32852 1.86367
3 1.92508 1.52759
4 1.34371 1.0996
5 1.04998 0.906779
6 0.838963 0.670256
7 0.803271 0.342908
8 0.567636 0.269082
9 0.398593 0.249948
10 0.218721 0.163461
1 1 0.212564 0.07233
12 0.140631 0.054165
13 0.07073 0.01 1848
14 0.059754 0.005174
15 0.033363 0.005089
16 0.01015 0
17 0.003563 0

 

118

Table 3.13. Principal component loadings for composite samples. In general, those
metrics with the highest loadings per axis were retained for the final protocol. When two
or more metrics had similarly high loadings, two were chosen for the final protocol. Bold
numbers indicate which metrics were retained from each axis. Some metrics with high
loadings were not retained due to inaccuracy (e.g., PER_OLIG). See Table 3.5 for metric

codes.

 

 

Axis 1 Axis 2 Axis 3 Axis 4 Axis 5
PRICH 0.428 0.486 -0. l 72 -0.467 -0.062
EPTRICH 0.586 0.108 -0.652 -0.070 -0.290
DIPRICH 0.500 -0.180 -0.569 0.410 0.083
RICH 0.682 -0.239 -0.625 0.084 -0.076
PERDMC -0.785 0.331 0.048 0.200 -0.345
PERDIP -O.764 0.170 -0.257 0.1 12 -0.253
PEROLIG -0.6 l 2 -0. 105 -0.122 -0. 149 0.619
PERP 0.430 0.588 0.239 -0.529 0.035
PERT 0.426 0.713 0.077 0.457 0.1 19 '
PEREPT 0.452 0.578 0.052 0.051 -0.266
LNSHD 0.560 -0.312 -0.025 0.085 -0.020
LNSCR 0.580 -0.493 0.285 -0.213 -0.377
CF 0.637 0.456 0.135 0.440 0.265
CG -O.853 0.184 -0.312 ~0.046 -0. 195
LNPRED 0.138 0.284 -0.578 -0.391 0.258
LNHSTAB 0.839 -0.01 1 0.362 0.052 -0.059
FFGDIV 0.905 -0.235 0.052 -0.103 0.093

 

119

Table 3.14. Principal component loadings for LWD samples. In general, those metrics
with the highest loadings per axis were retained for the final protocol. When two or more
metrics had similarly high loadings, 2-3 were chosen for the final protocol. Bold
numbers indicate which metrics were retained fiom each axis. Some metrics with high
loadings were not retained due to ambiguous response to human influences (e.g., LNSH)
See Table 3.5 for metric codes.

 

 

Axis 1 Axis 2 Axis 3 Axis 4

RICH 0.7826 -0.2029 -0.4415 0.1764
PRICH 0.5499 0.1234 -0.00821 -0.2297
EPTRICH 0.7727 -0.3737 -0.2902 0.01024
PERCG -0.6977 -0.4215 -0.3727 0.01761
DIPRICH 0.3951 -0.3218 -0.5372 0.5403
PERDOM -0.813 -0.2592 0.0859 0.05418
FFGDIV 0.8592 0.3726 -0.04755 0.1654
PEREPT 0.7988 -0.2562 0.1289 -0.2365
LNSC 0.6534 0.04705 -0.4381 -0.4052
LNCF 0.6487 -0.3167 0.5203 0.3262
LNHABSTAB 0.8653 0.07889 0.3857 -0.1 131
LNPRED 0.2839 0.6564 -0.3485 -0.2277
LNSH 0.1998 0.6841 0.08521 0.5824
LNTRICH 0.7677 0.4241 0.3334 -0.043 75

 

120

Table 3.15. Metrics retained and their corresponding weights in the final protocol
(composite samples only). Weights were determined by summing eigenvalues from
retained axes and calculating the total variance explained by each axis. Scores were then
approximated based on a 100-point scale. When 2 or more metrics were retained fiom
each axis, scores were divided equally based on the variance calculated for the entire
axis. See Table 3.5 for metric codes.

 

 

 

. . . Variance
Metnc Ans Eigenvalue (Prop [To tal) Score
FFG_DIV 1 25
HAB_ST AB 1 6.51766 0.50 25
PER_T 2 2.32852 0.18 20
EPT_RICH 3 8
RICH 3 1.92508 0.15 7
DIP_RICH 4 5
P_RICH 4 1.34371 0.10 5
PER_DOM 5 1 .04998 0.08 5
Total 13.16495 | 100

 

Table 3.16. Metrics retained and their corresponding weights in the final protocol (LWD
samples only). Weights were determined by summing eigenvalues from retained axes
and calculating the total variance explained by each axis. Scores were then approximated
based on a 100 point scale. When 2 or more metrics were retained from each axis, scores
were divided equally based on the variance calculated for the entire axis. See Table 3.5
for metric codes.

 

 

Metric Axis Eigenvalue “$71.1le; Score
HAB_STAB 1 20
F FG_DIV l 6.28627 0.58 20
PER_DOM 1 20
PER_T 2 1.86367 0.17 15
DIP_RICH 3 1.52759 0.14 15
SCR 4 1.09960 0.10 10

 

Total 10.77713 100

 

121

Table 3.18. Scoring criteria for biological monitoring of non-wadeable rivers using large
woody-debris samples only.

 

 

 

 

 

 

 

Biological Metric Poor Fair Good Excellent Metric Total
Habitat Stability Hab Stab Hab Stab Hab Stab Hab Stab
FFG Surrogate <0.08 0.08-0.19 0.20-0.56 >056

O 7 14 20
FFG Diversity FFG Diversity FFG Diversity FFG Diversity FFG Diversity

<O.76 0.76-1.09 1.10-1.33 >1 .33

0 7 14 20
% Dominant % Dominance % Dominance °/o Dominance % Dominance
Taxon >59 34-59 23-33 <23

0 7 14 20
% Trichoptera % Trichoptera % Trichoptera % Trichoptera % Trichoptera
Abundance <3 3-6 7-14 >14

0 5 10 15
Diptera Taxa Diptera Richness Diptera Richness Diptera Richness Diptera Richness
Richness <2 2—3 4-5 >5

0 5 10 15
% Scraper % Scrapers % Scrapers % Scrapers % Scrapers
Abundance <2 2-6 7-1 1 >1 1

O 4 7 10

 

Total 13er—

122

 

 

Table 3.17. Scoring criteria for biological monitoring of non-wadeable rivers using

composite samples with all habitats.

 

 

 

 

 

 

 

 

 

Biological Metric Poor Fair Good Excellent Metric Total
FFG Diversity FF G Div FF G Div FF G Div FF G Div
<0.95 0.95-1.41 1.42-1.70 >1 .71
0 8 16 25
Habitat Stability Hab Stab Hab Stab Hab Stab Hab Stab
FFG Surrogate <0.09 0.09-0.26 0.27-0.67 >0.67
0 8 16 25
% Trichoptera % Trichoptera % Trichoptera % Trichoptera % Trichoptera
<1.3 1.30-3.40 3.41-6.80 >680
0 7 14 20
EPT Richness EPT Rich EPT Rich EPT Rich EPT Rich
<4 4-6 7—9 >9
0 3 6 8
Total Richness Taxa Richness Taxa Richness Taxa Richness Taxa Richness
<15 15-18 19-24 >24
0 2 5 7
Diptera Richness Dip Richness Dip Richness Dip Richness Dip Richness
<2 2-3 4-5 >5
0 2 4 5
Plecoptera Plec. Richness Plec Richness Plec Richness Plec Richness
Richness
0 1 2 3
0 2 4 5
°/o Dominance % Dominance % Dominance % Dominance % Dominance
>60 47-60 35-46 <35
0 2 4 5

 

Total Point Score —

 

123

 

 

 

med

MME

eDZOU

.25.

205.8:

 

$3 -- 8:32 E0 0: E38: 220: :0:
as $32 0.33.. m: :95 gig: moan:
and -- -- -- :95 Bacon.

8: 3:0 .550 .E .z 50 3.58: n

«.3 -- 0:3 E .z 80 tags: :02 a:
«No -- @532 0.35 3.38: :02

a: -- -- -- 83.5 0338:

So -- -- 83.5 .9: 3.38m ..
5o -- -- $35 .53: 3.38: :2: Em
8: -- -- 8:0... 2 80 0838: 01:0:
86 -- -- am? $50 :8an:

one 8:? $332 83?. $35 338: u
a: -- -- :23 .5200 338: 9:0 0::
2: 3:0 .55.- .:200 :3 338: Zane":
S: :32 E0 .950 on Race: _

N: .x a: "x .x 02:»: 2.582

 

00:00 03:2?

3:028:60 SM 8.33 0030M. :5 3:00 0308 SM m.m 030,—. 00m .3.on £030 :0 :M .89qu £3, 003.:5, 38:09.... .modva
5:5 33:32, 88:09.. .0963 “00:80:05 8 00:88: 30:00N .0305: 35:: 9 00:80.0: 30:06. .20: :32? 0:: Acaioanv
3.580: 09 8 8:03:03 08:0 8 003093 e “mam 05 >30 0:800: 3508 533 9 83:9? 3:085:35 30.0%: 08020.9.
383088 Good 00:80:: mm M .ou0>oE0: 3 02.273 :ofimewa 003308 2:30.03 :5 286005203“ mm M .ouu0u:0 8 0:378
:ommm0uw0: 003305 2:38”— .m0E:_8> 35:88:60 53, 8308 0:08:80 Mo :ommm0aw0: 80:: 0322: Sea $38M .2.m 030,—.

124

 

05.0

005

03200

.25.

0.83:0:m

 

$3 -- -- -- :masz “.an0 Row

:50 *AHMHD :mkfiaz 1.95:3 02 POP vaigm IUEIEQ

:0 :25 1.0352 1035 .2 .8: 0528

m2 -- .000... .035 E: 03380 plum:

m2 -- am? .0300. E: 0338:

m3 -- 2.232 .035 .00 3.380 u

m; -- 20:32 .035 .00 0330: 200 :0:

mud £3302 :mm :5. .32 POP aniowm >_DIO...E

8.0 -- .232 .035 .0: 32¢:

30 -- .0032 .29.. ...z :0: 0338: u

”to -- -- mm? :masz 0338: 9:0 0 a z
n: i an "x .x 02002 9:32

 

.m0000 030i? 3808:8060 8.: omd: 0030:. 0:: 0.0000 00:08 8.: m.m 030:. 00m .3.qu £00.00 :0 E .25.ng

::3 330005 88:09:. .modv: 505 0030003 88:09.. .0305: 5:80: 00 030000: 30:09 .80: 950% 03 8:03:83
03080: 0: :0 8:02:03 .860 0: 003029, 0 we 05 >30 0:090: 0050.: :02? 0: 00300:? 3:085:30 0:20:30 00:00:00
00:52:00 Good 00520—9 mm _ .ou0>0E0: 00 0003-3 :0mmm0Hw0: 000536 0:03:00: 0:: 28.008803: mm_.ou:00:0 8 0:05

-3 :0mmm0uw0: 000,306 0350.: 003005, 08:08:20,:0 :00,» 0050:: QB: m0 :0mmmohw0: .805: 0322: 80: 0:83: .om.m 030:.

125

Table 3.21. Results fi'om the discriminant function analysis. Two analyses were done:
One in which approximately one-half of the sites from each classification type were used
to generate the model, while the other half were used to test the model. The other
analysis used jackknifmg to evaluate the model.

 

 

Classificaiton % Correct (test sites only) % Correct (jackknifed)
Composite LWD Composite LWD

Poor 75 100 88 78

Fair 33 25 50 75

Good 67 100 55 60
Excellent 100 33 67 so

 

126

Table 3.22. Site rankings based on CDG, RDG, and an overall ranking based on CDG
and RDG rankings combined with the number of transects with LWD (Mean Rank).
HI=Non-wadeable Habitat Index (Wilhelm 2002). Site scores for composite and LWD
assessments are also listed. These data were used to evaluate the NW-IBI sensitivity
(composite vs. LWD) to differing scales of human impacts.

 

NW IBI NW-IBI

 

Site CDG Rank RDG Rank Mean Rank HI Comp LWD
as_mth02 3O 20 1 7 47 69 47
as_whpOl 1 13 l 83 69 57
as_whp02 3 21 4 86 34 77
gd_cmr01 28 28 29 25 27 16
gd_cmr02 3 1 27 32 26 29 4
gd_gh02 16 17 19 43 2 5
gd_gld02 25 18 27 50 48 37
gd _gr01 29 31 28 27 27 27
gd_g102 32 3O 33 32 29 35
gd_ion01 13 11 22 51 42 21
gd_ion02 17 4 24 59 36 - 5
gd_jon01 14 l 13 52 63 54
gd_jon02 22 5 18 51 31 19
kz_cus01 1 1 2 1 1 66 62 59
ma_ctsOZ 4 22 5 85 84 80
ma_hbr01 8 l4 3 82 7O 69
ma_hbr02 9 6 6 85 69 89
ma_mnsOZ 33 25 21 28 58 19
ma_rbw01 19 12 10 66 74 95
me_kssOZ 12 19 12 58 45 71
me_stb02 5 23 7 69 77 89
mk_br02 l 8 7 15 49 9 32
mk_nwg02 10 8 14 6O 56 61
mk_thp01 20 15 8 78 65 65
mk_trkOl l 5 1 6 20 50 67 63
ra_dunOl 26 3 3O 38 17 31
ra_mon01 27 32 3 1 34 8 O
sg_zflOl 24 29 25 31 O 5
sh_ngl 2 33 16 63 22 .N/A
sj_mv102 23 9 26 66 79 64
sj_rvw02 34 24 34 39 20 38
tb_sagOl 21 26 23 42 22 4O
tq_nwb02 6 34 9 59 40 .N/A
tq_pdsOZ 7 10 2 59 40 63

 

127

Table 3.23. Regression data for composite assessments. In all cases, the composite NW-
IBI was the dependent variable. . Independent variables reflect differing scales of human
influence. CDG=Catchment Disturbance Gradient; RDG=Riparian Disturbance
Gradient; Mean Rank is based on catchment-wide, riparian, and in-stream habitat quality.
HI=Non-wadeable Habitat Index (Wilhelm 2002)

 

Indfiggzfi R2 P-valne
CDG 0.130 0.036

RDG 0.1 13 0052

Mean Rank 0.313 <0-001
HI 0.407 <0.001

 

Table 3.24. Regression data for LWD assessments. In all cases, the LWD NW-IBI was
the dependent variable. Independent variables reflect differing scales of human
influence. CDG=Catchment Disturbance Gradient; RDG=Riparian Disturbance
Gradient; Mean Rank is based on catchment-wide, riparian, and in-stream habitat quality.
H1=Non-wadeable Habitat Index (Wilhelm 2002)

 

Inde‘llr::i:;lll: R2 P—valne
CDG 0.338 <0.001

RDG 0.083 0110

Mean Rank 0.535 <0-001
HI 0.612 <0.001

 

128

FIGURES

129

'''''''''''''''''''''''''''''

 

'''''''''''''''''''''''''''''

IIIIIIIIIIIIIIIIIIIIIIIIIIIII

O'O'O'O'I'I'O'I'I'C'I'I'Illl

'''''''''''''''''''''''''''''

 

 

 

2000m
Figure 3.1. Diagram of a non-wadeable study reach. I chose a standard 2000m

reach, and sampled macroinvertebrates at transects placed every 200m.

 

4

130

Mean Richness (SE)

Mean H’ (SE)

1%} 1 01 —
5- i“ -

 

LWD Samples Only
20 I I I I I I I I I I I I I I I
a

 

1 . 1.

 

 

111111L111111L
110111213141518171846 7 8

 

L
9

 

4IIITIIITIIIIIII

 

 

 

l 1 J 1 1
11011121314151617184 6 7 8 9

, 1
1+— } 1—
o llllilll;l

25

20

15

1O

FPOM Samples Only

 

 

rIIIIIIIIIIIIIII

b

 

Jllllllllll

 

l 1 l 1 l
1101112131415161718194 6 7 8 9

 

 

IIIIIIIIITIIIIIII

 

1111111111

 

1 I 1 1L1
110111213141518171819 4 6 7 8 9

2001 Study Reach

Figure 3.2. Mean taxa richness and Shannon diversity (H’) fi'om selected 2001 study
sites. LWD (a and c) and F POM (b and (1) samples only. Note that despite coming
fi‘om the same habitat, richness and diversity still varied considerably in LWD
samples, while variability was less pronounced in FPOM samples. Numbers on x-axis

indicate different sites.

Eigenvalue

 

 

 

 

 

—O - Composite

JV--. +LWD

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

If T I I T

12 3 4 5 6 7 8 91011121314151617
PCAAxIs

Figure 3.3. Scree plot of eigenvalues (k) for each PCA axis. Axes 1-
5 were further examined for composite metric selection. Axes 1-4
were further examined for LWD metric selection. Criteria: D1.

132

 

 

 

 

 

6 mk
4a
N
m
31 2‘ 0d
<2 9d me
0d :sz
gd ('3 “hr; 9d Mk
0‘ as
e;
59 ma as ma me
“a... ma m.
as
-2-1 tq ma
tq
'4 I I 17 I I
-6 4 -2 o 2 4
Axis]

Figure 3.4. PCA site scores (axes 1 vs. 2) for the composite data. Site scores are
plotted with rivers identified (see Table 3.1 for river codes). This plot shows that
sites did not cluster by river or catchment.

133

 

 

 

 

 

 

134

a
6 .. NLP
4 -4
N
.2
X 2 ~ SLP
<1 SLP UP
SLP 5L LP
SLP SLP mfp SLP NLSL
0 n NLP UP SLP
SLP NLP NLP NLP
WP NLP NLP
-2 a up "LP NLP
up
'4 I I I I I
-6 -4 -2 o 2 4
Axis I
4
3 4 b ,9 UP
2 .
SLP
SLP
"w SLP NLP
- UP
'3 1 NLP gm
‘5; SLP
<1 UP
0 - NLF'fil'P
SLPNLP
LP
_1 . W, 31.116
mg? SL7”)
.2 . NLP SLP
'3 I I I I I
-4 -2 0 2 4 6
Axis 2

Figure 3.5. PCA site scores identified by ecoregion. (a) Axis I vs. Axis 2; (b) Axis 2 vs.
Axis 3. Sites clustered based on ecoregion along axis 1, but not along the other axes.
SLP: Southern Lower Peninsula; NLP: Northern Lower Peninsula; UP: Upper
Peninsula.

 

 

 

10

 

 

 

 

 

a a
:: 8-
55
t: 6.
'5
3% 1
o 4‘
o
g E
E 2-
Q
.5 i
a 04
E
SLP NLP UP
14
b

12-

10~ i

3..

6~

Catchment Disturbance Gradient (CDG) (SE)

 

l J l

 

 

SLP NLP UP

Ecoregion

Figure 3.6. (a) Mean RDG and (b) mean CDG by ecoregion (:h SE) (See
Wilhelm 2002 for information regarding the development of the RDG and
CDG).

135

Number of Sites

 

 

 

 

Metric Value

 

Figure 3.7. Theoretical example of how a 25 point and a 5
point metric were scored based on inter-quartile ranges.

 

 

 

 

 

 

 

 

 

 

 

 

 

I Comp
1:
El LWD
10
|
1
. _
6 ‘ l
4 -
z a
o , , T
Poor Fair Good Excellent

Figure 3.8. NW-IBI scores for all sites comparing composite and
LWD assessments. LWD assessments appeared to be more sensitive
than composite assessments.

136

ma_ctso2
sLjnw02
me_stbOZ
ma_rbw01
ma_hbr01
as_mth02
ma_hbr02
as_whp01
mk_trk01
mk_thp01
gd_jon01
kz_cusO1
ma_mns02
mk_nngZ
gd_gld02
me_kssOZ
gd_ion01
tq_pdsOZ
tq_nwb02
gd_ion02
as_whp02
96.10002 I Habltat Stablllty FFG Surrogate (25)
gd_cmr02 .
gd_gr02 . 1:3 % Trichoptera (20)
gd_cmr01
gd_gr01
5'15an I Total Richness (7)
tb_sagO1
sj_rvw02 I Diptera Richness (5)
ra_dun01
mk br02 I Plecoptera Richness (5)
ra_mon01
gd_gh02
sg_flKH

I FFG Dlverslty (25)

1:1 EPT Richness (8)

El °/o Dominance (5)

 

0 25 50 75 100
NW-IBI Score

Figure 3.9. Composite NW-IBI scores for each non-wadeable study site.
Individual metric scores are shown in each bar. This image is presented in color.
See Table 3.1 for site codes.

137

 

,4”, 1";

. m m
_; 3‘50." ”22
.- m —-'

own n
. - m.
’ q... a
.. c
; 'n'i'
so":
I. w
m
.n 41.6
‘ a“ l.-‘
I'
E
it 9
an -
an
1‘
. :5 I
z . _
'
-
.“.
a ..
1'4
9

~ I I -.¢-v‘0 as -‘-L

     

-l '.
., .7 a
., . .13; - mung”.

1 I

, u ”Jut'n‘dflu "

l .1 . J 'LIII'. Id

gé.

f

ma_rbw01
me_stb02
ma_hbr02
ma_ctsOZ
kz_vb01
as_whp02
me_kssOZ
ma_hbr01
mk_thp01
sj_mv|02
tq_pds02
mk_trk01
mk_nwg02
kz_cuso1
as_whp01
gd_jon01
as_mth02
tb_sagO1
sj_rvw02
gd_gld02

gd_gr02 I Habitat Stablllty FFG Surrogate (20)
mk_br02

ra_dun01 l FFG Dlverslty (20)
gd_gr01
gd_ion01 13 16 Domlnance (20)
ma_mnsoz
gd _j0f102 D % Trichoptera (15)
gd_cmr01
gd_gh02
gd_ion02
sg_zil01
gd_cmr02
ra_mon01

I Dlptera Rlchness (15)

I as Scapers (10)

 

O 25 50 75 100
NW-IBI Score (LWD)

Figure 3.10. LWD NW-IBI scores for each non-wadeable study site. Individual
metric scores are shown in each bar. This image is presented in color. See Table 3.1
for site codes.

138

 

1 00 I I I I I I l I
90 _ O O —'

Pearson Correlation: 0.76

LWD NW-IBI SCORE

 

 

 

0 0 Cl 1 1 I 1 1 1 1
010 20 3O 4O 5O 6O 70 80 90
COMP NW-IBI SCORE

 

Figure 3.11. Composite vs. LWD NW—IBI scores for non-wadeable river study
sites.

139

 

NW-IBI SCORE

 

90 l l l
80 " o

2
70 O R =0.31

 

 

 

 

0 1 O1 3 l
O 10 20 30
Mean Site Ranking

Figure 3.12. Mean Site Ranking vs. NW-IBI for composite assessments.

140

LWD NW-lBl SCORE

00 I I I

 

 

 

 

 

O 10 20 30 40
Mean Site Ranking

Figure 3.13. Mean Site Ranking vs. NW-IBI for LWD assessments.

141

 

Figure 3.14. Saginaw River @ Zilwaukee (sg_zilOl) riparian view. This site
scored lowest in the composite NW-IBI, and was classified as “poor” by both
types of assessment. This image is presented in color.

142

 

 

 

Figure 3.15. Grand River @ Ionia (gr_ion01, gr_ion02). This site was sampled
in both 2001 and 2002, receiving a score of “fair” each time (composite
assessment). This image is presented in color.

143

 

Figure 3.16. Manistee River @ High Bridge (ma_hbr). This site was scored as
“good” in both 2001 and 2002 (composite assessment). Inset: Note clear water and
slightly embedded coarse substrate. This image is presented in color.

 

Figure 3.17. Manistee River @ Coates Rd (ma_ctsOZ). This site scored “excellent”
in both assessment types. This image is presented in color.

145

# New Taxa

16

 

14+

12-

 

 

 

 

 

10 12

Number of Samples

Figure 3.18. Mean number of new taxa collected in successive LWD samples.
Note that alter 8 samples, there were never new taxa collected, indicating that 8
samples should be sufficient for LWD assessments. These data were tabulated
from 6 randomly chosen sites.

146

 

Are there at least 8

1 transects containing 1
I LWD? l

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

vies No l
LWD-only assessment. Sample all habitats at
Sample only large each transect.
woody debris at each _____________________ _> Combine into one
transect. Combine into composite sample
one sarlnple.
Preserve sample. Remove all Preserve sample. Subsample
macro invertebrates from the with a plankton splitter into
sample. Use Table 3.18 to quarters. Remove all
score site. macroinvertebrates from the
subsample. Keep the remaining
3 subsamples for further
analysis. Use Table 3.17 to
score site.

 

 

 

 

v

Enter raw data into the data sheet provided. Enter these raw numbers
into the appropriate sheet in assessor.xlt This will automatically
calculate metric values based on assessment type (composite or LWD).

 

 

 

 

Figure 3.19. Flowchart illustrating the steps involved in the biological assessment of
non-wadeable rivers in Michigan.

147

APPENDIX 1. COMPOSITE ASSESSMENT METRICS AND STRESSOR-

RESPONSE RELATIONSHIPS

148

 

APPENDIX 1. COMPOSITE ASSESSMENT METRICS AND STRESSOR-

RESPONSE RELATIONSHIPS

The following section provides more detailed information regarding specific
metrics’ response to various environmental parameters. For a summary of physical,
chemical and landuse variables, see Tables 2.1 and 2.2. Only significant relationships are
mentioned in this section. This information may be used when conducting composite
assessments to diagnose specific human influences causing ecological impairment. See
Table 3.19 for MLR results. In all cases the correlation coefficient was higher in
backward elimination MLR (Table 3.19).

Composite Metric 1. Functional Feeding Group Diversity (FFG_DIV) (25 pts.). In
addition to the parameters listed below, FFG diversity also shows a general response to
riparian landuse (Table 3.19).

- MLR Forward Stepwise:

l. Turbidity (-)
2. Percent natural riparian landuse (+)

o MLR Backward Elimination

Total Phosphorous (-)
Conductivity
Turbidity (-)
Suspended chlorophyll

99’1”!"

Composite Metric 2. Habitat Stability FFG Surrogate (HAB_STAB) [(# Scrapers+#Coll-
Filt)/(#Coll-Gath+#Shredders)] (25 pts.):
0 MLR Forward Stepwise:

1. Conductivity
2. Percent agricultural riparian 1anduse(+)

o MLR Backward Elimination

149

 

Percent urban landuse in the watershed (-)
Percent agricultural landuse in the watershed (-)
Percent natural landuse in the watershed (+)
Percent agricultural riparian landuse (-)

PPN?‘

Composite Metric 3. Percent Trichoptera Abundance (PER_T) (20 pts.)
0 MLR Forward Stepwise:
1. Percent agricultural riparian landuse
o MLR Backward Elimination

1. Percent agricultural riparian landuse

Composite Metric 4. EPT Richness (EPT_RICH) (8 pts.)
0 MLR Forward Stepwise:

1. Amount of LWD in the study reach
2. Percent urban landuse in the watershed

o MLR Backward Elimination

1. Amount ofLWD in the study reach
2. Percent urban landuse in the watershed

Composite Metric 5. Total Taxonomic Richness (RICH) (7 pts.)
0 MLR Forward Stepwise:
1. Percent urban landuse in the watershed

o MLR Backward Elimination: There were no significant relationships.

Composite Metric 6. Diptera Taxa Richness (DIP_RICH) (5 pts.)
- MLR Forward Stepwise:

1. Total nitrogen

150

o MLR Backward Elimination

Total nitrogen

pH

Turbidity

Suspended chlorophyll

99°F!“

Composite Metric 7. Plecoptera Taxa Richness (P_RICH) (5 pts.)
0 MLR Forward Stepwise:
1. Amount of LWD in the study reach
0 MLR Backward Elimination

1. Amount of LWD in the study reach

Composite Metric 8. Percent Dominance (PER_DOM) (5 pts.). In addition to the

parameters listed below, PER_DOM also showed significant correlations with riparian

and watershed landuse.
o MLR Forward Stepwise:
1. Percent natural riparian landuse
o MLR Backward Elimination

1. Total phosphorous
2. Conductivity

151

APPENDIX 2. LARGE WOODY DEBRIS (LWD) ASSESSMENT METRICS AND

STRESSOR-RESPONSE RELATIONSHIPS

152

 

APPENDIX 2. LARGE WOODY DEBRIS (LWD) ASSESSMENT METRICS

AND STRESSOR-RESPONSE RELATIONSHIPS

The following section provides more detailed information regarding specific
metrics’ response to various environmental parameters. For a summary of physical,
chemical and landuse variables, see Tables 2.1 and 2.2. Only significant relationships are
mentioned in this section. This information may be used when conducting LWD
assessments to diagnose specific human influences causing ecological impairment. A
complete list of environmental parameters included in MLR automatic stepwise
regression can be examined in Tables 2.1 and 2.2. In all cases, correlation coefficients
were equal or higher in the backward elimination method (Table 3.20).

LWD Metric 1. Habitat Stability FFG Surrogate (HAB_STAB) [(# Scrapers+#Coll-
Filt)/(#Coll-Gath+#Shredders)] (20 pts.):
0 MLR Forward Stepwise:
3. Percent natural landuse in the watershed
o MLR Backward Elimination
1. Total Nitrogen
2. Percent agricultural riparian landuse
3. Percent natural riparian landuse
LWD Metric 2. Functional Feeding Group Diversity (FFG_DIV) (20 pts.).
0 MLR Forward Stepwise

1. pH
2. Percent urban watershed-wide landuse

0 MLR Backward Elimination

Total nitrogen

Total phosphorous

pH

Percent natural riparian landuse

PP’NT‘

153

LWD Metric 3. Percent Dominant Taxon (PER_DOM) (20 pts.).
0 MLR Forward Stepwise

1. Dissolve oxygen
2. Percent urban landuse in the watershed
3. Percent natural riparian landuse

- MLR Backward Elimination

1. Dissolve oxygen
2. Percent urban landuse in the watershed
3. Percent natural riparian landuse

LWD Metric 4. Percent Trichoptera Abundance (PER_T) (15 pts).

0 MLR Forward Stepwise

1. pH
2. Percent agricultural landuse in the watershed
3. Percent agricultural riparian landuse

o MLR Backward Elimination

1. pH
2. Percent urban landuse in the watershed
3. Percent agricultural riparian landuse

LWD Metric 5. Diptera Taxa Richness (DIP_RICH) (15 pts.).

- MLR Forward Stepwise

1. Total nitrogen
2. Percent urban landuse in the watershed
3. Percent natural landuse in the watershed
4. Percent urban riparian landuse

o MLR Backward Elimination

Total nitrogen

Percent urban landuse in the watershed
Percent natural landuse in the watershed
Percent urban riparian landuse

PPN?‘

154

LWD Metric 6. Percent Scrapers (SCR) (10 pts.). In addition to the parameters listed
below, this metric has a broad response to riparian landuse.
o MLR Forward Stepwise
1. Percent natural landuse in the watershed
o MLR Backward Elimination
1. Total phosphorous

2. Conductivity
3. pH

155

APPENDIX 3:

FIELD MANUAL FOR THE BIOLOGICAL ASSESSMENT OF NON-
WADEABLE RIVERS IN MICHIGAN

156

I. GENERAL PROCEDURE

A. Use the equipment checklist to ensure all necessary equipment is brought along
for the assessment.

B. Locate the reach of interest. Assessment of non-wadeable rivers will be at the
reach scale. However, test reaches should be representative of the larger river and
catchment. Considerations of which reaches to evaluate include:

1. Proximity to urban centers (e.g., downstream from a metropolitan area or

intensively farmed area).

2. Ease of access. Can the crew get to the site with the needed equipment?

3. Specific stressors known to afi‘ect a certain area.

4. Motor to the downstream end of the study reach and mark this area. This is
Transect A. Additional transects are located z 200m upstream of each
subsequent transect. This should be done by the Habitat Assessment Crew.
Given a total reach size of 2000m, there are 11 total transect (A-K).
Determine randomly (e.g., flipping a coin) which bank to sample.

6. Sample all available habitats (or just the woody debris) within z 10m
upstream and downstream of the transect. Transects should be marked with
flagging and labeled (A-K) by the Habitat Crew in case they need to be
relocated at later times during the field portion of the assessment. Sampling
should take place in shoreline areas (<1m deep). See the next section for
detailed description of sampling procedures.

7. Using the taxonomic data sheet, record all taxa in the sample and the
abundance of each. This may need to be done at the lab for composite
samples. LWD samples may be processed in the field by experienced field
technicians.

5"

 

 

.-.-._.-._.-.-.-.-.-.-._.-....-..1
.—.—.-.-.--.—--—.—.-.—-—.-.-.1

i.—C-O-O-U-.-O-O-C-O-O-.-.-O-I-.J

.—o-o-o-o-o-o-o-o-o—o-o-o-o-I-o‘
0-.-.-0-.-.-O-o-O-o-o-o-o-o-o—Od
.-.-.--_.-.-.-.-.-.-.-.-0-.---.‘
‘-C-.-.-.-.-.-.-.-.-.-.-.-.-.-.‘
o-O-o-o-o-o-o-o-o-o-o-I-o-o-.-Id

b-0-u-o-o-o

 

A H I

O:
O
U

E

111
Q
a.
7‘!

Schematic diagram of a non-wadeable study reach. Total length is 2000m. Transects
are labeled A-K and are evenly spaced 200m apart. Macroinvertebrate sampling takes
place at each transect on randomly-chosen banks. Arrow indicates the direction of flow.

157

DETAILED SAMPLING PROCEDURES

Composite Samples: Use this method if large woody debris is not present or for a
more detailed assessment of the reach. Using this approach, the biological
assessment will reflect the available habitat as well as in-stream water quality.
This sampling procedure involves sampling all available habitats at each transect
and combining the individual samples into one composite for the entire reach. At
each transect:

1. Tally the individual habitat types. These include:

a) Fine particulate organic matter (FPOM)
b) Sand

c) Gravel

d) Cobble

e) Large woody debris (LWD)

t) Macrophytes

2. For each habitat type, take timed samples (15 seconds each) with a D-fi'ame
aquatic dip net with mesh size = 0.5mm. Habitat-specific considerations are
as follows:

a) FPOM: If there is flow through the sampling area, use kick
methods to reduce the amount of detritus in the sample. If there is no
flow, sweep the net along the bottom and make sure to wash as much
detritus from the net as possible before preserving the sample.

b) Sand: Same as above.

0) Gravel: Same as above.

d) Cobble: It is difficult to take timed sweeps of cobble
habitat. Therefore, try to choose a piece of cobble at least 15 cm in
diameter. Place the cobble in a bucket and brush organisms ofi‘ with a
toilet brush

e) Large Woody Debris (LWD): Sampling LWD presents
challenges, especially when the debris cannot be removed from the
river. We suggest using a toilet brush to dislodge organisms fiom the
LWD and following closely behind with the net. If there is high flow
in the area begin sampled, make sure the net opens into the current and
the brush is ‘upstream’ of the net. Do this for z 15 seconds.

i) Macrophytes: If there are macrophytes in the study reach,
take timed sweeps (z 15 seconds) of the stems to dislodge attached
macro invertebrates.

3. Empty the net into a white enamel pan filled with water. This allows you to
easily wash out the net (you may need to pick attached organisms from the net
with forceps).

4. Remove as much detritus as possible before pouring the sample into a 500 (.m
sieve to remove excess water.

5. Place each individual sample for each habitat at each transect into a bucket
and preserve in the field with 95% EtOH. Further processing will be done in
the laboratory.

158

.50

Habitat-Specific Sampling: If the study site contains sufficient amounts of LWD,

you may evaluate the reach by sampling only LWD habitats. This will

significantly reduce sample processing time and allow an evaluation of the

reach’s ecological integrity that is more independent of the habitat assessment.

Follow the procedures above to sample LWD. Because of the inherent variability

in non-wadeable systems, this should only be done if there is sufficient LWD

habitat (e.g., the number of transects with LWD habitat is 2 8).

Further Sample Processing

1. Composite Samples will be returned to the laboratory, subsampled (quarters),
and macroinvertebrates will be sorted and identified to family level.

2. LWD Samples may be sorted and identified in the field. If this is done, make
sure to enter raw data into the Bioassessment Field Data Sheet (back page).

QUALITY ASSURANCE/QUALITY CONTROL

For any macroinvertebrate identifications you are not sure about, place
representative specimens in vials containing preservative. This will be especially
important if you are sorting and counting invertebrate samples in the field, such as
when doing LWD habitat specific assessments.

Clearly label each specimen with site information and number of each ‘type’ in
the sample.

Take the specimens back to the laboratory for examination under a microscope.
For composite sample assessments, rettu'n one of the subsamples (1/4) to the
laboratory for storage. This will allow reassessment at a later time and
comparison of subsamples.

159

EQUIPMENT CHECKLIST FOR NON-WADEABLE RIVER BIOASSESSMENT

The following items should be included with the crew responsible for the collection of

 

 

 

 

 

 

 

 

 

 

 

 

 

 

macroinvertebrates:
\/ QNTY ITEM

2 500 int mesh D-frame aquatic dip nets (2)
5L 95% Ethanol

2 Standard toilet brushes (2)

1 Large bucket with lid for samples

2 Forceps (2) for picking organisms fi'om D-fiame net.

2 500nm sieves (2) for processing samples.

2 White enamel pans (2) for sorting organisms

10-20 Vials for voucher specimens (If sample processing is to be done in

the field)
Labels for voucher specimens
Color identification plates
Non-wadeable biological assessment data sheet (from and back)

1 Plankton splitter (If sample processing is to be done in the field)

2 Magnifying lenses (2) (If sample processing is to be done in the field)

 

 

 

 

 

 

 

 

160

 

BIOASSESSMENT FIELD DATA SHEET

 

 

 

 

 

 

 

 

 

 

 

 

 

(Front page)
DATE: CREW
RIVER: REACH LOCATION
GPS or Gazetteer Info Other information
Upstream 01' (City, Dam,
etc.)
Downstream
Other Notes:
Assessment COMPOSITE
TIP“ LWD
On the diagram below, mark the locations at which macroinvertebrate samples were
taken.
A B C D E F G H I J K

 

For composite assessments, note which macroinvertebrate habitats were present
at each transect.

 

 

 

 

 

 

 

 

 

 

 

A F Sa C Cb W M F Sa C Cb W M
B F Sa C Cb W M F Sa C Cb W M
C F Sa C Cb W M I F 811 C Cb W M
D F Sa C Cb W M J F Sa C Cb W M
E F 8:: C Cb W M K F Sa C Cb W M
F F Sa C Cb W M Total Samples:

 

F=FPOM; Sa=Sand; C=Coarse substrates; Cb=Cobee; W=Largc Woody Debris; M=Macrophytes

161

BIOASSESSMENT FIELD DATA SHEET

(Back page)

 

This data sheet allows you to quickly summarize your field data. Box

 

 

 

 

 

 

 

 

 

 

 

 

 

 

SAMPLE COMPOSITE numbers correspond to the MS Excel file (assessor.xlt) used for metric
TYPE LWD scoring and smnmary of ecological condition. When entered correctly,
scores are automatically calculated for each site.
Enter these values into corresponding box in Excel
-...... template (assessor.xlt). Box Number
Total Abundance 1
Total Richness 2
Number of Ephemeroptera 3
Families
Number of Plecoptera Families 4
Number of Trichoptera Families 5
Number of Diptera Taxa 6
Trichoptera Abundance 7
Abundance of Dominant Taxon 8
Shredder Abundance 9
Scraper Abundance 10
Coll-Filterer Abundance ll
Coll-Gath Abundance 12
Predator Abundance 13

 

 

 

 

 

162

 

APPENDIX 4
Record of Deposition of Voucher Specimens*
The specimens listed on the following sheet(s) have been deposited in the named museum(s) as samples of

those species or other taxa, which were used in this research. Voucher recognition labels bearing the
Voucher No. have been attached or included in fluid-preserved specimens.

Voucher No.: 200‘]! ’ 0 8

Title of thesis or dissertation (or other research projects):

 

BIOLOGICAL EVALUATION OF NON-WADEABLE RIVERS IN MICHIGAN

Museum(s) where deposited and abbreviations for table on following sheets:

Entomology Museum. Michigan State University (MSU)

Other Museums:

Investigator’s Name(s) (typed)
Kelly James Wessell

 

 

Date 12-16-2004

*Reference: Yoshimoto. C. M. 1978. Voucher Specimens for Entomology in North America.
Bull. Entomol. Soc. Amer. 24: 141-42.

Deposit as follows:
Original: Include as Appendix 1 in ribbon copy of thesis or dissertation.

Copies: Include as Appendix 1 in copies of thesis or dissertation.
Museum(s) files.
Research project files.

This form is available from and the Voucher No. is assigned by the Curator. Michigan State University
Entomology Museum.

163

Appendix 4

Voucher Specimen Data

965

of 12 Pa

Page 1

 

 

EB EN _.

230

 

 

SE; coo...

 

.2 22:8on 8%: gone 05 8283.

 

he $00M 025823

6098

=ommo>> 8an 5.8.

@0832 3239303.

Some-mom: z 88% 38:68 33

 

 

 

 

 

 

 

 

 

 

 

 

 

 

:92 F .1 3:20 .8 E3. :2 2-3 83822
30.2 F m 5923.2 8 28022 .5. 9.3 .3 Reef 883.8
:w: P .1 3:20 .oo sex :2 3-3 .3 £883 822$
:22 v .m ooficaz .00 $252 =2 3-3 .3 intense: 823.5
:92 P .m 9.20 .00 E3. =2 9-3 a. Sewage: Sufism
392 F .m 3263 do 26:53 =2 ~73 3 533m 3.8223
:ws. P .m 33222 .00 owe-”Ens. =2 3-3 an 38332 £886.30
:92 F .1 86.5.2 do $2.22 :2 9-3 .3 32105. 890.30
:92 r .m 3:80 .oo :5. =2 8-3 82682.5
:ws. P m Ease-as. 8 2822 =2 8-5 895520
392 F .m owe-Ems. .8 SEE: =2 3-3 .3 38°26 880820
:22 F m Swim 8 @252 =2 8-3 .8 3.5.-fin «82088.50
:22 F .m .333:- 00 26:33 =2 no-3 .8 unseat Sacaaofioo
:92 F m 9.80 oo 28.. =2 3-3 .3 888.28 2.8.888an
:92 F m Swim 8 08.8.2 :2 3-3 an .38 88.88%:0
:w: P m Swim 8 8.522 =2 No-3 an 838823. «£888.50
:92 F 1 539322 o0 39.52 =2 3-3 an :35... 08.03%

m N no 0. s 328300 ecu

m m M. m Pm m w W m % com: 8 382.8 2058on .8 Sec .33 :98. .050 .o .38on

m m a... a M M m m a a.

«o .3832

164

Appendix 4

Voucher Specimen Data

of 12 Pages

Page 2

230

.9950

 

. .anmss. 30.2..95

vofi ENF

030

 

5.92:: 2% 2.3.5.2 2.. s .388
.o. mcmEBoam new: o>onm 2.82831

o53> Ono..-

 

 

oz .ozo=o>

68....

=88; woe-.2. 2.3.

@0832 323.59...

3.388: .. £32m 6.5.283 83v

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2w... . .m 83%.... .0 cases... .5. 8-8 a sage .5322...
8.2 F .1 83mg... .8 288.2 .5. 8.8 a. wannabe ass-ea...
Ems. F .m commlxmzfi .oamimz E2 mo-No .3 2.255 ooc._.o.oEocqm
8.2 F .m 8.28.... .8 names... :2 3-8 a... 232828 82:80
Ems. F .m 5.2mm. .00 @0322 :2 mo-No .am «.330 82:30
:ws. . .m 82%.... .oo 8.9%... .s. 8.8 a... 328.32.. agate
:92 F m. 30:30:22 .00 ooEEocms. =2 3.3 .3. 338m 828.53
Ems. F .m gamma... do come. :2 3-3 am as... 82.2.:
:92 F .m 525. .8 8.8.2 =2 8.3 an 23... 82.2.:
322 F .m www.cms. 60 $6.35. =2 3.3 .3 38:5 82.3:
Ems. F .m 238.. .3 .00 Stem =2 v~-3 an 23...... 82.2.:
:92 F .m. $6.35. 60 $62.22 :2 8.3 an 2.8.2.. 82.2.:
:92 F .m www.cms .00 0235.2 :2 «mi. .3 38.5. 82.2.:
:92 F .m $5.35. .00 85.5.2 :2 FN-Fo a... 8.8.35 82:33
:22 F .m gamma... .8 88. :2 8-5 a fleece 32.5.5
392 F .1 531.82 .oflmmzoz =2 3.3 an 53.356 82.355
8.2 . .m Bee oo :3. =2 2.8 a eases 82.325
e 10 0+ 8

m m .m m m...- m m .m- w 5 02.832. new :98. .220 .o 3.0on

m. .an. W h m w W m. w w com: .0 882.8 25......on .o. 32. .83..

M w m d m A A P N m E

 

 

”.0 .8632

 

 

165

Appendix 4

Voucher Specimen Data

ges

of12 Pa

Page 3

200 .2050

 

33 {NF 050

 

. €02.35. >00_0Eo.:m
33.02:: 205 50.5.5. 2: :. 200000 2.55 000...
.0. 0:02.880 20.2. o>000 05 0020001 =0mmo>> 02:0... :3.

 

 

 

oz .2820> €098 @0502 9.203005.
308.80: .. 202.0 6:02.000 3:.

                        
 

F

F .nm 08.0.8020
F

F aw 08222.02:
F .% m:Ew.30£:<
F .00 02023
F .am 030.3033
F .3 0.20.88.
F .3 082.0205
F in 3.02.0.0,
F .00 30025.5.
F .3 0.000.033
F .% 0.:000301
F .00 0.2620on
F .00 0.0523m
F

00,332.:

00220.00
82.82828.
82.80
8822.80:
8258588
ofl2>o.8.E>_om
82.82853
88.22.82
82.50.38:
82.8.8.8...
82.8.8.8...
32.50200...
32.50050...
002.0Eocnm
om2.0:.ocaw
80......oEofim

 

    

02.880
.0 2202.00 20:52.0 .0. 200

     
 

:98. .220 .0 3.025

Appendix 4

Voucher Specimen Data

of 12 Pages

Page 4

250 .2830

 

33 EN F 800

 

pE=omsz BoBEoEm
5352:: 22w 59.2.2 as s 28%.. £55 88
.2 mcwEBam now: o>ona 2: 3283. =emmo>> moEw: =3.

 

 

 

 

oz hozoao> 68.8 @952 muofinfigs
3.888: = 98% 35:63 83

   
                                    

.3 3280.30 ou2E_m
.3 329828: 32Em
.am m3§c80u2 82:5
.3 £32.58 825m
.3 niggo 3256
.aw xxce‘cccx ou2E_m
.am xxcoioc< ou2E_m
.am .uanquotnoZ 323:8
.3 32383 328:5
.nm mitooufl 828:8
.3 «59.8an 320220
.3. wifibxl 828:5
.am 32:28:00 8232.5
.nm 3233300 32025
dm 35300 323:5

82:23.30

     
 

37,88

:98: $50 .o 8630
8 882.8 22:8QO .2 Sun

167

Appendix 4

Voucher Specimen Data

of 12 Pages

Page 5

050 503.50

 

33 SN F 200

 

. .Esomsi BoBEoEm
£235 220 $952.2 05 s 28%.. 222, 2.8
.8 mcoEBoam :06: 02.50 05 32301 .8000; 8an =3.

 

 

oz .053; €098 52:02 $303002.
30380: x 0.005 68:68 002

          
                               

2802203

.00 029800..
.3 035803
.3 02:85
.00 030900
.3 03023
.3 8$3=0a
.3 8:00:00
.3 029.0:
.00 022.0:
.90 0:220
.00 032.50
.% 350:5
.am 2.50:5
.am flE_0c0.w

mE_oc2m

002030:
8 880:8 m:0E_ooam .8 San

 

   

  

$228.2:
$228.2...
$228.1...
$228.1:
$225220»:
$3231:
32.9.0:
82.2.0:
325...“...
82.9.0:
oupciw
82ch
82ch
83ch
$25.0
$25.0

  

:98: .050 5 006000

168

Appendix 4

Voucher Specimen Data

of 12 Pages

Page 6

0.00 .2050

 

. 0.50022 30.060Em

£9020: 0.0.0 000225. 05 5 07.0000
.00 0000:0000 00.0: 0>000 05 002000”.
02 000030)

 

                   
   

  

FFFFFFFFI—F

.0 00000.80 0:0E_0000 .00 0.00

40%|

0.00

 

 

 

 

.00
.00
.00
.00

.00

00:00000

=0000>> 0080.. :3.
800.6 @0502 £809.02...
$000000: = 00000 0020000 003

05:5 000...

£03002
02001
00000000:
020080an50
0500002
02008.02
00300005
0000:”;

£0ES020Q
00500,
0000:0000
0.50am
0000:0080»:
00:50:08.5
0000200020.

 

  

03:038.).
00030:
0000.00
0000000
82:00
82:00
000.000
8300
002x80
000000.233
000200
0020020000
0020000000
002.202
832091:
000.2021:

  

00x2 .050 .0 006000

169

Appendix 4

Voucher Specimen Data

of 12 Pages

Page 7

 

Qan— .O«8:0

 

09m EN F 200

 

. 0500:: 30.088...»
5.0.020: 0.0.0 000.50.: 05 0. 000000 2.55 000...
.0. 000:..0000 00.0.. 0>000 05 0020001 ..0000>> 000.00 :00.

 

 

oz .0..0=o> F000»: @0802 01.203002.
9.000000: ._ 0.00:0 6:00.000 003.

       
                        
      

.00 0.09.3 000.0280.
.00 x05. 000.0200<
F .00 2.0.0 80200
F .00 0.00an 00030.0
F .00 0.0.0 0020.0
F .00 08232.0 80.00.08
F .00 0.500001 002.05..
F .00 580020.... 000.330
F .00 0.5.33 822$
F .00 0.3.802 08:32.32
F .00 00.00.000.00 0000005.).
.00 0.00) 0005.. 000...0>
.00 000) 9.0.3. 000___0>
.00 0.0.00 82200
.00 00.0002 000.20

00003 0000000202

  

02.00000

:98. .050 .0 00.000w
.0 00.02.00 0:0E.0000 .0. 0.00

170

Appendix 4

Voucher Specimen Data

995

ofj2_Pa

Page 8

 

0.00 .8050

 

. 05000.2 30.00.20m

0.0.025 220 000.00.... 05 0. 0.8000
.2 000000000 000.. 0500 05 0020000
02 .000:o>

 

      

Fi—FFI—FFFF‘I—V—FFT‘FI—

3% SN F 200

 

 

02.00000
.0 0082.8 000000000 .0. 200

 

0FE>> 00°...

__0000>> 00.00.. =00.
0.00.8 @0502 0.200009...
9.0000000 .F. 0.0000 000.0000 003

.00 00.200280

 
                                 

.00 0.00000
.00 000000000
.00 0000002
.00 0.50083
.00 00.00..
.00 0.00000.
.00 000%00000000
.00 0.9.x
.00 00000.0...

.00 500.0200
.00 0.00.00.50,

.00 003.00.
.00 00000.00
.00 00000008002
.00 0:200:

    
  

000.00.000.80
000.000
000.000
000.000
000.000
000200..
000.003.0080
000.02.02.80
03.00000080
80.002028
000.900.00.00
0023:0003
00020.80
80.00.80
002.528
82.328

08000 .050 .o 00.000m

171

Appendix 4

Voucher Specimen Data

Page 9 of 12 Pages

mafia gouge-O

 

.Esomsz >ao_oEoEm

£2025 22m... $9225. 05 5 «$82.
.8 mac—£890. 3E. o>8m c5 328$.
02 85:;

 

      
              
     

v-Fv-v-u-I—r'v-u-u-

Eugen
8 862.8 mcwEBoaw .8 8%

3?. {NF 2am.

 

 

was; 000...

 

:33? $62. =3.
63.6 @0652 eofigugs
9.838: a £85 38528 83

an £5525
a» sausage
.3 ~3§§
.3 £5»:
.am 3:23
.am siEeol
.3 2352332.:
.3 3033\be
.am ocoxmqenanco
.am ocuxoqooaol
.am ocuxaoofil
.nm £3821
.3 $3881
.3 maquoooicm
‘am mEcwEoE

«8:382:
5:322:
o8___ao§:
ogioix
Sgioi:
$259,821:
aficgaegx
08203921:
8252,8211
mecgmaegx
omgggaoga:
325332.01
$228886
mmuszowowaO
82803985
wmucfioquEm

  

:95 550 .o 36wa

172

Appendix 4

Voucher Specimen Data

Page 10 of

12 Pages

 

Sun. .2830

 

33 {NF 650

 

. .Ezmmss 30.0635

 

€525 22m c8225. 9: 5 :83.
.2 mcmevgm 8E. o>ona as 828?.

 

858%
5 88260 mcoEEmam .2 San

 

 

oz coco=o> A398

.8825 862.

2.55 cue...
=3.

@oEmz €089.82.
33800: t 98.7. .2368 82

.am
.am
.am
.am
.am
.am
.9.
.am
am
.am
.am
.am
dw
.am
.3

25°an magnoaozcoozoa

3:3ch unnocwgca
E330 o%_ea§o__§
E255 32%.822“.
ocuxwaocuxm wuBancES
28: 0858an
383m 035853
«.380 3308an
£980 03:08:34
ocuxmqeooz 03:30.9...
ofiimqotoz 32.3053
8208me 03:823..
quEmo 825833
.5830 30:88am:

mEemoEQS wacszofioEaB
$3322...

:95 850 .o 88wa

173

Appendix 4

Voucher Specimen Data

f 12 Pages

e110

 

Pag

Ban. ‘ .2230

 

 

.anmas. 30.2.8.5
5.22.5 28m $95.: a... c. .88..
.8 206.8% 3%.. m>3w 05 328mm

 

:38? 88!. =3.

33 {mp .250

2...; Utah

02 .o..o=o> €an AmvoEaz 9.23.605.
9.888.. = £82m .2368 33

.3
.nm
.3

.am

8:88..
.0 882.8 mcoEEeam .8 San

~.>.m>.m
”Seneca
«venom.
23.5055
€82.54.

85.3....
33:89.0
3.58.1...
xwxcqooz
33.05me
093

e331
maEEeobol
xaicqouoxcflnl
83.08302

 

83.22
8833.0
3863.0
3083.0
3083.0

Swarm

«2.85

3.85
3.5.0.5.

2.0.2.0...
Seizes»...
engaging
828.2282...
8283.83...
$288.28an
meuoao..coo>.om

:98. .050 .o 8.0on

174

Appendix 4

Voucher Specimen Data

San. . .8830

2:025. >8.oE3:m

3.29.5 28m 595.... 2.. c. .83..
.2 29:8on 808.. 023.. 85 80280”.
02 .2025 68.6

 

 

    
                    

ges

12 Pa

_—

f

 

e120

Pag

8.88%
.0 882.8 22.888. .2 Saw

 

3B {NF 650

2.55 28...
.8825 263. :3.
@0sz 223.802.

82...?on
382.20
382.20
382.20
389.20
382.20
382.20
382.20
382.20
«882.20

«.295

33302 = $028 .2058.“ 33

808602
23.88
82505.
«33:05.
«83:05.
83.22
82.85.
32.85.
802.22
82.85.
82.8.2
89522

  

:98. .28 .o 28on

    

175

Ilffimfi‘llflijufilflflflLfiflfllflflfifiu