3 .S‘
.unﬂv.
Km...

évgu

 

.. 5......
.2 .. .

 

Im11mlIllilﬁllllllnllllflujIll“

31293 01812 7

This is to certify that the
dissertation entitled

ECOLOGICAL IMPLICATIONS OF COARSE WOODY DEBRIS
IN LOW GRADIENT MIDWESTERN STREAMS

presented by

Lucinda Ballard Johnson

has been accepted towards fulﬁllment
of the requirements for

Ph.D. degree in Zoology

 

W.

Major professor

Date 8/0757 7 7
/ .7

MS U i: an Afﬁrmative Action /£qual Opportunity Institution 0-12771

 

 

LIBRARY

Michigan State

University

 

 

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1/98 mm.“

 

 

ECOLOGICAL IMPLICATIONS OF COARSE WOODY DEBRIS IN LOW
GRADIENT MIDWESTERN STREAMS

By

Lucinda Ballard Johnson

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Zoology

1 999

 

V!

Ci

ABSTRACT

ECOLOGICAL IMPLICATIONS OF COARSE WOODY DEBRIS IN LOW
GRADIENT MIDWESTERN STREAMS

By

Lucinda Ballard Johnson

Coarse woody debris (CWD) is an important component of small to medium
streams in forested regions, directly inﬂuencing stream geomorphology as well as many
ecosystem properties and processes. Little is known about the inﬂuence of the landscape
context on standing stocks of CWD. The goals of this dissertation are to: 1) characterize
the abundance, size, and distribution of CWD in low gradient streams of developed
watersheds; 2) quantify the relative inﬂuence of reach- and catchment-scale factors on the
abundance, distribution, and retention of CWD, and 3) examine the relationships between
CWD, channel form, habitat structure, macroinvertebrate community structure and
macroinvertebrate species traits.

CWD standing stocks in these Michigan streams are small compared with forested
streams, and strong interactions between land use and surﬁcial geology inﬂuence its
abundance and distribution. CWD accumulation density and distribution are well
predicted by the environmental variables measured in this study. Factors at the local
scale (e.g., bank-full width, percent of open canopy, and riparian vegetation type) have a

large inﬂuence on the density and distribution of debris accumulations, but only a

moderate inﬂuence on CWD abundance and volume. In contrast, landscape features
including link number, percent urban land use in the catchment, and topographic
heterogeneity, exert greater control over CWD abundance. The debris accumulations in
these Midwestern streams are smaller and contain fewer and smaller logs than streams of
forested regions. Debris accumulations do not play a physical role in modifying channel
morphology, as do debris accumulations in forested landscapes.

Although woody debris is not abundant in these streams, it is one of the most
important habitats for macroinvertebrates in the streams of the Saginaw Basin. Since
woody debris can occur in both fast and slack water, the taxa found in association with
wood habitats span a range of current preferences, as well as functional and habit traits.
The patterns in‘the distribution of habit and functional traits within wood habitats
suggests that these traits may vary with the location of woody debris in the channel
relative to the ﬂow regime.

Log retention and recruitment is in dynamic equilibrium, with the logs exhibiting
the greatest movement being the smaller logs. Approximately 50% of the logs at a site
were replaced between October 1995 and June 1996 following a ﬂood with a return
interval of approximately 5 years. Although the number of logs present before and after
the ﬂood remained approximately equal, log volume was greater before the ﬂood than
after. Management efforts to retain woody debris in streams must consider both local as

well as landscape-scale factors.

Dedication

To Randy, thank you for your love and support throughout this endeavor, and my

parents, Margaret and Martin Johnson for providing the incentive and love of learning.

F?

\i;
en.
su;
Sit.
m)

Era:

 

Acknowledgments

I sincerely acknowledge the contributions of Carl Richards and George Host for
their collaborative efforts and encouragement throughout this project. Roger Ham and
Dan Breneman contributed extensively to the collection of stream habitat data, and the
compilation / oversight of stream databases. I also wish to acknowledge the valuable
contribution of Tom Sampson in the collection of the woody debris data in 1996. Connie
Host, Dan Fitzpatrick, Shawn Boeser, and Gerry Sjerven were all instrumental in
assembling spatial data used in this project and performing relevant analyses when
necessary. Tom Hollenhorst and Amos Ziegler were extremely helpful in assisting with
preparation of graphics. I also gratefully acknowledge the assistance of Greg Grunwald
with the interpretation of the regression results. I am grateful to Dr. Gerald Niemi for
encouraging me to embark on this endeavor and providing the institutional and moral
support. I also wish to extend my appreciation to Thomas Burton (dissertation advisor),
Stuart Gage, Donald Hall, and Richard Merritt for their support and advice throughout
my program at Michigan State University. Funding for this project was provided by a
grant from the Environmental Protection Agency to Carl Richards, myself, and George
Host (Grant Number CR-822043-01-O), with additional monetary assistance from the
Department of Zoology at Michigan State University. The Natural Resources Research
Institute of the University of Minnesota, Duluth provided salary support during my
graduate program. Last, but not least, my husband Randall Hicks is gratefully

acknowledged for his love and support during this effort.

Table of Contents

LIST OF TABLES ......................................................................................................... viii
LIST OF FIGURES ....................................................................................................... xii
LIST OF ABBREVIATIONS AND SYMBOLS .......................................................... xiv
CHAPTER 1

Introduction ................................................................................................................... 1
CHAPTER 2

Channel, Riparian, And Landscape Features as Predictors of Coarse Woody Debris
Abundance in Midwestern Streams .............................................................................. 20
APPENDIX 2.]

Summary statistics of local / channel variables. N = 49 ............................................... 88
APPENDIX 2.2.

Characteristics of the riparian zone across 49 sites ....................................................... 88
APPENDIX 2.3.

Characteristics of catchments in the four landscape treatment groups .......................... 89
CHAPTER 3

Coarse Woody Debris Accumulations in Low Gradient Streams: Relation to Local And
Landscape Features ........................................................................................................ 90
CHAPTER 4

Coarse Woody Debris Retention And Recruitment in Low Gradient, Agricultural
Watersheds ..................................................................................................................... 1 34
APPENDIX 4-1.

Description of catchments and streams .......................................................................... 149
APPENDIX 4-2.

Description of channel-scale properties ......................................................................... 150
APPENDIX 4-3.

Wood debris abundance measures ................................................................................. 15]

vi

CHAPTER 5
Macroinvertebrate Community Structure and Function Associated with Coarse Woody
Debris in Low Gradient, Agricultural Streams .............................................................. 154

APPENDIX 5-1

Taxa closely associated with CWD (wood-associated or wood-dominant) and those that
are not found in CWD samples (wood-absent, wood-averse). Categories are described in
the text. Trophic relationships (collector - (gatherer or ﬁlterer), scraper, predator,
shredder), habit (clinging, climbing, swimming, sprawling, burrowing, planktonic), and
dominant habitat (erosional/depositional; erosional or depositional) also are listed....1 81

APPENDIX 5-2

Total number of taxa found at a site, in non-woody debris habitats, and in woody debris
habitats, and the number of unique taxa occurring only in wood habitats .................... 185
CONCLUSIONS ........................................................................................................... l 86
LITERATURE CITED .................................................................................................. 195

vii

fl ll

 

 

 

 

 

 

Ia'

Tat

Tat

List of Tables

Table 2-1. Summary of literature on coarse woody debris effects on streams and stream
ecosystems.

Table 2-2. Distribution of sample catchments and sample reaches (in parentheses) among
factor levels in the factorial design. Agricultural lands have a minimum of 60% of
land under production.

Table 2-3. Coarse woody debris variables measured during the study.

Table 2-4. Size classes of debris accumulations based on methods of Shields and Smith
(1992). X = channel width at the upstream point of the debris accumulation. The
sum of all debris accumulation sizes in each reach represents a measure of the
amount of channel covered by debris accumulations.

Table 2-5. Channel morphology and physical habitat variables measured during this
study. * Indicates the variable was used to predict the abundance and distribution
of CWD. Fast and slow units are deﬁned in Hawkins, et al. (1993) and represent
rifﬂe and pool habitats.

Table 2-6. Variables describing riparian zone structure and composition. (* indicates
variables used in most analyses.)

Table 2-7. Landscape variables used in analyses and spatial data used to derive data.
(U SFWS = US Fish and Wildlife Service, USGS = US Geological Survey.) (*
indicates variables used in most analyses; others are omitted due to high
correlations with other landscape variables.)

Table 2-8. Aggregation classes for Quaternary geology categories.

Table 2-9. Moran’s I statistic testing spatial autocorrelation among sites on the Bad River.
("' Indicates signiﬁcance at p = < 0.05 level with a Bonferroni adjustment).

Table 2-10. Mean : SE. of coarse woody debris measures across land use (agricultural
versus mixed) and Quaternary geology (lacustrine versus morainal) treatments.
Wood diameter and length measures are only for sites containing woody debris.
Number of sites where a metric is > 0 are also shown. (N = # sites). (*** =
significant effect due to LU x Geol; p < 0.001 ).

viii

Table 2-11. Comparison of discriminant functions and multiple regression models based
on different spatial scales for coarse woody debris abundance and distribution. (*
model applies only to sites where CWD volume was >0; n = 30). Direction of the
correlations are indicated next to each independent variable. The best model for
each variable is listed in full. Presence/absence refers to CWD volume > 0
(presence) or volume = 0 (absence); see text.

Table 2-12. Comparison of large woody debris densities for Saginaw Basin streams in
Michigan with published values for other low-gradient or disturbed streams.

Table 2-13. Two-sample t-test calculated for selected variables in lacustrine agricultural
(Lac/Ag) and morainal mixed (Mor/Mx; Group 1) versus lacustrine mixed
(Lac/Mx) and morainal agricultural (Mor/Ag; Group 2) catchments. Bonferroni
adjusted probabilities are for pooled variance.

Table 2-14. Associations between predictors at the local and riparian scale and
landscape, riparian and local scale variables across the entire study area (All), and
within lacustrine (Lac) and morainal (Mor) landforms. The variables % open
canopy, mean bankfull width, riparian vegetation height, and riparian zone width
are strong predictors of CWD in the regression models of CWD standing stocks
(discussed in the text). Symbols reﬂect Pearson correlation coefficients. -l-+-+- = p
<0.001; ++=p<0.01; += p<0.1

Table 2-15. Associations between predictors at the landscape scale and landscape,
riparian and local scale variables across the entire study area (All), and within
lacustrine (Lac) and morainal (Mor) landforms. The variables link number, urban
land use, S.D. elevation, and outwash sand and gravel are strong predictors of
CWD in the regression models discussed in the text. Symbols reﬂect Pearson
correlation coefﬁcients. n.a. indicates that there is no coarse till associated with
lacustrine landforms. -H-+ = p < 0.001; ++ = p < 0.01; + = p < 0.1.

Table 3-1. Description of debris accumulation types, aggregated classes, and attachment
points of debris accumulations in the channel. All accumulation are greater than l
m2 in area. Aggregated types were used to assess the combination of size and type
of debris accumulations. (Variable name in parenthesis.)

Table 3-2. Size classes of debris accumulations based on methods of Shields and Smith
(1992). X = channel width at the upstream point of the debris accumulation.
Classes 1-3 were aggregated to form the size category = small; classes 4-7 =
medium; classes 8-10 = large. The sum of all debris accumulation sizes in each
reach represents a measure of the amount of channel covered by debris

accumulations.

ix

Table 3-3. Coarse woody debris accumulation abundance and size variables measured
during the study.

Table 3-4. Summary of debris accumulation types across 49 stream reaches.

Table 3-5. Mean and S.E.M. of the debris accumulation types x size classes from 38 sites
with debris accumulations. N = number of debris accumulations of each type
encountered.

Table 3-6. Summary of debris accumulation attachment locations for each accumulation
type. Missing data indicate categories that were not likely to occur due to the
character of the debris accumulation type.

Table 3-7. Results of redundancy analysis; ﬁt as a fraction of the variance of woody
debris variables.

Table 3-8. Comparison of large woody debris densities for Saginaw Basin streams in
Michigan with published values for other low-gradient or disturbed streams.

Table 3-9. Summary of the mean number of debris accumulation and median size by
stream order and channel width. (Streams without debris dams are excluded from
this analysis; n = 39.)

Table 4-1. Descriptive statistics for the proportion of logs and log volumes ﬁ'om TI
retained (retention) and the proportion of newly recruited logs and log volumes
measured at T2. Ratios of logs at each sample period, recruited to exported, and
retained to recruited also are shown.

Table 4-2. Logistic regressions predicting whether or not a log is retained from log
dimensions and volume. CCR % = percent correctly classiﬁed. (N = 443 logs).

Table 4-3. Pearson correlation coefﬁcients for wood retention and recruitment versus
ﬂow and channel characteristics, including standing coarse woody debris per 100
m reach (see Johnson 1999a for a description of methods), and mean dimensions
of logs in the reach. The proportion of logs and log volume exported are inversely
related to the proportion retained and are not listed in this table. (Signiﬁcant
coefﬁcients in bold are corrected using Bonferroni methods.)

Table 5-1. Abundance of macroinvertebrates from ﬁve sample types. Average per
samples were derived by dividing the total number of individuals by the number
of sites in which the habitat was represented.

Table 5-2. Taxa forming the numerical dominant members of the macroinvertebrate
community present in woody debris samples.

Table 5-3. Functional feeding groups of taxa strictly associated with coarse woody debris,
and taxa averse to coarse woody debris habitats. Macrophyte piercers represented
less than 4 % of the total taxa. and taxa with unknown characteristics were about
1.5% of the total pool of taxa in these groups.

Table 5-4. Habit modes of taxa strictly associated with coarse woody debris, and taxa
averse to coarse woody debris habitats. Taxa with unknown characteristics
represent about 1.5% of the total pool of taxa in these groups.

Table 5-5. Pearson correlation coefﬁcient from number of unique taxa contributed to the
total taxa richness from wood and non-wood habitats (pools, rifﬂes, runs,
macrophytes) versus standing stocks of coarse woody debris measured as m / m2,
m / m3. and a measure of the amount of channel covered by debris accumulations.
Signiﬁcant values are adjusted by Bonferroni corrections.

Appendix 2.1. Summary statistics of local / channel variables. N = 49.

Appendix 2.2. Characteristics of the riparian zone across 49 sites.

Appendix 2.3. Characteristics of catchments in the four landscape treatment groups.

Appendix 4-1. Description of catchments and streams.

Appendix 4-2. Description of channel-scale properties.

Appendix 4-3. Wood debris abundance measures.

Appendix 5-1. Taxa closely associated with CWD (wood-associated or wood-dominant)
and those that are not found in CWD samples (wood-absent, wood-averse).
Categories are described in the text. Trophic relationships (collector - (gatherer or
ﬁlterer), scraper, predator, shredder), habit (clinging, climbing, swimming,
sprawling, burrowing, planktonic), and dominant habitat (erosional/depositional;

erosional or depositional) also are listed.

Appendix 5-2. Total number of taxa found at a site, in non-woody debris habitats, and in
woody debris habitats, and the number of unique taxa occurring only in wood
habitats.

xi

List of Figures
Figure 1-1 Study area in central Michigan, USA.

Figure 1-2. Quatemary geology in the Saginaw Basin. Michigan (from F arrand and Bell
1984).

Figure 1-3. Land use in the Saginaw Basin, Michigan (from the MI Department of
Natural Resources, MIRIS database).

Figure 1-4. Elevation across the Saginaw Basin, Michigan (from USGS Digital Elevation
Model data).

Figure 2-1. Distribution of log sizes.

Figure 3-1. Examples of four debris accumulations types. a). Vegetation plus trapped
debris; b) Root Wad plus trapped debris; c) Log / snag; (1) Loose Logs.

Figure 3-2. Mean and SE. of # debris accumulation/ 100m, debris accumulation size,
and the sum of the accumulation sizes (see text) across land use and geology
treatments. Results of two-way ANOVA are indicated.

Figure 3-3. Mean and SE. of # debris accumulations / 100m of four accumulation types
across land use and geology treatments. Results of two-way ANOVA are
indicated.

Figure 3-4. Mean and SE. of # debris accumulation attachment locations /100m across
land use and geology treatments. Results of two-way ANOVA are indicated.

Figure 3-5. Species-environment biplot derived from species scores from a redundancy
analysis of debris accumulation types and landscape, riparian and local variables
including catchment area (catch area), SD elevation (SD elev), stream density
(Strrn Dens), % urban (urban), % wetland (Wetland), Agriculture : F orest+Range
(Angor-arg), lacustrine sand (Lac Sand), lacustrine clay, coarse till + sand and
gravel (CT+SG), instream cover, ﬂood height (Fld Ht), % slow units, maximum
depth in slow units, mean bank-full width (BF W), % open canopy, and riparian
zone width (Rip Wid).

Figure 3-6. Mean and SEM of the number of debris accumulations across three channel
width classes. a) All accumulation types and sizes combined. b) Accumulations
grouped by size. c) ‘Log / snag’ accumulations grouped by size. Note change in
Y-axis scale between A and B, C.

xii

Figure 4-1. Regression predicting the proportion of logs retained and recruited from
log(ﬂood height).

Figure 4-2. Regression analysis predicting the proportion of logs retained and recruited
from bank-full width.

Figure 5-1. Map of study region.

Figure 5-2. a) One way analysis of variance results testing for differences in the
abundance of macroinvertebrates in woody debris samples among functional
categories. (Shown are means and SE). b) One way analysis of variance results
testing for differences in the abundance of macroinvertebrates in woody debris
samples among collector categories. (Shown are means and SE.)

Figure 5-3. One way analysis of variance results testing for differences in the abundance

of macroinvertebrates in woody debris samples among locomotor behavior types.
(Shown are means and SE).

xiii

List of Abbreviations and Symbols

CPOM = coarse articulate organic matter
CWD = coarse woody debris

Lac/Ag = lacustrine / agricultural
Lac/Mix = lacustrine / mixed

Mor/Ag = morainal / agricultural

Mor/Mix = morainal / mixed

xiv

 

Chapter 1
INTRODUCTION

For decades, streams managers regarded woody debris in the same vein as beavers
and old tires-- objects that must therefore be removed from stream channels because they
are either unsightly or impede ﬂow. Interest in angling and stream restoration has
highlighted the role of woody debris in controlling / inﬂuencing many physical, chemical,
and biological characteristics of stream ecosystems. Coarse woody debris (CWD;
deﬁned variously as wood 2 5 or _>_ 10 cm in diameter and 2 1 to 2 m in length), plays an
important role in shaping stream channel structure by altering channel pattern and
dimensions, and creating plunge pools and backwaters (e.g., Swanson, et al. 1976, Bilby
1984, Nakamura and Swanson 1993, Richmond and Fausch 1995). These alterations in
channel morphology affect ﬂow processes, thereby inﬂuencing hydraulic retention (e.g.,
Trotter 1990, Ehrrnan and Lamberti 1992, Raikow, et al. 1995), sediment transport and
storage (e.g., Bilby 1981, MacDonald and Keller 1987, Bilby and Ward 1989, Nakamura
and Swanson 1993), bank erosion (e.g., Murgatroyd and Teman 1983, Shields and Smith
1992, Smith, et al. 1993), and timing of peak ﬂood events (MacDonald, et al. 1982,
Gregory, et al. 1985). While most physical alterations by CWD are found in low-order
streams, historic records reveal signiﬁcant channel alterations in large rivers as well (e.g.,

Triska 1984).

Physical and chemical alterations to the channel and water column induce a
cascade of effects on ecosystem processes that affect primary and secondary production,

1

 

pr.
Sr.
Sig

 

(ll

DR.

era

and other trophic interactions. Debris accumulations trap organic matter. and their
removal results in a net export of dissolved organic carbon (DOC), ﬁne particulate
organic matter (F POM) and coarse particulate organic matter (CPOM) (Bilby and Likens
1980, Bilby 1981). Channels with cobbles and coarse woody debris (which behave as
periphyton substrates) show higher nutrient uptake (Aumen, et al. 1990) and thus,
presumably also have smaller spiraling distances (Newbold. et al. 1982). Fish,
invertebrates (Angenneier and Karr 1984, Lehtinen, et al. 1997) and the bioﬁlm
community (Shearer and Webster 1991, Hax and Golladay 1997) beneﬁt from increased
habitat heterogeneity in addition to CPOM retention. Fish respond positively to the
presence of coarse woody debris accumulations for cover, ﬂow and predation refugia, and
increased food availability (e.g., Angerrneier and Karr 1984, McMahon and Hartrnen
1989, Everett and Ruiz 1993, Culp et al. 1996). The invertebrate community responds to
the CPOM energy source by shifting feeding functional groups from scrapers and ﬁlter
feeders to collectors and predators (Anderson, et al. 1978, Benke, et al. 1984, Smock, et
al. 1989. Wallace, et al. 1995). Particularly in areas with unstable substrates, CWD
provides a stable substrate for both primary and secondary production (Benke, et al. 1984.
Smock, et al. 1989, Hax and Golladay 1997). Stream sections with added logs exhibit
signiﬁcantly greater secondary production than do sections without log additions
(Wallace, et al. 1995). In short, CWD positively inﬂuences many stream ecosystem
processes (see recent reviews by Harmon, et al. 1986, Gregory and Davis 1992, Gurnell,

et al. 1995; Table 2-1).

renter ..‘
transpo
clear d:
Mississ
ofEngi
C 0mm:

extensn

complex
and last;
delivers,
Origina.’
[he hisr;
and Cut-
Supply (:-
al. 1994. ,
land C0,.

praCllCef

 

In many regions of the United States coarse woody debris was historically a
prominent feature in streams. such that log jams stretched for kilometers on both small
and larger streams (Swanson. et al. 1976, Triska 1984, Maser and Sedell 1994). Debris
removal was originally initiated to provide unobstructed waterways for navigation and
transport corridors for harvested logs. In 1776 the US. Congress appropriated money to
clear driftwood from streams and rivers to improve navigation. beginning with the
Mississippi River. Woody debris removal remains an active role of the US. Army Corps
of Engineers (Harmon, et al. 1986), and is one of the primary roles of County Drain
Commissioners (locally elected ofﬁcials charged with creation and maintenance of an

extensive network of drainage ditches) in the state of Michigan.

Coarse woody debris abundance in temperate stream ecosystems is regulated by a
complex set of factors that act on the source of the wood itself, its delivery to the channel,
and lastly, on the myriad of factors that control its retention and mobility once it is
delivered to the channel. At regional scales, geomorphic features have regulated the
original vegetation of the landscape (Grimm 1984, Host and Pregitzer 1992), as well as
the historical and current land use/land cover patterns within the region. Both historic
and current land management factors in the riparian zone and the ﬂoodplain inﬂuence the
supply of the CWD (Hogan 1986, Murphy and Koski 1989, Evans, et al.1993, Ralph, et
al. 1994). Forested landscapes are increasingly fragmented by forest harvest as well as
land conversion for agricultural production and urban development. Silvicultural
practices alter species composition, number, and size distribution of trees in the upland,

3

retentrt
rip-aria:
maria-g:
cur-tent
highly r
Murph}

(Swans;

 

 

 

 

nulrients
lmPORar

Streams .

and thus modify the potential source and input rates of CWD, particularly in small to
medium-sized streams (Bilby 1984. McDade, et al. 1990, Fetherston, et al. 1995). In
agricultural and urban areas potential sources of woody debris as well as the stream
retention capacity are altered by management practices such as grazing, landscaping,
riparian vegetation thinning or removal, dredging and channelization. Although these
management activities can undoubtedly reduce the potential supply of in-stream CWD,
current riparian zone vegetation may not accurately reﬂect standing stocks of CWD. A
highly retentive stream channel can contain very old wood (Keller and Tally 1979,
Murphy and Koski 1989), which can predate the age of the current riparian vegetation

(Swanson and Lienkaemper 1978, Evans et al. 1993).

The riparian zone and the land-water ecotone mediate inputs of sediment,
nutrients. and particulate organic matter to streams, in addition to providing other
important ecosystem functions (Gregory, et al. 1991). The primary sources of CWD in
streams are derived from natural mortality. ﬁre, disease, insect damage, ice/snow loading,
and wind-throw damage to trees in the riparian zone or uplands adjacent to the stream
(Keller and Swanson 1979). Processes such as mass soil wasting, bank undercutting and
erosion, and ﬂooding transport this material into the stream. Beaver may be the primary
vector transporting large volumes of CWD to the channel in some systems (Naiman, et al.
1986, Maser and Sedell 1994). Alteration of the hydrologic regime resulting from stream
channelization, wetland ﬁlling, or urbanization frequently results in increased bank
erosion, one of the primary mechanisms of CWD input to low-gradient streams (Keller

4

hxdrol.

COESUL.

charms;
(Keller '.
initiated
to assim
Morpho.
fOTmatic
llll‘lUEnC.‘
formic;
('Braudm

Mention

Sueinns. 1'
ESSEmia] l
IOle of CC“

liable 2‘:

 

and Swanson 1979; Davis and Gregory 1994). Erosion processes are themselves
regulated by geologic factors (e.g., soil type. topography), vegetation cover type, and
hydrologic regime, in addition to anthropogenic factors including grazing, forest harvest,

construction, and agriculture.

Once wood is introduced to the channel it is either retained by obstructions in the
channel, or by the channel itself, if the log is large relative to the size of the channel
(Keller and Swanson, 1979, Keller and Talley 1979). A positive feedback loop is
initiated when wood is retained by an obstruction, resulting in debris jams that continue
to assimilate wood until the accumulation fails due to high ﬂow or some other factor.
Morphological changes in the channel that are attributed to CWD, including plunge pool
formation, lateral adjustments, sediment and organic matter retention, can themselves
inﬂuence CWD mobility and retention. For example, decreased ﬂow velocity due to pool
formation will result in decreased stream power and capacity to transport CWD
(Braudrick et al. 1997). Channel bars resulting from sediment and organic matter

retention themselves form obstacles for CWD.

Since CWD fundamentally inﬂuences both the structure and function of many
streams, identifying the myriad factors that regulate its abundance and distribution is
essential for understanding the fundamental factors regulating stream ecosystems. The
role of coarse woody debris in high- and low-gradient catchments has been well studied
(Table 2-1). Stream restoration activities that attempt to increase habitat heterogeneity to

5

calm
charm:-
cxacerr

10 EXLL'“.

Mldli'est

CondlllOf

 

enhance fish and invertebrate production frequently use log structures anchored in the
channel (Hunter, 1991). These structures are subject to failure. and can cause or
exacerbate existing problems (Beschta and Platts 1986). Recent studies have attempted
to examine the inﬂuence of log placement on the channel and the biota (Hilderbrand. et
al. 1997). Studies such as these fail to account for larger scale factors that inﬂuence both
the input and retention of CWD in streams. Landscape-scale factors such as land use and
surﬁcial geology inﬂuence the abundance of woody debris found in stream channels
(Ralph, et al. 1994; Richards, et al. 1996) and undoubtedly also play a role in mediating
the impact of disturbance events that inﬂuence the export of CWD and smaller organic
matter fragments. By examining the factors inﬂuencing large woody debris at a range of
spatial scales, the extent to which local and regional factors regulate the abundance and

distribution of CWD can be discriminated.

Context

This study is part of a larger study by researchers (Carl Richards, George Host,
and myself) at the University of Minnesota, Duluth to develop ecological indicators for
Midwestern streams. Ecological criteria are numeric or descriptive means by which the
condition of an ecosystem can be described with respect to designated water resources.
The use of ecological criteria is tied to the concept of ecological integrity, which is the
condition of aquatic ecosystems in unimpaired waterbodies. The concept of ecological
criteria is analogous to biocriteria (Barbour, et al. 1994) with the exception that
ecological criteria refer to the combined biological, physical, and chemical attributes of

6

any 02
ECOSVSIL

requires

IVDICGI K"

be ICSpC“:

ecologic;
inﬂ uencc
\t‘atershe~
determin.__

lnlcgm} ‘

dem0nsrr;
agrlCU‘ltur;

CommUni:

 

the impOr

 

ecosystems essential for maintaining sustained function, whereas biocriteria refer only to
the biotic communities that inhabit water bodies. Ecological criteria constitute a wide
array of parameters crucial to the assessment, maintenance. and restoration of aquatic
ecosystems. The development of ecological criteria for streams and their watersheds
requires the identiﬁcation of a select group of parameters that most strongly inﬂuence
stream ecosystems. Such criteria should be capable of deﬁning reference conditions
typical of unimpaired streams or streams with minimal anthropogenic stresses as well as

be responsive to ecological degradation found within the geographic region of interest.

The primary objective of the overall study was to develop watershed-scale
ecological criteria that quantify landscape and habitat factors that most strongly
inﬂuenced stream ecosystem integrity. In particular, we examined the inﬂuence of
watershed-scale attributes such as landuse/cover patterns and geomorphology as
determinants of the ﬁne-scale processes and conditions that impact the ecological

integrity of streams.

Previous work to identify biological criteria in the Saginaw Basin in Michigan
demonstrated the inﬂuence of Quaternary geology and land use (especially rowcrop
agriculture) as dominant landscape features that inﬂuenced habitat and macroinvertebrate
community structure (Richards, et al. 1996). In addition, CWD was identiﬁed as one of
the important reach-scale factors inﬂuencing the macroinvertebrate species traits
(Richards, et al. 1997). Since CWD was not a dominant feature in most of the streams in

7

1 _ a
31:3 sill-
‘1 v-
€\€.O.
. .1 _
lactoﬁ -

quantify

OVEI'Vlt'

based at
streams ,,
catchmer

examine

 

 

macroim

Emma

the study region. these ﬁnding were somewhat surprising. As a result, the project to
develop ecological criteria provided an opportunity to examine in greater detail the
factors contributing to the abundance and distribution of C WD in these streams. and to

quantify the relationship between CWD, habitat structure, and conununity composition.

Overview

The goals of the coarse woody debris project around which this dissertation is
based are to: l) characterize the abundance, size, and distribution of CWD in low gradient
streams of develOped watersheds; 2) quantify the relative inﬂuence of reach- and
catchment-scale factors on the abundance, distribution, and retention of CWD, and 3)
examine the relationships between CWD, channel form, habitat structure,
macroinvertebrate community structure and macroinvertebrate species traits. Speciﬁc

hypotheses that are addressed are:

1. CWD abundance and distribution is controlled primarily by local factors
(e.g., riparian zone structure and composition, channel features).
Local factors controlling CWD abundance and distribution are
themselves controlled by factors at larger spatial scales (e.g.,
dominant land use, Quaternary geology, topography, landscape
fragmentation).
2. The number, type, and size of debris accumulations are a function of
channel dimensions and topography. as well as landscape characteristics.

8

Debris accumulation type is a function of channel form, which is
controlled by landscape factors.
.3. Stocks of CWD within the stream reach positively inﬂuence channel form,
habitat structure, macroinvertebrate community structure, and
macroinvertebrate species traits.

4. CWD retention is inﬂuenced by log size, channel dimensions, and ﬂow.

This dissertation is composed of ﬁve chapters including the Introduction: 1)
Channel, Riparian, and Landscape Features as Predictors of Coarse Woody Debris
Abundance in Midwestern Streams, 2) Coarse Woody Debris Accumulations in Low
Gradient Streams: Relation to Local and Landscape Features, 3) Coarse Woody Debris
Retention and Recruitment in Low Gradient, Agricultural Watersheds, and 4)
Macroinvertebrate Community Structure and Function Associated with Coarse Woody
Debris in Low Gradient, Agricultural Streams. The goals of Chapter 2 are to:
characterize the abundance, size and distribution of CWD in low gradient streams within
streams in a highly developed landscape, and quantify the relative inﬂuence of reach-,
riparian-, and landscape-scale factors on the abundance and distribution of CWD. Coarse
woody debris has been studied primarily in mountainous streams of the Paciﬁc
Northwest; very few studies of CWD have been conducted in non-forested streams. This
study is intended to ﬁll this research gap, and to address hypotheses concerning the

regulation of CWD by factors operating at different spatial scales.

In Chapter 3 the number and distribution of four debris accumulations types were
quantiﬁed with respect to local, riparian and landscape variables. Debris accumulations,
rather than individual logs, are often the agent inﬂuencing channel geomorphology. In
contrast to forested streams. debris accumulations take on many different forms in the
Saginaw Basin. Many streams are characterized by thick overhanging vegetation in the
form of willow or alder. In many streams these structures are the only features within the
channel that can function as ﬂow refugia, habitat, or as geomorphic agents. Debris
accumulation types, including overhanging vegetation with trapped organic debris, root
wads with trapped organic debris, loose accumulations of logs, and log/snag jams were

examined in this chapter (Figure 3-2).

Chapter 4 reports the results of a log tagging experiment, conducted to examine
CWD retention in these streams. Individual logs within a reach were tagged and turnover
of these logs from the reach after a winter and associated spring ﬂoods was quantiﬁed.
Log size and volume. as well as channel dimensions and ﬂow characteristics were used to

predict log retention and movement.

The role of CWD in highly disturbed agricultural streams is largely unknown.
Chapter 5 assesses one potential role of CWD: the inﬂuence of CWD on the structure and
function of the macroinvertebrate community. The community of macroinvertebrates
unique to CWD were characterized, and the functional characteristics of the taxa that are

unique to the woody debris habitat were identiﬁed.

10

Si 14.1}

 

 

 

Study Area

The Saginaw Bay catchment of Lake Huron encompasses a 16,317 km2 region
(Figure 1-1), characterized by sand and clay-dominated lowlands rimmed by coarse-
textured glacial features such as ground moraines and outwash plains (Figure 1-2). The
study region is contained within two major ecoregions as deﬁned by Omernik and Gallant
(1986): the Southern Michigan/Northern Indiana Till Plains and the Huron/Erie Lake
Plain. Each is subdivided into two sub-regions. Soils in the lake plain are dominated by
medium- and ﬁne-textured loams ranging to clays. with sand in the outwash plains and
channels. These clay regions are extensively drained by artiﬁcial drainage and tile
systems. The periphery of the basin contains many coarse textured glacial features such
as ground moraines and outwash plains. The till plain exhibits the greatest variation in
basin topography and contains a high percentage of forested land intermingled with
agricultural land and old ﬁelds (Figure 1-3); elevations average about 278 m (Figure 1-4).
The entire drainage was logged for white pine and hemlock between 1840 and 1900, and

forests of the region now consist primarily of second growth hardwood species.

11

 

./\...I..~ ~ .::u¢.:~.u.-u.< ~:.-:7J.J -.- :35: .ﬁs::.f. .s s .t..:..r\.‘,.\

.<.m.D .cwwanz 3.550 E no.8 33m .2 8&3

12

 

 

 

./

A

L
L
(I.

 

_kv-«. -#\~\.

8x5 .822 \
8x5 25303 D
8556; 6:63:

23.30:? .2352

 

95:33 5.53m
65 manna—.36

E93 .55— 35.33

      
 

 

13

 

.a..._—.a.a....:_ .....aa:< .Antzac- :Lo— _1.:-:-:..:..- ......,.v :-1.:U.h/~ .:-?..;~ )¢::.:¢:.l. 3;. :.. >64. .T-Uhi .».:..:~-:::~.vv (I. ~ 3:...4...‘

.cozoohoa when? Ago. 20m 98 233m 88.: 5&2".va .Emnm Bacmwmm 05 E lac—cow $80830 N; 233m

14

                 

  

”1.8.9.... cm. to am

£5523 _IlL
.8: 9: sea New
.2 acne 85.87528...
2:3 i
.055 «.5 be? usﬁzuﬂ
=3 can in 3:583
.955 0cm gm c.3350 $889.09
36533 93 .996 can 83 59:8 .330 ME
5 _uonm 83in .3
a. 8.2.3.5236 3 $929,: new I
.2 3.2.2.95 .o 8559: 05 man...
9.8 8:0
.3 335 8.2.3.858 I

awe—cow
saiwam

15

 

.2. ~.—~....a.~..:.~ 7.54»: ’\ .«.d.l.~2:.£:<.
I—¥—_—/_ .2.J.4-.:.Z.4v- -:..::.2 up. -:.J.--..:-.JA— ——.)_ ta:- n...u-_sv ::ur«.:~.:—)L .:.-T.:¢~ >¢-2.~1:.f. .U.~s :.- .70)....v \ .41.: ~.:: ~ .~ I ~ .~:...\.~ \

\

£038.85 Eon? .AommnSaw
mam—2 .mooaomom :2an mo augmen— =z 2: 88.: 595:2 .Emam BuEmmm 05 E 8260 \ om: was .m-_ 232m

16

"m
9

sear—.22 8 om

  

    

 

uo>oU was

<H<G WEE

/
_...\\\\.O~Q>°U “UH-“q

BN:~W—wm

17

 

......:.J.u::n~ .r..7JC~/\ .2. .1th t (N x
np-pv-u Trivia-gun T...¢-—-.).J—.- .A-vuauua .id-vn-F.‘ n-..v.-~.I:J-..- —-n~.-.1-A - «I... v74. --.v.-. v .-.~u.‘.-.4-h.< v-114-n~ .Ia--.-i-..l. .J-- v.7.-~,J- --‘-~.r.v\w~ ‘4 \ ..d--V\.~..\

. .550qu £32 .8 wow - m2
80¢ 8ch weaves—m Emu .352 gouge—m 3:me mOmD :85 53:22 £623 Bacmwmm 2: 38% gouge—m .v; 835

18

on em "~— 0
2. .E 26.

 

.3395;—
keg—Rum

19

CHAPTER 2

CHANNEL, RIPARIAN AND LANDSCAPE FEATURES AS PREDICTORS OF

COARSE WOODY DEBRIS ABUNDANCE IN MIDWESTERN STREAMS

20

 

Abstru

LPQ'I ‘
gym“

area and
densin a
this Stud'

riparian \

 

 

Abstract

Coarse woody debris (CWD) is an important component of small to medium
streams in forested regions, directly inﬂuencing stream geomorphology as well as many
ecosystem properties and processes. Current land management practices can directly and
indirectly inﬂuence the abundance of wood in streams, especially in highly developed
regions. The goals of this chapter are to: 1) characterize the abundance, size, and
distribution of CWD in low gradient streams in developed landscapes, and 2) quantify the
relative inﬂuence of reach- and catchment-scale factors on the abundance and distribution

of CWD.

Strong interactions between land use and surﬁcial geology occur across the study
area and inﬂuence the standing stocks and distributions of CWD. CWD accumulation
density and distribution are well predicted by the environmental variables measured in
this study. Factors at the local scale (e.g., bank-full width, percent of open canopy, and
riparian vegetation type) have a large inﬂuence on the density and distribution of debris
accumulations, but only a moderate inﬂuence on CWD abundance and volume. In
contrast, landscape features including link number, percent urban land use in the
catchment, and topographic heterogeneity, exert greater control over CWD abundance.
The differences in the factors predicting CWD standing stocks versus accumulation
densities are probably related to the local-scale processes that entrain wood into debris

accumulations.

21

 

 

 

lntrod

streams
ecosx‘s:

t I
I

Grime.
habitats

irﬂuenc

depositi

by C \IT
interstic:
as invert
Sibley 1C
Oi’ipositi
im'erteb:
GOllaClay
1977) an;

ﬁlﬂCIIOn K

 

 

Introduction

Coarse woody debris (CWD) is an important component of small to medium size
streams in forested regions, directly inﬂuencing stream geomorphology as well as many
ecosystem properties and processes (e.g., Harmon, et al. 1986, Gregory and Davis 1992,
Gurnell, et al. 1995; Table 2-1). Woody debris exerts control over the structure of aquatic
habitats by impeding ﬂow, thereby increasing ﬂow heterogeneity in the channel,
inﬂuencing the pool-rifﬂe sequence, erosional processes. channel dimensions, and
deposition and retention of sediment and organic matter in the channel. Habitats created
by CWD are varied, including plunge pools, backwaters and eddies, as well as the
interstices of debris dams and individual logs. These habitats are critical for ﬁsh as well
as invertebrate species (Angerrneier and Karr 1984, Benke, et al. 1985, Beechie and
Sibley 1997), providing ﬂow and predation refugia for ﬁsh (Everett and Ruiz 1993),
oviposition and pupation sites (Dudley and Anderson 1982), a feeding platform for
invertebrates, and a substrate for bioﬁlm production (Shearer and Webster 1991, Hax and
Golladay 1997). The structure and dynamics of physical habitat in streams (Southwood
1977) and potential sources of colonizers (Gore I982) regulate the composition and
function of stream communities. Increased retention of particulate organic matter and
production of ﬁne particulate organic matter from decomposing logs alters nutrient ﬂuxes
through the biota and subsequently inﬂuences the functional response of the ﬁsh and
invertebrate communities within the stream (Minshall, et al. 1982, Sedell, et al. 1988). In
response to changes in organic matter storage, functional responses of invertebrate
communities, taxa abundance. and production are reported to vary between erosional and

22

 

deposit
1982. S
substra:

(Benke.

promine
small at".
Debris r'
and Iran:
Clear (in:
.‘tlississi;
Ann}- C C
COUnty I.

mainten;1

cOmplex ,
arid lasll}
deiiVered
original \;

the his,“

 

depositional habitats, as well as between reaches with and without debris dams (Molles
1982, Smock, et al. 1985, 1989), or logs (Wallace, et al. 1995). In regions with unstable
substrates, snags may support a large pr0portion of the insect biomass and production

(Benke, et al. 1984, 1985, Smock, et al. 1985).

In many regions of the United States coarse woody debris was historically a
prominent feature in streams, and in some cases log jams stretched for kilometers on both
small and larger streams (Swanson, et al. 1976, Triska 1984, Maser and Sedell 1994).
Debris removal was originally initiated to provide unobstructed waterways for navigation
and transportation of harvested logs. In 1776 the US. Congress appropriated money to
clear driftwood from streams and rivers to improve navigation, beginning with the
Mississippi River. Woody debris removal from rivers remains an active role of the US.
Army Corps of Engineers (Harmon, et al. 1986), and is one of the primary roles of
County Drain Commissioners (locally elected ofﬁcials charged with creation and

maintenance of an extensive network of drainage ditches) in the state of Michigan.

Coarse woody debris abundance in temperate stream ecosystems is regulated by a
complex set of factors that act on the source of the wood itself, its delivery to the channel,
and lastly, on the myriad of factors that control its retention and mobility once it is
delivered to the channel. At regional scales, geomorphic features have regulated the
original vegetation of the landscape (Grimm 1984, Host and Pregitzer 1992), as well as
the historical and current land use/land cover patterns within the region. Both historic

23

 

mﬁcc
surpl.‘
Forest;
conkers
SliViCL.
the up;
anaHte
”95). i
Sﬁﬂﬂnr
landscap
Ahhoug}
stream c
CM) A

I979.NIL

"teenrim

 

“1111181115. .
1mp‘mant
SUEamS aft

and current land management factors in the riparian zone and the ﬂoodplain inﬂuence the
supply of the CWD (Bilby and Ward 1991, Murphy and Koski 1989, Ralph, et al.1994).
Forested landscapes are increasingly fragmented by forest harvest as well as land
conversion for agricultural production and residential/commercial development.
Silvicultural practices alter species composition, number, and size distribution of trees in
the upland, and thus modify the potential source and input rates of CWD, particularly in
small to medium-sized streams (Bilby 1984, McDade, et al. 1990, F etherston, et al.
1995). In agricultural and urban areas potential sources of woody debris as well as the
stream retention capacity are altered by management practices such as grazing,
landscaping, riparian vegetation thinning or removal, dredging and channelization.
Although these management activities can undoubtedly reduce the potential supply of in-
stream CWD, current riparian vegetation may not accurately reﬂect standing stocks of
CWD. A highly retentive stream channel can contain very old wood (Keller and Tally
1979, Murphy and Koski 1989), which can predate the age of the current riparian

vegetation (Swanson and Lienkaemper 1978, Evans, et al. 1993).

The riparian zone and the land-water ecotone mediate inputs of sediment,
nutrients, and particulate organic matter to streams, in addition to providing other
important ecosystem functions (Gregory, et al. 1991). The primary sources of CWD in
streams are derived from natural mortality, ﬁre, disease, insect damage, ice/snow loading,
and wind-throw damage to trees in the riparian zone or uplands adjacent to the stream
(Keller and Swanson 1979). Processes such as mass soil wasting, bank undercutting and

24

 

 

l‘.

erosior
vector i
1086. 5
channe
erosion
and Six
regulate
hydrolo

COHSII'Ui

 

erosion, and ﬂooding transport this material into the stream. Beaver may be the primary
vector transporting large volumes of CWD to the channel in some systems (Naiman, et al.
1986, Maser and Sedell 1994). Alteration of the hydrologic regime from stream
channelization, wetland drainage, or urbanization frequently results in increased bank
erosion, one of the primary mechanisms of CWD input to low-gradient streams (Keller
and Swanson 1979; Davis and Gregory 1994). Erosion processes are themselves
regulated by geologic factors (e.g., soil type, topography), vegetation cover type, and
hydrologic regime, in addition to anthropogenic factors including grazing, forest harvest,

construction, and agriculture.

Once wood is introduced to the channel it is either retained by obstructions in the
channel, or by the channel itself, if the log is large relative to the size of the channel
(Keller and Swanson 1979, Keller and Tally 1979). A positive feedback loop is initiated
when wood is retained by an obstruction, resulting in debris jams that continue to
assimilate wood being transported from upstream. The stability of debris accumulations
depends on many factors, most important being the extent of burial, and the magnitude
and frequency of ﬂood events (Bilby 1984). Morphological changes in the channel that
are attributed to CWD, including plunge pool formation, lateral adjustments, sediment
and organic matter retention. can themselves inﬂuence CWD mobility and retention. For
example, decreased ﬂow velocity due to pool formation will result in decreased stream
power and capacity to transport CWD (Braudrick, et al. 1997). Channel bars resulting
from sediment and organic matter retention themselves form obstacles for CWD.

25

Since CWD fundamentally inﬂuences both the structure and function of streams
in historically forested regions, identifying the myriad factors that regulate its abundance
and distribution is essential for understanding how stream ecosystems are regulated.
Many studies have examined the role of coarse woody debris in high- and low-gradient
catchments (Table 2-1). However, few studies have attempted to quantify relationships
among landscape factors and the observed patterns in CWD abundance and distribution in
low gradient systems, particularly in landscapes that are not currently dominated by
forests. Stream restoration activities that attempt to increase habitat heterogeneity to
enhance ﬁsh and invertebrate production frequently use log structures anchored in the
channel (Hunter 1991). These structures are subject to failure, and can cause or
exacerbate existing problems (Beschta and Platts 1986). The inﬂuence of log placement
on the channel and the biota has been examined (e.g., Hilderbrand, et al. 1997) but did
not account for large-scale factors that inﬂuence both the input and retention of CWD in
streams. Landscape-scale factors such as land use and surﬁcial geology inﬂuence the
abundance of woody debris found in stream channels (Ralph, et al. 1994; Richards, et al.
1996) and also undoubtedly play a role in mediating the impact of disturbance events that
inﬂuence the export of CWD and smaller organic matter fragments. By examining the
factors inﬂuencing CWD at a range of spatial scales, the extent to which local and
regional factors regulate the abundance and distribution of CWD can be discriminated.
The goals of this paper are to: l) characterize the abundance, size, and distribution of
CWD in low gradient streams in developed landscapes, and 2) quantify the relative
inﬂuence of reach- and catchment-scale factors on the abundance and distribution of

26

Lsﬂt

 

l:

Slethtt

53w) .-

charac:
feature:
The stu-
Gailan:
Lake P;
dominat
Naunzr
lllC SFSI;
such as j
in bait:
agnCUlZg
logged r"

 

CWD. Parallel examination of the density and distribution of debris accumulations are
discussed elsewhere (Johnson 1999b) because the factors controlling the formation of

debris accumulations are believed to be local in scale.

Methods
Study Area

The Saginaw Bay catchment of Lake Huron encompasses a 16,317 km2 region,
characterized by sand and clay-dominated lowlands rimmed by coarse-textured glacial
features such as ground moraines and outwash plains (Figure 1-2; Johnson, et al. 1997).
The study region is contained within two major ecoregions as deﬁned by Omernik and
Gallant (1986): the Southern Michigan/Northern Indiana Till Plains and the Huron / Erie
Lake Plain. Each is subdivided into two sub-regions. Soils in the lake plain are
dominated by medium and ﬁne-textured loams ranging to clays, with sand in the outwash
plains and channels. These clay regions are extensively drained by artiﬁcial drainage and
tile systems. The periphery of the basin contains many coarse textured glacial features
such as ground moraines and outwash plains. The till plain exhibits the greatest variation
in basin topography and contains a high percentage of forested land intermingled with
agricultural land and old ﬁelds; elevations average about 278 m. The entire drainage was
logged for white pine and hemlock between 1840 and 1900 (Quinlan 1997), and forests

of the region now consist primarily of second growth hardwood species.

27

 

 

 

bout}
entirt
factor:

'land

lacustri
(Klor N
nunein
azetror
Onesue
IOIalOfE
CWilmer
morainal
l3inclu;

Iongnui

Cbaﬁeii

 

 

 

Previous work (Richards, et a1. 1996. 1997, Johnson, et al. 1997) has shown that
both land use patterns and Quaternary geology mediate the landscape's response to
environmental stress. The sampling design was chosen to reﬂect these factors. A 2x2
factorial design was used to investigate the inﬂuence of underlying geology and land use
/ land cover on CWD dynamics (Table 2-2; Figure 1-1). Each cell in the design included
3 replicate catchments/streams (n = 12 total), chosen from a pool of candidate
catchments. The treatments are designated as: lacustrine / agricultural (Lac/Ag),
lacustrine / mixed (Lac/Mix), morainal / agricultural (Mor/Ag) and morainal / mixed
(Mor/Mix). Three ﬁrst to third order reaches in each stream were selected to quantify
some internal variation within streams, resulting in a total of 36 subcatchments ranging in
size from 712 to 23,448 ha. These sites are collectively referred to as the “core” sites.
One site was impounded by beaver during the second year of this study resulting in a
total of 35 sites. Longitudinal gradients in streams in a lacustrine agricultural (Lac/Ag)
catchment (Bad River, n = 11 including “core” sites), and a catchment dominated by
morainal geology and mixed agricultural/forested land use (South Branch Flint River, 11 =
12 including “core sites) also were examined. These sites are collectively referred to as

“longitudinal” sites.

Coarse Woody Debris Abundance, Distribution and Size

Coarse woody debris assessments were performed during low ﬂow conditions
during the summer of 1995 (Table 2-3). CWD volume was measured using the line
transect method (De Vries 1974, Wallace and Benke 1984). Three random transects

28

a

 

 

to

3,3

1m:

S
s

 

 

across the channel were established in each ﬂow regime (e.g., fast, turbulent; fast,
non-turbulent; slow; as per Hawkins, et al. 1993) represented within a 100 m reach.
Diameter measurements were obtained for all wood _>_ 0.02 m and 0.25 m in length that
intersected the transect and fell within the bank-full channel. Wood volume per unit area

was calculated based on the formula:

)2. = (7:2/81.) {- d'2

where L = transect length, and d = stem diameter intercepted by the transect (Wallace and
Benke 1984). Volume per unit area was calculated for each transect and summed for
each reach. Volume data are reported for size fractions of 2 0.02 m diameter and >0.25
m length, 2 0.05 m diameter and 0.5 m length, and 2 0.10 m diameter and 0.5 m length.
The majority of the analyses were performed using _>_ 0.05 m diameter and 0.5 m length
data. (Wood density measured by the line-transect method will be referred to throughout

as wood volume.)

In addition to volume measurements, counts of the total length of CWD 2 0.05 m
diameter and z 1 m in length were made at 10 m intervals within the reach and
summarized as total meters of wood per m2 of stream bottom (m / m2) for each site. (This
assessment method is hereafter referred to as the “linear estimation method”, and the data

generated by this technique will be referred to as ‘wood abundance’ throughout the text.)

29

 

 

 

.- y .. _
1“ at". -

organz.
log tie.”
photo;
of the .
accurr...
“ere s.

retleetir

the surr

Charm.

SUﬁicie:
20 time.
COmpre:
at “11hr:

habitar‘

R1170 r 1.0}

cOmPOSi

Obtaineq‘

 

Debris accumulations were deﬁned broadly to include: vegetation with trapped
organic debris, root wads with trapped organic debris, loose accumulations of logs, and
log debris dams. Debris accumulations _>_ 1 m2 in area were counted, mapped, and
photographed. Debris accumulations were assigned a size class based on the dimensions
of the debris accumulation relative to the channel width at the upstream point of the
accumulation location (Table 2-4; Shields and Smith 1992). Debris accumulation data
were summarized by total number of debris accumulations per 100 m reach, and a metric
reﬂecting the amount of stream channel covered with debris accumulations, derived from

the sum of all debris accumulation sizes per reach (= Z(accum size); Table 2-3).

Channel morphology and habitat structure

At each site, a stream reach of approximately 100 m was sampled. This is usually
sufficient to incorporate more than one riffle-pool sequence and represented between 10 -
20 times the stream width (Richards 1982, Bisson and Montgomery 1996). A
comprehensive set of parameters commonly evaluated in stream surveys were measured
at within each stream reach, representing factors associated with the channel morphology,

habitat, and riparian conditions (Table 2-5).

Riparian structure

Measurements and observations of riparian width, riparian SIOpe, vegetative
composition and height, riparian and ﬂoodplain land use, and ﬂoodplain slope were
obtained from each bank at three points along the reach (Table 2-6). Riparian zone width,

30

.uu.

 

 

reﬂel

six V3

y ’ v
Lands:

quantz' ft
thesany
[0P0 Ezra;

dam “‘35

 

vegetation height, and SIOpe were encoded separately for the left and right banks, and the
six values were averaged to derive a mean value for each site. Riparian zone slope
(perpendicular to the channel) was measured directly at six points along the reach using a
clinometer. Riparian vegetation height classes (0 = paved, 1 = lawn, 2 = grasses/herbs, 3
= shrubs, 4 = trees) reﬂected an increasing potential to serve as a source of CWD.

Riparian vegetation height values were highly correlated with riparian vegetation cover.

Landscape Structure

Land use, hydrography, Quaternary geology, and elevation databases were used to
quantify several aspects of landscape structure (Table 2-7). Catchment boundaries above
the sample points were delineated manually and digitized from USGS 1:24,000
topographic maps. Digital elevation data were used to verify boundaries. Hydrography
data was derived from digital line graph (DLG) data at a scale of 1:100,000 (USGS).
Stream orders (Strahler 1964) and link numbers (Shreve 1966) were assigned as an
attribute of the stream data ﬁle derived from the USGS as Digital Line Graph ﬁles. Mean
catchment elevation and slope were derived from 30-second digital elevation models at a
scale of 1:100,000. Topography in the region is relatively ﬂat, therefore the standard
deviation in elevation was used to represent topographic heterogeneity. Slope was
derived from elevation data using ARC/INFO algorithms. Stream density was calculated
as the total length of all streams divided by catchment area (km / kmz). All spatial

databases were transformed into a common digital format, projected onto a common

31

-..

 

 

approxn

digital d

 

 

accurate
Chances

large prt

coordinate system (Albers) and analyzed in ARC/INFO as vectors unless otherwise

speciﬁed.

Land use - land cover data and patch density (a measure of landscape
fragmentation; F onnan and Godron 1986) reﬂect the extent of human intervention in
shaping the landscape. Land use data for the study area were obtained from the Michigan
Department of Natural Resources (Michigan Resource Information System database),
based on aerial photography dated in the late 1970's. Mapping resolution for these data is
approximately 1 ha, with a minimum lateral dimension of 61 m. Comparison of the
digital data with 1987-1988 photographs revealed that the digital data was about 90%
accurate over the upper Flint River catchment. The majority of observed land use
changes reﬂected wetland habitat loss near urban centers. The Flint catchment covers a
large proportion of the total study area and contains the largest concentration of urban
areas, and therefore represents the extreme in land use conversion in the basin. and
increases conﬁdence that the land use data reﬂect conditions during the water quality
sampling program. The classiﬁcation of land use categories was based on a modiﬁed
version of the Anderson, et al. ( 1976) scheme, which was constructed speciﬁcally for
natural resource applications. Based on the areal extent of certain land use classes and
previous work (Johnson, et al. 1997; Richards, et al. 1997), land use categories were
aggregated into ﬁve classes: urban, row crop agriculture, forest, range, and wetlands in
most of the analyses. Mixed land use was designated as less than 50% agriculture in the
catchment. Agricultural land was highly negatively correlated with both forest and range

32

land, therefore, a derived variable consisting of a ratio of total agriculture: forest + range
(AG : FOR + RNG) was substituted for the individual classes. Range lands in this region
are predominantly abandoned ﬁelds with shrub or herbaceous cover types. Open water
was not included as a land cover type in these analyses. Land use values are reported and
analyzed as the proportion of total catchment area. Land use patch density was calculated

from land use / land cover data as the number of patches per square kilometer.

Quaternary geology data were digitized from Farrand and Bell (1984; Table 2-8),
and also were reported as a proportion of the total catchment area. Based on areal extent
of minor categories and previous work (Richards, et al. 1996, 1997; Johnson, et al 1997)
geological categories were aggregated by particle size or omitted due to the small areas
encompassed by that geologic class. Coarse till plus sand and gravel variables were
combined because they were highly correlated with one another, and the combined

variable enabled us to reduce the total number of geologic variables.

Data Analysis

Distributional properties of all variables were assessed on the raw data and
appropriate transformations were applied to non-normal variables. Box-Cox plots were
examined to determine the best transformations to achieve normality. Variables not
passing the Wilke-Shapiro test for normality were transformed as follows: square root
transformations were performed on debris dam abundance and 2(accum size), and log
transformations were performed on wood abundance (rn / m2 ), wood volume (m3/m2 ),

33

 

 

location
Spatiall}
Data fro
Flint R11
betWeen
exhibits

Branch ( .

 

 

watershed area, and urban and residential land use values by taking the natural log of the
datum plus '/2 the lowest non-zero values (arcsine transformations were not performed on
the land use proportions because of the relatively small range of the data). Pearson
correlations were performed for each discrete data set (e. g., landscape, riparian, channel-
habitat) to assess the degree of intercorrelation among variables. Highly correlated
variables were not included in the analyses, although separate regression analyses were
performed to assess the ability of some variables (e. g., urban versus residential and non-
residential urban; wetland versus forested and non-forested wetland; coarse till +
sand/gravel versus coarse till and outwash sand and gravel) to improve regression

models.

The sampling protocol for this study included multiple samples at different
locations on the same stream. Samples from adjacent sites on the same river could be
spatially autocorrelated, and would therefore not be considered statistically independent.
Data from the longitudinal series on the Bad River (n=10) and the South Branch of the
Flint River (n=9) were used to calculate Moran’s I, a measure of the interdependence
between data at adjoining locations (Odland 1988). None of the CWD measures
exhibited signiﬁcant spatial autocorrelation between adjacent sites on either the South
Branch of the Flint or the Bad River (Table 2-9). These results were used to justify the
assumption that the three sites sampled on each river could be treated as independent

samples, and that the longitudinal sites could be grouped with core sites.

34

 

 

relation
dlSlrlbu

Viinflow

 

 

 

The hypothesis that were tested were:

1. C WD abundance and distribution are controlled primarily by local factors
(e.g., riparian zone structure and composition. channel features).

2.. Local factors controlling C WD abundance and distribution are themselves
controlled by factors at larger spatial scales (e. g.. dominant land use,

Quaternary geology, topography, Iandscapeﬁagmentation).

To test these hypotheses analyses were performed to 1) describe patterns in the
distribution of coarse woody debris abundance across the study area. 2) quantify the
effects of land use and Quaternary geology on CWD standing stocks, 3) predict CWD
standing stock from local, riparian and landscape features, and 4) identify hierarchical
relationships among landscape, riparian and local factors inﬂuencing the abundance and
distribution of CWD in disturbed streams. Analyses were performed using SAS v. 6.1 for

Windows and SigmaStat v. 2.03 for Windows unless otherwise speciﬁed.

A two-way AN OVA was conducted to determine the effects of land use and
Quaternary geology on the number of CWD abundance and volume. To predict
abundance (log m / m2), (sq. rt.) number of debris accumulations / 100 m and (sq. rt.)
£(accum size), multiple regressions were conducted separately with local, riparian and
landscape variables. A combined data set, consisting of previously identiﬁed predictors
at each spatial scale, was used to predict the CWD abundance variables from data at
multiple spatial scales. For C WD abundance and volume, regressions were conducted

35

 

 

1981

the be
\‘a'iab
partial
Wilkes
residua
1981 1.

examini

analyses

 

 

separately on morainal and lacustrine sites in an attempt to identify a signiﬁcant model

when none were found for the full data set.

All regression models were examined using the Cp statistic (Draper and Smith,
1981) and R2 values. Variance inﬂation factors. and condition indices were examined for
the best candidate models to assess the degree of collinearity among independent
variables (Belsley, et al. 1980). The candidate models were further examined using
partial regression leverage plots, plots of residuals vs independent variables and the
Wilkes-Shapiro statistic to examine the assumption of a normal distribution of the
residuals. Inﬂuential outliers were identiﬁed using Cook's Distance (Draper and Smith,
1981). Hierarchical relationships among local and regional factors were inferred by

examining the relative strength of the models at each spatial scale.

Since CWD volume based on logs 2 5 cm was 0 at 19 of 49 sites, regression
analyses were not performed; instead, predictions of wood volume ((log) vol 2 5 cm)
were made using a two stage analysis. The ﬁrst set of procedures was intended to
identify predictor variables distinguishing between sites with and without wood _>_ 5 cm
on the transects (volume = 0 versus volume > O; hereafter referred to as ‘CWD volume
presence / absence’). A second analyses, performed only for sites where vol _>_ 5 cm was

greater than 0 (n = 30), was intended to predict actual CWD volume from three sets of

predictor variables.

36

 

I.“ 'm“. K V

 

 

Result

I t
abuna.

 

A discriminant function was identiﬁed to predict CWD volume presence /
absence, using a stepwise discriminant function analysis. This procedure selected a
subset of predictor variables at each spatial scale which were subsequently used to
develop a discriminant criterion to classify each site based on the CWD volume presence
/ absence, and to estimate the accuracy of the classiﬁcation. The second stage of the
analysis predicted actual CWD volume (>0) using multiple regression. with local,

riparian and landscape factors as independent variables, as described above.

Results
Abundance and Size of Coarse Woody Debris Across the Region

In general, coarse woody debris in the study streams was not abundant and the
mean size of individual logs was small (Table 2-10). Wood abundance at 10% of sites
was zero, and 36% of sites had less than 0.1 m / m2. Wood volume of logs 2 0.10 m
diameter was 0 at 55% of sites, 0 at 39% of sites when logs _>_ 0.05 m diameter were
included, and 0 at only 14% of sites when wood _>_ 0.02 m diameter was considered. The
mean volume of CWD _>_ 0.10 m diameter across all sites was 0.0017 1 0.0005 m3/ m2,
and 0.0024 1 0.0007 m3 / m2 for wood 2 0.05 m diameter. The mean size of logs across
these streams also was small. When wood _>_ 0.05 m diameter was considered, mean log
diameter was 0.093 : 0.007 m and length was 2.5 j: 0.23 m. Since large wood was rare, a

smaller minimum log size was used for this study, despite the fact that this would over-

represent the density of wood at these sites, compared with other studies.

37

 

 

Lani;

 

than.
domir‘.
and g;
abun:
and I;

transet

 

Land Use and Surﬁcial Geology Effects on C W D and Accumulation Abundance

A two-way analysis of variance was performed testing the hypothesis that the
abundance and distribution of coarse woody debris differed across sites stratiﬁed by
dominant land use and Quaternary geology. Signiﬁcant interactions effects of land use
and geology on wood abundance were detected (m / m2; p < 0.001; Table 2-10). CWD
abundance was signiﬁcantly greater in Mor/Mix and Lac/Ag catchments than Mor/Ag
and Lac/Mix catchments. Although differences in CWD volume, numbers of logs on the
transects, and log sizes were not signiﬁcant, some strong trends were evident. Mor/Mix
and Lac/Ag sites had the largest mean log diameter and length; mean wood volume was
large, but highly variable in the Mor/Mix/mixed sites, and was approximately equal in the
other three catchment types. Numbers of logs 2 5 cm diameter encountered on the
transects was greater in morainal than lacustrine sites, but this trend was not apparent for

the data set that included small logs (2 2 cm diameter; Table 2-10).

A mixed model ANOVA was performed testing LU, GEOL, LU*GEOL treatment
effects while accounting for the random effects of the basins. No signiﬁcant differences
resulting from the treatments were observed when variation due to basins was accounted
for in the error term of the model. Differences due to error (basin) effects, however, were
signiﬁcant for all CWD measures with the exception of diameter and length. This
suggests that either within-basin variation masked the variation due to land use and
geology, or there was insufﬁcient power to detect treatment differences due to the
reduction in the degrees of freedom from 45 to 10.

38

II. M.“

 

 

TEST;
~

ofCV

cadu

dime

ddno

hei in

"Q

all 800?.

PTEdic:
Oblajn:

(Table

 

01 land.-

elel'am

 

Environmental Factors Inﬂuencing C WD Standing Stocks

CWD presence / absence at a site was predicted by a discriminant function
consisting of local-scale variables including % open canopy and mean bank-full width.
The presence of wood volume > 0 was predicted with an accuracy of 97%, while the sites
where wood volume = 0 was predicted with an accuracy of 79%. No significant
regression model was found using local-scale variables for predicting the actual volume
of CWD at the 30 sites where wood volume > 0 (Table 2-11); however, when lacustrine
catchments were examined separately, the percent of slow units in a reach predicted 36%
of the variance in woody debris volume. A separate analysis of the morainal catchments

did not result in a signiﬁcant model.

At the intermediate scale a discriminant function consisting of riparian vegetation
height and riparian zone width successfully predicted sites where wood volume > 0 with
an accuracy of 90%, while sites where wood volume = 0 were predicted successfully only
79% of the time. As with the local-scale variables, no signiﬁcant regression was found to
predict CWD volume from riparian-scale variables, and no signiﬁcant models were
obtained when data were analyses separately for lacustrine and morainal landforms

(Table 2-11).

The discriminant function predicting CWD presence / absence consisted entirely
of landscape-scale variables included link number. (log) residential urban land, S.D.
elevation and the ratio of agricultural to forest + range land. This discriminant function

39

 

had;

“rut.

 

 

Separ.

h0\VC\

 

outu'a
S’J‘éan
predic

idenu

I01a] -

had an overall prediction rate of 81%; the sites where wood volume > 0 were predicted
with a 0% error rate, but sites where wood volume = 0 were very poorly predicted (39%
error). In contrast to the local- and riparian-scale variables, a signiﬁcant model was
found to predict CWD volume at the 30 sites with non-zero values. Predictors were: %
outwash sand / gravel. % wetland. and % lacustrine clay. An R2 = 0.30 indicates that
there is a large amount of variance not accounted for by this model (Table 2-11).
Separate analyses for each landforrn did not improve the predictive power of the model,
however, differences in the predictor variables were evident. On lacustrine landforms,
outwash sand and gravel and (log) residential urban were identiﬁed as predictors, while
stream density, the ratio of agriculture to forest land, and coarse tills were signiﬁcant
predictors on morainal landforms. The multi-scale model including both landforms was

identical to the landscape model.

CWD abundance (m / m2) was poorly predicted by local-scale variables, unlike
the number of debris accumulations / 100 m and the 2(accum size) metric (Table 2-11).
The % open canopy was the only signiﬁcant predictors of wood abundance. On
lacustrine landforms, percent open canopy was a signiﬁcant predictor of CWD
abundance, and explained 36% of the variance, compared to 19% for both landforms
combined. No signiﬁcant models were found for morainal sites. Riparian vegetation
height (= riparian vegetation type) predicted both CWD abundance and the presence /
absence of wood volume (>0). On morainal landforms, this variable predicted 71% of the
total CWD abundance variance when one outlier was eliminated (p = .0001 ). The best

40

ﬂ

 

 

I" I VII-3‘51: u:
I I.-

C \‘tD

C8001".

accum

OlllWaf

numb:

 

landsc'
Vduef

on but}

landscape- scale model predicted CWD abundance from link number, (log) urban land,

and SD. elevation. The multi-scale model was identical to the landscape model.

The number of debris accumulations / 100 m was equally well predicted by the
local and landscape variables. but was best predicted by the multi-scale model (Table 2-
11). The local scale predictors for density of debris accumulations, 2(accum size), and
CWD volume presence/absence were identical: mean bankfull width and percent open
canOpy. The riparian vegetation height metric was the best predictor of debris
accumulation density at the intermediate scale. Link number, (log) residential land use,
outwash sand and gravel, and medium till were the best landscape-scale predictors of the
number of debris accumulations / 100m. The multi-scale model contained both local and
landscape variables including mean bankfull width and (log) residential land use; the R2
value for this model was 0.61, which was the largest of all of the predictive models based

on both landforms.

The metric representing the amount of stream channel covered by debris
accumulations, 2(accum size), behaved similarly to the number of debris accumulations /
100 m, with the exception that the landscape scale model was best predicted by the SD.
elevation, link number, (log) urban land use, and coarse till (Table 2-11). The multi-scale
model contained independent variables from each data set: mean bank-full width, riparian

vegetation height, and (log) urban land use.

41

 

 

 

 

l—i—

 

Disc

Discussion

Several local and regional factors account for the patterns in CWD abundance and
distribution observed in streams across a disturbed landscape: however. these
relationships are complicated by the underlying structure of the landscape and the

disturbance history of the region.

Regional patterns in wood abundance and size

Streams of the Saginaw Basin, Michigan contain a lower abundance of coarse
woody debris and smaller logs in comparison with forested streams in the United States
and elsewhere (see review by Gurnell, et al. 1995). Direct comparison among studies is
difﬁcult, however. due to inconsistencies in the minimum size of logs considered, and a
lack of studies in similar streams. Previous studies have deﬁned coarse woody debris as
logs ranging from 0.05 to 0.20 m in diameter and 1 to 2 m in length. Differences also
exist in the use of geometric means versus non-geometric means. Logs with diameters 2
0.10 m (the most common deﬁnition of CWD cited in the literature) are rare in the
Michigan streams studied (Figure 2-2). Logs 2 0.10 m on the wood volume transects
were encountered at only 45 % of the sites. A comparison of the density of CWD in
developed versus forested landscapes only underscores the scarcity of wood in the
Michigan streams (Table 2-12). The volume of CWD observed in this study is
comparable to reaches in an agricultural stream in Tennessee that had recently been
cleared of CWD (Shields and Smith 1992), to a basin with mixed land use in Hampshire
England (Gregory, et al. 1993). and disturbed reaches of some Rocky Mountain streams

42

.ql

 

 

(Ru:

 

ﬂood:
Bank.
in the
comp.

Gun:

hlount

sues.t

 

and ler

ileum:

(Richmond and F ausch 1995). In contrast. two low-gradient streams with forested
ﬂoodplains in Georgia had C WD volumes of 0.0148 and 0.0167 m3/ m2 (Wallace and
Benke 1984). compared to 0.0007 and 0.0038 m3 / m2 in the mixed land use catchments
in these Michigan streams. Woody debris standing stocks in the Michigan streams are
comparable only to the lowest values reported for old growth forest catchments

(summarized by Gumell. et al. 1995).

Wood volume in these Michigan streams was comparable to the disturbed Rocky
Mountain sites, however, the mean log diameter was smaller. Across all of the Michigan
sites, the mean CWD diameter (based on diameter 2 0.05 in) across all sites was 0.093 m,
and length was 2.5 m (Table 2-10). Wood diameter at the disturbed sites in the Rocky
Mountain study (Richmond and F ausch 1995) ranged from 0.15 to 0.20 m. The low
standing stocks of C WD in this study are consistent with the disturbance history of the
region. which was logged of native white pine and hemlock from 1840 through 1900, and
then subjected to widespread ﬁres followed by extensive soil erosion (Quinlan 1997).
While coarse woody debris density is low, the number of debris accumulations / 100 m in
this study area is similar to those encountered in some forested streams (Table 2-12).
Debris accumulations were deﬁned broadly in this study to include overhanging
vegetation and root wads that trapped organic debris (Johnson 1999b). The density of
debris accumulations consisting only of logs and snags, however. were comparable to a
wide variety of stream types including Iowa streams (Zimmer and Bachman 1976 in
Shields and Smith 1992), an uncleared agricultural stream in Tennessee (Shields and

43

 

 

5113 ~
11121.".

9".'\
~11.

Sili’d'

14.
“M.”

thelﬁ

hgh;

dtblh

 

Lani
suean
hBSI
banks
Obser
“‘asu
Vet}- .
“00C

llS.

Smith 1992). a managed stream with mixed land use in England (Gregory et al. 1993).
and a forested stream in New Hampshire (Bilby and Likens 1980). Debris accumulation
sizes are difﬁcult to compare across studies due to inconsistencies in measurements
methods. The fact that there are large differences in standing stocks of CWD, but similar
densities of debris accumulations suggests that the debris accumulations in the Michigan

streams are smaller in size and are composed of smaller logs (Johnson 1999b).

Aside from disturbance history, differences in C WD abundance among the higher
gradient streams, where many studies have taken place, and the lower gradient streams of
the Midwest could also be attributed to CWD input mechanisms. Woody debris inputs in
high gradient systems are largely due to whole tree or tree-mp blowdown, debris slides,
debris avalanches. and mass soil movement from adjacent hillsides (e.g., Swanson and
Lienkaemper 1978, Lienkaemper and Swanson 1987). Woody debris enters low-gradient
streams through blowdown, bank erosion, and ice loading (Keller and Swanson 1979). In
this study area many downed trees were observed in the streams resulting from undercut
banks and bank erosion. In addition, numerous new limbs and tree fragments were
observed in the streams following intense summer storms. In comparison to hillside mass
wasting and avalanches which move large volumes of debris into the stream channel in a
very short period of time. bank erosion and storm damage deliver smaller amounts of
woody debris to the channel. Ice storms such as one that occurred in the northeastern

US. and Canada in February 1998 have the potential to deliver large quantities of wood

44

x-rrn" II. 1"" “-

 

 

.f‘r'r
I"

relax:
patter
abungi
llor 7

denst

 

over a very short time frame. In contrast to mass wasting. however. the effects on the

channel are not as dramatic.

Effects of land use and Quaternary geology
The direct effects of land use and landform in these Michigan streams were

mixed. Instead. historic forest harvest, along with current land management practices,
probably account for the low abundance and the small size of logs in Michigan streams,
relative to less disturbed forested systems (e.g.. Richmond and F ausch 1995). Land use
patterns and landform at the scale of catchments, had a signiﬁcant effect on CWD
abundance (m / m3) and the number of debris accumulations / 100m (Table 2-10);
Mor/Mix and Lac/Ag catchments had a signiﬁcantly greater abundance of CWD and
density of debris accumulations than did Lac/Mix or Mor/Ag catchments. Aside from the
treatment differences in land use and Quaternary geology. Mor/Mix and Lac/Ag
catchment types differed from the other land use / geology classes by having lower ﬂood
heights, steeper catchment gradients with more topographic heterogeneity. and more
range and forest land cover (Table 2-13). Lower ﬂood height implies lower stream power
and shear stress to transport woody debris out of the reach. while land cover differences
point to an absence of agricultural activities (the non-agricultural land use in common
across these catchments is range land, as opposed to urban or forested land; Tables 2-14,
2-15). The prevalence of debris accumulations characterized as “loose log’ and ‘log/snag’
at these sites, rather than ‘overhanging vegetation with trapped debris’ or ‘root wad with
trapped debris’ (Johnson 1999b), suggests that either woody debris that enters the

45

 

 

Elli.

qua:
onst
cond
Beet
1989

Zeak

strea:
abut,
and I"
lfan
SITE? r
Cl‘t; .
as U. 5

50m.

 

 

channels of these streams is retained due to lower ﬂood heights. or that there is a
sufﬁcient supply of CWD upstream or in the riparian zone to replace wood that is

transported out of the system.

Other regionally extensive studies of CWD in streams have focused on
quantifying the impacts of forest harvest activities and/or identifying associated impacts
on stream channel morphology. Extensive stream surveys of this nature have been
conducted in western Washington state (Bilby and Ward 1989, 1991, Ralph, et al. 1994,
Beechie and Sibley 1997), Oregon (Carlson, et al. 1990), Alaska (Murphy and Koski
1989), the Rocky Mountains in Colorado (Richmond and F ausch 1995), and New
Zealand (Evans, et al. 1993). In western Washington, intensive forest harvest did not
affect the abundance of CWD, however, basins that had undergone intensive forest
harvest had smaller logs that were located near the channel margins. These changes were
associated with a decrease in pool area and depth (Ralph et al. 1994). Rocky Mountain
streams with past disturbances in the riparian zone (harvested prior to 1900) had less
abundant, smaller logs than did streams associated with old-growth forests (Richmond
and F ausch 1995). In contrast, Bilby and Ward (1991) documented a rapid (within 5
years of harvest) change in the species mix. a decrease in both CWD abundance and (in
streams _>_ 10 m wide) average CWD size following harvest. (Bilby and Ward deﬁned
CWD as logs 3 10 cm diameter and 2 m in length.) Variability in streams within a study,
as well as larger minimum log length compared to other studies, may partially account for
some of the discrepancies in forest impact data.

46

 

i
pie.

 

 

Sin-t:

 

lllel}.

 

 

 

 

Influence of Local-Scale Features

Local-scale factors (especially bank-full width and % open canopy) are better
predictors of debris accumulation density and distribution than of CWD density (Table
2-1 1). Debris accumulation density and distribution are better predicted because debris
dam formation requires a physical obstruction such as a downed tree. boulder, point bar
or island in the channel. The presence of such obstacles is generally related to the
morphology of the channel and the geomorphology of the valley. Logs that are not
associated with a debris accumulation (and are not buried) are more likely to be
mobilized and transported out of the reach. Wood density measures reﬂect both the pool
of unentrained (mobile) wood as well as that which is entrained in debris accumulations.
The abundance of highly mobile wood in the channel, therefore. may bear no relation to
local-scale conditions. Bank-full width was positively correlated with debris
accumulation density and Z(accum size), and accounted for the majority of the variability
in accumulation density and distribution. Gregory and colleagues (1993) explained 19%
of the variance in debris accumulation density from the similar features, including
distance downstream. and percentages of deciduous and coniferous trees in the reach.
Since their catchment lies within one landform, riparian and catchment land use is most
likely to exert a relatively greater inﬂuence on debris accumulations compared to

Saginaw Basin streams.

Harmon. et al. (1986) suggested that the distribution of CWD along a longitudinal
gradient is due to a combination of both ﬂuvial and terrestrial factors. In small streams

47

the location of debris accumulations along the longitudinal gradient is regulated by the
spatial pattern of log input, since wood is relatively immobile in small charmels. In
intermediate-sized streams. stable structures such as debris accumulations and boulders
entrain CWD in the channel (Swanson and Lienkaemper 1978, Keller and Tally 1979).
Channel morphology, including sinuosity, width and depth. and presence of point bars
and islands are the most important factors regulating the location of debris accumulations
in larger channels (or intermediate channels with smaller CWD). The streams in this
study range from 3.6 to 12.6 m wide. Debris accumulations were most frequently found
to be associated with the banks, however, other structures which trapped debris included
root wads, point bars and islands and snags: many accumulations were not associated
with any visible structure in the channel (Johnson 1999b). Due to the small average
length of CWD in these streams, wood is relatively mobile, compared to other
comparable-sized systems. As a result, even in small streams. the mechanisms leading to
debris accumulation formation in these Michigan streams are more likely to be similar to

those of intermediate or large streams in forested landscapes.

The number and distribution of debris accumulation were well predicted by bank-
full width alone; however, only 8% of the variance in wood abundance was explained by
bank-full width alone (data not shown), and the relationship between volume (at sites
with volume > 0) and bank-full width was not signiﬁcant. Strong positive relationships
between CWD volume and channel width have been found in some studies (Bilby and
Likens 1980, Bilby and Ward 1989, Murphy and Koski 1989, Robison and Beschta

48

 

 

 

 

31011:

 

 

 

1989), but not others (Ralph, et al. 1994, Richmond and Fausch 1995, Beechie and Sibley
(1997). There was no association between number of logs / m and channel width,
however, a strong inverse relationship was found when CWD abundance was expressed
on an area basis (Beechie and Sibley 1997). Unlike the previously mentioned studies,
however, there was no clear relationship between the size of logs and channel width in
this study (Table 2-11). Bilby and Ward (1989) predicted 85% of the variance in wood
volume, and 79% in wood diameter and length from channel width alone, with CWD
volume, diameter and length increasing with channel width. This strong relationship, in
contrast to that seen in data in the Saginaw Basin streams, is more than likely due to the
fact that their study was located in undisturbed old-growth forest, where streams were
within 100 km of one another (and were therefore probably similar in geomorphology
and channel form), and had a similar history of discharge patterns. Streams studied by
Murphy and Koski (1989) were chosen for their distinct channel, geomorphic, vegetative
and hydrologic features, however, all streams were located in undisturbed old growth
forests. In contrast, Ralph, et al. (1994) studied catchments in western Washington state
that ranged from intensively harvested to pristine, while those studied by Richmond and
Fausch (1995) in the Rocky Mountains were either unharvested, or had been harvested
around 1900. The Saginaw Basin streams were selected from two contrasting land forms
and dominant land use patterns. Furthermore, they have been exposed to numerous large-
scale disturbances ranging from forest harvest and ﬁre early late in the 1800's to
channelization and other land management practices in modern time. Channel width is
largely controlled by catchment area (Richards 1982, Church 1992). however,

49

anthropogenic factors such as stream channelization and land use factors disrupt the
natural hydrologic regime and artiﬁcially widen and deepen a stream channel. The weak
relationships between stream width and CWD size and abundance may be confounded by
the combined effects of hydrologic regime and the cumulative effects of land

management practices.

In addition to patterns of abundance of debris accumulations along the
longitudinal proﬁle of a river, many researchers have reported strong associations
between coarse woody debris and plunge pool formation (e.g., Andrus, et al. 1988, Bilby
and Ward 1991, Hilderbrand, et al. 1997), lateral adjustment in the channel (e.g.,
Nakamura and Swanson 1993, Richmond and Fausch 1995), changes in the longitudinal
proﬁle of a river (e.g., Beechie and Sibley 1997, Smith, et al. 1993), and channel width
(Bilby and Ward 1989; Gregory, et al. 1993.). The lack of strong associations between
CWD accumulations and channel features in this study is probably due to two interacting
factors: logs in these systems are smaller than those encountered in most previous studies
of CWD-channel interactions, and these streams have been subjected to a range of
management practices that includes debris removal and channelization. Both of these
management practices would eliminate any evidence of a structural role for CWD in these
streams. In addition. smaller logs are more mobile and are therefore less likely than
larger logs to exert control over channel morphology by forming plunge pools and

altering channel widths.

50

 

 

 

lﬂﬁr

that.

 

abse:
[liar l

0116 If

 

Neither pool frequency nor maximum depth of pools was related to the abundance
or distribution of CWD in the Saginaw Basin streams, and correlations between CWD
variables and percent of reach with pools or maximum depth of pools were not
signiﬁcant. Only on lacustrine landforms. at sites where volume > 0. was the maximum
depth of pools identiﬁed as a predictor of wood volume. Along with channel width, the
number, location, and volume of pools in a reach have been shown to be very closely tied
to the presence of CWD (e.g., Carlson, et al. 1990, F ausch and Northcote 1992,
Richmond and Fausch 1995). Beechie and Sibley (1997) noted that pool forming factors
in low-gradient streams appear to be formed by mechanisms other than CWD. The
history of channelization in this region, and the small average size of the logs in the
streams generally preclude CWD from functioning as a pool-forrning agent in this region
(Johnson 1999b). Furthermore, it appears that CWD standing stocks are only partially
controlled by the local-scale features measured in this study. The local-scale factors
important to CWD abundance (i.e., bank full width, percent open canopy) are themselves

inﬂuenced by larger-scale factors (Table 2-14).

Inﬂuence of Riparian Features

The factors inﬂuencing the absence of wood at a site are more difficult to predict
than those inﬂuencing its presence. The discriminant function predicting the presence /
absence of CWD volume (>0) was better able to predict sites with CWD >0 much better
than CWD volume = 0. In many areas woody riparian vegetation has been preserved on
one ownership block and totally removed on an adjacent parcel (unpublished

Sl

abuf

lnur

 

 

Predl.

 

butt,
unit
”pan;
for C\

bel'On

 

observations). Clearly. stewardship practices of individual landowners are important
factors controlling the abundance of CWD in streams, and both economic as well as
social/ethical issues come into play when a farmer chooses a particular management
practice (Ryan, et al. 1999). These issues appear to vary in importance from one
landform to the other. For example, a very strong relationship was found between CWD
abundance and riparian vegetation on morainal but not on lacustrine landforms.
Interestingly, riparian vegetation height is negatively correlated with rowcrop agriculture
on morainal but not on lacustrine landforms (Table 2-14). It is likely that social and/or
economic issues, (perhaps related to soil productivity) account for the lower variation in
riparian vegetation height and riparian zone width on the lacustrine landforms versus
morainal landforms. Regardless of the underlying factors governing management
decisions, the result is that the riparian vegetation structure of Lac/Ag catchments is more
similar to those in the Lac/Mix. compared to the riparian vegetation in the two land use

types on morainal landforms.

Riparian vegetation type is more important than riparian zone width when
predicting CWD density in Midwestern streams. In addition, the width of the riparian
buffer strip is independent of vegetation type (Table 2-14). This relationship is most
striking in Mor/Ag catchments, where relatively wide riparian zones coincide with
riparian vegetation heights indicative of herbaceous vegetation. The source-distance area
for CWD in a stream is less than about 20-30 m (about two tree lengths); therefore,
beyond some threshold distance, riparian vegetation does not behave as a source of CWD

52

.. Hg

 

11

CV

and
in 51
may
tesid
hec
blur;
Wooc
Bani
agriu
Veget
10 5111

lOggt

 

for the stream (e.g., McDade, et al. 1990). The magnitude of this distance is dependent
upon geomorphic factors such as slope, soil type. age and species composition of the

riparian vegetation.

Modiﬁcation of riparian vegetation can rapidly inﬂuence the characteristics of the
CWD pool in a stream. Bilby and Ward (1991) reported a change in the log sizes as well
as the species composition within 5 years of harvest. These disturbance effects are
detectable for a very long time, as evidenced by lower wood volumes and log sizes in
Rocky Mountain sites with riparian disturbances that occurred around 1900 (Richmond
and F ausch 1995). Recovery to preharvest levels is predicted to take more than 250 years
in some Alaska streams (Murphy and Koski 1989). Although the effects of past harvest
may result in long-term changes in abundance and size distribution, logs may remain
resident in the stream for many years and continue to perform ecosystem functions within
the channel, unless they are mechanically removed or transported downstream (e. g.,
Murphy and Koski 1989, Evans, et al. 1993). Riparian vegetation conversion from
woody vegetation to herbs and shrubs probably took place in two stages in the Saginaw
Basin; during the intensive forest harvest activities late last century, and then again when
agricultural production intensiﬁed in the region. Unfortunately, the effects of riparian
vegetation harvest and conversion are exacerbated by channelization, which, in addition
to enlarging the channel, mechanically removes roughness elements such as boulders and

logs to enhance drainage from adjacent farm ﬁelds. The result is complete removal of all

53

remnant and modem-day CWD from the channel and a functional simpliﬁcation of the

stream channel.

Influence of Landscape Features

Land use and Quaternary geology are integrally linked on this landscape due to
the strong interaction between hydrologic processes and soil porosity. Highly
impermeable soils associated with lacustrine regions, especially lacustrine clays, are
dominated by surface water ﬂows (Wiley, et al. 1997). The resulting “ﬂashy” ﬂow
regime may have greater power to transport CWD through the system. In contrast, the
more porous soils associated with morainal deposits result in groundwater-dominated
systems with relatively more stable ﬂow regimes. which are likely to be more retentive of
CWD. The Saginaw basin was historically covered by extensive wetlands, the great
majority of which have been drained and are now under agricultural production (Comer,
et al. 1993). Within the study area, agricultural production is pervasive but is most
intensive in the lacustrine clay regions; wetlands and range land, and to a lesser extent,
urban lands. are most prevalent on coarse till or outwash sand and gravel. Forest land
cover is found in the ﬂoodplains of the larger rivers, and in association with low

productivity soils such as lacustrine sands (Table 2-15).

Urban land use. link number, and the SD. elevation were the best predictors of
CWD abundance. debris accumulation density, £(accum size), and the CWD presence /
absence (Table 2-11). In each case, either urban land use or link number explained the

54

 

 

'1'?
l 11

C07

VCg§

river

altht

 

 

lou :

greatest amount of variation in the CWD variables. Link number is most closely
associated with catchment size and stream density (Table 2-15), and is also highly
correlated with channel morphology, particularly, stream bank-full width, the percentage
of the reach with slow units. and the maximum depth of pools (Table 2—14). These
relationships are more pronounced on lacustrine landforms, where link number is
positively correlated with the presence of outwash sand and gravel lacustrine clay soils,
and agricultural crop land. Topographic heterogeneity in this landform is most likely to
be associated with the larger river valleys. The association between CWD and link
number is reﬂected through its control on channel morphology. Larger rivers are less
likely to have been channelized. and are more likely to be associated with a ﬂoodplain
(which behaves as a buffer to agricultural land use and urban development). A negative
correlation with % open canopy indicates that these systems have woody riparian
vegetation that can provide a source of CWD to the channel. Lastly, these lacustrine
rivers are characterized by having a deeper pools and a larger percentage of their reach in
pool habitat. Such features are known to be associated with the presence of CWD,
although the mechanisms forming these pools are independent of the presence of CWD in

low gradient streams (Beechie and Sibley 1997).

Urban land use also exhibits some interesting patterns with respect to other
landscape variables (Table 2-15). Across the two dominant landforms, urban land use is
correlated negatively with agricultural land and positively with range land. Whereas land
use x geology interactions were more common on lacustrine landforms with respect to

55

.....,_,,_..1

 

In inr-

 

sueu
assoc
infra:
more
teget
0n la
Whicl
Sllllt‘l
detel
the ca

that 11

the at

how:-
metric
morai
Charm
011 la:

Chara:

link number. the interactions with urban land use are most prevalent on morainal
landforms. On this landform urban land occurs in association with hilly regions where
stream densities are low, soils are dominated by coarse ﬁlls, and land use is strongly
associated with range, forest and wetland land cover in a catchment. Urban land use
inﬂuences CWD indirectly through a lack of agricultural land use in the catchment, and
more directly through positive correlations with riparian vegetation height (woody
vegetation), wider riparian zones, comparatively closed canopies and stable ﬂow regimes.
On lacustrine landforms urban land use is highly correlated only with % range land,
which has relatively wide riparian buffer strips and a highly variable riparian vegetation
structure. Urban land use is primarily associated with residential, rather than commercial
development, and is not very abundant across the study region. The median is 5.7% of
the catchment on morainal versus 2.7% on lacustrine landforms, which again suggests
that the factors such as topographic relief and soil type, in conjunction with land use, are

the actual factors controlling CWD volumes in these streams.

The SD. of elevation variable is common to three of the ﬁve CWD models,
however, it contributes greatly to the explanatory power of only the 2(accum size)
metric. Topographic heterogeneity is not tightly associated with either lacustrine or
morainal landforms (Table 2-15). Some moraines with inclusions of outwash are
characterized by very low relief, while lacustrine sand dunes have high t0pographic relief.
On lacustrine landforms low topographic relief is chieﬂy associated with streams
characterized by small catchment size, low link number and stream density on lacustrine

56

 

li-

sani 51'
uuhlt
“00d:
18161111
would
relief (
urban.

can-op:

predic
51165 \I
must i
gravel
Study
diflen
Wetlar
mOrai
P051111
catch)
”ran;

Suppb

sand soils. with narrow bankfull widths and shallow channels. Land cover is associated
with forests and forested wetlands. These characteristics suggest that there is a supply of
wood for the channel. along with small channel dimensions that may be relatively more
retentive of woody debris. In higher relief areas on lacustrine landforms. the conditions
would mirror those described above in conjunction with link number. High topographic
relief on morainal landforms is most highly correlated with low stream density, wetland,
urban, range and forest land cover, on coarse tills; stable ﬂows and relatively closed

canopies would both supply wood and retain it in the channel.

As previously mentioned, the presence of CWD in the stream is more accurately
predicted than its absence. Landscape characteristics that account for CWD volume at
sites where volume > 0 were different than the landscape variables discussed above. The
most important predictors of C WD volume include wetland land cover, outwash sand and
gravel and lacustrine clay. Like urban land use, the total acreage of wetlands across the
study area is small, never exceeding 13% of a single catchment. There are striking
differences between lacustrine and morainal landforms in the median proportion of
wetlands, with the median over all lacustrine catchments being 0.3% versus 3.9% in
morainal catchments. Wetlands are negatively associated with agricultural crop land, and
positively associated with forested lands, which implies a potential supply of CWD in the
catchment. A signiﬁcant correlation also was observed among wetland land cover,
riparian vegetation height and riparian zone width. which further suggests that there is a
supply of CWD associated with the wider riparian zones in the stream corridor. Lastly, a

57

stable ﬂow regime, indicated by a negative correlation with ﬂood height, suggests that
these systems are capable of retaining the CWD that is delivered to the channel. The
positive relationship between wetlands and CWD abundance probably stems from both
the association with forest land cover and the role of wetlands as a factor moderating peak
ﬂows. Lacustrine clays, another predictor of CWD volume, are highly associated with
agricultural and use and ﬂashy ﬂow regimes, both of which negatively inﬂuence the

supply and retention of CWD in streams of this region.

Outwash sand and gravel was an important predictor of both CWD volume and
the density of debris accumulations. On morainal catchments outwash sands and gravels
are strongly negatively correlated with agricultural land, and positively correlated with
range, forest, wetland, and urban land use / covers. The inﬂuence of outwash areas on
CWD density stems is related to channel morphology (i.e., wide channels, large percent
of slow units, presence of deep pools), with wide riparian buffer strips and stable ﬂow
regimes. Overall, the absence of agricultural land use and its associated impacts probably
accounts for the positive inﬂuence of this variable on CWD volume and the density of
debris accumulations. On lacustrine landforms, outwash sand and gravels are associated
with large river ﬂoodplains, areas which are protected from agricultural and urban land
uses, and which generally retain woody vegetation in the riparian zone which can

contribute CWD to the channel.

58

 

Hlt’rdft'll.

Operating
b} landsc.
Oil) abu
predicted
canon p1
combined
metrics. i
that are 5'
relationsl

there is 1

density
approxi
1 l l. an:

Lastly,

 

Hierarchical Relationships Among Factors Inﬂuencing C WD Standing Stocks

The abundance of CWD in a reach was hypothesized to be controlled by factors
operating at local scales, however. local scale factors would be hierarchically controlled
by landscape-scale factors. These hypotheses were accepted to be true with respect to
CWD abundance and volume. C WD abundance and presence /absence were poorly
predicted by local scale factors including percent open canopy in both cases, and open
canopy plus bank-full width in the case of CWD presence/absence. The model which
combined data from all three spatial scales, did not include either of these two local-scale
metrics, but did contain metrics (e.g., link number and to a certain extent, S.D elevation)
that are strongly correlated with bank-full width, in particular. As previously stated, this
relationship is well recognized in the literature, and lends support to the hypotheses that

there is hierarchical control.

These hypotheses are not as well substantiated for the debris accumulation
density. First, the local scale model of debris accumulation density accounts for
approximately the same amount of variation as did the landscape scale model (Table 2-
11), and second, the multi-scale model incorporates both local and landscape variables.
Lastly, the predictor variables in the multi-scale model do not reﬂect the extent of large-
scale control over local processes seen in the models predicting abundance and volume.
The models predicting the distribution of debris accumulations are reﬂect both local and
landscape-scale controls. The abundance and distribution of debris accumulations in the
channel are controlled by within-channel features, as well as the potential supply of CWD

59

from the

relations

Ella“ (l

densities
rel ect t1
wood \‘0
Wood to
that inch
have strc
not amer
qualitath
measures
Were the
Ptrform

011161 Stu

Val-13110 n

analyses I

from the riparian zone and upstream sources. The predictive models reﬂect these

relationships well.

Effects of data type

Wood volume is the measure that is most commonly used for studies of CWD
densities in streams. Values derived from the line transect method do not adequately
reﬂect the abundance of wood throughout a reach, and appear to underestimate actual
wood volume in the Saginaw streams. In contrast, the line-transect method overestimated
wood volume in a lowland Australian river (Gippel, et al. 1996). Wood volume measures
that include a set number of logs per reach, or are measured for all logs in a reach may
have stronger associations to local and landscape features, however. these techniques are
not amenable for use across many sites. The lineal estimation method, which is a more
qualitative measure, better reﬂects the abundance of CWD in a reach. The other two
measures of CWD abundance that were well-predicted by the multiple regression models
were the number of debris accumulations and 2(accum size). These metrics are easy to
perform and are easily reproduced between technicians, and may be viable candidates for

other studies of CWD across large numbers of sites.

The low R2 value of models predicting woody debris volume indicates that either
three transects in each habitat type per 100 m reach were not sufﬁcient to capture the
variation in wood volume in the reach, or the variables included in the regression
analyses did not explain the variation that was present. Another study that measured

60

 

tdu:
oire
111535

nunu

SUear
each:
“00d

lanai

huphc

Rkhm
uteri
COmmt
manna
Eusocha
expecte
mahm
the lam
lahdSCa;

hldIOlOE

volume along a transect (e.g., Wallace and Benke 1984) used fewer transects per length
of reach, but sampled longer reaches. Most other studies quantifying wood abundance
measured individual pieces of wood throughout the reach, but sampled a restricted
number of reaches (e.g., Robison and Beschta 1989, Murphy and Koski 1989. Carlson, et
al. 1990, O’Connor 1992). Bilby and Ward (1989, 1991) sampled a large number of
stream reaches. but restricted their survey to 50 logs (> 10 cm diameter and 2 m long) at
each site. It is impossible to say at this time whether the lack of predictive power for
wood volume in this study is due to a sampling issue or to one or more environmental

variables that were not measured.

Implications for stream ecology

The most striking outcome of this study and others in these catchments (e.g.,
Richards, et 31.19%, 1997; Johnson, et al. 1997) is the pervasive effect of land form
rather than land use on many aspects of the habitat (including woody debris density),
community structure, and water chemistry. Agricultural land cover was expected to be
associated with lower densities of CWD, while forested land use were expected to be
associated with greater densities. Furthermore, local- and riparian-scale factors were
expected to play a large role in regulating the overall abundance and distribution of CWD
in a stream. Contrary to original expectations, the data indicate that between 40- 50% of
the variability in wood abundance (m / m2) and volume (m3 / m3) can be accounted for by
landscape-scale features (predominantly those associated with the regulation of
hydrology and channel dimensions), and less by land use and land cover in the catchment.

61

 

 

forest

3506C

to lac
lmple
Clean;
‘0 the
l’l’hich

dOmin

 

The abundance and distribution of debris accumulations, however, does follow the
original expectations by responding to the structural characteristics of the channel and the
potential sources of CWD. Most studies of stream ecosystems in highly developed
catchments have been designed to characterize the inﬂuence of a particular stressor (e.g.,
forest harvest. agricultural management practices, urban run-off). As a result, many

aspects of the structure and function of these ecosystems have not been well documented.

Although debris accumulation density and distributions are well explained by the
local, riparian and landscape variables, the relatively low power of both local and
landscape characteristics to predict CWD abundance and volume is surprising. These
results are probably due to three factors: the ﬁrst is the lack of direct measurements of
hydrologic patterns (e.g., peak ﬂow and duration); streams in this region are largely
ungauged, therefore appropriate data describing the hydrograph in the study area were not
available. There are currently only three stream gauges in the basin, and these are located
in the lower reaches of the basin and therefore do not reﬂect the intensity and duration of
the peak ﬂows associated with the ﬂashy streams of the region. The second factor relates
to lack of data regarding the social and economic variables that inﬂuence the
implementation of management practices, such as the extent and frequency of stream
clearing, and the extent of riparian vegetation conversion. The third limitation is related
to the complex interactions between land use and Quaternary geology in this region,
which are best illustrated by the situation on lacustrine landforms. Catchments
dominated by agricultural land uses on lacustrine landforms have among the highest

62

abundance of CWD and largest number of debris accumulations. However. the
proportion of agricultural land use is negatively correlated with CWD abundance over the
entire study area. This highlights the role of local conditions in the channel and the
riparian zone in lacustrine regions as a controlling factor of CWD density and

distribution.

Across the region local-scale factors are important for explaining the distribution
of CWD in the reach and the types of debris accumulations that deveIOp (Johnson 1999b);
the local factor that is the most important predictor, bank-full width, is largely controlled
by larger-scale features. Hydrologic regimes are understood to control many aspects of
stream ecosystem structure that ultimately regulate the biotic communities (Poff and
Ward 1989). The hydrologic regime is itself regulated by catchment climate, topography,
geology, soils, and land use. Land use is the most visible of these factors, and has
potential to alter many of the physical and chemical attributes of streams ecosystems. As
a result, land form is frequently ignored as a controlling factor in studies examining
interactions between landscape-scale features and stream ecosystems. The current study
serves to illustrate the importance of land form in the hierarchy of factors controlling

stream ecosystems, especially in developed catchments.

63

 

0.3m EC 1.0.4.20 m_u£..—~ :11-----

.MLPF-aduxshwmrundnd FP-nwnuhumw —UP-nw MF-L-N

ailnlin . ..
. . 5..- q.

 

.32 .a a .32... use romeo .05 :85 $2 97.5

.832
wee—83 ”28—53 8:38

 

.32 8.855
can «amt... due 59:08—55 wee :oEeBm :3— ._e 8 .EESWE £23 .3 8 508022.

£28 3982: «888

 

. dog 539,—. ”No2 Exam use mEoEm

mum—Ewan:
$535 8322:

 

.32 a s £5552: ”82 can Ea

 

0383. ”go— :omaam use ccoEcoE 602 .130 £02.30 £3. .38 £322. ”SE .5332 Econ owns—q E8
.82 Suﬁsm Ea no»: ”32
.835 use oEooom ”Mai ._m we .nzﬁm 3&0— ..m Ho .bomouo mam— :OmemBm was «5:8me oEoE

Soo— Snomom can 53301 awa— ._m .o £03m mama .3 Lo 9:92 Kwﬂ .3 5 .835
6mg ammo: Hui ._e 3 .bowoumv mama .3 Ho 2035 53— his... new 8.3% ”SE >282

356335. use 8:268
burr—con mavens—LE

 

.82 a a 65.98:: ”32

conga Ea 28852 ”no? ._e B .£_Em ”moi cememam 28 83:8er 692 ._e we £8:on
as: use: 6% 280m Ba 8833. an: .53 2a 3.5 $2 a .0 .2222. $2
Swen £2 :32: $2 3.5 "32 Ex Ea 58:85.. ”SE :25. as 3233::
use a e .82; £2 35 an: 8820 Be 8.2.; as. cesam Ea .23

mono. 85¢. new ems—BM 63— .3 .o .c8ce3m £52 .130 58.692. Koa— ._a Lo deacon—EN

 

8282:? 98 80:8
.2520 whose ”Egon
.523 6535 828.55

 

>OOAOIAEOE quZ<IU

 

 

mUZm—Mmmmm mmaémbq

map—km...“

 

 

.mEQmmmooo 835 can £523 :0 acute mince .3095 838 :o 2388: no DES—Sm .TN 2an

64

..:.&

 

Ng— ..o:=oU,O .vwe 5M 3:3 838.092.

32.3:
8892.35 388.05

 

mmERm—mOMm thm>moom

 

 

.33 .3 «o .bowBO

9:8:
.33 woo: 3253::

 

.32 bowEO 3:3 225 .29 acmcwgm 3:3 8:3—

3.25 33336

 

.32 22.20 3.3 3.25 .32
.3 3 ....em .82 5.5 3.3 3.33.3 .32 .3533 3.3 3232:: .22 38.3% .23 3:3.

89.88..
.8398 v3.3 8053:...

 

moo. acmcaam 3:3 358332 63— .225 3.3 cam—332 63— :8392 3.3
3.3.. .22 .333 3.3 3.3. .32 3.... .23 2.9.... .32 3.82 .52 3.3 .22 5.38m

093.03 E2533. 388.05

 

.82 .3 3 6.5
.82 83.3% 3.3 22.3.32 .52 3.3V. 3.3 23:88.). .32 332). .22 5.33m

tang:
.5868 80:25:.

 

.32 .3 3 $9.3”. .82 .3233. 23 52.5.... .82 33o; .22 .3 3 52.20

55:82
0:33.33 82.255

 

.82 £35 3.3 3.3.2.3 .82 5.38m 3.3 83.33. .22 .23.. .32 .33.... .32 83.;
.32 :3. 3.3 33:535. .32 30.82 .22 23.. .33 3.30. .82 .3 3 5:32:53

 

5353.83:
26: 33805

 

mmmmmOOMm \ det<m 30.:

 

 

32.538. .-N 2.33

 

 

65

 

.22 5580.0 .82 ._3 .o .583. .82 3.03820 3:3 33—th

3:053: .8 3:... 82.6.:

 

.82 83...... 3.3 33383 .82 .3 3
£8.33 .82 5...... 3.3 3333.3 .82 3.2.33. 3.3 .335... .82 .3383 3.3 32.3. .382
3:9. .82 .3 3 .3233 .82 333 3.3 83m. .82 .33. 3.3 .3338... .82 8.... 3.3
338.: .82 8.... .82 3.3.... 3.3 333 .22 8.3.. 3.3 3:3. .22 .3333 3.3 .3332

ow3.o.m
.333... 2:39: 33380:.

 

.82 333.30 3.3 .3: .82 33.33 3.3 332.3 .82 .3 3 .3333

83.898 6503 2.2.6.:

 

.32 3.3— 3:3
:o.o>m .32 58:33 3:3 2353353.. .32 3.03880 3:3 33—3.; .mp2 .o>.3Z 3:3 3.3.3.5

Eiwao. 38:330.:
3:3 Bo: 8223.:

 

.32 ._3 .3 .5...qu .332 ._3 .0 582m .332 :3— 3:3 .22...qu

2.56.88 53... 38:33.5

 

82 3.3.3: 3.3
33.3232 .82 .3 3 .3333 .38. .3 3 5.3.32 .82 33.320 3.3 3.33.3 .332 33.53:

32:08—33
8.. .833: w:_.3o. $330.:

 

.82 8.3.3 3.3 2.383 .82 38.3... 3.3 333.33... .82 28.332 3.3 33.33. .82
.3 3 .3333 .38. .3 3 .8932 .82 .38 3.3 33.3333... .82 33. 3.3 33.3: .82
.82 3333 .82 .82 .3 3 .3833. .82 383.3 .38. 33.380 3.3 3.3.; .22 8.33
3:3 8:3— .wn2 ._3 .o 4.8.02.3. .32 83M 35 3.38.00 £52 59:03:55 3:3 :Om:33m

.833: :3... 38308:.

 

.32 ._3 .3 6335.325 .32 ._3 .o 803:3? .32 8580.0 .82 ._3 8 £83.30 .22

.3 3 .3233 .82 8.5.... 3.3 333 .82 83.33 3.3 3b.... .82 .3 3 .3333 .82
e333 3.3 303.3.... .82 83.33 3.3 3...... .82 :3. 3.3 33533.2 .82 8.32 .82
._3 B .EosBoZ .22 mag 3:3 358830 £52 ._3 .3 68.222 .232 0.0:...34 3:3 .822

Summon—Eco 23.5888
83.93.35 38:26:.

 

.32 ._3 .o £88m .32 ._3 .o 63—55

530230.:
3352.3... 83322..

 

 

82.5.3.8.

 

TN 033...

 

 

66

Table 2-2. Distribution of sample catchments and sample reaches (in parentheses) among
factor levels in the factorial design. Agricultural lands have a minimum of 60% of land
under production.

 

 

 

 

Land Use
Agriculture Mixed forest +
Agriculture
Geology Morainal 3 (9) 3 (9)
Lacustrine 3 (9) 3 (9)

 

 

 

 

 

 

67

 

‘1
D ",

 

Table 2-3. Coarse woody debris variables measured during the study.

 

CWD Variable Description Units

Name

Abmdanc;

m/m2 cumulative CWD length per unit area, m/m2
wood 2 5 cm diam and 1 m in length

Vol(2cm) CWD volume per unit area, (wood > 2 cm mil/m2
diameter; .25 m in length)

Vol(Scm) CWD volume per unit area, (wood > 5cm m3/m2
diameter; .5 m in length)

Vol(lO cm) CWD volume per unit area, (wood > m3/m2
10cm diameter; .5 m in length)

Count (2) # logs counted per transect (wood > 2 cm
diameter)

Count(5) # logs counted per transect (wood > 5 cm
diameter)

M2:

Diam(5) Mean CWD diameter, wood > 5 cm m
included

Length (05) Mean CWD length, wood > 0.05 m m
included

Debris (from Johnson !999b)

Ammulatism

# Accum Number of debris accumulations / 100m

Sum Accum Size

Derived measure reﬂecting the amount of
stream bottom covered by debris
accumulations

68

Table 2-4. Size classes of debris accumulations based on methods of Shields and Smith
(1992). X = channel width at the upstream point of the debris accumulation. The sum of
all debris accumulation sizes in each reach represents a measure of the amount of channel
covered by debris accumulations.

 

 

 

 

 

 

Size Size of Accumulation in Direction Parallel to Flow
Perpendicular

to Flow < 0.25 X 0.25 - 0.5 X 0.5 - X > X
< 0.25 X 1 2 4 5
0.25 - 0.5 X 2 3 6 7
0.5 - X 4 6 8 9
> X 5 7 9 10

 

 

 

 

 

 

69

 

Table 2-5. Channel morphology and physical habitat variables measured during this
study. * Indicates the variable was used to predict the abundance and distribution of
CWD. Fast and slow units are defined in Hawkins, et al. (1993) and represent riﬁle and
pool habitats.

 

Variable Description Method
Boulders, cobbles, Proportion of substrate particles in Osborne, et al. 1991;
gravel, sand, clay, silt each class Platts, et al. 1983.
# Fast Units, % Fast Number of fast units per reach and Hawkins, et al. 1993
Units“ (% fast) proportion of wetted area with fast

units

# Slow Units, % Slow Number of slow units per reach and Hawkins, et al. 1993

Units (% slow)* proportion of wetted area with slow
units
Maximum depth in fast Greatest depth recorded in the fast Hawkins, et al. 1993
unit (maxfast) units in the reach
Maximum depth in slow Greatest depth recorded in the slow Hawkins, et al. 1993
unit“ (maxslow) units in the reach
Mean bank-full width” Mean bank-full width Osborne, et al. 1991;
(abankwd) Platts, et al. 1983.
Mean bank-full depth“ Mean bank-full depth Osborne, et al. 1991;
(abankdep) Platts, et al. 1983.
Wetted Width (width) Mean width of wetted channel at Osborne, et al. 1991;
low ﬂow Platts, et al. 1983.
Wetted Depth (depth) Mean depth of wetted channel at Osborne, et al. 1991;
low ﬂow Platts, et al. 1983.

Habitat area (habarea) Mean width "‘ mean depth
Habitat volume (habvol) Mean width“ mean depth * reach

length
Flood height“ (fldht) Maximum bank-full depth
Instream cover amount Percent of wetted area with cover Armour, et al. 1983
(incovamt) (e.g., macrophytes, overhanging

bank)
% Open Canopy Proportion of wetted area not shaded Armour, et al. 1983
Coverage“ (campy) by riparian vegetation

7O

Table I
ranabi
Vanai

\Vidtl

Slope
Veg-e
\"ege
°.-"b R
0/0 T

050 \

Flt

Table 2-6. Variables describing riparian zone structure and composition. (* indicates
variables used in most analyses.)

 

Variable Description

Width“ (ripwidth) width of the covertype immediately adjacent to the
river

Slope“ (ripslope) mean slope within riparian zone

Vegetation type (riptype) vegetation cover type

Vegetation height "‘ (ripht) vegetation cover type height

% Row crop" (rrowcrop) percentage of riparian zone with rowcrop agriculture
% Forest (rforest) percentage of riparian zone with trees and shrubs
% Urban“ (rurban) percentage of riparian zone with urban/residential

land use adjacent to stream

Floodplain slope mean slope within the ﬂoodplain.

7l

Table 2-7. Landscape variables used in analyses and spatial data used to derive data.
(U SFWS = US Fish and Wildlife Service, USGS = US Geological Survey.) (*
indicates variables used in most analyses; others are omitted due to high correlations

with other landscape variables.)

Landscape Variable Data Set

Data Source

 

standard deviation of elevation“ elevation
(sdelev);
mean catchment slope" (slope)

proportion of land use classes in land use / land cover
catchment; (see text for variables used)
land use patch density (ptchden)
stream density“ (strmden) hydrography;
digital elevation model
proportion of surﬁcial geology Quaternary geology
in catchment“ (see Table 2-9 for variables
used)
Link number * Number of ﬁrst order streams
entering above this site
Stream order
catchment area“ (log wshed station location;
area) topographic map; digital

elevation

72

USGS digital
elevation model

MIRIS database (MI,
DNR)

USGS, digital line
graph

Farrand and Bell
1 984

Shreve, 1966

Strahler, 1964

GPS readings; ﬁeld
notes; USGS topo

Table 2-8. Aggregation classes for Quaternary geology categories.

 

Aggregated Class Quaternary Geology Particle Type
Coarse till Coarse-textured glacial till,
End moraines of coarse-textured till
Medium till“ End moraines of medium-textured till,
Medium-textured glacial till
Fine till End moraines of fme-textured till,
F inc-textured glacial till
Sand and gravel* Glacial outwash sand and gravel and postglacial
alluvium;
Ice-contact outwash sand and gravel
Clay and silt“ Lacustrine clay and silt
Lac Sand“ Lacustrine sand and gravel, Dune sand

Peat and muck

Coarse till + sand and
gravel“

Peat and muck

SCC tCXt

73

:..lt‘ -....l...:.: . :.lc...€.< C h. {7‘55

 

 

74

:2 and- NS- .93 Rd- mg 58.2.83 a

RS 86 m3- 39o- So- N2 awe“: ac:

83 :3. NS- _ 3- :5- 86. E5 :8:

E _ .o :3 mod. and .33 so- :5 3o:

:2 m _ .o- :3- ad :5. 53. ~82 30:

3.20

w v m N _ 857.5

3325 w a h h a Ea .oz
nanoseou 2 _N 2 : m E: 28m 85in

AEoEHmsFm Betcha—om m .23 .32 36
v u a E cacao—mama 32865 L HEM 3m 05 so mean wcoEm cote—obooousm 33an 9:68 0:39,.” _ Maﬁa—z .m-~ 2an

 

 

98 H 3 8.8 H _H 8.8 H 2 98 H 2 mm... H m.” 353 £83
_ 8.8 H 83 88.8 H _88 88.8 H E; 28.8 H 88.8 83. H 8.8 €53 .55
:uz muz T2 anz cmnz 333
E H 2 mm H 3. .2 H 2 8.8 H E 8.8 H S. 8 893 28%
in H _d S H a: 3 H 8.2 _H H 2 S H 2: NH :83 22;.
I 88.8 ..
N _8. + 88.8 :8. H m .88 88. H 28.8 88. H 28.8 H :88 9. E85 :5
n 88.8 :88 88.8 888 88.8 I
+ 88.8 H 88.8 H H _88 H w 8.8 H 88.8 8 803 :5
n 88.8 m :88 88.8 88.8 I
+ 88.8 H 88.8 H 88.8 H 88.8 88.8 H 88.8 mm :53 :5
t... _ S + 2.8 8.8 H 8.0 8.8 + :8 .........8.8 H .2 8.8 H 8.8 a. Na \ a
gamma
ﬁuzv 8qu auzv €qu 8205
£3302 9:52 53:03 9&3; :32 595 Mo 82:52 2382 030

75

.Cood v a :80 x D: 2 26 “onto «505%? u it .886 m n 7c .8505 8? 0.8 o A ﬂ 0508 a 20:3 3% .«o 55:52
25% @003 335.80 mozm he Eco as 85308 :ch2 new 36:86 woo? 85:58: 29:89: 333 octagon:
awe—com 35830 can :5me 389» .8330th om: Ea— mmoSm 8582: £53 383 3.88 .«o .m.m H :82 .oTN 29m...

 

ow H :3

13¢.— H We—
2 "2

ms H @6—

wdH ON
auZ

_.w H odm

v; + oé

m6 H 06m

:3“.— H 56
2 NZ

3 H 3m 32m 582 N

:52

8 H E 3.88%
aeuz

382 8222 82.: gnaw—

Gosgaoe o E 28p

76

 

A098 + 88.05 88:50th 85
- £88828: mo: 36 + 88888 .Q.@ .86 - C880: 0.8: 86 + 3. - u 88888885

85820 .D.m
888 + 80:8 H 0tmw

 

888 .828: $0: 523 0:0N .8: 30:8 :80 .x.
898:: 08: 3: 8038.88 .8: 8.23 .5383
88.8 n a 88.8 n a M88.8 n a 8 u a
58 u 8 _3. 8.8 u 8 :2 :8 u “a _3, 85338588
8888— 8 088 ”88888885 808888885 ”008888885 AH8\n8v 0820> woq
.0002 .880 058884 8885 _804 Boar—3’ 0003

 

.82 08 A8883 0 u 2:20,» :0 8088:: o A 0823 930 8 8050: 88888885

:8 8 :08: mm 0388 :08 :05 .808 :8: BS. 838:? 888588 :08 0. .8: 8.888 8 80:20:00

05 50 8:085 .Sm u : 6A 8B 0820> 930 08:3 8% 3 3:0 8:58 .808 L 808558: :8 8:80:88 £58 .8003
8880 :05 888 38% 808558 :0 88a 2808 88888 28:08 88 80:08.5 888888: 50 888800 ._ TN 055.

77

 

888 .08 08.8 + cones: 85 85 + €35 we: N: + an - u use 830

7888898 0:
8 u a 8582.
Ea: so:
58:8 88:8
838:: 05::

i n a .830 u a
62. u E .8
lejqu: 0a a

3: 88388 8808

2808 .m8 0:
88mg

6N u : .805 u 5 T3080 :80 .x.

 

885.0 .08 :3 u 8 .8 in a .89"; 6m.
:82. E 8.8% u w. .3 9883
M88: 80: .E 838%? 888.. .3080 850 .x.
:88 u a 2880 u a mmood n a E. u 5
3888 a 25.9. ”S. u "a .8 KS n E .8 a; u R .8 use me:
A cm H :v
2883888 88300 .5 8.0 + 920 8:88:02 o\cv 5.0 - 88:03 £9 3.0 - m.m- n 0833 Rod
:808 0: 488.885 :808 .w8 0:
:88. 2. 33335 8qu
>20 0:88:02 5808 .w8 0: 8:8: 308 .x.
888888 58380 ”Raga mm 7": ”No.0 u:
688 :85 .8 8 :2 u 8 .8
830 L 8834_ 83334.
08088— 8 08mm wow. u N5 :8 _808 «80588 0: .808 885:me 0: cm H :
H... 0820> H... 8830> {0823/ {0823/ :bEkEV 0823 wou

 

888808

:-~ 28,—.

 

.2238 038-338

2:8: 2 won: Bo? 225.: 2.83::— :5 5:8: .80— 05 :_ £8062: Emu—mama 203 35 333.5» 332%
$30 8538— .x. 6:3 8:63:— ..\o :5 8:62: _x. A_o>8w :5 :53 595:5 + :5 838 ..\o :8 33:33am :835:

.x. :5 :3 3.88 o\.. Avg—83 37.28 .x. :8 @5303 .x. A595 3:523: ..\o :8 :35 .x. 038 .x.+ +622 o\c. . 3523:?
.x. 56:2. 8:26 £2353 E2538 dd .30? «58:88 :38 Anna 60:50.8 wo_ :8 Spas: 2:: ”3%..—

.umohom 5 0:8 5:83: ..\o

.535 :_ 0:0: 5:39: _.\o £8032 :_ 0:3 55:: c\.. .2203 5:303.» 552:: duo—m 0:8 5:89: .523 5:3: Mg
.523 .3325: :88 .53: =&-x:3 :35 “8E: 30? 5 5:0: 8:822: .220: coca £25: 32m .x. c3053 :3: ..\° 3de
”20:9: 2: :m 832:3 Eggnog?

 

A:: :5
8: + 3; 855%? as n; + 323 :55“. 53.5 E. + no - u 3% :58an Eve
:3 8:30
am: :5: :35 Amos 82353 .D.m .

a: :o_§owo> E: Um: :52 :35 30:3 5:: .x.
5.23 _Eéan :88 ac: 5:82 Va: .2 85%? a: ”523 38.2 2% n 5
88. u a 88. u a 38. n a 88. u : A85
i. u N: _3. 5.. n N: _3. an. n N: .3 am. n N: .3 882% o: 5

 

2.23 :55: 2.: + as; .2528. $2 a :8 + on: - u :52 35:22:82 u $35

595:: 2::
£23 5:253 322w \ :53 5:35: 3:53 :30 ..\°
“:35 €028: Awe: ”:35 4:023: Awe: .E :o_§owu> dc ”£23 =82§n Aav .I. 5
M58: u a :83 u a :83 u : m58.0 u a e 2: 28%
id u N: .e< is n N: 3 63 n R “as u N: .a< 250: a 9:5

 

62.5.53 ._ TN 2%:

79

 

$28 380350 on
.88m :38 S - o

@685 SEO 29 m

$885 620 23 3

.886
Sod
$86
$56

Sod - wood

8382 302
.U.m £020 .0on
.U.m £020 {962
<6 1080 x85

.<0 53% oonooowo

53:22 £80393
v82: £38m was?»

2288mm 262

82 5:5

$2 .25:

$2 5%:

$2 8.5m as 8233
$2 3.5m Ea 8%?

dwa— xomEEU Ea 8.th
”S .83 Raﬁ—pas: £35m

one 2.8.: Ba 35
82 5:5 Ea £35 5

80

 

3H - cod - mEmobm «32 03— .cmﬁnomm Ea BEEN
53o - 8 :53
H 32:. “Baa :3 m a
awe A» :3 - 8 88¢ “~32
md H wé H Coed ASE—u :52 MV gwanz :853. can 35% £5
E 8: Esoo< £500 ANFEMEV 562% GB: cots?— 853.3%

.3826 c3566 5 EoEEwéo— 35o
8.“ 329» won—£32— 53; §wE£2 E 382% £55 Baﬁwmw .8 852% Esau €003 puma. .8 saga—Eco .NTN 2an

_.o-o

w.m - co

w.m - Wm

mvoo - _oo.o

©N_o.o
m _o.o H omoo
wooo H hmoo
vomo H NKo
noﬁo H coo.—
hzoo - good

98% Eu m A
mooo -oooooo

_ooo.o - oooo

woo. - mvoo

85m Samaria?

E £886 $28 895?“:

oEQ So .59» o—
ozﬁ: So a?» o—
nion: Bo SPA om—
o>uac cusp—mo?
252% BanN 302

OD 3:826 ooh—56%

Eocwcg 3ch
Emma owmcmﬁv :chﬁrﬁ

E .35.
:05? .VE .m £82 @283

E .55. 832
.xm .m £38 3320::

82 556 Ba 258m
no? .3 5

5255 E .83 35:93::

coon—23:55 can bowohoo

82 .3 a .8?m
32 saga Ea 25552

moi bowEO Ea =oE=0
”mag .3 we .bowEO

moo. :zEm was mEoEm

~62 :zEm .28 mEoEm

 

@8553 NZ 22¢

81

 

wood S; H EH 2: H a? £92 Boa
moo v.3 H _Nv com H no 98m .x.
:3 E H _H n: H n: as. ”was .x.
voooo _Nmo H oo._ vo_.o H M: no 37.2% 885
oooo Him H ow w.m H 5.: gage—o .Q.m
Sod _ a. H 8% a3. H $3 Mao: 55:38
.D.w H :32 dd H :82
a Bum—Ana N 9:90 _ 9.20 oEmtm>

82

.8593 “yo—con 8o 2m 82—338“. 353.2%
_cotoocom .mEoEﬁEo AN 9.86 ”mats—2o 3.3—act? 3582: 28 Axgomd @838 05:88. may? 2 @280 metoEo
c058 3582: 28 @30on 35:33“ 05532 E 83589» 3828 .8 ovum—=28 38.. 038863... .m TN Bomb

 

 

 

 

 

 

 

owﬁwx

 

 

3830M

 

23K . '9“,

880m

 

 

EH EBB—2

 

95w 0:59.81.

 

 

 

 

V++

.35 05.5884

 

 

.++

 

E>SO>Emm
£8350

 

_E 858

 

ﬁasco Emotm

 

 

 

 

32m E05580

 

 

 

 

+

+

+

 

82

 

25

 

:<

~02 own— =<

 

 

502 on:— =<

 

Essa .om

 

3:52 is

 

mvu< «Cogoummv

 

no: UNJ =<

 

 

 

 

523 25M 5235

Emmi wo> Sakai

 

 

£23 =a€£

 

30ch SEC .x.

 

 

 

Svauisdvau

i. :86 v a n +++ .mEumquon. cons—oboe Eamon 80:2 235$ .98“ 05 E woman—86v 838m mcﬁcﬁm 930 he £32:
. coﬁmﬂwﬂ 2: E 030 we 280%2n 9.8% Ba £23 econ .8ch EB Jaws: cozﬂow? 5:8? .523 =£x§n 53:.
3058 :30 .x. 833—9 2:. .mncew5_ C2): 3582: can ﬁnd 2:538— :23? Ba .913 83 35%. 0.55 95 $28
83%.; 038 _82 Ba 55% 69835. 23 28m grant was .82 2: «a $8285 c833 mcozﬂoOmm‘v .3-N 035,—.

83

 

 

30ch :25 .X.

 

imam Bo:

 

5&5 _Exﬁm

 

52>? =€x§m

 

 

so; £95 522

 

 

 

E5 32w .x.

 

 

mohoaom SEEM

 

 

is:
.wo> $ng

 

£23 5:25.

 

c533

 

53.2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

€288 VTN 28¢

 

 

84

 

- + - - omcmm .x.

 

mohzsom .x.

 

 

- amp—om o\o

 

 

         

:C. €3on o\o

95m 2:52.05 .3

 

 

 

 

 

>30 2:333 e\o

 

 

_uizwhéam
53350 .x.

a... + a... 2:. 8809A.
x + . Us A. ‘ , ++ z 4 . +_ , -- + ++ + bucoﬂ Emubm
. . , ..I. + + 82m E25330
++ ‘ . . m. ++ ” . gage—m .Q.m

- - 58:22 xcﬁ

+ . +.+ .. + - . m2< EoEﬁEU

 

 

 

 

 

6.: v a...

 

 

 

 

 

 

 

 

 

 

 

 

 

 

82 05 =< 82 03 =< SE 03 .2 SE 03 .2

 

 

 

 

_o>SO\w§m 593:0 gage—m dm om: v53 :wﬁD 598:2 x54

 

 

 

 

 

 

a

Svauisdvani ._8.¢vau+++

.mgomu§_ encasem— §B @3308? :3 8.88 o: m_ 965 :2: 826%E .md £56508 8:288“. coma—“om 80:2 238%
4x8 2: E @8386 £258 552? 05 E Q30 .8 £92qu Q55. 03 622w EB 38 £3330 can 5283.0 dd

.3: was :35 558:: V2: moi—ES, 2F .mczomvsﬂ Cozy _aEEoE Ea Gad 055:8— :EE» can .83 85 3:6 2:5
05 $28 83%? 28m _82 can Swish—t “amowv§_ can 28» 2389:: 05 a 808:.55 :8an mcozﬂoomm< .m_-N 058‘

85

 

30ch :30 .x.

 

 

5%: Bo:

 

 

 

 

 

 

 

 

 

 

 

 

 
 

 

5&5 Essa

523 Evian

.8; save $2
:5 32m _x.

 

8533* «stung

 

 

 

 

Ems:
wo> 518E

 

 

 

 

£23 .553

 

 

«Em—63 .X.

 

 

 

 

:35 .x.

 

 

 

 

 

 

 

 

 

 

 

9:53 W; 2%;

 

 

86

 

 

Frequency
3.
O
i

 

 

 

 

100 -
50 l
1 o L, _ILII I I _ _ _ ;
i 0.02 006 0.1 0.14 0.18 0.22 0.26 0.3 0.5 ‘
CWD Diameter (m)
’ 300

i 250 r-

200 r.

Frequency
Q

_l
O
O

 

 

 

8 9 10 >10
cwo Length (m)

 

Figure 2-1. Frequency distribution of log diameters and log lengths measured in the
Saginaw Basin streams.

87

Appendix 2-1. Summary statistics of local / channel variables. N = 49.

 

Local Variable Median Mean 3; SEM Range
Flood Height 1.3 1.7 i 0.18 0.2 -5.2
% fast units 0.72 0.54 i 0.06 0 - 1.0
% slow units 0.27 0.46 i 0.061 0- 1.0
Max Depth of Pool 0.6 0.55 i 0.05 0- 1.2
Mean bank-full width 6.6 7.3 i 0.4 3.6 - 12.6
Mean bank-full height 0.59 0.65 i 0.5 0-2

% open canopy 0.70 0.64 i 0.05 0.02 - 1.0

Appendix 2-2. Characteristics of the riparian zone across 49 sites.

 

Riparian Zone Mean SEM Range
Variable

Width 25.17 1.75 2 - > 40m
Vegetation Ht 5.12 0.54 0.5 - 10m
% Rowcrop 0.32 0.06 0 - 1.0
% Residential Urban 0.21 0.05 0 - 1.0
% Forest 0.00 0.00 0 - .2

88

Appendix 2-3. Characteristics of catchments in the four landscape treatment groups.

 

Variable Lacustrine Lacustrine Morainal Morainal
Agriculture Mixed Agriculture Mixed
(n=1 6) (n=9) (n=9) (n=1 5)

Catchment 11,238 i 2004 2,923 i 367 8,785 : 2,129 9,935 : 1,232

area (ha)

Slope (%) 0.42 i 0.03 0.32 i 0.08 0.59 i 0.02 1.50 i 0.11

Elevation (m) 217.6 : 2.6 206.7 : 3.7 247.9 3: 2.9 307.7 _-t 5.44

Patch Density 6.67 i 1.86 8.31 i .33 3.88 i 0.18 11.7 i 1.64

(# / kmz)

StreamDensity 0.93 i 0.03 0.77 i 0.02 1.13 i 0.07 0.56 i 0.02

(km/kmz)

% 80.1 i 3.8 34.9 _+_- 0.65 81.9 j; 2.8 35.7 i 2.93

Agricultural.

crop

% Forest 11.3 i 2.1 35.7 i 6.2 9.3 i 2.0 22.2 i 0.85

% Range 3.2 i 0.7 11.9 i 4.0 4.7 i 0.7 22.7 i 1.26

% Urban 4.1 _+_ 1.1 9.5 :1; 3.4 1.0 i 0.2 9.50 i 1.30

% Wetland 0.21 i 0.04 2.2 i 0.42 2.2 i 0.4 5.8 i 0.48

% Coarse Till 0.0 i 0.0 0.0 3; 0.0 14.4 :1: 3.6 50.9 i 6.5

% Medium Till 24.1 i 5.3 7.6 i 0.4 73.4 i; 6.2 16.0 :1; 8.1

% Lacust Sand 21.5 i 5.8 88.7 i 5.5 0.0 i 0.0 0.0 i 0.0

% Outwash 5.4 i 1.6 0.0 _+_; 0.0 11.7 i 3.3 29.5 i 2.8

Sand Gravel

% Lacust Clay 56.5 i 4.7 3.7 i 1.5 0.0 _+_ 0.0 3.96 i 2.0

89

CHAPTER 3

COARSE WOODY DEBRIS ACCUMULATIONS IN LOW GRADIENT STREAMS:

RELATIONSHIPS WITH LOCAL AND LANDSCAPE FEATURES

90

Abstract

The abundance and distribution of four types of debris accumulation were
examined across 49 stream reaches in a highly developed region of central Michigan,
USA. This study was conducted to characterize the distribution of debris accumulations
with respect to land use, Quaternary geology, riparian, and channel-scale features.
Coarse woody debris (CWD) standing stocks in these streams are low when compared to
forested streams, however, the density of debris accumulations / 100m is similar. This
suggests that the debris accumulations in these streams are smaller and contain fewer and
smaller logs than streams of forested regions. More stable ﬂow regimes and less
extensive channelization are associated with larger densities of ‘log / snag’ debris
accumulation types in the morainal catchments with mixed land use. Active channel
management from dredging and removal of riparian vegetation contributes to a small
supply of quite mobile CWD in agricultural catchments. As a result, ‘loose log’
accumulation types are closely tied to bank-full width, which in turn is a function of
catchment area. ‘Root wad’ and ‘overhanging vegetation’ with trapped debris
accumulation types are not well accounted for by landscape- or channel-scale factors.
Debris accumulations in these Michigan streams do not play a physical role in modifying
channel morphology, as do debris accumulations in forested landscapes. The current
study illustrates the importance of land form and ﬂow regime and the secondary role of
local-scale conditions in the hierarchy of factors controlling stream ecosystems,
especially in catchments impacted by chronic disturbances such as agricultural
management practices and channelization.

91

Introduction

Until the late 1970's coarse woody debris (CWD) in streams was regarded
primarily as an obstruction to navigation and ﬁsh migration in streams and rivers. Many
important ecological functions of coarse woody debris in streams have been identiﬁed
(see reviews by Harmon, et al. 1986, Sedell, et al. 1988, and Gurnell, et al. 1995). The
role of woody debris in streams spans a broad range of physical, chemical, and biological
functions that, in turn, regulate many ecosystem properties. Channel dimensions and
structures are strongly altered in the presence of CWD, increasing channel width, altering
the longitudinal proﬁle of the river, forming plunge pools, and increasing channel
roughness (e.g., Zimmerman et al., 1967, Swanson, et al. 1976, Keller and Talley 1979,
Bilby and Ward 1989, Nakamura and Swanson 1993). In consequence, ﬂow
heterogeneity is increased, along with hydraulic retention, which promotes sediment and
organic matter storage (e.g., Gregory, et al. 1985, Ehrrnan and Lamberti 1992, Nakamura
and Swanson 1993). These physical changes in channel morphology, ﬂow regime, and
retention dynamics are associated with increases in the amount of suitable habitat, ﬂow
refugia, and food for ﬁsh and invertebrates (Angermeier and Karr 1984, Benke, et al.
1984, 1985, Bisson, et al. 1982, 1988, Richmond and Fausch 1995, Beechie and Sibley
1997). Despite the wide-spread recognition of the important role of CWD in streams,
citizen-based clean-up activities in urban and agricultural areas still remove woody debris

from stream channels.

92

Few studies have attempted to quantify relationships between landscape factors
and the observed patterns of CWD abundance and distribution in low gradient systems,
particularly in landscapes that have undergone extensive deforestation. Landscape-scale
factors, such as land use and Quaternary geology, undoubtedly play a role in mediating
the impact of disturbance events (e. g., ﬂoods, channelization, forest harvest, alteration of
natural riparian communities) that inﬂuence the export and retention of CWD and smaller
organic matter fragments. Streams in central Michigan have been subjected to a large
number of chronic stressors, beginning in the 1840's with forest harvest activities,
followed by devastating ﬁres. The denuded landscape was then subject to massive
erosion that swept large volumes of sediment into the streams (Quinlan 1997). The
current landscape is largely shaped by intensive agricultural practices and residential
development. Low-order streams in the Saginaw Basin of Michigan are extensively
channelized to enhance agricultural production in low-lying areas that were previously
wetlands, leading to reduced substrate and habitat heterogeneity while removing existing
pools of CWD. The historic and current land management practices and land use have
severely diminished the supply of coarse woody debris, resulting in low standing stocks
(Johnson 1999a) that are comparable only to disturbed streams in the Rocky Mountains
(Richmond and Fausch 1995) and agricultural streams recently cleared of woody debris
in Tennessee (Shields and Smith 1992). Under these conditions the dominant structural
elements providing in-stream cover, habitat, and organic matter retention consists of

overhanging vegetation and root wads of tree stumps or living trees.

93

Large particulate organic matter and CWD are derived from natural mortality,
ﬁre, disease, insect damage, ice/snow loading, and wind-throw damage to trees in the
riparian zone or uplands adjacent to the stream (Swanson and Lienkaemper 1978, Keller
and Swanson 1979). These materials are transported into streams by mass soil wasting,
bank undercutting and erosion, ice and snow damage, and ﬂooding. In some systems
beaver may be the primary vector transporting large volumes of CWD to the channel
(Naiman, et al. 1986, Maser and Sedell 1994). The processes controlling CWD input to
streams are inﬂuenced locally by tree species, stand age, soil stability, local topography,
and human intervention (e. g., forest harvest and riparian zone clearing), and regionally by
geology, climate, valley geomorphology and land use patterns. Once in the channel,
CWD mobility and retention is controlled primarily by channel morphology, especially
channel width relative to the size of the logs, and other obstructions (e.g., boulders, root
masses; Keller and Swanson 1979, Bilby 1984). There exists a complex relationships
between those factors inﬂuencing the input of CWD to streams on one hand, and those
regulating its retention, on the other. Channel morphology regulates CWD retention, yet
CWD can have dramatic effects on channel width, pool-riﬁle sequence (e.g., Keller and
Swanson 1979, Gregory, et al. 1985, MacDonald and Keller 1987, McKenney, et al.
1995), suggesting a feedback loop that inﬂuences the abundance and distribution of CWD
in streams. Debris accumulations are important habitats which support a large proportion
of the total taxa richness at a given site (Johnson 1999(1), however, little is known about
these habitats in Midwestern streams. This study was conducted to characterize the
distribution of debris accumulations with respect to land use, Quaternary geology,

94

riparian, and channel-scale features. The goals of this chapter are to: 1) quantify and
characterize the distribution of four woody debris accumulations types in low gradient
streams in developed landscapes; and 2) quantify the relative inﬂuence of reach- and
landscape-scale factors on the abundance and characteristics of debris accumulations.
Debris accumulations are broadly deﬁned here to include log dams and loose log
assemblages, as well as overhanging vegetation and root wads that function to trap coarse

particulate organic matter, including woody debris.

Methods
Study Area

The Saginaw Bay catchment of Lake Huron encompasses a 16,317 km2 region,
characterized by sand and clay-dominated lowlands rimmed by coarse-textured glacial
features such as ground moraines and outwash plains (Figure 1-1; Johnson, et al. 1997).
Previous work (Richards, et al. 1996, 1997, Johnson, et al. 1997) has shown that both
land use patterns and Quaternary geology mediate the landscape's response to
environmental stress, therefore the sampling design was chosen to reﬂect these factors. A
2x2 factorial design was used to investigate the inﬂuence of underlying geology and land
use / land cover on CWD dynamics (Table 2-2). Each cell in the design included 3
replicate catchments/streams, chosen from a pool of candidate catchments. Three ﬁrst to
third order reaches within each stream were selected to quantify internal variation within
the 12 streams, resulting in a total of 36 subcatchments ranging in size from 712 to
23,448 ha. These sites are collectively referred to as the “core” sites. One site was

95

impounded by beaver during the second year of this study, resulting in a total of 35
“core” sites. Sites along a longitudinal gradient also were examined in a lacustrine
agricultural (Lac/Ag) catchment (Bad River, 11 = 11 including “core” sites), and a
catchment dominated by morainal geology and mixed agricultural/forested (Mor/Mx)
land use (South Branch Flint River, 11 = 12 including “core” sites). These sites are
collectively referred to as “longitudinal” sites. A total of 49 sites are included in this

study.

Coarse Woody Debris Abundance, Distribution and Size

Coarse woody debris assessments-were performed at base ﬂow conditions during
the summer of 1995. Debris accumulations _>_ 1 m2 in area were classiﬁed by type (Table
3-1; Figure 3-1), counted, mapped, and photographed. The obstacle causing the
formation of each debris accumulation also was recorded (e.g., overhanging vegetation,
root wad, bank, point bar or island, snag or log, rifﬂe, or no apparent obstacle). Debris
accumulations were assigned a size class based on the dimensions of the accumulation
relative to the channel width at the upstream point of the accumulation location (Table 3-
2; Shields and Smith 1992). Debris accumulation data were summarized by total number
of debris accumulations / 100 m reach, median accumulation size, and a metric reﬂecting
the amount of stream channel covered with debris accumulations, derived from the sum
of all debris accumulation sizes (Table 3-3) per reach. For statistical analysis of

accumulation size x type, nine potential size classes were collapsed into three to reduce

96

Channel morphology and habitat structure

A stream reach of approximately 100 m was sampled at each site. This was
usually sufﬁcient to incorporate more than one rifﬂe-pool sequence and represented
between 10 - 20 times the stream width (Richards 1982, Bisson and Montgomery 1996).
A comprehensive set of parameters commonly evaluated in stream surveys were
measured within each stream reach, representing factors associated with the channel
morphology (e. g., bank-full width and depth) and habitat (percent of reach with slow
units, maximum depth in slow units (as per Hawkins, et al. 1993), ﬂood height, percent
open canopy, percent in-stream cover, substrate composition; Appendix 2-1; Richards et

al. 1996,1997).

Riparian structure

Measurements and observations of riparian width, riparian slope, vegetative
composition and height, riparian and ﬂoodplain land use, and ﬂoodplain slope were
obtained at three points along the reach (refer to methods in Johnson 1999a). Riparian
zone width, vegetation height, and slope were encoded separately for the left and right
banks, and the six values were averaged to derive a mean value for each site. Riparian
zone slope was measured directly at six points along the reach using a clinometer.
Riparian vegetation height values (0 = paved, 1 = mowed lawn, 2 = herbaceous, 3 =
shrubs, 4 = trees) reﬂected an increasing potential to serve as a source of CWD

(Appendix 2-2).

97

shrubs, 4 = trees) reﬂected an increasing potential to serve as a source of CWD

(Appendix 2-2).

Landscape Structure

Land use, hydrography, Quaternary geology, and elevation databases were used to
quantify several aspects of landscape structure (methods in Johnson 1999a; Johnson, et
al. 1997, Richards, et al. 1997). In addition, catchment boundaries above the sample
points were delineated manually and digitized from USGS 1:24,000 topographic maps.
Digital elevation data were used to verify watershed boundaries. Topography in the
region is relatively ﬂat, except in morainal regions; therefore the standard deviation in
elevation was used to represent topographic heterogeneity. Stream density was calculated
as the total length of all streams divided by catchment area (km/km’). All spatial
databases were transformed into a common digital format, projected onto a common
coordinate system (Albers) and analyzed in ARC/INF O as vectors unless otherwise
speciﬁed. Based on the areal extent of certain land use classes and previous work
(Johnson, et al. 1997, Richards, et al. 1997), land use categories were aggregated into ﬁve
classes: urban, row crop agriculture, forest, range, and wetlands in most of the analyses.
Open water was not included as a land cover type in these analyses. Land use and
Quaternary geology values are reported and analyzed as proportion of total catchment

area (Appendix 2-3).

98

Data Analysis

Distributional properties of all variables were assessed on raw data and
appropriate transformations were applied to non-norrnal variables. Box-Cox plots were
examined to determine the best transformations to achieve normality. Square root
transformations were performed on debris accumulation abundances not passing the
Wilke-Shapiro test for normality. Pearson correlations were performed for landscape,
riparian, channel-habitat to assess the degree of intercorrelation between variables.

Highly correlated variables were not included together in the redundancy analyses.

The hypothesis that the number, type, and size of debris accumulations are a
primarily controlled by local factors such as channel dimensions was tested. To test this
hypotheses analyses were performed to: 1) describe patterns in the distribution of debris
accumulation types across the study area, 2) quantify the effects of land use and
Quaternary geology on debris accumulation number, size, and type, and the debris
attachment point, 3) to identify factors inﬂuencing the number, size, and type of debris
accumulations, and 4) assess the potential role of CWD debris accumulations as an agent
inﬂuencing channel features. Analyses were performed using SAS (SAS Institute 1989)

and SigmaStat v. 2.03 for windows unless otherwise speciﬁed.

To examine the distribution of debris dam types across the study region, a Chi
Square analysis was conducted to test the presence / absence across all catchments of
individual debris accumulation types, and debris accumulation size x type classes. A

99

two-way AN OVA was conducted to determine the effects of land use and Quaternary
geology on the number of debris accumulation types / 100m and the number of

accumulation type x size / 100m.

Redundancy analysis (RDA) was used to quantify relationships between the
number of accumulation / 100m and environmental variables characterizing local,
riparian and landscape conditions (ter Braak 1995), using Canoco v. 4 software (ter Braak
and Smilauer 1998). Redundancy analysis is a direct gradient analysis technique used to
detect patterns of variation in the response data set as a function of a predictor or
independent data set. RDA selects linear combinations of environmental data that
maximize the dispersion of the response data. The pattern of variation in response
composition and the relations between independent variables and response variables can
be derived from this analysis. Square root transformations were performed on the
accumulation type data A Monte Carlo permutation test was used to determine the
statistical validity of the RDA. Tests were conducted by randomly pennuting the site
numbers in the landscape (or predictor) variables (Johnson, et al. 1997, Richards, et al.
1996). The predictor data were randomly assigned to the response data (debris
accumulation type/size) and a new ordination was calculated. This procedure was
repeated 200 times to develop a population of eigenvalues. If the debris accumulation
type variables respond to the environmental variables, then the test statistic calculated

from the observed data will be larger than the data derived from most of the random

100

simulations. An association was considered signiﬁcant if the observed eigenvalue was

within the ﬁve largest simulated values (p< 0.05).

The interactions between CWD and channel structure are complex. On one hand,
in-stream structures trap woody debris and create woody debris accumulations. These
accumulations alter ﬂow in the channel, which in turn inﬂuences channel width and
depth. Many studies have identiﬁed strong relationships between channel width and
wood abundance and number of debris accumulations (e.g., Zimmerman, et al. 1967,
Keller and Swanson 1979, Beschta 1983, Gregory, et al. 1985, MacDonald and Keller
1987, Beechie and Sibley 1997). Regression analyses predicting CWD abundance from
channel width in these streams resulted in models with low R2 values (Johnson 1999a).
An alternative approach was used here testing the possible effect of channel width on
debris darn abundance using a one-way analysis of variance based on channel width,
where streams were classiﬁed as small (3-6.9 m), medium (7 - 9.9 m) and large (2 10 m).
In addition, Spearrnan Rank correlation coefﬁcients between debris accumulation types
and bank-full width, bank-full depth, percent slow units, maximum depth of pools, and
substrate composition were examined. The association between the maximum depth in
pools and the presence / absence of medium- and large-size debris accumulations of all
types was examined using a Kruskal-Wallis test. These analyses were conducted on a

reduced data set including only sites having debris accumulations (n = 38).

101

Results
Distribution of Debris Accumulation Types

Three hundred and eighteen debris accumulations were counted and characterized
across the 49 stream reaches in the study area. There was an average of 6.7 i 0.8 debris
accumulations / 100 m across all sites (Figure 3-2). The median accumulation size was
5.6, exceeding the width of the channel in one dimension (either lateral or perpendicular
to ﬂow; see Table 3-2). Sum accumulation size reﬂects the extent to which the channel is
covered with debris accumulations. The range was 0 to 93, and the mean was 33,
mirroring the low abundance and numbers of accumulations / 100m. ‘Logs/snag’
accumulations were the most abundant type encountered (Table 34). ‘Root wad’,
‘vegetation’, and ‘1oose log’ accumulations were about equally abundant. Medium and
large ‘log / snag’ were the most abundant of the type x size classes of accumulations
(Table 3-5). Most ‘log / snag’ accumulations were associated with the bank (65%),
followed by snags (16%; Table 3-6). ‘Loose log accumulations generally were not
associated with any speciﬁc attachment obstacle (61%). ‘Vegetation’ and ‘root wad’

accumulations occurred in association with stream banks.

The null hypothesis was that each accumulation type had an equal probability to
occur at a site. No signiﬁcant difference in the presence / absence of debris accumulation
types at the sites was found (p = 0.165; Table 3-4). The occurrence of the three size
classes of accumulations across the reaches also was not signiﬁcantly different than
predicted. However, when the distribution of the four accumulation types classiﬁed by

102

three size classes was examined, the resulting type x size classes were not randomly
distributed across sites (p = 0.001). There were fewer than expected reaches with large
‘vegetation’, large ‘loose logs’, large ‘root wads’ and small ‘loose log’ accumulation
types, and a greater than expected number of reaches with medium ‘log / snag’

accumulations.

Although they do not constitute debris accumulations, the most prevalent
structural element in the stream channel across the 49 sites was overhanging vegetation
and root wads without trapped debris. Overhanging vegetation was present at 45% of the
sites, and root wads were present at 59% of the sites. At four sites, root wads and
overhanging vegetation were the only structural habitat elements in the channel, aside
from the sediments. At an additional ﬁve sites, the only debris accumulation types
present were overhanging vegetation or root wads with trapped debris. Therefore, at 18%
of the sites the only apparent habitat structure was in the form of overhanging vegetation

or root wads, with or without associated organic debris.

Eﬂects of Land Use and Quaternary Geology

A two-way analysis of variance was performed testing the hypothesis that the
distribution of coarse woody debris did not differ across sites stratiﬁed by dominant land
use and Quaternary geology. Signiﬁcant differences due to interactions between land use
and surﬁcial geology were found in the number of accumulations / 100m (p < 0.001;
Figure 3-2). There also were signiﬁcant interaction effects (LU * GEOL) on the

103

abundance of ‘loose logs’ and ‘log / snags’ accumulations (p = 0.01, 0.002; Figure 3-3).
The number of accumulations / 100m as well as ‘loose log’ and ‘log / snag’ accumulation
types were present in greater abundance in the Lac/Ag and Mor/Mix catchments than in
the other two land use/geology treatment categories. Signiﬁcant main effects were
observed on the abundance of the ‘vegetation’ accumulations types due to land use (p =
0.006). The ‘vegetation’ accumulation type occurred in greater abundance in mixed land

use catchments than in agricultural catchments.

Attachment points of the debris accumulation types were also examined with
respect to landscape characteristics. A signiﬁcant interaction effect was seen on ‘no
attachment point’ (p = 0.04; Figure 3-4), and a land use effect on ‘bank’ attachment
points (p = 0.02). Bank attachment points were more prevalent in mixed than agricultural
land use catchments. Debris accumulations with no apparent attachment point were more

abundant in Moer and Lac/Ag than in the Mor/Ag and Lac/Mix catchments

Landscape and Local Predictors of Debris Accumulation Types

A redundancy analysis (RDA) was conducted to identify the local, riparian, and
landscape factors that could account for the distribution of debris accumulation types. All
environmental variables combined accounted for 56% of the total variance in the
accumulation type data (p = 0.005). ‘Log / snag’ accumulation types and the total
number of accumulations / 100 m were best explained by the ﬁrst RDA axis (Table 3-7),
which was positively correlated with stream density, and ﬂood height, and negatively

104

correlated with SD. elevation, coarse till + outwash sand and gravel, bank full width,
catchment size and ﬂood height (Figure 3-5). ‘Loose log’ accumulations were negatively
correlated with % lacustrine sand and positively correlated with bank-full width and
catchment area. ‘Vegetation’ accumulations types were relatively well explained by the
inverse of the same variables (Table 3-7). ‘Root wad’ accumulations were best accounted
for by the third axis representing urban land use and a negative association with the

percent of slow units in the channel, and bank-full width.

Interaction Between Debris Accumulations and Channel Characteristics

There was no signiﬁcant difference between density of accumulations in three
channel width classes (Figure 3-6a). At bank-full widths between 7 - 9.9 m there were
fewer accumulations / 100 than in smaller and larger sized channels. The same results
were observed when all debris accumulations were grouped into small, medium, and
large size classes (Figure 3-6b). However, a moderate correlation was observed between
channel width and the number of medium-sized ‘vegetation’ accumulations. When only
‘log / snag’ accumulations were considered (consistent with debris dam types discussed
in other studies), medium-sized accumulations increased in number with increasing
channel size, and small accumulations decreased with increasing channel size (Figure 3-

6c); however, these differences were not statistically signiﬁcant.

No signiﬁcant differences in maximum pool depth between sites with and without
medium- and large-size debris accumulations of all types were found. Spearman rank

105

correlations between channel and substrate characteristics showed a negative correlation
between bank-full width and ‘vegetation’ accumulation types (r = -0.44). In addition,
‘root wad’ accumulations were positively correlated with the percent of slow units in the
reach (r = -0.46). When similar analyses were performed for streams < 10m wide (n =
28), ‘vegetation’ accumulations were negatively correlated with bank-full depth (r = -
0.56). The strongest correlation between channel characteristics and debris
accumulations was for medium ‘vegetation’ types (r = -0.62). Streams greater than 7m
wide (11 = 21) exhibited positive correlations between ‘vegetation’ debris types and the

maximum depth of slow units (r = 0.56).

Discussion
Regional patterns in debris accumulation type and size

The most common debris accumulation type in the stream channel across the 49
sites in the study region was the ‘log / snag’, occurring with a mean abundance of 3.1
accumulations / 100m. Although these accumulation types were abundant across the
study region and were found at 25 of the 49 sites, overhanging vegetation and root wads
without associated debris were encountered at all but 5 of the 49 sites in the study area.
In the absence of other structural elements in the channel, overhanging vegetation and the
cavities formed by root wads may be important ﬂow refugia and habitat for the ﬁsh and
non-burrowing benthic invertebrates in the stream, particularly during high-ﬂow events.
The role of bankside vegetation on channel morphogenesis was explored by McKenney
and colleagues (1995) in Ozark streams. Depending on the size of the channel and the

106

characteristics of the valley, the function of bankside vegetation varies widely. To my
knowledge, no one has quantiﬁed the role of this habitat with respect to its potential
contribution to secondary production and functional diversity in stream channels with few

stable substrates or structural elements.

While the number of accumulations / 100 m may be comparable to those reported
in other studies, CWD standing stocks are extremely low and low sizes are small in these
streams (Johnson 1999a) compared to those reported in the literature. This suggests that
the size of debris accumulations in the Saginaw basin are probably smaller than those
found in other regions. The density of debris accumulations observed in these Michigan
streams is comparable to that observed in an agricultural catchment in Tennessee (Shields
and Smith 1992), and streams in Washington State (Sedell, et al. 1988), but are well
below those observed for small streams in New Hampshire (Bilby 1979) and Virginia
(Smock, et al. 1989; Table 3-8), particularly when only the ‘log / snag’ accumulation
types typically studied are considered. Unfortunately, the size and abundance of debris
accumulation are difficult to compare across studies, since standard methods for

quantifying debris accumulation size do not exist.

Landscape, riparian, and local influences on debris accumulations

Overall, the low abundance and size of CWD in the Michigan streams is
consistent with the fact that the region was completely logged of its native white pine and
hemlock from 1840 - 1900, and what little second growth forest persists in this landscape

107

is restricted to fairly small patches on the landscape. Standing stocks of CWD are most
greatly inﬂuenced by land use and landscape-scale features that regulate channel size and
morphology (Johnson 1999a). The density and distribution of debris accumulations
responds more directly to local-scale features such as bank-full width and percent of open
canopy; however, other factors including riparian vegetation type, surﬁcial geology, and

land use also appear to inﬂuence the density and distribution of debris accumulations.

The interaction between land use and land form had a signiﬁcant effect on the
type of debris accumulations at a site. This interaction was evident from the results of
both the analysis of variance and the redundancy analysis of accumulation types (Figures
3.2, 3.5). The abundance of all debris accumulations, especially the ‘log / snag’ types,
were well explained by the entire set of environmental variables, but in particular, by
those associated strongly with landforms and topography, (e.g., high S.D. elevation and
percentage of coarse till plus outwash sand and gravel, and low stream density and ﬂood
heights). This combination of characters is associated with morainal landforms, where
the largest number of accumulations / 100 m and the largest standing stocks of CWD
were observed, or in the ﬂoodplains of the larger rivers on lacustrine landforms (Johnson
1999a). Coarse till plus outwash sand and gravel is highly correlated with range land,
which is primarily unproductive land, dominated by abandoned farms and old ﬁelds
(Table 2.15). Riparian zones in the catchments with a large percentage of coarse till or
outwash sand and gravel are among the widest in the study area, however, the vegetative
composition of those zones is highly variable. With large amounts of CWD and large

108

numbers of accumulations, these results suggest that there is ample source material in the
river upstream of the sample sites with non-forested riparian zones, or that the stream is
highly retentive of existing CWD. In the South Branch of the Flint River, there is
evidence of a great deal of past beaver activity. These sites described above currently
have few trees in the riparian zone and many logs appear to be quite old, therefore stable

ﬂow regimes associated with morainal landforms appear to be highly retentive of CWD.

In contrast,‘loose log’ accumulations are more closely associated with catchment
area and bank-full width. Catchment area exerts strong control over many aspects of
channel morphology, including bank full width and bank full depth (Richards, et al.
1996), suggesting that landscape-scale features have the greatest inﬂuence on the
abundance of ‘loose log’ accumulations / 100m. The abundance of ‘loose log’
accumulations is negatively correlated with the percentage of lacustrine sand, and
positively correlated with lacustrine clays, suggesting a strong association with
agricultural land uses. Streams in agricultural regions of the study area are the most
likely to be channelized. Therefore, loose assemblages of logs in the channel are a
natural outcome of the channelization process, since obstacles that would normally trap
woody debris have been removed or reduced in number in those streams. Furthermore,
ﬂashy ﬂows associated with surface-water dominated ﬂow regimes on lacustrine
landforms are likely to mobilize CWD to a greater extent than the ﬂows associated with

ground-water dominated systems, such as those in the morainal systems.

109

‘Vegetation’ accumulations are associated with contrasting environments
compared with ‘loose log’ accumulations. In the case of ‘vegetation’ accumulation types,
there is a strong positive association with lacustrine sand, which is associated with
forested land covers. In addition, there is a weaker association with wetlands. The
relationship between ‘log / snag’ and ‘loose log’ accumulation types and environmental
variables may be stronger than the vegetation-based types because logs and snags are
frequently removed by farmers and the Drain Commission of Michigan, whereas
overhanging vegetation is ubiquitously distributed across the study region. Although
removal of overhanging willow and alder from the stream banks has been observed, this

practice appeared to be less pervasive than debris removal.

Woody debris abundance and size relative to channel dimensions and structure

The pattern of decreasing number and increasing size of accumulations with
increasing stream size reported in the literature are contradicted by the results of this
study (Figure 3-6). Although a positive relationship between bank-full width and the
total abundance of CWD in the channel was observed (Johnson 1999b), no signiﬁcant
difference between the number of accumulations / 100 m and channel width was seen
when all accumulations were lumped by both size and type (Figure 3-6a), or when
accumulations were grouped into small, medium, and large size classes (Figure 3-6b). A
trend suggesting an increase in the number of debris accumulations with stream order
(Table 3-9) was also observed; however, this trend is opposite that found by Bilby (1979)
and Smock and colleagues (1989). Bilby and Ward (1989) and Gregory, et al. (1993)

110

reported larger numbers of small aggregations in small streams, and fewer large
aggregations in larger streams. Bilby (1979) reported 20-40 debris accumulations / 100
m in ﬁrst order streams, 10-15 in second order streams and 1-6 in third order streams in
New Hampshire, and Smock, et al. (1989) reported between 8 and 13 accumulations / 100
m in two ﬁrst order, low-gradient coastal streams in Virginia (Table 3-8). First order
streams in the current study averaged 6.5 accumulations / 100 m (median size = 6.2),
second order streams averaged 6.6 (median size = 6.3), and third order streams averaged
9.5 accumulations / 100 m (median size = 5.2; Table 3-9). Since accumulation sizes are
based on a proportion of channel width in the current study (Table 3-2), these results may
underestimate the size of large accumulations in large streams and overestimate the size
of accumulations in smaller streams. (A large accumulation in a small stream may be
classiﬁed as a small accumulation in a larger stream in the classiﬁcation system used in
the current study.) In a subset of sites in Mor/Mix land use catchments (most resembling
forested streams from other studies), a moderate correlation between link number (r=
0.46), mean bank-ﬁll] width (r = 0.58) and accumulation abundance was observed. This
correlation did not hold for Lac/Ag catchments, which contained the second largest CWD
standing stocks and numbers of accumulations. These results suggest that underlying
factors not associated with wood supply are effecting some control over the abundance of
CWD and accumulations in this region. The major difference between the two catchment

types lies in the degree to which groundwater dominates ﬂow.

111

There are at least two possible explanations for the reversed trend in the number
of accumulations relative to stream size observed in the Michigan streams. First, many
ﬁrst and second order streams have been channelized and cleared by the County Drain
Commission Ofﬁce. Historic data to determine whether and when the streams had been
channelized are lacking. To address this issue 49 stream reaches were examined with the
assumption that channelized streams would have smaller charmel widthzdepth ratios than
unchannelized streams. The widthzdepth ratio was signiﬁcantly larger in Mor/Mix
catchments than the other land use / geology types in the study area. The streams that did
not appear to have undergone extensive channelization were also more likely to resemble
forested streams examined in other studies. In the Mor/Mix streams, no signiﬁcant
correlation was seen between the number of accumulations and stream order (r = 0.21)
and bank-full depth (r = - 0.11), and only moderate correlations with link number (r=

0.46) and mean bank-full width (r = 0.58).

A second explanation for the pattern in accumulation number with respect to
channel size is that headwater streams have narrower ﬂoodplains, causing them to be
more vulnerable to development than the downstream reaches with extensive ﬂoodplains.
This would alter the availability of source material in the headwater relative to the larger
reaches. In all probability, however, the real explanation for the reversed trend in the
number of accumulations with channel size is probably a combination of the two factors:

land use patterns in the headwaters, along with active debris clearing.

112

CWD accumulations in these Michigan streams do not appear to inﬂuence
channel morphology to the extent seen in forested streams. The lack of strong
associations between CWD accumulations and channel features is probably due to three
interacting factors: 1) logs in these systems are smaller than those encountered in most
previous studies of CWD-channel interactions, 2) these streams have been subjected to a
range of management practices that includes debris removal and channelization, and 3)
ﬂashy ﬂow regimes on lacustrine land forms rapidly transport wood out of the reach.
Many researchers have reported strong associations between coarse woody debris and
plunge pool formation (e.g., Andrus, et al. 1988, Bilby and Ward 1991 , Hilderbrand, et al.
1997), lateral adjustment in the channel (e.g., Nakamura and Swanson 1993, Richmond
and Fausch 1995), and changes in the longitudinal proﬁle of a river (e.g., Beechie and
Sibley 1997, Smith, et a1. 1993). In low gradient streams, however, the mechanisms
related to pool formation are thought to be related more to valley geomorphology than to
the presence of CWD (Beechie and Sibley 1997). Aside from potential geomorphic
factors, smaller logs are more mobile, therefore, they are less likely than larger logs to
exert control over channel morphology by forming plunge pools and altering channel

widths in this study area.

The spatial distribution and structure of debris accumulations in small streams
reﬂects the input mechanisms for that region, as well as channel shape and dimensions.
Debris in small streams rarely moves except perhaps under large ﬂow events. Wood
mobility is constrained by the size of the log relative to the channel dimensions (Bilby

113

1984, Bryant 1983), log orientation and geometry (e.g., presence of branches; Bilby
1984), extent of burial (Bilby 1984), presence of obstructions in the channel, and ﬂow
conditions. In intermediate streams debris accumulations tend to form behind boulders,
large pieces of CWD, and other stable structures in the channel (Keller and Swanson
1979; Keller and Tally 1979). In larger streams, debris accumulations form at stream
conﬂuences, at upstream point of islands and point bars, and in depositional zones of the
channel (Harmon, et al. 1986, Abbe and Montgomery 1997). Since logs in the Michigan
streams are small relative to the size of the stream channels, they are readily mobilized,
resulting in a distribution pattern similar to larger streams (e.g., Keller and Swanson

1979)

Retention structure and dynamics vary with channel dimensions. In these
Michigan streams, the most retentive structure in the channel was associated with the
bank or stream margin. This is not surprising, given the types of the debris
accumulations studied, (e.g., root wads and overhanging vegetation with trapped debris,
in addition to log dams). However, woody debris in ‘log / snag’ accumulations in the
current study was most frequently associated with the bank, followed by vegetation (root
wads, snags, and overhanging vegetation). ‘Loose log’ accumulations were seldom
associated with a particular obstruction. Debris dams were the largest retention devices
for dowel rods in a 3rd order, low gradient woodland stream, followed by single CWD
pieces and roots (Ehrrnan and Lamberti 1992). However, dowel rods are much smaller
than the CWD considered in most woody debris studies, and time frames for the

114

experimental releases were in the order of minutes to hours. In a ﬁrst order low gradient
stream, Smock and colleagues (1989) reported that 52% of the debris dams were
associated with root masses. In higher-gradient streams in the Rocky Mountains
(Richmond and Fausch 199.5), and larger low-gradient streams in Alaska and
Washington state (e.g., Bilby and Ward 1989; Robison and Beschta, 1989) woody debris
was frequently associated with the stream margin. Wood is more readily mobilized and
redistributed by ﬂuvial processes in larger streams. These results highlight the effect of
the simpliﬁed channel morphology in these streams with respect to the process of debris
dam formation. Stream channelization removes structures such as point bars and islands,

resulting in a pool of highly mobile CWD.

Many debris accumulations in these Michigan streams were not associated with a
speciﬁc retention structure (‘no apparent’ obstacle). The lack of structural heterogeneity
in many of these streams and the prevalence of debris accumulations not associated with
a speciﬁc retention structure probably account for the large pool of CWD that remains
mobile (Johnson 1999c). Therefore, the function of CWD as ﬂow refugia, retention
structures for small particulate organic matter, and a habitat component of the stream
ecosystem may be limited in these Michigan streams. The prevalence of loose logs and
accumulations that were not associated with a visible retention structure not unexpected.
Debris accumulation surveys were conducted during base ﬂow conditions. When
discharge decreases, particulate organic matter drops out of the water column and larger
pieces in the channel form obstacles for other material moving downstream. These

115

particles are among the ﬁrst to be mobilized when discharge increases (Braudrick, et al
1997). Abbe and Montgomery (1996) classiﬁed debris dam types based on their relative
location in the channel. In their system, bar top jams were the most similar to the
deﬁnition of ‘loose log’ accumulations used in the current study. Bar apex jams formed
barriers to ﬂow and thus potentially had a large inﬂuence on channel morphology.
Meander jams were deposited at the upstream head of a point bar and functioned to armor
the banks. These types were rare in the Michigan streams for reasons that relate both to

land form and land use patterns in the basin.

Conclusions

The Saginaw Basin streams in this study contained very low standing stocks of
CWD compared to those reported from less disturbed streams (Johnson 1999a).
Interestingly, the number of debris accumulations / 100m was comparable to values
reported for other regions with much higher standing stocks of CWD. This suggests that
the debris accumulations in the Michigan streams are smaller, and entrap fewer and
smaller logs than in other streams. A safe, standardized method for measuring the size of

debris accumulations is needed to verify this assertion.

The size and abundance of debris accumulations in these highly disturbed streams
do not follow the same pattern of decreased abundance and increasing size with
increasing channel size, nor do they play a physical role in modifying channel
morphology, as do debris accumulations in forested landscapes (e.g., Andrus, et al. 1988,

116

Bilby and Ward 1989, Bilby and Ward 1991, Nakamura and Swanson 1993, Richmond
and Fausch 1995, Gregory, et al. 1993, Smith, et al. 1993, Beechie and Sibley 1997,
Hilderbrand, et al. 1997). Streams in Mor/Mix land forms, however, display trends that
are more similar to those reported in the literature, with moderate correlations between

debris accumulation abundance and channel width.

Active channel management from dredging and removal of riparian vegetation in
the agricultural catchments contributes to a poor supply of CWD, in conjunction with few
in-stream obstacles to retain logs in the channel. Stable ﬂows and less channelization
positively inﬂuence the standing stocks and number of accumulations in the morianal
catchments with mixed land use. Thus, both land form and the association of land use
practices with those land forms, inﬂuence the abundance and distribution of CWD in
these highly impacted streams. Land form is frequently ignored as a controlling factor in
studies examining interactions between landscape-scale features and stream ecosystems.
The current study illustrates the importance of land form and ﬂow regime in the hierarchy
of factors controlling stream ecosystems, especially in catchments impacted by chronic
disturbances such as agricultural management practices and channelization. While bank-
full width was moderately associated with ‘loose log’ and ‘log / snag’ accumulations,
catchment area plays a strong role in controlling channel dimensions (Richards 1982).
The combined presence of landscape and local variable in the statistical analyses
probably overshadows the effects of channel-scale features reported in the literature.
Clearly management activities such as channelization and agricultural production,

117

combined with the historic harvest have had a long-lasting effect on the input and
retention of C WD in these streams. Factors that appear to inﬂuence the abundance and
distribution of four different debris accumulation types have been identiﬁed, however,
further studies to examine the role of these accumulations in low-gradient, highly
developed catchments are warranted, since little is known about their relative importance

with respect to ﬁsh and macroinvertebrate community structure and productivity.

118

Table 3-1. Description of debris accumulation types, aggregated classes, and attachment
points of debris accumulations in the channel. All accumulation are greater than 1 rn’z in
area. Aggregated types were used to assess the combination of size and type of debris
accumulations. (Variable name in parenthesis.)

 

Debris Description Aggregated types Attachment Points
Accumulatio

11 Type

Vegetation Overhanging Vegetation w/ Bank

w/ trapped vegetation with trapped trapped debris

debris organic matter (size of = ‘Vegetation’

(vegetation) twigs or larger)

Root wad w/ Root wads on bank Root wad w/ ‘ Point Bar or Island

trapped with trapped organic trapped debris

debris matter (size of twigs or = ‘Root Wad’

(root wad) larger)

Logs on Logs greater than 5 cm Logs and snags . Downed tree/Snag
point bar, diameter in an = ‘Log / snag’

island, bank, aggregation. This

channel category is aggregated

(log dam) from 3 separate

categories based on
location (e.g., bank,
channel, or point bar).

3 Overhanging

Snag Downed tree, with or Logs and snags
(snag) without an associated ‘ vegetation, including
debris dam 1 root wads

- No apparent
' attachment point

Logs loosely Logs greater than 5 cm Loose Logs
aggregated diameter in a loose = ‘Loose log’
(loose logs) aggregation

 

 

119

Table 3-2. Size classes of debris accumulations based on methods of Shields and Smith
(1992). X = channel width at the upstream point of the debris accumulation. Classes 1-3
were aggregated to form the size category = small; classes 4-7 = medium; classes 8-10 =
large. The sum of all debris accumulation sizes in each reach represents a measure of the
amount of channel covered by debris accumulations.

 

 

 

 

 

 

 

Size Size of Accumulation in Direction Parallel to Flow

Perpendicular *

to Flow <.25 X .25-.5X .S-X >X

<.25x 1 2 4 ' 5
SMALL Mil/I’m

.25-.5X 2 3 6 7

.5-X 4 6 8 9
ME; 1U LARGE

>X 5 7 9 10

 

 

120

 

Table 3-3. Coarse woody debris accumulation abundance and size variables measured
during the study.

CWD Variable Name Description

# Accum Number of debris accumulations per 100 m
Median Accum Size Median value of debris accumulation size classes per reach
Sum Accmn Size The amount of stream bottom covered by debris

accumulations, derived from the sum of all accumulation
sizes in a reach.

121

 

23.5

to one 2 a e \3 5:98;
N; .2 S a 2 m3 82:
34.5

no «No S 8 2 E 23 zoom
m _.o _3 E 2: mm was \ won
m:o_§=E=oo< mezm 09C. :o_.a_=E=oo<

owned 2mm ozm con :82 me .3852 :38. Co 38:32 £500

3:32 835m av 388 8a.: cote—aﬁzuon $53 no @5885 41m 633.

122

 

2509
g one 2 cm 2 E 8:93.;
mi. 3.6 N; mm 2 woq 884
£500

wé «No m.. cm 2 \3 an? Bod
26 _md ﬁm :— mN wmcm \ wou—
mcozﬂsgouxx 33m 33. =o_§_:§oo<

owcmm 2.m.m ozm Ba :32 mo 83:52 _Soh mo 63:52 $500

$2.82 8.85 av $28 womb corn—2:33 mine.” mo bugsm 41m 2an

123

Table 3-6. Summary of debris accumulation attachment locations for each accumulation
type. Missing data indicate categories that were not likely to occur due to the character of
the debris accumulation type.

 

Debris Vegetation w/ Root Wad Loose Logs/ Log / snag/
Accumulation debris/ 100m w/ debris/ 100m 100m 100m
Attachment Point

Bank 52 (96%) 59 (98%) 8 (14%) 96 (65%)
Point Bar/ Island 2 (4%) 0 (0%) 4 (7%) 14 (9%)
Snag - l (2%) 1 (2%) 23 (16%)
Vegetation - - 7 (5%) 6 (4%)
Rifﬂe - - 2 (3%) O (0%)
No Apparent - - 35 (61%) 8 (5%)

124

Table 3-7. Results of redundancy analysis; ﬁt as a fraction of the variance of woody
debris variables.

 

Wood Variable RDA 1' RDAZ2 RDA33 Total Variance

Explained (%)
(all axes)

Log / snag 0.54 0.01 0.04 66.1

# Accum/ 0.60 0.03 0.01 63.8

100m

Loose Log 0.11 0.36 0.00 53.1

RootWad 0.50 0.07 0.38 51 .3

Vegetation 0.00 .24 0.12 44.2

' RDA Axis I is positively correlated with stream density and ﬂood height and negatively
correlated with SD. elevation, coarse till + outwash sand and gravel, mean bank-full
width, catchment area, and ﬂood height.

2 RDA Axis 2 is positively correlated with mean bank-full width and catchment area, and
negatively correlated with lacustrine sand.

3 RDA Axis 3 is positively correlated with urban land use and negatively correlated with
% slow units in the reach.

125

Table 3-8. Comparison of large woody debris densities for Saginaw Basin streams in
Michigan with published values for other low-gradient or disturbed streams.

 

Debris Location Reference

Accumulations /

100 m

6.7 :1; 0.8 central Michigan (49 stream this study

(0-13) reaches)

0.06 - 3.4 Iowa streams Zimmer and Bachman, 1976

(in Shields and Smith 1992)

2-8 Washington Sedell et al. 1988

3.5 - 5.8 uncleared reach, S. Fk. Albion Shields and Smith 1992
River, TN

0.6 - 5.8 cleared reach, S. Fk. Albion Shields and Smith 1992
River, TN

0 - 2.4 Lymington drainage basin, Gumell and Gregory 1995
United Kingdom

0-10 aspen forest, New Mexico Trotter 1990

8-13 15‘ order stream, Virginia Smock et al. 1989

20-40 1“ order stream, New Bilby 1979
Hampshire

10-15 2"" order stream Bilby 1979

1-6 3rd order stream Bilby 1979

126

Table 3-9. Summary of the mean number of debris accumulation and median size by
stream order and channel width. (Streams without debris dams are excluded from this

analysis; n = 39.)

 

Stream order (n) mean # debris median size size range
accum/ 100 m
1 (2) 6.5 i 2.5 6.2 5.5 - 7
2(14) 6.6: 1.4 6.3 4-9
3 (21) 9.5 i 0.92 5.2 2.5 - 8
4 (1) 13 5.0 5

127

.mtnov woman: 51$ 33 “com Am

a

 

.mm2 8004 a $9330; AU

mtnoc woman: 33 =o_§owo> 2 £25 coca—383% £50m. So.“ we moi—Seam .7

m unswE

128

 

‘
‘

 

LU ‘ Gaol effect -o— Lacusm

p < 0.001 X + Mar-mot

a
N
A

 

 

 

.5
O
J

Accumulations I 100 m
‘ O

 

 

 

 

 

 

 

 

 

 

 

Am “1300
so
+mm
40 +uaw
g
‘0
§=°‘
g
5 /
< 2°‘
E
3
(0
1o«
0 ,
W m

 

 

 

 

 

 

Median Accumulation Size
0

 

 

 

Agnew Mind

Land Use

Figure 3-2. Mean and SE. of # debris accumulation/ 100m, debris accumulation size, and
the sum of the accumulation sizes (see text) across land use and geology treatments.
Results of two-way AN OVA are indicated.

129

on: 9:...

00x. 2
P

32.3

P

 

 

 

 

 

 

 

 

 

 

 

 

. o
. .
v N
v n
v C
= «8... u a . a
35802 [1| ﬁ «00:0 .000 c 3..
05:25.; + OQCW\OOJ . o
h
0»: new;
3...: 33.3
v o
\ v w
\

 

.oseoa nil
2.3.3... IOI

 

 

 

258 5.; 2.2, .8:

L.

 

 

“1001300“me p 8

.3306”: 2a <>OZ< .3362; he 3.3.3— £6253:
awe—cow 28 8: was $83 893 noun—=838a So.“ .3 88: macaw—seas“ mine—o n .«o .m.m Ba :82 .m-m oSwE

 

 

 

 

 

 

 

 

 

 

 

on: we...
con: .33.:
. o
. —
\
m \
. u
. n
v 1
Ed n a a
hug”... H 80.3 .30 . 3..
30.. 030.. o
s
on: we...
3a.: 382
. o
\ I ¢
. a
. a
1 v
°°°.° u a 1 m

.0593! +
0:523... 1.1

 

 

 

60:0 0...: 2:3

258 5.3 5:23; .

 

 

n300 mmmmw )0 O

mootnuomnwmwn I

130

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

7 Bank T
‘ + 7 PointBar/lsland
J Land use effect + m. 1 : mm
6 p<0.008 / a]
8 5. 8 s
.1: r / a l
2 “ s '
g 3 g 34
t:
: 24 : 2.
’+ 1‘ \
0< o< o——"""—
LandUee LandUee
I I
71
Snag + 7‘ VW +M
,. +Wu a +W

 

 

 

 

 

 

‘AW/‘lm
u D
lAtbchmentltOO
u .

 

 

 

 

 

 

 

 

 

 

 

 

 

 

LmdUee WU“
e
6‘ Lu'Geol +“""'
<0.04
2.3 .. "
'57 4‘
5.
g
e: 3‘
14 ‘4
o e”
Avian Jon
LmdUee

Figure 3-4. Mean and SE. of # debris accumulation attachment locations /100m across
land use and geology treatments. Results of two-way AN OVA are indicated.

131

 

 

 

 

 

 

Loose Log .
o BFW
‘ Catch Are i
Angor + Rng ° ° Stn'n Dans
. Rip Wld j
0 CT+SG ;
1 SD Elevo 3 o FldHt
' Accum ‘— O 1
/
LogISnag
RW ; o Wetland
Urban 0 i
Veg ° Lac Sand

 

 

 

Figure 3-5. Species-environment biplot derived from species scores from a redundancy
analysis of debris accumulation types and landscape, riparian and local variables
including catchment area (catch area), SD elevation (SD elev), stream density (Strm
Dens), % urban (urban), % wetland (Wetland), Agriculture:Forest+Range (AngoH-Rng),
lacustrine sand (Lac Sand), lacustrine clay, coarse till + sand and gravel (CT+SG),
instream cover, ﬂood height (F 1d Ht), % slow units, maximum depth in slow units, mean
bank-full width (BF W), % open canopy, and riparian zone width (Rip Wid).

132

 

.a
N

 

# Accumulations I 100m
OdNUAmOVOOS:
J

 

 

3-6.9m 7-9.9m 10-12m
Channel Width

 

A L-' Medium

5 _ + Small
—l- Large '

 
 

# Accumulations/100m

 

’\,
\+
1-1

3-6.9m 7-9.9m 10-12m
Channel Width

 

 

 

 

+ Small Log/Shit!
5 4 A» Medium Lou/Snail
.. L.rg. Log/Snag

Y Data
0)

 

 

 

 

3-e.9n 7-9.0rn 10-12m
Channel Width

Figure 3-6. A. Mean and SEM of the number of debris accumulations across three
channel width classes. A. All accumulation types and sizes combined. B. Accumulations
grouped by size. C. ‘Log / snag’ accumulations grouped by size. Note change in Y-axis
scale between A and B, C.

133

CHAPTER 4

COARSE WOODY DEBRIS RETENTION AND RECRUITMENT IN LOW

GRADIENT, MIDWESTERN WATERSHEDS

134

Abstract

Coarse woody debris (CWD) is an important component of many small to
medium-size temperate streams, directly inﬂuencing many ecosystem properties and
processes. Much is known about the importance of wood in mediating hydrological,
geomorphological, and ecological processes of forested streams, however, retention and
recruitment dynamics of coarse woody debris in streams are poorly understood. CWD
retention and (upstream) recruitment were examined in 10 low-gradient Midwestern
streams before and after a 5-year ﬂood event. Effects of channel morphology and log
size on retention and recruitment also were examined. Although there was a turnover of
approximately 50% of the logs at a site, the number of logs present before and after the
ﬂood remained approximately equal. However, log volume was greater before the ﬂood
than after. The ratio of retained to recruited logs was <1, indicating that the population of
logs after the ﬂood was dominated by recruited rather than retained logs. Log retention
and recruitment is in dynamic equilibrium, and the logs exhibiting the greatest movement
are the smaller logs. CWD retention and recruitment were successfully predicted from
logistic regression models from log dimensions with concordance values ranging from 62
- 68%. Flood height was positively correlated with recruitment and negatively correlated
with retention. Retention and recruitment, however, were not correlated with estimates of
stream power and shear stress. Bankfull width was negatively correlated with the
proportion of logs retained, and positively correlated with the proportion exported and
recruited. It appears that rough estimates of these processes can be derived from an

estimate of ﬂood height.

135

Introduction

Coarse woody debris (CWD) is an important component of many small to
medium-size temperate streams, directly inﬂuencing stream geomorphology as well as
many ecosystem properties and processes (see reviews by Harmon, et al. 1986, Gurnell,
et al. 1995). In non-forested catchments potential inputs of CWD are reduced by
disturbances to the riparian vegetation and stream clearing activities which eliminate
CWD from the channel. Land use conversions, installation of drain tiles in agricultural
ﬁelds, wetland drainage, and stream channelization alter the hydrologic regime, resulting
in higher peak ﬂows and ‘ﬂashy” hydroperiods. Interactions between management
practices and hydrologic factors contribute to low CWD standing stocks (Johnson 1999a)

and potentially to reduced retention in streams impacted by agricultural land use.

Retention and recruitment dynamics of coarse woody debris in streams have been
poorly studied, despite the fact that much is known about the importance of wood in
mediating hydrological, geomorphological, and ecological processes of forested streams.
The hydraulic signiﬁcance of large woody debris in stream channels has been well-
studied (see reviews by Harmon, et al. 1986, Gurnell, et al. 1995), but the dynamics of
wood movement have not. Retention and recruitment of woody debris are important
processes that must be characterized to fully understand the factors controlling habitat
and biotic community structure of these stream ecosystems. Recent studies by Braudrick,
et al. (1997) and Braudrick (1997) have characterized wood movement under different
ﬂow regimes, and have attempted to model wood transport as a function of discharge.

136

Bilby (1984), Lienkaemper and Swanson (1987), and Berg, et al. (1998) have reported
interactions between channel width or structure (e.g., channel roughness) and log
dimensions with respect to CWD movement. Others have simulated transport and
retention of wood using dowel rods or equivalents over short (Ehrrnan and Lamberti
1992, Hax and Golladay 1998), and longer time frames (Jones and Smock 1991, Webster,
et al. 1994). Retention is closely linked to the number and type of obstacles in the
channel, and to the size of the log relative to the size of the channel (Bilby 1984, Webster,
et al. 1994). In the dowel rod release experiments conducted at the Coweeta Hydrologic
laboratory, dowel rods remained generally stable after a period of initial movement.
Similarly, wood released in the channels of two low gradient streams in Virginia were
recruited into the ﬂoodplain the ﬁrst time there was bank overﬂow. Interestingly,
movement distances in the ﬂoodplain exceeded those in the channel in one of the two

streams because of retention in debris dams within the channel (Jones and Smock 1991).

Streams in the Saginaw Basin in Michigan are characterized by low standing
stocks of CWD, and moderate ntunbers of debris accumulations (Johnson 1999a).
Because of the generally homogeneous nature of the stream channels of many of these
streams, woody debris accumulations are associated with the banks, rather than structures
such as boulders and root wads (Johnson 1999b). As a result, debris accumulations
appear to be transient. During three years of ﬁeld work in this region, several large debris
accumulations present at the onset of this work were dismantled following a storm event
with a 5-year return interval (http: \waterdata. USGS. gov). Further, many large branches

137

were delivered to the stream channel following intense summer storms. These branches
remained loosely distributed throughout the stream reaches throughout the open water
season in 1996. These observations suggest that CWD is much more mobile in these
streams compared to those studied in higher gradient, forested systems (e.g., Bilby 1984,
Webster et al. 1994). This study examined CWD retention and recruitment in 10 stream
reaches in an agricultural landscape in the Midwestern USA and addressed two
hypotheses: 1) CWD movement in streams of non-forested landscapes is inﬂuenced by
log size, channel dimensions and ﬂow characteristics, and 2) the location, orientation, and

spatial relationship with other woody debris inﬂuences CWD movement.

Study Region

The Saginaw Bay catchment of Lake Huron encompasses a 16,317 km2 region,
characterized by sand and clay-dominated lowlands rimmed by coarse-textured glacial
features such as ground moraines and outwash plains. The study region is contained
within two major ecoregions as defrned by Omernik and Gallant (1986): the Southern
Michigan/Northern Indiana Till Plains and the Huron/Erie Lake Plain. Each is
subdivided into two sub-regions. Soils in the lake plain are dominated by medium and
ﬁne-textured loams ranging to clays, with sand in the outwash plains and channels.
These clay regions are extensively drained by artiﬁcial drainage and tile systems. The
periphery of the basin contains many coarse-textured glacial features such as ground
moraines and outwash plains. The till plain exhibits the greatest variation in basin
topography and contains a high percentage of forested land intermingled with agricultural

138

land and old ﬁelds; elevations average about 278 m. The entire drainage was logged for
white pine and hemlock between 1830 and 1900, and forests of the region now consist
primarily of second growth hardwood species. Current land use is dominated by
agriculture. Ten streams reaches in 8 rivers were selected to span a range of land use and
Quaternary geology typical of this region (Appendix 4-1, 4-2). Substrates are quite
homogeneous and are composed mainly of sands and silts. One site has large boulders at

one end of the reach; however, all are submerged and were never observed to trap CWD.

Methods

Woody debris retention and recruitment were measured in ten stream reaches.
Logs greater than 5 cm diameter and 1 m length were individually measured and marked
with pre-numbered tags during base ﬂow conditions in October 1995. Location of each
log was recorded with respect to a known location in the reach. Ancillary information
including log orientation with respect to ﬂow (parallel, perpendicular, diagonal), location
with respect to the banks (leﬁ, right, center, spanning channel), and whether the log was
included in a debris dam was also recorded. The goal was to tag at least 50 logs in each
stream reach. If less than 50 logs were present in the reach, all logs in the reach were
tagged. When a study reach contained more than 50 suitable logs, 5 m stream segments
were randomly sampled until at least 50 logs were tagged. A total of 394 logs were
tagged across 10 reaches. During the second visit following the spring ﬂood, the location
and orientation of each tagged log was noted, along with the location, size, and
orientation of newly recruited (untagged) logs. If a log was found within 5m of its

139

original location in the channel it was considered to be ‘retained’. Logs were considered
‘exported’ if they had moved more than 5m from their original location. New logs within
a reach, including logs exported from upstream, were considered to be ‘recruited’.
During the post-ﬂood period in June 1996, 376 logs were measured. Woody debris data
were summarized as 1) proportion of original number of logs and wood volume either
retained or exported, 2) proportion of newly recruited log number and volume, 3) the
ratio of log number and volume at '1‘1 and T2, and 4) the ratio of log number and volume
retained to recruited. Arcsin transformations were used for proportional data (sqrt (x /
100)) where necessary; log transformations were performed on other non-normal
variables by taking the natural log of the datum plus '/2 the lowest non-zero value.
Differences between means of log number and log volume retained versus exported were

tested using a paired T-test (Table 4-1).

In addition to woody debris, channel features including bank-full width, bank-full
depth, and percent of channel with fast and slow units were measured (Appendix 4-2).
Standing stocks of coarse woody debris in the reach were measured as length of CWD per
unit area (m/mz), volume (m3/m2), and number of debris accumulations / 100m (see
Johnson 19990 for details on measurement methods; Appendix 2-3). Mean log diameter
and length of all wood 2 5 cm diameter were recorded. Flow characteristics including an
estimate of stream power and shear stress at ﬂood stage (Gordon, et al. 1992), and an
estimate of ﬂood height were measured. A logistic regression was performed to predict
whether or not a log is retained or recruited based on log dimensions and volume. Linear

140

regression models were used to predict the actual pr0portion of retention or recruitment
using independent variables identiﬁed by the logistic regression (above). To identify
those features contributing to the dynamics of wood transport, a Pearson correlation was
performed between retention/recruitment variables and 1) ﬂow characteristics, 2) channel
morphology, and 3) standing stocks and dimensions of CWD throughout the reach (Table

42).

Results and Discussion
Retention and Recruitment

Approximately 60 % of tagged logs, representing about 50 % of the log volume,
moved more than 5 m between the two sampling periods (Table 4-1). Although the
proportion of the original logs retained was not signiﬁcantly different from the proportion
of logs exported or recruited, the ratio of retained to recruited logs was less than one,
indicating that the population of logs after the ﬂood (T2) was dominated by recruited
rather than retained logs. This suggests that CWD retention and recruitment is in a
dynamic equilibrium, characterized by a highly mobile pool of logs that are replaced with
wood from upstream sources. In contrast, the proportion of log volume per reach was
greater before the ﬂood (T l) than after (T 2), and the ratio of retained to recruited log
volumes following the ﬂood was high (Table 4.1), suggesting that larger logs were
retained, while smaller logs were recruited. Not surprisingly, the logs exhibiting the
greatest movement are the smaller logs. These results are consistent with ﬂume studies
of Braudrick, et al. (1997) and observations from ﬁeld studies (Bilby 1984, Berg, et al.

141

1998). Studies with small pieces of artiﬁcial woody debris (dowel rods), however,
showed that there is little movement of these pieces after the initial pulse (Webster, et al.
1994). Releases in that study took place in very small forested streams in the
Appalachian Mountains where, in contrast to the current study, streams contained large

standing stocks of CWD and CPOM.

Predicting Log Transport

The probability that a log was retained at its original location was successfully
predicted from log dimensions. (Table 4-2). Using a logistic regression log(diameter),
log(length), and log(volume) successfully predicted whether a log was retained, with
concordance values of 62-68%, based on 443 observations. These models are consistent
with those of Bilby (1984), who also found that the probability of movement was related
to the length and diameter of the log. The geometry of the log (e.g., branching patterns),
as well as the extent of burial were confounding factors inﬂuencing wood movement.

The distance traveled also was related to log length, but not to log diameter.

The current study was not designed to explicitly test the issue of transport
distance; however, a sufﬁcient number of logs were recovered to test for the eﬂ'ect of
channel features on transport distance. A signiﬁcant regression model was obtained from
measures of steam power and substrate characteristics; however, R2 values were very low
(2% of variance explained). Bilby (1984) suggested that log geometry plays a role in
transport distance. Log geometry is likely to play a larger role in the Saginaw streams

142

than elsewhere because many of the “logs” in the stream are actually large branches that
have recently fallen from overhanging trees during summer storms. These “logs” have
complex geometries, with many branches still intact. These data may therefore not
predict transport distance well because of the complex structure of the logs being
transported. Clearly, a more robust analysis of transport distance would require peak
ﬂow measures over the time period between the two sampling events, and more frequent
sampling to capture transport as a function of individual events. A lack of stream gages
on these rivers precludes obtaining accurate discharge measurements. Such information
would be invaluable for developing accurate models of transport and retention dynamics

for CWD.

Log dimensions interact with channel morphology and ﬂow regime in the
regulation of log transport/retention dynamics. To identify those features contributing to
the dynamics of wood transport, a Pearson correlation was performed between
retention/recruitment variables and 1) ﬂow characteristics, 2) channel morphology, and 3)
standing stocks and dimensions of CWD throughout the reach (Table 4-3). Flood height
was the only ﬂow variable that was signiﬁcantly correlated with retention/recruitment
data. A linear regression predicting the proportion of logs retained from log(ﬂood height)
explained approximately 52% of the variance (Figure 4-1a). A similar analysis predicted
65% of the variance in the proportion of logs recruited from log(ﬂood height) (Figure 4-
1b). Webster and colleagues (1994) found a signiﬁcant relationship between stream
depth and distance which artiﬁcial leaves traveled. The effect of discharge on retention

143

was believed to operate through differences in depth, rather than differences in stream
power. However, no signiﬁcant correlations were observed between retention or
recruitment and depth variables in the current study. Poor relationships between stream
power and CPOM retention have been observed by others (e.g., Naiman 1982, Minshall,

et al. 1992).

Bankfull width (BFW) was the best predictor of retention/ recruitment dynamics
(Figure 4-la,b); indeed, the proportion of number of logs retained, exported, and recruited
were all signiﬁcantly correlated with BFW (Table 4-3). The negative correlation between
retention and BFW is consistent with previous work (e.g., Bilby and Ward 1989, Gregory
et al. 1993) conﬁrming that smaller channels are more retentive of CWD than larger
channels. In contrast, the proportion of log volume retained and exported were poorly
predicted by bank-full width and ﬂood height, but were signiﬁcantly correlated with the
standing stocks of CWD and log dimensions. Log volume recruited, however, was the
only signiﬁcantly correlated with BFW. Similarly, the ratio of log number at TI to T2,
and the ratio of the number of logs retained to recruited was correlated with BFW,
whereas log volume ratios were primarily correlated with CWD standing stocks and
wood dimensions, implying that sites with larger standing stocks experience less CWD
turnover. Log volume measures all exhibited much larger coefficients of variation (c.v.)
than did measures based on log number. These data suggest that there are large variations
across sites with respect to CWD transport and retention dynamics that are not related to

channel width or ﬂow variables.

144

The location, orientation in the channel, and inclusion in a debris darn also did not
inﬂuence whether a log was retained in the reach. Ehrrnan and Lamberti (1992) found
that in stream reaches with little CWD, dowel rods were retained by root wads, stream
banks, and single pieces of woody debris in the channel. Three of the ten streams in the
current study had been channelized; as a result ﬂow patterns are homogeneous across the
reach and have relatively little resident CWD. This is in contrast to other sites with well-
developed pool rif’ﬂe sequences, and moderate accumulations of CWD that enhance

retention.

Conclusions

Retention and recruitment dynamics of coarse woody debris in low gradient,
agricultural streams were examined with respect to channel and ﬂow characteristics.
These sites have modest standing stocks of relatively small coarse woody debris and, in
general, exhibit stream proﬁles consistent with disturbed systems. CWD in these streams
appears to be highly mobile, with approximately 50% of the logs turning over between
the two sampling periods before and aﬂer modest spring ﬂoods. As anticipated, larger
volume logs are less mobile than smaller logs. Retention and recruitment, surprisingly,
was not signiﬁcantly correlated with estimates of stream power and shear stress during
ﬂood conditions. Rather, these processes were most highly correlated with ﬂood height
and bankfull width. Channel dimensions have long been known to inﬂuence the transport
dynamics of coarse woody debris. It now appears that rough estimates of these processes
can also be derived from a simple estimate of ﬂood height. Managers who are

145

considering restoration methods involving addition of CWD to the stream may be
interested in predicting the amount of wood that is expected to be retained or transported
under different ﬂow conditions. Additional work, including deriving better measures of

peak ﬂow on these ungauged streams should enhance our predictive powers.

146

Table 4—1. Descriptive statistics for the proportion of logs and log volumes ﬁom Tl
retained (retention) and the proportion of newly recruited logs and log volumes measured
at T2. Ratios of logs at each sample period, recruited to exported, and retained to
recruited also are shown.

 

Variable Mean i S.E.M. Range
Proportion Retained # Logs 0.42 i 0.06 0.15 - 0.73
Volume (m3/m2) 0.52 i 0.09 0.07 - 0.88
Proportion Recruited # Logs 0.58 i 0.05 0.37 - 0.75
Volume (m3/m2) 0.36 i 0.05 0.12 - 0.69
Proportion Exported # Logs 0.58 i 0.06 0.06 - 0.44
Volume (m3/m2) 0.48 i 0.09 0.12 - 0.92
T1 : T2 # Logs 1.03 i 0.11 0.63 - 1.63
Volume (m3/m2) 1.46 :_l-_ 0.33 0.69 - 4.23
Retain : Recruit # Logs 0.80 i 0.18 0.33 - 1.80
Volume (m3/m2) 2.12 i 0.66 0.45 - 7.62

Table 4-2. Logistic regressions predicting whether or not a log is retained from log
dimensions and volume. CCR % = percent correctly classiﬁed. (N = 443 logs).

 

Variable Parameter S.E.M. adj R2 CCR(%)
Estimate
log (length) 0.72 0.15 0.07 62.4
log(diameter) 1.12 0.22 0.09 63.0
log(volume) 0.42 0.08 0.10 66.6

147

 

148

2...- 2....- ..m...- 5....- NN...- 0.....- 2.5.3,
2....- .......- on..- .5.- m....- E...- 03 .. E53. “ :53.
0o... 2... 2... mm... 2... 3... 25.3,
0.....- 2... 9.... .6... 2... em... 03 a S H E
0...... 3... 8... 5.... 0m... 8... 2.5.3,
8... 2... 2... F... N... .0... as. .. 02.38.. 8.8%...
2...- 6...- S...- m.....- .. . ...- 3....- 2.5.;
S...- N.....- s...- 0.....- R...- 2...- n.a.. .. .0553. 8.082..
.523
59.0.. .2255 E so. .3359 2:5 93.0... 8 .u:.
we. :82 mo. 5...: \ 030 e 2. 32m .x. 08:. .6. 2.25., 083

3.0052: .coh0..:om mew... 00.02.00 0... 20.. a. 0.00.0508 2.3.-:35. .030. 0.... a. 00.0.. .2. 0... 0:...

0050.0. 22:02.0... 2... 0. 00.0.0. 50202.. 0.0 00.2.9.0 0523 we. .05. mm... .0 2.2.8002. 2C- ...000. 0... 5 $0. .0 2.0.0205...
:02: .05. .3552: .0 2.0.3.6000 0 8.. 030. =8:..0.. 00m. £000. 8 co. .2. 0.2.2. .8003 00.000 2.65% 25.20:.
.m0..m..0.00.0...0 .2820 .2... 30... 2.23 2.0.5.300. 0:0 :0..:0.0. .0003 2... 02.0.0503 02.0.0200 000.00.. .m-v 0.2.0...

 

m. Nwwd SEN v... .o a N 2.20 .N . m
mN moN. w.omN 30... N m w.moNN. m ..N
m.m MhN. 9mg ﬁnd 9. m o.oow.N ..0N
hm... NE... QEN .8... N N Wmmwm 3...
N.. mde as... 0mm. w . adv-No mmm
N. . .mod v.mNN 966 N N Em .h. mNm
N.. 3...... QEN 20... m m m.mmmm .Nm
m.m MN... ..mON 3.... m N w.mNmN mo
0.». 0.0.0 0N0. 3N... N. N 02-2. .0
mN won... wNNN .36 N. m 22m.- .m
05.00.... .8. 8.05... .2. 2.20
.25 3.2.3. 2.2.2.200 2.25.200 0... 02¢.
2.30.. .50.“. 2.02% 2.003. :02). .22.... Z 0...... .220 2.02% 2.022.200 2.m

22.0020 ....0 02.25.200 .2. 20......000D ..-v 5.2.2.3.

149

 

omd odd dmd 3+ ddd mud dd. _ odd _m _ m
mod n _ d. m m d 3d cod and dd. _ odd QVN
mud 2».— _ dmd m~d_ ddd 3d dd; odd _ :&
end dad odd dim Ed 3d 3d mwd mom
mod tum _vd Ed nmd wad de and mmm
mad mod cmd 2; dvd ddd cod 8.. mmm
mvd wmd mvd m _ .m and odd odd dd. _ _mm
wwd mod omd mm.m and 2: 2d 5d 3
mod dad and 2: and 3d Ed 3d 3
mwd odd 3d mod mmd 3d Rd mud _m
:5 d5 Ed .255 :5 3:55 Ed

530 __=,._ £23 :5 @283 3:03 AEV E5 “mam «ED 32m 3:5

-xcmm :32 -xcmm :32 539 w>< 52>» w>< 530 5:2 Suva ~82 32m °\o 3:5 EEK. ozm

.momtoqoa 038-3585 do Santana .Né 528%?

150

.moﬁmtg 82: mo cog—:28 2: do mzﬁou “8 93¢ C coma—*3 8m .._

 

w m; on; 88.0 2:3 as
v .2 ~86 $85 on; 2%
5 CE 085 356 $2 :ﬁ

0 on. $3. 886 8% gm

o 23 an; 286 63 mm

o 2; S 3 $85 9.8 mm

92 RM 83 $86 89o _Nm

5 a: 336. 886 $3 8

w ”3 BS «Bod 82. a

2 SN 33 :86 53 _m

.,,,58 E58< c5354 c5 maxed 2%:

£an a mod :32 38:85 woq :32 0829, 930 3.825% D30 26

.3582: oogcgnm mined woo? .mé deeming

151

€92

cooEmo. m5 .0 cozoca a ma 3:202 ucm BEEP. mmo. do 53.895 9: 9:2qu £2.98 8386mm. .5. 2:9".

2%: 80m?

 

 

0.0

2%: 8°98.

 

 

v8. on a S on N:
2292 uSEuoa. m «a + m 9." 8:28.". 3.:

 

 

 

mod n a “no u E
8:92 “.858. .39 - 85 u 8532 was

palmsﬁcnpmwodmd

 

 

8—

152

.523 =33.ch Eo: 8:209 ucm 853m. mac. .0 52.895 ms. mczoﬁma 29:98 commmamm .Nd 059....

 

 

  

\ So u a go H mm
2.2; _Egcam . me + of u 8.5.8”. c2585

 

 

8—

N—

;ﬁeS =2“..me

: o— m m h o
. _ _ _ F _

 

 

«8.0 u a .Ed u «m
3.22, __a-v_cmm . we - 32 H 852mm SE89;
/

 

 

 

CHAPTER 5

MACROINVERTEBRATE COMMUNITY STRUCTURE AND FUNCTION

ASSOCIATED WITH COARSE WOODY DEBRIS IN LOW GRADIENT,
MIDWESTERN STREAMS

154

Abstract

Coarse woody debris (CWD) plays a number of important roles in forested stream
ecosystems, including providing habitats for ﬁsh and invertebrates, ﬂow refugia, and a
site of bioﬁlm production that serves as food for grazing organisms. Logs added to
streams are rapidly colonized by invertebrates, and the changes in associated habitats are
accompanied by changes in community composition and functional attributes. A
multiple habitat, qualitative sampling approach was employed to evaluate
macroinvertebrate communities associated with woody debris accumulations in 36 stream
reaches in low gradient Midwestern streams. Taxa were classiﬁed with respect to habit
(e. g., sprawler, clinger, swimmer) as well as trophic/feeding characteristics. These traits
were used to examine community structure as a function of coarse woody debris
abundance and distribution. Two taxa, the amphipod Hyallela and chironomid
Polypedilum, made up 23% of the total abundance of organisms found in association with
woody debris. The mayﬂy, Caem's and elmid Dubiraphia made up an additional 11% of
the individuals in the woody debris community. Individuals belonging to the most
common 9 taxa made up 52% of the woody debris community. Although woody debris
is not abundant in these streams, it is one of the most important habitats for
macroinvertebrates in the streams of the Saginaw Basin. Since woody debris can occur in
both fast and slack water, the taxa found in association with wood habitats span a range
of current preferences, as well as functional and habit traits. The patterns in the
distribution of habit and functional traits within wood habitats suggests that these traits
may vary with the location of woody debris in the channel relative to the flow regime.

155

Introduction

Macroinvertebrate community structure and function are inﬂuenced by a complex
array of abiotic and biotic factors that interact over a range of spatial and temporal scales
(Carter, et al. 1996, Richards, et al. 1996, 1997). At the reach scale, abiotic factors such
as local ﬂow regime (Stratzner and Higler 1986, Brown and Brussock 1991), substrate
composition (Erman and Erman 1984, Wood and Armitage 1997), substrate stability
(Cobbs, et a1. 1992, Death 1995), and the presence of coarse woody debris (Benke, et al.
1984, 1985, Wallace, et al. 1995) play a strong role in structuring macroinvertebrate
community structure and function. Coarse woody debris (CWD) plays an important role
in forested stream ecosystems as a geomorphological agent, increasing ﬂow
heterogeneity through retardation of flow and creation of plunge pools (Keller and Tally
1979; Robison and Beschta 1990), changing channel depth and form (Nakamura and
Swanson 1993; Richmond and F ausch 1995), and increasing organic and inorganic matter
retention (Smock, et al. 1989; Beechie and Sibley 1997). CWD also plays an important
role as nursery habitat for salmonids (Murphy, et al. 1986; McMahon and Hartman 1989),
perching habitat for invertebrates (Angermeier and Karr 1984; O’Connor 1992), and a
site of bioﬁlm production that serves as food for grazing invertebrates (Hax and Golladay
1997, Bowen, et al. 1998). Logs added to streams are rapidly colonized by a wide range
of invertebrates (Nilsen and Larimore 1973, O’Connor 1991, Hilderbrand, et al. 1997),
and the change in associated habitats (i.e., pools formed ﬁom the erosive action of stream
ﬂow around the logs) is accompanied by changes in community composition and
functional attributes. Ephemeroptera abundance increased in pools associated with log

156

additions, while Plecoptera, Coleoptera, Trichoptera and Oligochaete abundance
decreased (Hilderbrand, et al. 1997). Large changes in biomass and abundance of
functional groups followed the addition of logs to rifﬂes in a high-gradient stream
(Wallace, et a1. 1995). In the littoral zone of two Canadian lakes, differences in
functional groups on natural and introduced logs were attributed to bioﬁlm chlorophyll 0

concentrations (Bowen, et al. 1998).

In a study relating catchment and reach-scale characteristics of 58 catchments in
central Michigan, to macroinvertebrate taxon traits, the presence of coarse woody debris
was positively associated with the presence of large bodied insects and clinging
macroinvertebrates (Richards, et al. 1997). Feeding traits (shredders) and habit modes
(swimmers, climbers, sprawlers) responded negatively to the percent of open canopy
cover, which also is correlated with the amount of CWD in the stream (Johnson 1999a).
In a related study, the presence of CWD, percent of reach with deep pools and the percent
of ﬁne sediments in the substrate were positively associated with the presence of
macroinvertebrate predator taxa found exclusively in depositional habitats, and negatively
associated with the proportion of taxa found exclusively in erosional habitats, and the
proportion of the taxa belonging to Ephemeroptera, Plecoptera, or Trichoptera (EP'I)

families (Richards, et al. 1996).

Coarse woody debris is being widely used in stream restoration to improve ﬁsh
habitat (Hunter 1995); however, little is known about the role that CWD may play in

157

structuring the macroinvertebrate community in highly altered streams with few stable,
hard substrates. The focus of the above studies by Richards, et al. (1996, 1997) was to
quantify associations among macroinvertebrate taxa or macroinvertebrate taxon traits and
a suite of landscape and habitat-scale variables. These studies identiﬁed woody debris as
a potentially important local and regional factor inﬂuencing macroinvertebrate
community structure and function. The current study was designed to speciﬁcally
address the relationships between CWD abundance and distribution with respect to the
compositional and functional attributes of the macroinvertebrate community. The goals
of this study are to: 1) identify compositional, life history, and behavioral habits of the
macroinvertebrate community that are most closely associated with CWD; 2) contrast
structural differences in the communities on wood habitats versus those of other habitats;
and 3) quantify the effect of CWD abundance and distribution on macroinvertebrate

community structure and function in non-forested streams.

Methods
Study Area

The Saginaw Bay catchment of Lake Huron encompasses a 16,317 km2 region,
characterized by sand and clay-dominated lowlands rimmed by coarse-textured glacial
features such as ground moraines and outwash plains (Figure l-l). The study region is
contained within two major ecoregions as deﬁned by Omernik and Gallant (1986): the
Southern Michigan/Northern Indiana Till Plains and the Huron/Erie Lake Plain. Each is
subdivided into two sub-regions. Soils in the lake plain are dominated by medium and

158

ﬁne-textured loams ranging to clays, with sand in the outwash plains and channels.
These clay regions are extensively drained by artiﬁcial drainage and tile systems. The
periphery of the basin contains many coarse textured glacial features such as ground
moraines and outwash plains. The till plain exhibits the greatest variation in basin
topography and contains a high percentage of forested land intermingled with agricultural
land and old ﬁelds; elevations average about 278 m. The entire drainage was logged for
white pine and hemlock between 1830 and 1900, and forests of the region now consist
primarily of second growth hardwood species. Three ﬁrst to third order reaches in each
of 12 catchments were selected to quantify some internal variation in dominant land use
and surﬁcial geology within streams. A total of 36 subcatchments ranging in size from
712 to 23,448 ha were studied (Figure 5-1). Sample reaches were located at least 50 m

upstream of bridges and culverts.

Coarse Woody Debris

Coarse woody debris assessments were performed during low ﬂow conditions
during the summer of 1995. CWD volume was measured using the line transect method
(Wallace and Benke 1984). Volume per unit area was calculated for each transect and
summed for each reach (see Johnson 1999a for details of sampling methods). In addition
to volume measurements, counts of the total length of CWD z 0.05 m diameter and 2 1 m
in length were made at 10 m intervals within the reach and summarized as total meters of
wood per m2 of stream bottom (m/m’) for each site. Debris accumulations z 1 m2 in area
were counted and summarized as the total number of debris accumulations / 100 m reach.

159

Debris accumulations were broadly deﬁned to include overhanging vegetation with
trapped debris, root wads with attached debris, loose log accumulations, as well as debris
jams (see Johnson 1999b for details). Debris accumulations were assigned a size class
based on the dimensions of the accumulation relative to the width of the stream (Shields
and Smith 1992); the amount of stream channel covered by debris accumulations also

was recorded (= sum accumulation size; see Johnson 1999a).

Macroinvertebrate Community

A multiple habitat approach as described by Lenat (1988) was employed to
evaluate macroinvertebrate communities associated with major subhabitats including
erosional and depositional areas, shorelines, leaf packs, and woody debris. This
technique reduces bias encountered in single habitat sampling techniques (Kearns, et al.
1992) and allows a larger number of sites to be evaluated. The approach employs a
variety of sampling devices (timed kick net samples, Ekman samplers) and effectively
samples a high diversity of organisms across size categories (Lenat 1988). Five different
habitats types were examined in this analysis: pools, runs, riﬁles, macrophyte beds, and
woody debris accumulations. Three replicate samples were obtained from each habitat
type. Sampling was conducted once during late summer or early autumn when the largest
number of invertebrates (with the exception of shredders) were of sufﬁcient size for clear
identiﬁcation to generic level. Sampling in the wood habitats was achieved by

vigorously agitating wood accumulations and capturing displaced organisms with the

160

dipnet. This methodology not only captured organisms associated with the wood itself,

but also captured planktonic organisms in slack water associated with log accumulations.

Samples were preserved and returned to the laboratory for processing and
identiﬁcation. Large pieces of inorganic and organic matter were removed and the
remaining material was spread over shallow trays with grid lines. Organisms were
removed from randomly selected grids until 100 individuals were obtained or the entire
sample was processed. Taxa were classiﬁed with respect to trophic/feeding categories, as
well as habit (as per Merritt and Cummins 1996). These traits were used to examine
community structure as a function of coarse woody debris abundance and distribution (as

per methods in Richards, et al. 1997).

To assess the compositional and functional aspects of the macroinvertebrate
community associated with CWD, taxa were classiﬁed according to their affinity for the
woody debris habitat as “wood-associated” (found exclusively in wood habitat samples),
“wood-dominant” (> 90% of individuals encountered in the wood habitat samples),
“wood-averse” (less than 10% of individuals associated with woody samples), and
“wood-absent” (taxon never found in association with woody samples). Taxon traits
were summarized as a proportion of the pool of wood-associated/dominant or wood-
averse/absent taxa (n = 55 taxa). To determine whether there were signiﬁcant

differences in taxon traits within the wood habitat samples, a one way analysis of

161

variance was performed on the proportion of individuals associated with each class of

traits (e.g., habit and functional feeding).

The contribution of CWD habitats to the macroinvertebrate community at each
site compared to all other habitats was examined by comparing the number of taxa
associated with woody debris habitats with taxa richness values derived from all other
habitat types combined (Appendix 5-2). The number of unique taxa contributed by wood
was derived ﬁ'om the difference between the number of taxa contributed by the CWD
habitats and the taxa richness from all other habitat types combined. A Pearson Product
Moment Correlation test was performed between the number of unique taxa contributed
by CWD habitats and all other habitats and CWD abundance metrics, log (m/m2), log
(volume), number of debris accumulations / 100 m, and sum accumulation size (see

Johnson 1999a for a description of methods).

Results

Wood debris habitats were present at 31 of the 36 stream reaches sampled (Table
5-1). Twenty four wood-associated taxa and eleven wood-dominant taxa were found
across the study area (Appendix 5-1). The wood-dominant taxa each occurred in much
greater abundance than the wood-associated taxa. The taxa that are most closely
associated with woody debris are distributed among eight insect orders, in addition to
nematode and naidid worms. The most abundant wood-associated taxa found were:
Enallagma (Coenagrionidae), Anopheles (Culicidae), Matus (Dytiscidae), Belostoma

162

(Belostomatidae), C yphon (Helodidae), Hydraena (Hydraenidae), Paraponyx (Pyralidae),
and Boyeria (Aeschnidae). Most of the wood-associated taxa were very rare in the study
area; only ﬁve taxa, Plea (Pleidae), Brillia (Chironomidae), Platycentropus and
Limnephilus (Limnephilidae), and Lype (Psychomyiidae) were represented by more than
5 individuals in the entire collection. Twenty one wood-averse taxa and four wood-
absent taxa occurred in the study area (Appendix 5-1). Wood-averse taxa with the

exception of Atherix sp. (Athericidae) were rare.

The most common taxa found in the woody debris habitat samples were the
amphipod Hyallela and chironomid Polypedilum, composing 23% of the total abundance
of organisms in this habitat (Table 5-2). The elmid, Dubiraphia, and mayﬂy, Caenis
composed an additional 11% of the individuals in the woody debris habitat. Individuals

belonging to the nine most common taxa compose 52% of the woody debris community.

Traits Associated with Wood-Dominant and Wood-A verse T axa

Of the 55 taxa that were either wood-associated/averse or wood-absent/averse,
45% percent were classiﬁed as predators, 25% as collectors, and 18% as shredders, and
5% as scrapers (Table 5-3). Collector taxa were approximately equally represented
between the wood-associated/dominant and wood-averse/absent groups. In contrast,
there were more scraper taxa in the wood-absent/averse group, while there were more
shredders and predators in the wood-associated/dorninant group. The ﬁve habits were
approximately equally represented across these 55 taxa. The most common habit modes

163

of the wood-associated/dominant taxa were climbers and clingers, with swimmers and
burrows having somewhat smaller representation (Table 5-4). Sprawlers were most
common and climbers were absent amongst the wood-averse/absent taxa. The proportion
of borrowing and clinging taxa were approximately equal between the wood-averse and

wood-associated groups.

Of the nine numerically dominant taxa associated with woody debris habitats,
four are collectors (three are collector-gatherers, one is a ﬁlter feeder), three are predators,
one each are grazers and shredders. Across the entire community associated with woody
debris there was a signiﬁcantly greater proportion of collectors than scrapers, shredders
and predators (Figure 5-2a). The proportion of collector-gatherers was greater than
collector-ﬁlterers (Figure 5-2b). In addition, there was a signiﬁcantly greater proportion
of individuals whose behavior was characterized as clinging forms compared with that of

swimming and climbing forms (Figure 5-3).

Inﬂuence of C WD Standing Stocks on Macroinvertebrate Community

At 23 sites, the presence of the CWD habitat contributed a mean of 1 1 i 2.3
(range 1 to 28) unique taxa to the total taxa richness (Appendix 5-2). There was a
signiﬁcant correlation between the number of taxa contributed by woody debris habitats
and the standing stocks of CWD at a site, as measured by log(m/mz), and log(volume)

(Table 5-5). The log (m/mz) and sum accumulation size (the metric reﬂecting the amount

164

of stream channel covered by CWD accumulations) were both negatively correlated with

the number of taxa contributed by non-wood habitats.

Discussion

Coarse woody debris is not abundant in these Saginaw basin streams; despite this,
woody debris is one of the most important habitats for macroinvertebrates in these
streams. Of 150 total taxa encountered across the study region, 130 were found in
association with coarse woody debris. Of these 130 taxa, 24 were found only in the
woody debris samples, and another 11 taxa were disproportionately represented in the
wood samples compared with other sample types. Despite the smaller standing stocks of
CWD and the apparent smaller size of debris accumulations compared with forested
regions (J ohnsonl 999b), woody debris habitats are important contributors to the taxa
richness of these streams. At more than half of the sites (22 of 35), the CWD habitat
contributes a mean of 11 unique taxa (with a range of between 1 and 28) to the total pool
of taxa. Benke, et al. (1984) and Smock, et al. (1985) both report increased production in
association with snag habitats in coastal plain rivers where shitting sand substrates
otherwise provide few stable habitats. Many streams in this region are severely degraded
due to disturbances that began when the region was harvested of white pine and hemlock
from 1840-1900. Following this period of intensive harvest, ﬁres burned much of the
regions, resulting in a denuded landscape that was then subject to massive soil erosion.
Conversion to agricultural land followed in the 1930's when large expanses of wetlands
were drained and placed under production (Comer, et al. 1993). Inherent in agricultural

165

production are management practices that involve stream channelization, replacement of
woody riparian vegetation with grasses, and stream clearing to remove any obstacles to
ﬂow. Superimposed upon these management regimes, which themselves have a tendency
to reduce stream habitat heterogeneity, is the effect of the Quaternary geology on the
hydrologic regime. Regions dominated by lacustrine sediments, particularly lacustrine
clays, tend to have ﬂashy ﬂow regimes due to the intensive tiling in agricultural regions,
and the dominance by surface water rather than ground water ﬂow regimes. Even in
catchments dominated by mixed, rather than agricultural land use, and morainal geology,
the streams are relatively homogeneous compared with more pristine streams in the
northwestern and northern part of the state. This homogeneity clearly inﬂuences the low
standing stocks of CWD (see Johnson, 1999a), the paucity of hard surface habitats such

as rifﬂes.

At sites where woody debris habitats contributed unique taxa to the overall taxa
pool, total taxa richness was lower compared to sites where woody debris did not
contribute unique taxa to the pool. The inﬂuence of woody debris habitats therefore
appears to be greatest when overall taxa richness is low, probably due to low overall
habitat heterogeneity or speciﬁcally to the absence of other hard substrate habitats at the
site. When overall taxa richness is high, the relative contribution of woody debris to the
overall taxa pool is lower, due to high habitat heterogeneity which creates the potential
for unique taxa to be contributed from a variety of habitat types. This interpretation is
reinforced by the negative correlation between CWD abundance standing stocks and the

166

number of unique taxa added to the taxa pool from non-wood habitats (Table 5-5).

Where woody debris is present in abundance, generalist taxa associated with wood are
also found on other habitats, therefore few unique taxa are added when additional habitats
are available. Introduction of unique habitats with habitat-speciﬁc taxa are necessary to

increase the pool of unique taxa.

The number of unique taxa added to the total taxa pool from wood habitats was
positively correlated with standing stocks of CWD at a site (Table 5-5). Such a pattern is
expected based on species-area curves (as per MacArthur and Wilson 1967, Barbour and
Brown 1974), if one considers increasing abundance of CWD to be equivalent to
increasing habitat area. Number of unique taxa associated with wood habitats are also
positively correlated with mean bank-full width, but not with substrate characteristics or
other characteristics of channel structure (e.g., depth, ﬂow characteristics; data not
shown). It is not surprising that ﬂow characteristics would not be related to taxa richness.
Rather, this feature of streams is more likely to inﬂuence the functional and
compositional response of the community, through alteration of life history traits (Poff

and Ward 1989).

Community Structure and Function

The majority of the 11 numerically dominant taxa found in association with
woody debris habitats are habitat generalists and were found in all or most of the sample
types studied; Asellus, Anopheles, and Calopteryx was not found in rifﬂe samples, and

167

Rheotanytarsus, Paratanytarsus, and Hydropsyche were either rare or absent in pool
habitats. With the exception of the coenagrionid, Calopteryx, the eleven most common
taxa or close relatives, as well as the wood-associated and wood-dominant taxa, have
previously been reported to occurr in association with coarse woody debris (N ilsen and
Larimore 1973, Dudley and Anderson 1982, Benke, et al. 1984, Smock, et al. 1985,
Phillips and Kilarnbi 1994a, b, Bowen, et al. 1998). Calopteryx is a widespread genus
found in the margins of lotic habitats. This taxon was found exclusively in slack water

habitats, including macrophyte beds, pools, and in association with woody debris.

One of the roles ascribed to CWD is that it provides a surface for bioﬁlm
development, and thereby serves as a food source for scrapers (Hax and Golladay 1997,
Bowen, et al. 1998). To a limited number of xylophagous taxa, the wood itself is a
source of nutrition. For the most part, taxa associated with woody debris use it as a
resting and feeding platform (Dudley and Anderson 1982). The complexity of the wood
bark increases the potential surface area of this habitat, providing a ﬂow and predation
refugium (O’Connor 1992). Predators, grazer/scrapers, and ﬁlter-feeders could
potentially use a hard substrate habitat such as CWD. Among the nine dominant taxa
associated with CWD in our streams, three taxa were collector-gatherers; one was a ﬁlter-
feeder, two were predators, and one each were shredders and grazers. Few of the wood-

associated/dominant taxa were designated as collectors; most were predators or shredders.

168

 

Since woody debris can occur in both fast and slack water, the taxa found in
association with wood habitats span a range of current preferences, as well as functional
and habit traits. Within the CWD habitat, there were signiﬁcant differences in abundance
between gatherers and ﬁlterers, with gatherers being more abundant than ﬁlterers. These
patterns are similar to those observed by Wallace, et al. (1995), who reported decreases in
biomass and abundance of ﬁlterers in log-inﬂuenced sections of rifﬂes compared to those
with no log additions, whereas collectors (gatherers) increased in abundance and
biomass. The effect of the log addition was to decrease ﬂow rates and increase standing
stocks of organic matter. The riffle section became more pool-like in its character with
the addition of logs to deﬂect ﬂow. Benke, et al. (1984) and Smock, et al. (1985) also
reported high production associated with ﬁlter-feeding and collector-gatherer taxa,
compared with benthic production and production in the muddy stream banks of lowland
rivers. 1n the Saginaw Basin streams, the variance associated with the abundance of
ﬁlterers in the woody debris habitats was greater than that for gatherers. This may be
related to the ﬂow conditions surrounding the CWD accumulations. Since woody debris
can accumulate in slack water, as well as in faster current, ﬁlter feeding
macroinvertebrates may be responding to the current, as well as the presence of hard

substrate.

Among the numerically dominant taxa in this study were the chironomids
Polypedilum and Thienemannimyia, and the coenagrionid Calopteryx. The damselﬂy
would potentially account for a large pr0portion of biomass, due to its large size.

169

However, this taxa, and Polypedilum are associated with slack water habitats, therefore,
ﬂow conditions may again be an important determinant of the distribution of this

functional group.

Macroinvertebrates have evolved many morphological and habit characters to deal
with the effects of ﬂow. The habit related to locomotion would be expressed most
strongly across gradients of ﬂow regimes, with pool habitats and macrophytes beds at one
end of the spectrum and rifﬂe and run habitats at the other end. Depending on the
location of CWD in the channel, the organisms found in association with that habitat
might express a range of locomotor traits. Within the woody debris habitat there were
signiﬁcantly more clingers than climbers and swimmers (Figure 5-3). A clinging habit is
commonly found among ﬂattened species found in rapid currents, and is commonly
associated with organisms living in riffles (Allan 1995). Again, these patterns suggest
that the functional/habit composition of the macroinvertebrate community on CWD may

vary with its location in the channel relative to the ﬂow regime.

Conclusions

Coarse woody debris in these low gradient, Midwestern streams represents a very
important habitat for macroinvertebrates, even though it is not abundant. Much of the
overall taxa richness can be attributed to the presence of CWD at a site by providing a
pool of taxa that are unique to that habitat. As CWD abundance increases, the potential
contribution of taxa from other habitat types decreases. This reﬂects the tendency of

170

 

 

 

most taxa to be habitat generalists- in the presence of CWD, taxa occupy that habitat as

well as others, reducing the potential contribution of unique taxa ﬁ'om other habitat types.

The community of taxa that occupy CWD habitats are diverse in their behavior
and trophic status. While the taxa found only in association with CWD were numerically
uncommon, a large proportion of these taxa were predators and shredders. In contrast,
the numerical dominants of the community were largely composed of collectors. The
turbid nature of the streams in the study area and the lack of other hard substrates may

cause these taxa to use CWD disproportionately compared to other types of streams.

Despite the low standing stocks of CWD, the wood that is present appears to play
an important role in structuring the composition and the function of the macroinvertebrate
community. Many management practices employed across the basin are directed at
removing woody debris and other structures that represent obstacles to ﬂow, such as
overhanging vegetation. These habitats are important constituents of the stream
ecosystem and contribute a large pool of taxa to the overall richness. Attempts should be

made to moderate stream clearing practices that reduce structural heterogeneity.

171

Table 5-1. Abundance of macroinvertebrates from ﬁve sample types. Average per
samples were derived by dividing the total number of individuals by the number of sites
in which the habitat was represented.

 

Abundance Metric Wood Macrophyte Run Riﬁle Pool
Total # of Taxa 130 80 82 60 80
Total # Individuals 6809 2415 2265 2132 4140
# Sites 31 9 14 10 26

172

1802383 .823 .8233 u 39>.on 3:02.883 03930 n monon— 38233 23330 n m8m
3885a u 83m 3.38% u 38% 38:”. u 950 238% u :80

32088 n 83m 2828“— u do:— 338 u 80 3823;803:200 H 8.200 3358-38230 H 58.30
82283 n >830 832.53 u 2.30 86283 n 8oz

 

 

aonoam 238m 35 >830 3.6di 02 8880 3288820 Eaﬁwzzosoamaﬁ
m8m 8:0 2.0-30 «Son 3.; .80 com 8085 3288820 8.383883%
58.8
3983 was 62am 8: 83$ em 828 828885 .3885
noemocm p8=0 235 >830 3&de Rm 8830 32wboaoﬁ0 508380
38m
m8m 826 838.00 880 3%di wen EoEanEoF—qm 3250anqu uwdwEQSmESok
038m .
3D :3an E3830 =on 33.30 Gm EodnoaoEozam 32:30 £230
m8m 8:0 58.30 280D @1de wmv 8038—00 328m oEQohdaQ
8:0 Saw
8a 586 63 25:0 89$ 8m 8&5 825825 55338
h8:0
moo 823m 8380 2on 33.30 Rd 30¢Em8< 328:3... 3335
mecca—om £58m g. 88:00
323: :3: 288B 288,—. .232: a 380 383...— «3...

.3388

253 .383 2 8808 582888 8890838838 2: do 83808 832820 .3588 2: 8288 88% .m-w 23...

173

Table 5-3. Functional feeding groups of taxa strictly associated with coarse woody debris,
and taxa averse to coarse woody debris habitats. MacrOphyte piercers represented less
than 4 % of the total taxa, and taxa with unknown characteristics were about 1.5% of the
total pool of taxa in these groups.

 

 

 

AssociationWith Feeding Functional Group
Wood Habitats

Shredder Collector Scraper Predator
Overall Total 18.0% 25.0% 5.0% 45.0%
(n = 55)
wood-absent 1 6% 21 % 1 0% 42%
(n = 19)
wood-averse 0% 33% 33% 33%
(n = 3)
wood-dominant 20% 20% 10% 50%
(n = 10)
wood-associated 22% 22% 4% 48%
(n = 23)

174

Table 5-4. Habit modes of taxa strictly associated with coarse woody debris, and taxa
averse to coarse woody debris habitats. Taxa with unknown characteristics represent
about 1.5% of the total pool of taxa in these groups.

 

 

 

Habit
Association With
Wood Habitats burrow climb cling sprawl swim
Overall Total 17.9% 19.6% 23.2% 17.9% 14.3%
(n = 56)
wood-absent 7. 1% 0.0% 8.9% 8.9% 5 .4%
(n = 19)
wood-averse 1 .8% 0.0% 1 .8% 3.6% 0.0%
(n = 4)
wood-dominant 3.6% 7. 1% 3.6% 0.0% 1 .8%
(n = 11)
wood-associated 5.4% 12.5% 8.9% 5.4% 7.1%

(n = 22)

175

 

Table 5-5. Pearson correlation coefficient from number of unique taxa contributed to the
total taxa richness from wood and non-wood habitats (pools, rifﬂes, rtms, macrophytes)
versus standing stocks of coarse woody debris measured as m/m2, m3/m2, and a measure
of the amount of channel covered by debris accumulations. Signiﬁcant values are
adjusted by Bonferroni corrections.

 

Source of Taxa Log (m/m2) Log (m3/m2) Sth (# debris Sth (2 accum
accum/ 100m) size)

 

# Unique Taxa -0.557 -0.559 -O.437 -0.514

on Non-Wood (p = 0.012) (p = 0.012) n.s. (p = 0.037)
Habitat

# Unique Taxa 0.462 0.508 0.31 0.32
Contributed by n.s. (p = 0.043) n.s. n.s.

Wood Habitat

 

176

82w8 88m .8 32 ._.m 8sz

177

 

8238A 885m
Ea 3:25.85

83m 32m Baiwam _ -

178

1.0

 

0.8 -

0.6 ~

0.4 d

Proportion

0.2 i

0.0 -'

 

 

 

1.0

0.8

0.6

Proportion

0.4

0.2

0.0

Figure 5-2a. One way analysis of variance results testing for differences in the abundance
of macroinvertebrates in woody debris samples among functional categories. (Shown are

shredder predator

Trait

collector scraper

 

d

 

Functional Group

collector > shredder.
predator. scraper
(p < 0.05)

scraper > shredder
(p < 0.05)

predator > shredder
(p < 0.05)

Collectors

gatherer > fltterer
(p < 0.001)

 

 

T T

gatherer ﬂlterer

Trait

means and SE.) 2b. One way analysis of variance results testing for differences in the
abundance of macroinvertebrates in woody debris samples among colector categories.
(Shown are means and SE.)

179

 

 

 

 

 

 

 

0.35
0.30 -i
0.25 a
c .
o
"E Habit
a. 0.20 -
8 cling > swlm
o. (p < 0.001)
0‘15 ‘ cling > climb
(p < 0.001)
i i
0.05 v . r
swim climb cllng
Trait

Figure 5-3a. One way analysis of variance results testing for differences in the abundance
of macroinvertebrates in woody debris samples among locomotor behavior types. (Shown
are means and SE.)

180

 

 

 

20: 08:0 0028 3203880030: 8082.20..- 8082 083-0003 382200522.
2..-.00
28. M28 .08: 3208880020: 88:022....- 8082 083-0003 3.28209.
2.0. w2.0 :88 32288.08: 82:022.; 808:. 083-0003 0&3
”.388 Sam-.00
£02 w2.038:0 .0020 30:28:85 88:020.; 808:. 083-0003 3.2328...
:88
28. w::0 3....-.00 32.800235 88:022.; 808:. 083-0003 322000.08m
28:0
280. ”8.26 08: 320.: 808.80.. 808:. 083-0003 :0:
8.38
:8: 88:0 08: 32380820.. 80.280... 808:. 083-0003 3000020..
280. .388 08: 328083.00 80.2802 808:. 083-0003 28083.00
28. w2.0 08: 32.8: 88:000.: 808:. 083-0003 02.0233:
58. 08:0 08: 302288000 88000 808:. 083-0003 :2me
280. 28:0 08: 328.36. 88000 808:. 083-0003 08.3.\
28. m2.0 ﬂaw-.00 3005802....- 80.:8080::m. 808:. 083-0003 32:28:04
323.02
.8202 .20.. 02:8... 2.83“. 80.5 80.0 0:»... m3..-

.w2m28 800

u . . .008: 03 8.3 3802880 8 .8288 230288238880. .828. 88.800 03 82023.: .3.3822 a3:288
$288.30. $222.0 .w2w2.0. .20.. 6.00088 .8308: 8:88 A8802: 8 080233 - 8.02.00. 32323.8 02:8 ..-

.20. 02 2 0020800 03 3.8330 A3008» u 8.83-0003 .33-0003 "883-0003. 3.:88 030 2 0:00.. .0: 03 .80
80:. 03 .800.0003 M 832800-0003 ”083-0003 .1. 003.083-0003. DB0 23 003.083 380.0 83..- ..-m x.0:0::<

181

 

 

 

28. 08:0 00.: 0308030: 338000 3.08:. 800I0003 0.208:
0.00 08:0 08: 032380.828. 808.80: 3.08:. 80010003 080.33%
0.00 080.083.: ..:-.00 030.280 808.0 3.08:. 800I0003 00.8302:
28. m2.0 :88 ..00 0328.0 88:00.00 3.08:. 80010003 3888053
8.38 00.: 80.:
0.00 .3880 .:8038 3003802 083-0003 300.3802
0.8. 3880 0.38.00 030.80:820 88:5 3.08:. 0830003 0080:088k
.388
0.80. 38:0 0.38.00 030.80:820 88:5 3.08:. 0830003 08000800.:
8.38
0.00 0388 038.00 30.380085 88:2. 3.08:. 0830003 30:80:20.0
.388 038.00
0.8. .3880 .0808 030.80:820 88:5 3.08:. 0830003 0.52%
0>.0 8.38
0.00 38:0 00.: 030:2:803. 88:00.00 3.08:. 0830003 3.28003
08:0
0.00 8.38 00.: 030.822. 88:00.00 3.08:. 0830003 3.08%.
0.00 08:0 00.08 032.2800»... 88:00.00 3.08:. 083-0003 30080.3
08:0
0.00 ”8.38 08: 0328.80 88:00.00 3.08:. 0830003 3.000085
08: 030.0830 88:00.00 3.08:. 0830003 030.0830
08:0 :88
0.00 0388 .0803 30:28:85 88:000.; 3.08:. 0083-0003 0008:0003.
8002800. Tm x.0:0::<

182

 

 

 

 

0:2 887. 38 Swan; 88880: 8085 maul—woo? 83>cma$~
£3 mam—o 08:8 Rufﬁan—8:03 889808... 8085 maul—goo? 8338:»:
£80— wczo 38m 32830 88383 888— mnmlvoozr c.8838;
£80— 368? 38 823.335 8880 8085 mnmlvoes 88383
032 2685 38 03583—8800 8880 8035 maul—goo? 88833380
conga
£8. 2685 83980 82808820 8880 88m:— >mln¢oB wzﬁczcéxoottm
83 158% . 82808820 8880 38m:— >wlwo9$ SBSNNXESM
988
£82 mam—o 8mm-_oo 82808820 8880 8085 $1803 chumactmapi
3085
53 338% 38 8288830 88me 803.:— >mlv¢oB 8883398890
833.
£82 58:0 387. 26:83 8080295 888— EOvlvooa 388.8%
Eon 2685 .8 «2532qu 802825 831803 8288 Z
Eon 3% ”839m 38 320880 8088.00 888— 80ulvo03 .38:
98%
Son ”maze 88 0380833 8080200 808:— 82ulcooa 86335
5mm.._oo
8:0 3:38
53 ”>355 608? $288830 8880 88m:— EO—VI—uooa mmﬁhzﬁoibw
Eon 98:0 88 828029880 8880 88m:— EOcIuooa uﬁmazunm
282 “8:0 826201 8880200 888— 881803 8833
€283:on Tm 88583

183

 

800 3885

800 886
8:0

080. 3885
800 858%
0880— 8:0

800 38.80
080— .388

800 0>8 886
8:0
080— 238%

800 0>8 886
£58

”—38%

8800 88:0

8:0
800 2388

 

32828820 883888

8.38 828083
808
_8

_00 82808880

808 82082
8.200
8080

“88.80 828080880
8.300

88-80 82808880

808 8228088.

. 828008080:
88-80
8080808

8888 82.8888:

808 82828080

808 820285

8080 808 828000885
88-80
808

”00800 8288802

8085
8085
8085

8085
8085
8080080

80800.00
80800.00
8080080

808082;

8080825.

80880—08

80888.82
8088088
8008—
8088

800085

8008—
80008—
80005

80008

8088—
8008—

8088—

8008—

38.8003
maul—0003
0840003
08.8003
3818003

3818003
0840003
3818003
maul—0003
0818003
088003
388003

088003

0818003

0828082
828088820
80800800
88888082
82082

88080830088
888880880
#2088
88000880882
800.80%

8038

808MV

0.80000

8880:

 

808888000

2 8828.88

184

Appendix 5-2. Total number of taxa found at a site, in non-woody debris habitats, and in
woody debris habitats, and the number of unique taxa occurring only in wood habitats.

 

 

SAGCODE Total Taxa Non-Wood Wood Unique Wood
51 32 4 29 25

52 32 32 O

53 38 18 31 13

91 28 15 19 4

92 47 28 3O 2

93 41 26 29 3

232 35 8 30 22 E
233 38 39 26 l
234 46 46 0

242 60 37 31

243 41 41 1

244 44 24 34 10

311 39 32 16

312 43 23 29 6 '
313 37 37 0

321 38 24 29 5

322 25 5 22 17

323 17 5 14 9

351 28 9 21 12

352 28 12 21 9

353 22 16 7

361 51 26 41 15

362 43 26 29 3

363 49 31 35 4

375 49 33 29

376 57 43 37

377 44 32 33 1

2411 41 13 35 22

2412 42 18 32 14

2413 43 10 38 28

5111 54 34 37 3

5112 39 33 15

5113 38 31 9

5121 45 24 33 9

5122 33 10 28 18

5123 31 31 0

185

CONCLUSIONS

 

186

The Saginaw Basin in central Michigan is structurally diverse with respect to its
underlying Quaternary geology and land use. Soil productivity has been one of the major
factors inﬂuencing current land use and land cover patterns in this region, resulting in a
cascade of effects that have had a profound inﬂuence on both the structure and function
of stream ecosystems. Huge expanses of wetlands located on the historic lake bed
sediments were drained and placed under intensive agricultural production. Land
management practices including channelization and riparian vegetation conversion to
grasses have led to a simpliﬁed channel structure that poorly retains nutrients, sediments,
and coarse woody debris. Abandoned farmland in areas with low soil productivity has ,
slowly succeeded to old ﬁelds with savannah-like vegetation (now called range land).
Stream channels in these regions have slow regained habitat heterogeneity as riparian
vegetation invades the channels. Remnant second grth forest patches currently are
concentrated in regions of unproductive, sandy soils that are unsuited for farming, but
which have recently become attractive sites for low density residential development.
Hydrologic patterns controlled by the underlying geology-- groundwater inﬁltration on
morainal landforms and surface water phenomenon on lacustrine soils, interact with (and
exacerbate) the negative effects of land management practices. Tile drainage in
conjunction with wetland draining, and stream channelization result in ﬂashy ﬂow
regimes that transport CWD out of the reach, and have erosive effects on the channel.
Removing woody riparian vegetation and plowing to the edge of the stream bank results
in increased sediment ﬂow, and chemical inputs, and decreased inputs of coarse woody
debris to the stream. Deciphering the effects of these complex interactions on the stream

187

ecosystem would not have been possible without two sets of tools, geographic
information systems and multivariate statistical techniques, and the widespread
availability of spatial databases describing elevation, land cover, and hydrography in the
region. When used in conjunction with ﬁeld data, these tools have allowed me to
quantify patterns of association and infer the effects of a variety of local and regional
factors on stream ecosystems, particularly with reference to the standing stocks and

distribution of coarse woody debris.

A combination of ﬁeld-collected data, quantifying in-stream and riparian
conditions across 12 catchments in the Saginaw Basin, and spatial data, quantifying the
location and type of land use and Quaternary geology, were used to quantify the standing
stocks and distribution of in-strearn coarse woody debris. The long disturbance history of
the region is reﬂected in modem-day standing stocks of CWD, which, along with the size
of the logs, were much smaller in comparison to most other streams studied, especially
those in high gradient, forested ecosystems (Table 2-12). Land use and surﬁcial geology,
by themselves, did not have an effect on most measures of wood abundance or size;
however, interactions between land use and geology were evident with respect to their
effects of the abundance of CWD and the density of debris accumulations. Highest
standing stocks were found in association with Lac/Ag and Mor/Mix catchments,

compared to catchments dominated by Lac/Mix and Mor/Ag land use and geologies.

188

 

At the channel scale the factors that appeared to have the greatest inﬂuence over
CWD were the channel bank-full width and the percent of open canopy (Table 2-11).
These two local-scale characteristics of the stream were the best predictors of the number
and distribution of debris accumulations. Bank-full width was positively correlated with
number of debris accumulations per 100 m and the Z accum size metric (reﬂecting the

extent of channel covered by debris accumulations). Percent of open canopy was

 

negatively correlated with those two CWD measures. Whereas many other studies E
(reviewed by Harmon, et al. 1986, Gurnell, et al. 1995) have found strong interactions

between CWD and channel morphology, there did not appear to be any measurable effect

of CWD on channel features such as the bank-full width or depth and extent or depth of r

pools and rifﬂes.

Riparian vegetation and riparian width are determined by an individual
landowner’s preference in agricultural settings. As a result, factors inﬂuencing the
absence of wood at a site are more difﬁcult to predict than those inﬂuencing the presence
of CWD. From a management perspective, the presence of woody vegetation in the
riparian zone is more important than the width of the riparian zone in predicting CWD
standing stocks, suggesting that landowner education may be instrumental in helping to
restore riparian zone function. This speaks only to the role of the riparian zone in
contributing CWD to the stream, however, and does not consider the role of herbaceous

vegetation as a ﬁlter strip for sediments and anthropogenic chemicals.

189

Landscape features, including urban land use, link number, the SD. elevation, and
percent coarse till + sand/gravel were the best predictors of CWD density and debris

accumulation density and distribution (Table 2-11). These landscape-scale predictors

 

underscored the role of hydrology in the retention of CWD in the stream channel, and
indirectly pointed to the negative inﬂuence of agricultural land use, especially on

morainal landforms (Tables 2-14, 2-15). When predicting the actual volume (excluding

 

volume = 0) of CWD, even small amounts of wetlands in a catchment had an effect on F“
the regional stability of ﬂow, which was instrumental in retaining CWD in the channel

once it is delivered to the stream. Relatively rare land use types such as urban areas and

wetlands were surprisingly inﬂuential as predictors of CWD abundance and distribution, L

and probably reﬂected the absence of agricultural land use more than anything else. The
combination of local and landscape variables that best predicted debris accumulation
density and distribution highlighted the interaction between channel-scale factors that
inﬂuenced the entrainment of wood into a debris dam, and the landscape-scale factors

that inﬂuenced retention.

When the composition of debris accumulations are broadly deﬁned, their density

 

was greater than that reported in other studies (Table 3-8). When only ‘log/snag’
accumulation types are considered, density were similar to those of forested streams.
There is reason to believe that debris dams composed of logs and snags are smaller in the
Saginaw Basin than elsewhere; however, standard methods for quantifying debris
accumulation size are lacking, making direct comparisons difﬁcult.

190

In these highly disturbed streams, the most prevalent structural element in the
stream channel across the 49 sites was overhanging vegetation and root wads without
trapped debris. The majority of debris accumulations were associated with the bank,
rather than other types of obstructions such as root wads, point bars or islands, and there
were surprisingly large numbers of debris accumulations for which no visible obstruction
could be identiﬁed. These debris accumulations represent a pool of relatively mobile

CWD and coarse particulate organic matter (CPOM) in the channel.

The debris accumulation types exhibited different responses to the landscape and
local-scale predictors; ‘log/snag’ and ‘loose log’ accumulations were well explained by
the landscape and local variables, whereas, ‘root wad’ and ‘overhanging vegetation’ types
were not well explained at all. ‘Log/snag’ accumulations were best explained by factors
such as (low) stream density and (low) ﬂood height, higher topographic relief, larger
bank-full widths and larger catchment areas (Figure 3-6). ‘Loose log’ accumulations
were best explained by (large) bank-full width and catchment areas, and lower
proportions of lacustrine sand soils. These results parallel those obtained for total
number of debris accumulations and the Z accum size, and again reﬂect the inﬂuence of
landform on hydrologic processes (e.g., ﬂood height). No effect on channel features,
such as pool frequency and depth, were observed when debris dam type were examined

separately.

191

Perhaps the most important conclusion of this dissertation is that the inﬂuence of
landform cannot be ignored when attempting to understand the linkage between
terrestrial and aquatic ecosystems. Although land use and land cover are important and
their effects appear to be unambiguous, this study and others (Richards, et al. 1996, 1997;
Wiley, et al. 1997) demonstrate that land use effects are mediated by their underlying

landforms and the inﬂuence of landform on the hydrologic regime.

Some of the mechanisms regulating CWD retention and export from a stream
reach were examined by tracking tagged logs at two time periods separated by a
winter/spring season and a ﬂood with a return interval of about 5 years. In terms of
absolute numbers of logs, the turnover was very large; more than 50% of the logs at a
given location were exported and replaced from the fall through the following June
following a ﬂood with a 5-year return interval (Table 4-1). The proportion of retained log
volume after the ﬂood was two times greater than the recruited volume, while the
proportion of individual logs retained was less than the proportion of recruited logs. This
suggests that larger volume logs were retained, while smaller, more mobile logs were

recruited.

High ﬂood height was negatively correlated with the probably of retention, and
positively correlated with recruitment (Figure 4-1); bank-full width had a similar
response (Figure 4-2). Neither the original orientation, location of the log in the channel,
nor a log’s association with a debris accumulation inﬂuenced whether or not a log was

192

 

retained. Nor could the distance that a log moved be predicted from either ﬂow or
channel characteristics, possibly because the retention and movement patterns of logs

with branches differs from logs with a simple geometry.

Bank-full width is a good predictors of debris accumulation density, and
speciﬁcally of ‘log/snag’ debris accumulation densities. In addition, low ﬂood height
was one of the factors explaining the abundance of ‘log/snag’ accumulations (Figure 3-5).
These are relatively easy metrics to acquire in the ﬁeld, and can potentially provide
important information about CWD mobility as well as the density of CWD
accumulations.

Due to the small standing stocks and sizes of logs (and debris accumulations) in
the Saginaw streams, CWD does not play an important role in structuring the stream
channel, although it is proported to play many different roles in forest stream ecosystems
(Table 2-1). This leaves open the question: does CWD play a role in structuring the
macroinvertebrate community, and if so, how? Despite low standing stocks of CWD
across the basin, this habitat type supported 87% of the total macroinvertebrate taxa
found; of those, 25 % were found only on wood, or wood was their dominant habitat
(>90% of individuals were found on wood compared to other habitats). The importance
of this habitat in contributing to taxa richness is illustrated by the fact that at almost 50%
of the sites, wood habitats contributed an average of 11 unique taxa to the taxa richness.
The most abundant taxa (23% of the total individuals) found in association with wood
were amphipods and a chironomid. The other two dominants were the mayﬂy Caenis,

193

 

and the elmid beetle, Dubiraphia. The functional and behavioral attributes of the
community were diverse, suggesting that CWD serves as a hard substrate habitat whose
community composition probably varies with the ﬂow regime in which the wood is

located.

Overall, coarse woody debris is not abundant across this region, and the factors
that control the abundance are complex. At the landscape scale, both landform and land
use are important predictors of CWD, but interactions between landform and land use
result in complex patterns of association that are difﬁcult to interpret. Factors that
inﬂuence the hydrologic regime are particularly important for explaining the patterns of
abundance and distribution of CWD. Relatively rare land use types such as urban areas
and wetlands are surprisingly inﬂuential as predictors of CWD abundance and
distribution. Channel width and riparian vegetation types are among the most important
predictors of CWD at the local and riparian scale. Channel width also is an important
predictor of CWD retention, along with ﬂood height. Lastly, the role of CWD in this
region is conﬁned to biological, rather than physical effects. Geographic information
systems and multivariate statistical techniques are indispensable tools for unraveling

complex patterns across ecosystems.

194

LITERATURE CITED

195

 

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 and
Management 12:201-221.

Allan, J .D. 1995. Stream Ecology. Chapman and Hall, London. 388 pp.

Anderson, J. R., E. E. Hardy, J. T. Roach, and R.W. Witrner. 1976. A land use and land
cover classiﬁcation system for use with remote sensor data. USGS Professional
Paper 964. 28 p.

Anderson, N. H., J. R. Sedell, L. M. Roberts, and F.J. Triska. 1978. Role of aquatic
invertebrates in processing wood debris in coniferous forest streams. American
Midland Naturalist 100, 64-82.

Andrus, C. W., B. A. Long, and H. A. Froehlich. 1988. Woody debris and its
contribution to pool formation in a coastal stream 50 years after logging.
Canadian Journal of Fisheries and Aquatic Sciences 45:2080-2086.

Angerrneier, P. L. and J. R. Karr. 1984. Relationships between woody debris and ﬁsh
habitat in a small warmwater stream. Transactions of the American Fisheries
Society 115:716-726.

Armour, C. L., K. P. Bumham, and W. S. Platts. 1983. Field methods and statistical
analysis for monitoring small salmonid streams. US Fish and Wildlife Service.
FWS/OBS-83/33. Allan, J .D. 1995. Stream Ecology. Chapman and Hall, London.
388 pp.

Aumen, N. G., C. P. Hawkins, and S. V. Gregory. 1990. Inﬂuence of woody debris on
nutrient retention in catastrophically disturbed streams. Hydrobiologia 190:182-
192.

Barbour, C. D. and J. H. Brown. 1974. Fish species diversity in lakes. American
Naturalist 108:473-489.

Barbour, M. T., J. B. Stribling, J. Gerritsen, and J .R. Karr. 1994. Biological Criteria.

Technical Guidance for Streams and Small Rivers. US. Environmental Protection
Agency. EPA 822-B-94-001.

196

 

 

Beechie, T. J. and T. H. Sibley. 1997. Relationships between channel characteristics,
woody debris, and ﬁsh habitat in Northwestem Washington streams. Transactions
of the American Fisheries Society 126:217-229.

Belsley, D.A., E. Kuh, and RE. Welsch. 1980. Regression Diagnostics. John Wiley &
Sons, N.Y.

Benke, A. C., R. L. Henry, D. M. Gillespie, and R. J. Hunter. 1985. Importance of snag
habitat for animal production in Southeastern streams. Fisheries 10:8-13.

Benke, A. C. and D. I. Jacobi. 1984. Production dynamics and resource utilization of
snag-dwelling mayﬂies in a blackwater river. Ecology 75: 1219-1232.

Benke, A. C., T.C. Van Aarsdall, Jr., and D.M. Gillespie. 1984. Invertebrate productivity
in a subtropical blackwater river: The importance of habitat and life history.
Ecological Monographs 54:25-63.

Berg, N., A. Carlson, and D. Azurna. 1998. Function and dynamics of woody debris in
stream reaches in the central Sierra Nevada, California. Canadian Journal of
Fisheries and Aquatic Sciences 55:1807-1820.

Beschta, R. L. 1979. Debris removal and its effects on sedimentation in an Oregon
coastal range stream. Northwest Science 53:71-77.

Beschta, R. L. 1983. The effects of large organic debris upon channel morphology: a
ﬂume study. pp 8.63-8.78, In: Proceedings, D.B. Sirnons Symposium on Erosion
and Sedimentation. Fort Collins, CO.

Beschta, R. L. and W. S. Platts. 1986. Morphological features of small streams:
signiﬁcance and function. Water Resources Bulletin 22:369-379.

Bilby, R. E 1979. The function and distribution of organic debris dams in forest stream
ecosystems. Dissertation, Cornell University, Ithaca, NY.

Bilby, R. E. 1981. Role of organic debris dams in regulating the export of dissolved and
particulate matter from a forested watershed. Ecology 62:1234-1243.

Bilby, R. E. 1984. Removal of woody debris may affect stream channel stability.
Journal of Forestry 82:609-613.

Bilby, R. E. and G. E. Likens. 1980. Importance of organic debris dams in the structure
and function of stream ecosystems. Ecology 61:1107-1113.

197

 

Bilby, R. E. and J. V. Ward. 1989. Changes in characteristics and function of woody
debris with increasing size of streams in Western Washington. Transactions of the
American Fisheries Society 118:368-378.

Bilby, R. E. and J. V. Ward. 1991. Characteristics and function of large woody debris in
streams draining old-growth, clear-cut, and second-growth forests in southwestern
Washington. Canadian Journal of Fisheries and Aquatic Sciences 48:249-2508.

Bisson, P. A. 1982. Salmonid populations in streams in clearcut vs. old-growth forests of
Western Washington. In: W. R. Meehan,, T. R. Merrell, Jr, and T. A. Hanley,
(ed.). Fish and Habitat Relationships in Old-growth Forests: Proceedings of a
Symposium. American Institute of Fishery Research Biologists.

Bisson, P.A., J. L. Nielson, R. A. Palmason, and L. E. Grove. 1982. A system of naming
habitat types in small streams, with examples of habitat utilization by salmonids
during low streamﬂow. In: N.B. Arrnantrout (ed.) Acquisition and utilization of
aquatic habitat inventory information, p. 62-73. Western Division, American
Fisheries Society, Portland, OR. -

Bisson, P. A., R. E. Bilby, M. D. Bryant, D. M. Gillespie, G. B. Grette, R. A. House, M.
L. Murphy, K. V. Koski, and J. R. Sedell. 1987. Large woody debris in forested
streams in the Paciﬁc Northwest: Past, present and future. In: E. O. Salo, and T.
W. Cundy, (ed.). Streamside Management: Forestry and Fishery interactions,
Proc. Symp., 12-14 February 1986. University of Washington Press Seattle, WA.

Bisson, P.A., K. Sullivan, and J. L Nielsen. 1988. Channel hydraulics, habitat use, and
body form of juvenile coho salmon, steelhead, and cutthroat trout in streams.
Transactions of the American Fisheries Society 117:262-273.

Bisson P. A., and D. R. Montgomery. 1996. Valley segments, stream reaches, and
channel units. Pp. 23-52, In: F .R. Hauer, and GA. Lamberti (eds.) Methods in
Stream Ecology. Academic Press, San Diego.

Bowen, K. L., N. K. Kaushik, and A. M. Gordon. 1998. Macroinvertebrate communities
and bioﬁlm chlorophyll on woody debris in two Canadian oligotrophic lakes.
Archiv ﬁir Hydrobiologie 1412257-281.

Braudrick, C. A. 1997. Entrainment, transport, and deposition of large woody debris in

streams: Results ﬁom a series of ﬂume experiments. Oregon State University.
Master’s Thesis. 87 pp.

198

 

Braudrick, C. A., Grant, G. E., Ishikawa, Y., and Ikeda, H. 1997. Dynamics of wood
transport in streams: A ﬂume experiment. Earth Surface Processes and
Landforms 22:669-683.

Brown, A. V. and P. B. Brussock. 1991. Comparisons of benthic invertebrates between
rifﬂes and pools. Hydrobiologia 220:99-108.

Bryant, M. D. 1982. Organic debris in salmonid habitat in southwest Alaska:
measurement and effects. In: W. B. Armantrout, (ed.). Proceedings of a
symposium on the Acquisition and Utilization of Aquatic Habitat Inventory
Information. Western Division of the American Fisheries Society Portland,
Oregon. .3

Bryant, M. D. 1983. The role and management of woody debris in west coast salmonid
nursery streams. North American Journal of Fisheries Management 3:322-330.

Bustard, D. R. and D. W. Narver. 1975. Aspects of the winter ecology of juvenile coho
salmon (Oncorhynchus kisutch) and steelhead trout (Salmo gairdneri). Journal of
the Fisheries Research Board of Canada 32:667-680. L

 

Carlson, J. Y., C. W. Andrus, and H. A. Froehlich. 1990. Woody debris, channel features
and macroinvertebrates of streams with logged and undisturbed riparian timber in
Northeastern Oregon, USA. Canadian Journal of Fisheries and Aquatic Sciences
47:1103-1111.

Carter, J. L., S. V. Fend, and S. S. Kennelly. 1996. The relationship among three habitat
scales and stream benthic invertebrate community structure. Freshwater Biology
35:109-124.

Church, M. 1992. Channel morphology and typology. Chapter 6. In: P. Calow and GE.
Petts (eds). The Rivers Handbook, Volume 1. Blackwell Scientiﬁc Publication.
Oxford, 526 pp.

Cobb, D. G., T. D. Galloway, and J. F. Flannagan. 1992. Effects of discharge and
substrate stability on density and species composition of stream insects. Canadian
Journal of Fisheries and Aquatic Sciences 49:1788-1795.

Comer, P., A. Albert, T. Leibﬁied, H. Wells, B. Hart, M. Austin. 1993. Historical
wetlands of the Saginaw Bay watershed. Michigan Department of Environmental

Quality. 14 pp.

Cumming, K. W. and M. J. Klug. 1979. Feeding ecology of stream invertebrates.
Annual Review of Ecology and Systematics 10:147-172.

199

Culp, J. M., G. J. Scrimgeour, and G. D. Townsend. 1996. Simulated ﬁne woody debris
accumulations in a stream increase rainbow trout fry abundance. Transactions of
the American Fisheries Society 125:472-479.

Davis, R. J. and K. J. Gregory. 1994. A new distinct mechanism of river bank erosion in
a forested catchment. Journal of Hydrology 157:1-11.

Death, R. G. 1995. Spatial patterns in benthic invertebrate community structure: products
of habitat stability or are they habitat speciﬁc? Freshwater Biology 33:455-467.

DeVries, P. G. 1974. Multi-stage line transect sampling. Forest Science 20:129-133.

Draper N. R. and H. Smith. 1981. Applied Regression Analysis. John Wiley & Sons,
New York. 709 pp.

Dudley, T and N. H. Anderson. 1982. A survey of aquatic invertebrates associated with
wood debris in aquatic habitats. Technical Paper 6419. Pp 1-21 In: Oregon
Agricultural Experiment Station.

Ehrman, T. P. and G. A. Lamberti. 1992. Hydraulic and particulate matter retention in a
3rd order Indiana stream. Journal of the North American Benthological Society
11:341-349.

Erman, D. C. and N. A. Errnan. 1984. The response of stream macroinvertebrates to
substrate size and heterogeneity. Hydrobiologia 108:75-82.

Evans, B. F., C. R. Townsend, and T. A. Crow]. 1993. Distribution and abundance of
coarse woody debris in some southern New Zealand streams from contrasting
forest catchments. New Zealand Journal of Marine and Freshwater Research
27:227-23 9.

Everett, R. A. and G. M. Ruiz. 1993. Coarse woody debris as a refuge from predation in
aquatic communities. Oecologia 93:475-486.

Farrand W. R. and D. L. Bell. 1984. Quaternary Geology of Southern Michigan.
University of Michigan, Ann Arbor.

Fausch, K. D. and T. G. Northcote. 1992. Large woody debris and salmonid habitat in a
small coastal British Columbia stream. Canadian Journal of Fisheries and Aquatic
Sciences 49:682-693.

200

 

F etherston, K. L., R. J. Naiman, and RE. Bilby. 1995. Large woody debris, physical
process, and riparian forest development montane river networks of the Paciﬁc
Northwest. Geomorphology 12: 133-144.

Franklin, J. F ., K. Cromack, and W. Denison. 1981. Ecological characteristics of old-
grth Douglas-fir forests. USDA Forest Service, Paciﬁc Northwest Forest and
Range Experiment Station. PNW-118, 1-48. Portland, OR.

Forrnan, R. T. T. and M. Godron. 1986. Landscape Ecology. John Wiley and Sons, New
York.

Gippel, C. J ., B. L. Finlayson, and I. C. O'Niell. 1996. Distribution and hydraulic
signiﬁcance of large woody debris in a lowland Australian river. Hydrobiologia
318:179-194.

Gordon, N.D., T.A. McMahon, and BL. Finlayson. 1992. Stream Hydrology. An
Introduction for Ecologists. John Wiley, Chichester.

Gore, J. A. 1982. Benthic invertebrate colonization: source-distance effects on
community composition. Hydrobiologia 94: 1 83-193.

Gorman, O. T. and J. R. Karr. 1978. Habitat structure and ﬁsh communities. Ecology
59:507-515.

Gregory, K. J. and R. J. Davis. 1992. Coarse woody debris in stream channels in relation
to river channel management in woodland areas. Regulated Rivers: Research and
Management 72117-136.

Gregory, K. J ., R. J. Davis, and S. Tooth. 1993. Spatial distribution of coarse woody
debris dams in the Lymington Basin, Hampshire UK. Geomorphology 6:207-224.

Gregory, K. J., A. M. Gurnell, and C. T. Hill. 1985. The permanence of debris dams
related to river channel processes. Hydrological Sciences Journal 30:371-381.

Gregory, K. J ., A. M Gurnell, C. T. Hill, and S. Tooth. 1994. Stability of the pool-riﬂle
sequence in changing river channels. Regulated Rivers: Research and
Management 9:35-43.

Gregory, S.V., F. J. Swanson, W. A. McKee, and K. W. Cummins. 1991. An ecosystem
perspective of riparian zones. BioScience 41:540-551.

Grimm, E. C. 1984. Fire and other factors controlling the Big Woods vegetation of
Minnesota in the mid-nineteenth century. Ecological Monographs 54:291-311.

201

 

Gurnell, A. M. and K. J. Gregory. 1987. Vegetation characteristics and the prediction of
runoff: analysis of an experiment in the New Forest, Hampshire. Hydrological
Processes 1:125-142.

Gurnell, A. M and K. J. Gregory. 1995. Interactions between semi-natural vegetation and
hydrogeomorphological processes. Biogeomorphology 13:49-69.

Gurnell, A. M, K. J. Gregory, and G. E. Petts. 1995. The role of coarse woody debris in
forest aquatic habitats: implications for management. Aquatic Conservation:
Marine and Freshwater Ecosystems 52143-166.

Gurtz, M. E. and J. B. Wallace. 1984. Substrate-mediated response of stream
invertebrates to disturbance. Ecology 65:1556-69.

Harmon, M. E., J. F. Franklin, G. W. Lienkaemper, K. Cromack, and K. W. Cummins.
1986. Ecology of coarse woody debris in temperate ecosystems. Advances in
Ecological Research 15: 133-302.

Hartman, G. F. 1965. The role of behavior in the ecology and interaction of
underyearling coho salmon (Oncorhynchus kisutch) and steelhead trout (Salmo
gairdneri). Journal of the Fisheries Research Board of Canada 22:1035-1081.

Hauer, F. R. 1989. Organic matter transport and retention in a blackwater stream
recovering from ﬂow augmentation and thermal discharge. Regulated Rivers:
Research and Management 4:371-380.

Hawkins, C. P., J. L Kershner, P. A Bisson, M. D. Bryant, L. M. Decker, S. V. Gregory,
D. A. McCullough, C. A. Overton, G. H. Reeves, R. J. Steedman, and M. K.
Young. 1993. A hierarchical approach to classifying stream habitat features.
Fisheries 18:3-12.

Hax, C. L. and S. W. Golladay. 1997. Macroinvertebrate colonization and bioﬁlm
development on leaves and wood in a boreal river. Freshwater Biology 29:79-87.

Hax, C. L. and S. W. Golladay. 1998. Flow disturbance of macroinvertebrates inhabiting
sediments and woody debris in a prairie stream. American Midland Naturalist
139:210-223.

Hicken, E. J. 1984. Vegetation and river channel dynamics. Canadian Geographer
28:111-126.

202

 

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

Hogan, D. L. 1986. Channel morphology of unlogged and logged debris torrented
streams in the Queen Charlotte Islands. 4, pp 1-94. In: British Columbia Ministry
of Forests and Lands. Land Management Report.

Hortle, K. G. and P. S. Lake. 1982. Macroinvertebrate assemblages in channelized and
unchannelized sections of the Bunyip River, Victoria. Australian Journal of
Marine and Freshwater Research 33:1071-1082.

Host, G. E. and K. S. Pregitzer. 1992. Geomorphic inﬂuences on ground-ﬂora and

overstory composition in upland forests of northwestern Lower Michigan.
Canadian Journal of Forest Research 22:1547-1555.

Hunter, C. J. 1991. Better Trout Habitat. Island Press, Washington, DC 320 pp.

Huryn, A. D. and J. B. Wallace. 1987. Local geomorphology as a determinant of
macrofaunal production in a mountain stream. Ecology 68:1932-1942.

Johnson, L. B., C. Richards, G. E. Host and J. W. Arthur. 1997. Landscape inﬂuences on
water chemistry in Midwestern stream ecosystems. Freshwater Biology 37: 193-
208.

Johnson, L. B. l999a. Channel, riparian, and landscape features as predictors of coarse
woody debris abundance in Midwestem streams. Chapter 2, In: Ecological
Implications of Coarse Woody Debris is Low Gradient, Midwestern Streams.
PhD Dissertation, Michigan State University, East Lansing, MI.

Johnson, L. B. 1999b. Coarse woody debris accumulations in low gradient streams:
relationships with local and landscape features. Chapter 3, In: Ecological
Implications of Coarse Woody Debris in Low Gradient Midwestern Streams.
PhD Dissertation, Michigan State University, East Lansing, MI.

Johnson, L.B. 1999c. Coarse Woody Debris Retention and Recruitment in Low Gradient,
Midwestern Watersheds. Chapter 4, In: Ecological Implications of Coarse
Woody Debris in Low Gradient Midwestern Streams. PhD Dissertation,
Michigan State University, East Lansing, MI.

203

Johnson, L.B. 1999d. Macroinvertebrate Community Structure and Function Associated
with Coarse Woody Debris in Low Gradient, Midwestern Streams. Chapter 5,
In: Ecological Implications of Coarse Woody Debris in Low Gradient Midwestern
Streams. PhD Dissertation, Michigan State University, East Lansing, MI.

Jones, J. B. and L. A. Smock. 1991. Transport and retention of particulate organic
matter in two low-gradient headwater streams. Journal of the North American
Benthological Society 10:1 15-126.

Keams, B.L., J .R. Karr, and SA. Ahlsteadt. 1992. Aquatic invertebrate assemblages:
spatial and temporal differences among sampling protocols. Journal of the North
American Benthological Society 11:377-390.

Keller, E. A. and F. J. Swanson. 1979. Effects of large organic debris on channel form
and ﬂuvial process. Earth Surface Processes 4:361-380.

Keller, E. A. and T. Tally. 1979. Effects of large organic debris on channel form and
ﬂuvial processes in the coastal redwood environment. In: D. D. Rhodes and G. P.
Williams (ed.). Adjustments of the Fluvial System, Proceedings of the 10th
Annual Geomorphology Symposium, SUNY Bingharnpton, N.Y. Kendall/Hunt
Publishing Co.

Lehtinen, R. M., N. D. Mundahl, and J. C. Madejczyk. 1997. Autumn use of woody
snags by ﬁshes in backwater and channel border habitats of a large river.
Environmental Biology of Fishes 49:7-19.

Lenat, D. R. 1988. Water quality assessment using qualitative a collection method for
benthic macroinvertebrates. Journal of the North American Benthological Society
71222-233.

Lienkaemper, G. W. and F. J. Swanson. 1987. Dynamics of large woody debris in
streams in old-growth Douglas-ﬁr forests. Canadian Journal of Forest Research
17:150-156.

Likens, G. E. and R. E. Bilby. 1982. Development, maintenance and role of organic
debris dams in New England streams. Pp 122 - 129, In: F. J. Swanson, R. J.
Janda, T. Dunne, and D. N. Swanston, (eds.). Sediment budgets and routing in
forested drainage basins. U.S.D.A. Forest Service Research Paper PNW-141.

204

MacDonald, A., E. A. Keller, and T. Tally. 1982. The role of large organic debris on
stream channels draining redwood forests in northwestern California’. Pp. 226-
245, In: D. K. Harden, D.C. Marran, and A. MacDonald (eds.) Friends of the
pleistocene Paciﬁc cell ﬁeld trip guidebook. Late Cenozoic History and Forestry
Geomorphology of Humbold Co., California.

MacArthur, R. H. and E. O. Wilson. 1967. The Theory of Island Biogeography.
Princeton University Press, Princeton, NJ.

MacDonald, A. and E. A. Keller. 1987. Stream channel response to the removal of large
woody debris, Larry Damm Creek, northwestern California. Intemational
Association of hydrological Sciences Publication 165:405-406.

Malanson, G. P. and D. R. Butler. 1990. Woody debris, sediment, and riparian
vegetation of a subalpine river, Montana, USA. Arctic and Alpine Research
22:183-194.

Marston, R. A. 1982. The geomorphic signiﬁcance of log steps in forest streams. Annals
of the Association of American Geographers 72:99-108. Maser, C. and J. R.
Sedell. 1994. From the Forest to the Sea. St. Lucie Press. Delray Beach. 200 pp.

Maser, C. and J. R. Sedell. 1994. From the Forest to the Sea. St. Lucie Press. Delray
Beach. 200 pp.

McDade, M. H., F. J. Swanson, W. A. McKee, J. F. Franklin, and J. VanSickle. 1990.
Source distances for coarse woody debris entering small streams in western
Oregon and Washington. Canadian Journal of Forest Research 20:326-330.

McKenney, R., R. B. Jacobson, and R. C. Wertheimer. 1995. Woody vegetation and
channel morphogenesis in low-gradient gravel-bed streams in the Ozark Plateaus,
Missouri and Arkansas. Geomorphology 13:175-198.

McMahon, T. E. and G. F. Hartman. 1989. Inﬂuence of cover complexity and current
velocity in winter habitat use by juvenile coho salmon (Oncorhynchus kisutch).
Canadian Journal of Fisheries and Aquatic Sciences 46: 1 551-1557.

Merritt, R.W. and K.W. Cummins. 1996. An Introduction to the Aquatic Insects of
North America. 3“1 edition. Kendall-Hunt Publisher.

Minshall, G. W., J. T. Brook, and T. W. LaPoint. 1982. Characterization and dynamics
of benthic organic matter and invertebrate functional feeding group relationships
in the Upper'Salmon River, Idaho (USA). lntemationale Revue der gesamten
Hydrobiologie 67:793-820.

205

 

Minshall, G.W., and J. N. Minshall. 1977. Microdistribution of benthic invertebrates in a
Rocky Mountain, (U.S.A.) stream. Hydrobiologia 55:231-249.

Minshall, G.W., R.C. Petersen, T.L. Bott, C.E. Cushing, K.W. Cummins, R.L. Vannote,
and J .R. Sedell. 1992. Stream ecosystem dynamics of the Salmon river, Idaho: an
8‘h order system. Journal of the North American Benthological Society 11:111-
137.

Molles, M.C., Jr. 1982. Trichopteran communities of streams associated with aspen and
conifer forests: long-term structural changes. Ecology 63:1-6.

Mosley, M. P. 1981. The inﬂuence of organic debris on channel morphology and
bedload transport in a New Zealand stream. Earth Surface Processes and
Landforms 62571-579.

Murgatroyd, A. L. and J. L. Teman. 1983. The impact of afforestation on stream bank
erosion and channel form. Earth Surface Processes and Landforms 82357-369.

Murphy, M. L., J. Heifetz, S. W. Johnson, K. V. Koski, and J. F. Thedinga. 1986. Effects
of clearcut logging with and without buffer strips on juvenile salmonids in
Alaskan streams. Canadian Journal of Fisheries and Aquatic Sciences 43: 1521-
1533.

Murphy, M. L. and K. V. Koski. 1989. Input and depletion of woody debris in Alaska
streams and implications for strearnside management. North American Journal of
Fisheries Management 92427-436.

Myers, T. and S. Swanson. 1997. Variability of pool characteristics with pool type and
formative feature on small Great Basin rangeland streams. Journal of Hydrology
201:62-81.

Naiman, R.J. 1982. Characteristics of sediment and organic carbon export from pristine
boreal forest catchments. Canadian Journal of Fisheries and Aquatic Sciences
39:1699-1718.

Naiman, R. J. and J. R. Sedell. 1979. Benthic organic matter as a function of stream
order in Oregon. Archiv ﬁir Hydrobiologie 87:404-422.

Naiman, R. J ., J. M. Melillo, and J. E. Hobbie. 1986. Ecosystem alteration of boreal
forest streams by beaver (Castor canadensis). Ecology 67:1254-1269.

206

Nakamura, F. and F. J. Swanson. 1993. Effects of coarse woody debris on morphology
and sediment storage of a mountain stream system in western Oregon. Earth
Surface Processes and Landforms 18:43-61.

Newbold, J. D., D. C. Erman, and K. B. Roby. 1980. Effects of logging on
macroinvertebrates in streams with and without buffer strips. Canadian Journal of
Fisheries and Aquatic Sciences 37:1076-1085.

Newbold, J. D., R.V. O’Neill J. W. Elwood, and W. VanWinkle. 1982. Nutrient spiraling
in streams: implications for nutrient limitation and invertebrate activity. Am. Nat.
120:628-652.

Nilsen, H. C. and R. W. Larimore. 1973. Establishment of invertebrate communities on
log substrates in the Kaskaskia River, Illinois. Ecology 54:366-374.

Naiman, R. J. and J. R. Sedell. 1979. Benthic organic matter as a function of stream
order in Oregon. Archiv ﬁir Hydrobiologie 87 1404-422.

O'Connor, N. A. 1991. The effects of habitat complexity on the macroinvertebrates
colonising wood substrates in a lowland stream. Oecologia 85:504-512.

O'Connor, N. A. 1992. Quantiﬁcation of submerged wood in a lowland Australian
stream system. Freshwater Biology 27:387-395.

Odland, J. 1988. Spatial Autocorrelation. Sage Publications, Newbury Park, Ca. 87 pp.

Omernik J. M., and A. L. Gallant. 1986. Ecoregions of the Upper Midwest States.
EPA/600/3-88-O37, 56 pp.

Osborne, L. L., B. Dickson, M. Ebbers, R. Ford, J. Lyons, D. Kline, E. Rankin, D. Ross,
R. Sauer, P. Seelback, C. Speas, T. Stefanavage, J. Waite, and S. Walker. 1991.
Stream habitat assessment programs in states of the AF S North Central Division.
Fisheries 16:28-35.

Phillips, E. C. and R. V. Kilambi. 1994a. Use of coarse woody debris by Diptera in
Ozark streams, Arkansas. Journal of the North American Benthological Society
13:151-159.

Phillips, E. C. and R. V. Kilambi. 1994b. Utilization of coarse woody debris by

Ephemeroptera in three Ozark streams of Arkanses. The Southwestern Naturalist
39:58-62.

207

 

Piégay, H. and A. M Gumell. 1997. Large woody debris and river geomorphological
pattern: examples from SE. France and S. England. Geomorphology 19299-116.

Platts, W. S., W. F. Megahan and G. W. Minshall. 1983. Methods for evaluating stream,
riparian, and biotic conditions. INT-138, 1. Ogden, UT, U.S.D.A. Forest Service,
Intermountain Forest and Range Experiment Station. General Technical Report.

Poff, N. L. and J. V. Ward. 1989. Implications of stream ﬂow variability and
predictability for lotic community structure: a regional analysis of strearnﬂow
patterns. Canadian Journal of Fisheries and Aquatic Sciences 46:1805-1817.

Potts, D. F. and B. K. M. Anderson. 1990. Organic debris and the management of small
stream channels. Western Journal of Applied Forestry 5225-3 8.

Quinlan, M. 1997. Lumbering in Michigan. Available http://www.sos.state.mi.us/
history/teclrstuf/lumber/Lumhinfgﬁml, November 13, 1997.

Raikow, D. F., S. A. Grubbs, and K. W. Cummins. 1995. Debris dam dynamics and
coarse particulate organic matter retention in an Appalachian Mountain stream.
Journal of the North American Benthological Society 14:535-546.

Ralph, S. C., G. C. Poole, L. L. Conquest, and R. J. Naiman. 1994. Stream channel
morphology and woody debris in logged and unlogged basins of Western
Washington. Canadian Journal of Fisheries and Aquatic Sciences 51:37-51.

Richards, K. 1982. Rivers. Form and Process in Alluvial Channels. Methuen, London.

Richards, C., L. B. Johnson, and G. E. Host. 1996. Landscape-scale inﬂuences on stream
habitats and biota. Canadian Journal of Fisheries and Aquatic Sciences 53(Suppl
1):295-311.

Richards, C., R. J. Haro, L. B. Johnson, and G. E. Host. 1997. Catchment and reach-
scale proporties as indicators of macroinvertebrate species traits. Freshwater
Biology 37:219-230.

Richmond, A. D. and K. D. Fausch. 1995. Characteristics and function of large woody
debris in subalpine Rocky Mountain streams in northern Colorado. Canadian
Journal of Fisheries and Aquatic Sciences 52:1789-1802.

Robison, E.G., and R. L. Beschta. 1989. Coarse woody debris and channel morphology
of low-gradient streams in southeast Alaska, U.S.A. Pp. 371-380, In: W.W.
Woessner, and DP. Potts, (ed.) Symposium on Headwaters Hydrology. Missoula.
American Water Resources Association. Bethesda, MD.

208

Robison, E. G. and R. L. Beschta. 1990. Coarse woody debris and channel morphology
interactions for undisturbed streams in southeast Alaska. Earth Surface Processes
and Landforms 15:149-1565.

Ryan, R. D. Erickson, and R. DeYoung. 1999. Understanding farmer’s motivations for
improving ecological health. Abstract. Pg 132, In: Landscape Ecology. The
Science and the Action. 5th World Congress of the International Association for
Landscape Ecology, Snowmass Village, CO, USA. July 29- August 4, 1999.

SAS Institute Inc. 1989. SAS/STAT User's Guide Version 6, 4th Ed., Volume 2, Cary,
NC. SAS Institute Inc. 846pp.

Sedell, J. R., P. A. Bisson, F. J. Swanson, and S. V. Gregory. 1988. What we know
about large trees that fall into streams and rivers. Pp 47-81, In: C. Maser, R. F.
Tarrent, and J. F. Franklin (eds.) From the Forest to the Seas: A Story of Fallen
Trees. USDA Forest Service, Paciﬁc Northwest Station. General Technical
Report. Portland, Oregon, PNW-GTR-229.PNW-GTR-229.

Sedell, J. R. , F. J. Swanson, and S.V. Gregory. 1984. Evaluating ﬁsh response to woody
debris. pp 191-221. In: T. J. Hassler (ed.). Proceedings of the Paciﬁc Northwest
Streams Habitat Management Workshop. American Fisheries Society, Humbolt
State University, Arcata.

Shearer, C. A. and J. Webster. 1991. Aquatic Hyphomycete communities in the River
Teign. IV. Twig colonization. Mycological Research 95 2413-420.

Shields, F. D. and R. H. Smith. 1992. Effects of large woody debris removal on physical
characteristics of a sand-bed river. Aquatic Conservation: Marine and Freshwater
Ecosystems 2:145-163.

Shreve, R.L. 1966. Statistical law of stream numbers. Journal of Geology 74:17-37.

Smith, R. D, R. C. Sidle, and P. E. Porter. 1993. Effects on bedload transport of
experimental removal of woody debris from a forest gravel bed stream. Earth
Surface Processes and Landforms 18:455-468.

Smock, L. A., E. Gilinsky, and D. L. Stonebumer. 1985. Macroinvertebrate production
in a southeastern United States blackwater stream. Ecology 66:1491-1503.

Smock, L. A., G. M. Metzler, and J. E. Gladden. 1989. The role of debris dams in the
structure and functioning of low gradient headwater streams. Ecology 70:764-

775.

209

Southwood, T. R. E. 1977. Habitat, the template for ecological strategies. Journal of
Animal Ecology 46:337-365.

Spencer, T., 1. Douglas, T. Greeer, and W. Sinun. 1990. Vegetation and ﬂuvial
geomorphic processes in south-east Asian tropical rainforests. Pp. 451-470, In:
Thornes, J .B. (Ed.) Vegetation and Erosion. Wiley.

Strahler, A.N. 1964. Quantitative geomorphology of drainage basins and channel
networks. Sec 4.40-4.74, In: V.T. Chow (ed.) Handbook of Applied Hydrology.
McGraw Hill, NY.

Stratzner, B. and B. Higler. 1986. Stream hydraulics as a major determinant of benthic
invertebrate zonation patterns. Freshwater Biology 16:127-139.

Swanson, F. J ., G. W. Lienkaemper, and J. R. Sedell. 1976. History, physical effects, and
management implications of large woody debris in western Oregon streams. Gen.
Tech. Rep. PNW-S6. USDA Forest Service, Paciﬁc Northwest Forest and Range
Experiment Station. 15 pp.

Swanson, F. J. and G. W. Lienkaemper. 1978. Physical consequences of large organic
matter in Paciﬁc northwest streams. PNW-69, pp 1-12, In: USDA Forest Service,
Paciﬁc Northwest Forest and Range Experiment Station. General Technical
Report. Portland, Oregon.

ter Braak, C. J. F. and P. Smilauer. 1998. CANOCO Reference Manual and User’s Guide
to Canoco for Windows: Software for Canonical Community Ordination, version
4. Centre for Biometry, Microcomputer Power. Ithaca, NY. 352 pp.

ter Braak, C. J. F. 1995. Ordination. pp. 212. In: R.H. G. Jongman, C.J. F. ter Braak, and
O. F. R. van Togeren. Data Analysis in Community and Landscape Ecology,
Cambridge University Press, Cambridge. Agricultural Mathematics Group,
Wageningen, The Netherlands.

Triska, F. J. 1984. Role of wood debris in modifying channel geomorphology and
riparian areas of a large lowland river under pristine conditions: A historical case
study. Verhandlungen der lntemationalen Vereingung fur theoretische und
angewandte Limnologie 2221876-1 892.

Triska, F. J. and K. Cromack, 1980. The role of wood debris in forests and streams. in:
Waring, R. H. (editor). Forests: Fresh Perspectives From Ecosystem Analysis.
Oregon State University, Corvallis, Oregon

210

 

Triska, F. J ., J. R. Sedell, and S. V. Gregory. 1982. Coniferous forest streams. pp 292-
332. In: R. L. Edmonds (Ed.) Analysis of coniferous forest ecosystems in the
Western United States. US/IBP Synthesis Ser. 14. Stroudsburg, PA. Hutchinson
Ross Publishing Co.

Trotter, E. H. 1990. Woody debris, forest-stream succession, and catchment
geomorphology. Journal of the North American Benthological Society 9:141-
156.

Tschaplinski, P. J. and G. F. Hartman. 1983. Winter distribution of juvenile coho
salmon (Oncrhynchus kisutch) before and after logging in Carnation Creek,
British Columbia, and some implications for overwinter survival. Canadian
Journal of Fisheries and Aquatic Sciences 40:452-461.

Wallace, J. B. and A. C. Benke. 1984. Quantiﬁcation of wood habitat in subtropical
coastal plain streams. Canadian Journal of Fisheries and Aquatic Sciences
41:1643-1652.

Wallace, J. B., J. R. Webster, and J. L. Meyer. 1995. Inﬂuence of log additions on
physical and biotic characteristics of a mountain stream. Canadian Journal of
Fisheries and Aquatic Sciences 52:2120-2137.

Ward, G. M. and N. G. Aumen. 1987. Woody debris as a source of ﬁne particulate
organic matter in coniferous forest stream ecosystems. Canadian Journal of
Fisheries and Aquatic Sciences 43:1635-1642.

Webster, J. R., A. P. Covich, J. L. Tank, and T. V. Crockett. 1994. Retention of coarse
organic particles in streams in southern Appalachian Mountains. Journal of the
North American Benthological Society 13:140-150.

Wiley, M. J ., S. L. Kohler, and P. W. Seelbach. 1997. Reconciling landscape and local
views of aquatic communities: lessons from Michigan trout streams. Freshwater
Biology 37:133-148.

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

Zimmerman, R. S., J. C. Goodlett, and G. H. Comer. 1967. The inﬂuence of vegetation

on channel form of small streams. Symposium on River Morphology,
International Association of Scientiﬁc Hydrology, Publication 75:255-275.

211