INFLUENCES OF TIMBER MANAGEMENT AND NATURAL LANDSCAPE FACTORS
ON ANADROMOUS STREAMS OF SOUTHEASTERN ALASKA: RELATING LOCAL
AND CATCHMENT FACTORS TO AQUATIC HABITAT
By
Jared A. Ross

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

ABSTRACT
INFLUENCES OF TIMBER MANAGEMENT AND NATURAL LANDSCAPE FACTORS
ON ANADROMOUS STREAMS OF SOUTHEASTERN ALASKA: RELATING LOCAL
AND CATCHMENT FACTORS TO AQUATIC HABITAT
By
Jared A. Ross
Timber harvest is the primary disturbance to streams of Southeast Alaska. Timber management
pre-1980 occurred with few rules restricting harvest near streams, leading to degraded habitat in
streams near harvested areas. In response, regulations were established in the late 1970’s to
protect both terrestrial and aquatic resources. Initial best management practices (BMPs)
restricted use of heavy equipment and regulated harvest practices, though did not define how
close harvest could occur near streams. Trees would be taken directly up to banks, potentially
decreasing future sources of large wood and increasing bank erosion. Further, both historical
and current harvest through watersheds can have additional effects on streams, including altering
catchment hydrology and increasing streambed sedimentation. Understanding how current and
historical timber harvest may affect the region’s streams is critical for their successful
management. In this study, I characterize current condition of streams harvested under the initial
BMPs and explore relationships between natural and timber harvest-related landscape factors at
local and catchment scales, with results indicating relationships between landscape factors and
stream habitat at both scales. Local channel morphology and riparian characteristics were
strongly associated with habitat, with smaller median particle sizes detected in streams with
harvested vs. unharvested riparian zones. At the catchment scale, timber harvest factors
explained more variation than natural landscape factors in measures of large wood abundance
and distribution. Results from this work will aid managers of both timber and fisheries industries
in protecting and restoring streams and forests of the region.

To all my human and non-human friends and family.

iii

ACKNOWLEDGEMENTS
This research would not have been possible without the financial support of Sealaska
Corporation, Schrems West Michigan Chapter of Trout Unlimited, The National Fish Habitat
Partnership, Dr. William Taylor, Mark Rey, and Rick Harris. I would also like to thank John and
Marilyn Hillman for their hospitality and for providing so much advice and historical knowledge
on streams and timber management in the region.

I would like to thank my advisor Dr. Dana Infante for all of your support and assistance over the
years. I would like to thank Dr. Doug Martin for his role as a committee member and for giving
me multiple opportunities to work with him collecting stream data in Alaska. I would also like
to thank Mark Rey for his role as a committee member and for linking me in with Sealaska
Corporation. Emil Tucker and John McDonell from the Tongass National Forest provided data
that made much of this work possible. Numerous data were also collected from the Southeast
Alaska GIS Library. Dr. Rich Grotefendt offered many hours of support in study design and
identifying suitable streams to sample. Additionally I want to give my deepest gratitude to
numerous individuals who have assisted in the completion of field work and GIS analysis as well
as providing personal and professional support: Kyle Herreman, Arthur Cooper, Isaac Cadiente,
Allyson Hughes, Jo Ann Patterson, Ron Wolfe, Brian Kleinhenz, The ALE Lab, and the MIRTH
Lab. A big thanks to my Mother, Father and entire family, especially the love and support
provided by Elizabeth Throckmorton and all of her furry friends.

iv

TABLE OF CONTENTS
LIST OF TABLES ....................................................................................................................... viii
LIST OF FIGURES .........................................................................................................................x
INTRODUCTION ...........................................................................................................................1
LITERATURE CITED ........................................................................................................6
CHAPTER 1
THE INFLUENCE OF GEOMORPHIC, RIPARIAN, AND TIMBER HARVEST FACTORS
ON AQUATIC HABITAT IN ANADROMOUS STREAMS OF SOUTHEAST ALASKA
Abstract ................................................................................................................................8
Introduction ..........................................................................................................................9
Methods
Study region ...........................................................................................................13
Site selection ..........................................................................................................14
Field data collection ...............................................................................................16
Geomorphic and riparian factors ...............................................................17
Channel morphology......................................................................17
Riparian condition ..........................................................................18
Aquatic habitat factors ...............................................................................18
Habitat complexity .........................................................................18
Substrate.........................................................................................19
Large wood ....................................................................................19
Data analyses .........................................................................................................20
Data reduction ............................................................................................20
Principal components analysis ...................................................................21
Understanding influences of geomorphic and riparian factors
on aquatic habitat .......................................................................................22
Understanding influences of timber harvest on aquatic habitat .................23
Results
Geomorphic, riparian, and timber harvest characteristics .....................................23
Regional trends in timber harvest ..............................................................24
Geomorphic factors ....................................................................................24
Riparian factors ..........................................................................................24
Aquatic habitat characteristics ...............................................................................25
Habitat complexity, substrate, and large wood ..........................................25
Understanding influences of natural factors on aquatic habitat .............................27
Principal components analysis ...................................................................27
Stepwise multiple regressions ....................................................................28
Understanding influences of timber harvest on aquatic habitat .............................28
Cumulative frequencies .............................................................................28
Discriminant function analysis ..................................................................29
Analysis of Covariance (ANCOVA) .........................................................29
Discussion
v

Geomorphic influences on aquatic habitat .............................................................31
Pool characteristics ....................................................................................31
Substrate characteristics .............................................................................32
Large wood characteristics ........................................................................32
Riparian influences on aquatic habitat ...................................................................33
Timber harvest influences on aquatic habitat ........................................................34
Management implications
Stream size and channel type constraints...............................................................37
Legacy of actions on the landscape .......................................................................38
Riparian stand composition........................................................................38
Stream substrate composition ....................................................................39
Silviculture as a tool to manage for future large wood ..........................................39
APPENDIX ........................................................................................................................41
LITERATURE CITED ......................................................................................................79
CHAPTER 2
EFFECTS OF LANDSCAPE FACTORS ON LARGE WOOD IN STREAMS DRAINING
SECOND GROWTH WATERSHEDS OF SOUTHEAST ALASKA: COMPARING
NATURAL AND TIMBER HARVEST-RELATED INFLUENCES
Abstract ..............................................................................................................................88
Introduction ........................................................................................................................89
Methods
Study area, catchment creation, and landscape data ..............................................93
Study site description .............................................................................................94
Timber harvest data................................................................................................94
Large wood data .....................................................................................................95
Statistical methods .................................................................................................96
Data preparation .........................................................................................96
Best subsets multiple regression ................................................................98
Constrained ordination analysis .................................................................98
Results
Regional trends of natural landscape factors in Southeast Alaska watersheds ......99
Trends in natural landscape factors of study streams ..........................................100
Trends in timber harvest-related factors of study streams ...................................100
Explaining variation in large wood habitat factors using natural and timber
harvest related landscape factors .........................................................................101
Average large wood length ......................................................................101
Large wood recruits located within the active channel vs. above the
channel .....................................................................................................102
Medium-sized large wood density ...........................................................102
Large and extra-large-sized large wood density ......................................103
Assessing the relative importance of natural vs. harvest-related landscape factors
in explaining variation in large wood habitat factors...........................................103
Discussion
Natural landscape factors .....................................................................................105
Catchment area.........................................................................................105
vi

Catchment slope .......................................................................................106
Deciduous forests .....................................................................................107
Forested wetlands.....................................................................................108
Timber harvest-related factors .............................................................................109
Historical timber harvest (1980-1990) .....................................................109
Current timber harvest (post-1990)..........................................................113
Logging road and stream crossings..........................................................114
Comparing landscape influences on large wood habitat factors ..........................115
Management implications
Landscape considerations in managing the region’s streams ..............................116
Future research .....................................................................................................117
APPENDIX ......................................................................................................................119
LITERATURE CITED ....................................................................................................131
CHAPTER 3
CONCLUSIONS: TIMBER HARVEST, NATURAL LANDSCAPE FACTORS, AND
RELATIONSHIPS WITH FLUVIAL HABITAT; FROM REACHES TO WATERSHEDS OF
SOUTHEASTERN ALASKA – SUMMARY OF FINDINGS, MANAGEMENT
IMPLICATIONS, AND STUDY LIMITATIONS
Principal findings
Chapter 1: Aquatic habitat and influences from reach scale geomorphic, riparian,
and timber harvest factors of Chichagof and Prince of Wales Islands, Alaska,
U.S.A....................................................................................................................139
Chapter 2: Large wood and influences from the landscape – natural and timber
harvest factors ......................................................................................................141
Management implications
Natural landscape considerations .........................................................................144
Timber harvest considerations .............................................................................146
Study limitations ..............................................................................................................148
LITERATURE CITED ....................................................................................................152

vii

LIST OF TABLES
Table 1.1
Pearson's correlations among geomorphic, riparian, and aquatic habitat variables
used in analyses. An asterisk next to correlation coefficient indicates significance at a 0.05 level
while two asterisks indicate significance at a 0.01 level. Bold variable codes indicate a
transformed variable (See Methods for specific transformations and Table 1.2 for variable
codes).............. ...............................................................................................................................42
Table 1.2
Mean, minimum, maximum, and standard deviation (SD) of calculated variables
used to describe the current aquatic and riparian habitat condition of study sites .........................45
Table 1.3
Pearson's correlations between key response variables used in comparison
between harvested and unharvested streams. An asterisk next to a correlation coefficient
indicates significance at the 0.05 level and two asterisks indicates significance at 0.01 level.
Width to depth ratio (WD) is the ratio between bankfull width and bankfull depth, average pool
area (APA) was in m², pool spacing (SPACE) is measured in # channel widths per pool, and pool
length (RPL) is the proportion of reach length classified as pool. All other variable codes can be
found in Table 1.2 ..........................................................................................................................46
Table 1.4
Total area harvested and percent of total land area harvested for all of Sealaska
owned land organized for three time periods. Time periods correspond to changes to Alaska's
forest resource regulations .............................................................................................................47
Table 1.5
Mean and standard deviation (SD) of study reaches draining land owned by
Sealaska Corporation. Sites are grouped by gradient as either low (< 2%) or moderate gradient
(2-6 %)............................. ..............................................................................................................48
Table 1.6
Results of multiple linear regressions predicting each aquatic habitat variable for
28 streams on Sealaska timberlands. Independent variables are listed in order of their influence
on the model, based on standardized coefficients, which are located in parentheses. Sig. indicates
the p-value for each hypothesis test on the coefficient of the corresponding term in the linear
model. Proportion of variance explained is indicated by R² for one variable models and R²adj for
two variable models. A variance inflation factor value of 4 (VIF< 4) was used to ensure little
multicollinearity .............................................................................................................................49
Table 1.7
Classification results organized by count and percent correctly classified from the
discriminant function analysis for both known (Original) and unknown membership (Crossvalidated)........................................................................................................................................50
Table 1.8
ANCOVA adjusted means and standard errors of variables reported for both low
and moderate gradient reaches. Differences are significant (P < 0.05) when the 95% confidence
intervals do not include 0 and an asterisk denotes variables with significant differences .............51
Table 1.9
ANCOVA adjusted means and standard errors of variables reported for both
unharvested and harvested reaches. Differences are significant (P < 0.05) when the 95%
confidence intervals do not include 0 and an asterisk denotes variables with significant
differences....... ...............................................................................................................................52

viii

Table 1.10
Large wood density (pieces/km) in streams of Southeast Alaska organized by
channel type....... ............................................................................................................................53
Table 1.11
Data characterizing aquatic and riparian habitat in and along 28 study sites on
Sealaska-owned land ......................................................................................................................54
Table 1.12
Data characterizing habitat in streams of the Tongass National Forest having
unharvested riparian zones .............................................................................................................58
Table 2.1
Mean, min, max, and standard deviation (SD) of natural landscape variables for
28 sampled stream reaches draining second growth watersheds .................................................120
Table 2.2
Mean, min, max, and standard deviation (SD) of timber harvest-related variables
for 28 sampled stream reaches draining second growth watersheds ...........................................121
Table 2.3
Pearson's correlations between timber harvest-related and other landscape factors
used in analyses. An asterisk indicates significance at a 0.05 level ............................................122
Table 2.4
Natural landscape factors for streams in Southeast Alaska summarized for local
and network catchments ...............................................................................................................123
Table 2.5
Results from best subsets regression for LW length. Models are in order of most
variance explained, predictor variables are ordered by their influence on the model based on
standardized beta coefficients, and ∆ AIC represents the difference when compared to catchment
area alone........................... ..........................................................................................................124
Table 2.6
Results from best subsets regression for the ratio of LW recruits that are in:above
the channel. Models are in order of most variance explained, predictor variables are ordered by
their influence on the model based on standardized beta coefficients, and ∆ AIC represents the
difference when compared to catchment area alone ....................................................................125
Table 2.7
Results from best subsets regression for medium sized LW density. Models are in
order of most variance explained, predictor variables are ordered by their influence on the model
based on standardized beta coefficients, and ∆ AIC represents the difference when compared to
catchment area alone ....................................................................................................................126
Table 2.8
Results from best subsets regression for large and extra-large sized LW density.
Models are in order of most variance explained, predictor variables are ordered by their influence
on the model based on standardized beta coefficients, and ∆ AIC represents the difference when
compared to catchment area alone ...............................................................................................127

ix

LIST OF FIGURES
Figure 1.1.
Maps of (A) Chichagof Island and (B) Prince of Wales Island showing locations
of stream reaches sampled (n=28) on timberlands owned by Sealaska Corporation during the
summer of 2011 .............................................................................................................................61
Figure 1.2
Cumulative frequencies of earliest Sealaska Corporation harvest year organized
by number of banks that experienced riparian timber harvest .......................................................62
Figure 1.3
Frequency of 876 riparian trees organized by size class and vegetation type.
Black bars represent coniferous trees, light gray bars represent deciduous trees, and dark gray
bars represent standing snags .........................................................................................................63
Figure 1.4
Relative amount (%) of riparian trees organized by low vs. moderate gradient
streams. Black bars represent small (10-30 cm dbh), light gray bars represent medium (30-60 cm
dbh), and dark gray bars represent large and extra-large (60-90 & > 90 cm dbh) riparian trees...64
Figure 1.5
Relative amount (%) of riparian trees organized by number of banks harvested.
Black bars represent small (10-30 cm dbh), light gray bars represent medium (30-60 cm dbh),
and dark gray bars represent large and extra-large (60-90 & > 90 cm dbh) riparian trees ............65
Figure 1.6
Relative amount (%) of riparian trees organized by low vs. moderate gradient.
Black bars represent conifers, light gray bars represent deciduous, and dark gray bars represent
snags................... ............................................................................................................................66
Figure 1.7
Relative amount (%) of riparian trees organized by number of banks harvested.
Black bars represent conifers, light gray bars represent deciduous, and dark gray bars represent
snags................. ..............................................................................................................................67
Figure 1.8
Percent distribution frequencies of (a) diameter and (b) length of 2237 LW pieces
measured across all 28 Sealaska stream reaches (minimum diameter ≥ 10 cm and ≥ 2 m length).
X-axis values equal to minimum length requirement for 3 m length classes ................................68
Figure 1.9 Relative amount (%) of large wood pieces organized by low vs. moderate gradient
streams. Black bars represent small (10-30 cm dbh), light gray bars represent medium (30-60 cm
dbh), and dark gray bars represent large and extra-large (60-90 & > 90 cm dbh) pieces of wood69
Figure 1.10. Relative amount (%) of large wood pieces organized by number of banks
harvested. Black bars represent small (10-30 cm dbh), light gray bars represent medium (30-60
cm dbh), and dark gray bars represent large and extra-large (60-90 & > 90 cm dbh) pieces of
wood...............................................................................................................................................70
Figure 1.11. Box plots showing decay class of LW recruits organized by size class. See
Methods for decay class and size class descriptions ......................................................................71
Figure 1.12. Frequency of LW recruits organized by decay class. Darker bars represent
recruits that functioned while lighter bars represent recruits that did not function .......................72

x

Figure 1.13. Frequency of LW recruits organized by size class. Darker bars represent recruits
that function while lighter bars represent recruits that did not function ........................................73
Figure 1.14. Results of the PCA of 17 habitat measures. The four components explained
72.93% of the variation in the habitat data. Axis 1 explained 30.48%, Axis 2, 16.60%, Axis 3,
13.44%, and Axis 4, 12.42%. The y-axes show the weights on each variable for the individual
components. Black bars indicate variables with weights with an absolute value greater than 0.55.
Variable descriptions are included in Table 1.2.............................................................................74
Figure 1.15. Cumulative percent distributions of (a) D50, (b) RPD, (c) POOLKM, and (d)
RPA. Black triangles represent unharvested streams and white boxes represent harvested
streams............. ..............................................................................................................................77
Figure 2.1.
National Land Cover Database (2001) for Southeast Alaska. For interpretation of
the references to color in this and all other figures, the reader is referred to the electronic version
of this thesis ..............................................................................................................................128
Figure 2.2.
Results of variance partitioning from the CCA. A total of 44% of variation in LW
habitat factors is partitioned into timber harvest-related factors in gray, natural factors in black,
and the interaction in white ..........................................................................................................129
Figure 2.3
Graph of CCA axis one vs. CCA axis 2. Black arrows are natural landscape
factors, red arrows are timber harvest-related factors, and black squares are LW habitat factors.
LWLENGTH is average LW length, MEDLWKM and LXLWKM are densities of medium sized
and large and extra-large sized LW, while INABOVE is the ratio of LW recruits that are located
within vs. positioned above the active channel ............................................................................130

xi

INTRODUCTION
In the harsh, rugged, and remote landscape of Southeast Alaska, anadromous salmonids
and old growth forests have supported human populations for thousands of years. Abundant
wood resources allowed for houses to be built and for people to settle in permanent villages. A
seemingly inexhaustible supply of fish, wildlife, berries, mushrooms, and other wild edibles also
supported this subsistence way of life for the earliest Alaskans, and rural populations today still
rely on subsistence living. The economic base in Southeast Alaska has also historically revolved
around natural resources, including the commercial fishing and timber industries. Currently, it is
estimated that salmon provide nearly $1B annually and are linked to 10% of the jobs in the
region, while the timber industry is responsible for another 2% of jobs (TCW Economics 2010).
Salmon and timber, while crucial for humans in the region, also share a mutually
beneficial relationship with one another. For longer than humans have been in North America,
the coastal temperate rainforests of Southeast Alaska have relied upon marine subsidies as mass
migrations of anadromous fishes move into freshwater to spawn and die annually, leaving
gametes, excrements, and carcasses to be transported, used, and cycled throughout the regions’
forest ecosystems. Streams and the fishes they support in turn are influenced by the forested
landscapes within river basins of natal streams as they regulate stream flow by intercepting
precipitation and supporting increases in shallow groundwater, slowing delivery of precipitation
to streams and encouraging stable stream flows. Natural vegetation also functions to stabilize
steep slopes and reduces the risk of landslides and mass wasting events which can lead to excess
sediment being delivered to stream channels. Forested riparian zones also provide allochthonous
inputs in the form of leaf litter and terrestrial insects to streams. Riparian vegetation in particular

1

helps to stabilize stream banks and regulate light input and temperature fluctuations through
shading.
The immediate riparian zone and the vegetation it supports is the source of the majority
of large wood to streams, and this input is of particular importance to salmonids for multiple
reasons. First, large wood in stream channels serves as habitat in the form of cover. This is
important for numerous life stages of salmon; young fishes will hide from predators within the
structure provided by large wood, and feeding fish of many ages use cover to search for drifting
invertebrates. Large wood can create low velocity habitats utilized by adult salmonids for
resting purposes, as thousands of miles may be traveled before spawning occurs. Aside from
directly providing physical cover, large wood also creates cover by influencing channel
morphology and forming pools. Scour pools are common and formed when water is forced over
and/or under wood structures, scouring the streambed and creating deep, low velocity habitats.
Dam pools can also form when a log or accumulation of logs act to hold water back, creating
habitats similar to those formed behind a beaver dam. In terms of sediment, large wood retains
coarse substrates important for spawning, and it also stabilizes banks and reduces erosion.
While large wood is clearly an important resource to the region’s streams and salmonids,
it is also vulnerable to conflicting management strategies from the fisheries and timber industries
in Southeast Alaska. This is evident as initial best management practices occurring from 19781990 for timber harvest activities in the region did not require the retention of any trees directly
on the stream bank, even though a one-tree buffer was suggested as part of management
practices. It was instead more common to harvest anything of significant economic value directly
up to stream banks (Douglas Martin, personal communication).

2

Alaska’s Forest Resources and Practices Act (FRPA) was established in 1978 and is the
governing document on how access, harvest, and reforestation can occur on state, private, and
municipal land, while federal standards must meet or exceed those of the state. The Act is
intended to provide for commercial harvest by both fishing and timber operations while
protecting fish habitat, water quality, and productive forests to ensure sustainable use of forest
and aquatic resources into the future. With no standards regarding harvest in the riparian zone
before 1990, certain streams that salmon depended on for spawning may have been deprived of
wood inputs, and research from Alaska and the Pacific Northwest have documented reductions in
production of stream fishes, and most notably of Pacific salmon, in watersheds extensively
harvested for timber prior to statewide implementation of ecosystem management programs.
Amendments to the 1990 FRPA were intended to address that potential challenge by requiring 20
m wide riparian buffers be left along anadromous streams on private timberlands in Southeast
Alaska. While effectiveness monitoring of the regulations has occurred since the early 1990’s,
little research has occurred to empirically describe current condition of streams in the region, and
much is assumed about current habitat in streams harvested under the BMP’s in place in
Southeast Alaska from 1978-1990.
The goal of this research was to determine the current condition of streams harvested
under the initial BMPs and explore relationships between natural and timber harvest related
landscape factors at local and catchment scales. A better understanding of streams with second
growth riparian stands will inform managers in the region regarding if and how habitat is
degraded in stands and streams after being harvested under the early BMP’s. We also wish to
understand whether or not second growth stands are currently providing woody material so we
can better understand the amount of time needed after harvest before managed riparian zones
3

will support large enough trees to provide ecologically-functioning large wood to streams.
Addressing these knowledge gaps is crucial as decision makers and resource managers will be
tasked with prioritizing restoration efforts or transitioning to second growth timber harvest in
Southeast Alaska, and characterization of natural and anthropogenic factors influence on habitat
in such streams will support decision making in the region.
To understand the effects of early BMPs on habitat requires knowledge of both natural
and anthropogenic factors operating at the local reach as well as additional spatial scales
including watersheds (Allan 2004; Wang et al. 2006). Also, stemming from a historically
terrestrial focus, the field of landscape ecology has recently been taken “into the water” (Wiens
2002), and rather than characterizing aquatic systems as separate bodies of water, this evolution
in thinking considers them as a connected piece of the landscape, affecting and being affected by
a myriad of other factors operating across landscapes.
It is currently a challenge in Alaska to examine streams in a landscape context as
consistent data representing fluvial resources and their watersheds over large spatial extents are
limited. As part of a project supported by the National Fish Habitat Partnership (NFHP), local
catchments, defined as the land area draining to a single stream reach, were created for every
stream reach in Southeast Alaska. This allows for landscape characteristics occurring within this
land area to be summarized and their influence on stream reaches to be assessed. Additionally, a
tool was developed that allowed for summarizing landscape factors occurring in the entire
upstream network catchment draining to any stream reach of interest. In support of the following
research, this spatial framework was constructed for Southeast Alaska, and landscape-scale
factors including forested land cover and human influences including timber management history
were summarized within local and network catchments of my study streams of the region. I have
4

associated these landscape scale factors to metrics characterizing large wood abundance and
distribution in coastal streams of Southeast Alaska, adding to our general understanding of
landscape influences summarized over different spatial extents on large wood, providing a richer
understanding of influences of the landscape on the region’s streams.

5

LITERATURE CITED

6

LITERATURE CITED
Allan, J. D. 2004. Landscapes and riverscapes: The influence of land use on stream ecosystems.
Annual Review of Ecology Evolution and Systematics 35:257-284.
TCW Economics. 2010. Economic contributions and impacts of salmonid resources in southeast
Alaska. Final Report prepared for Trout Unlimited Alaska Program, Juneau, Alaska.
Wang, L., P. W. Seelbach, and R. M. Hughes. 2006. Introduction to landscape influences on
stream habitats and biological assemblages. Pages 1-23 in R.M. Hughes, L. Wang, and P.
W. Seelbach, editors. Landscape influences on stream habitats and biological
assemblages. American Fisheries Society, Symposium 48, Bethesda, Maryland.
Wiens, J. A. 2002. Riverine landscapes: taking landscape ecology into the water. Freshwater
Biology 47:501

7

CHAPTER 1
THE INFLUENCE OF GEOMORPHIC, RIPARIAN, AND TIMBER HARVEST FACTORS
ON AQUATIC HABITAT IN ANADROMOUS STREAMS OF SOUTHEAST ALASKA

Abstract
Early logging is known to have negatively affected stream habitat and biota in the Pacific
Northwest and Alaska. Large-scale harvest was historically done with no consideration for
stream systems. Altered hydrology and connectivity, unstable banks, increased erosion,
sedimentation of stream beds, and simplified channels have resulted from logging. As timber
management is widely accepted as having the potential to alter stream habitat, understanding the
influence of natural controls is also critical to define habitat potential for streams in a given
region. The goal of this study is to compare geomorphic, riparian, and timber harvest-related
influences on aquatic habitat in streams of Southeast Alaska with riparian zones harvested
between 1980 and 1990 to better account for natural controls vs. the legacy of human impacts to
the region’s streams. Results show that aquatic habitat in sampled streams is tightly linked to
both natural and anthropogenic land uses, and that geomorphic and riparian characteristics were
associated with many aspects of stream habitat. Pool, substrate, and large wood characteristics
were related with geomorphic factors including bankfull width, channel gradient, and channel
confinement, while pools formed by wood and large wood density were also related to riparian
composition. In this study, size of substrates was found to be related to historical timber
management practices; harvested streams were found to have smaller median particle sizes than
substrate in unharvested streams. These findings suggest that riparian and geomorphic factors
strongly influence fluvial habitat at the reach scale and reinforce the importance of understanding
natural controls on aquatic habitat to effectively manage streams in Southeast Alaska.

8

Introduction
Stream characteristics are influenced by properties of the landscapes they drain (e.g.,
Hynes 1975, Frissell et al. 1986, Wiley et al. 1997, Allan 2004), and physical and biological
conditions within streams are in part created and constrained by hydrologic and geomorphic
processes operating within their watersheds (Wang et al 2006). Stream habitats and their biology
have also been shown to be differentially affected by landscape factors operating over different
spatial scales including riparian zones and catchments (Frissell et al. 1986, Townsend et al. 2003,
Cohen and Brown 2007), with that differential influence resulting from specific mechanisms by
which landscape factors influence stream conditions (Allan 2004). For example, wholecatchment land cover may be more important for controlling hydrologic factors and delivery of
nutrients through river networks (Boyer et al. 2002, Allan 2004). Riparian condition, on the
other hand, including deforested riparian zones, may be associated with inputs of fine sediment
(Jones et al. 1999, Sutherland et al. 2002), large wood (LW) (Murphy and Koski 1989), and
organic litter (Hetrick et al. 1998, Allan et al. 2003). Stream temperature is also influenced by
riparian vegetation (e.g., Imholt et al. 2013). The ability to discern between landscape influences
and the scale at which they most strongly affect stream characteristics is critical for protecting
and managing stream systems from human impacts.
In Southeast Alaska, interrelationships between streams and landscape characteristics of
their watersheds are acknowledged as important in supporting anadromous fishes that migrate
from the ocean into streams annually. Forested landscapes within river basins of Alaska and the
Pacific Northwest regulate flow by intercepting precipitation and encouraging increases in
shallow groundwater, slowing delivery of precipitation to streams and providing flood protection
(Hicks et al. 1991, Jones and Grant 1996). Natural vegetation also stabilizes steep slopes and
9

reduces the risk of landslides and mass wasting events that can lead to excess sediment being
delivered to stream channels (Wu et al. 1978, Swanson and Marion 1991). Forests provide
allochthonous inputs to streams in the form of leaf litter and terrestrial insects, a benefit that is
especially pronounced in forested riparian zones of stream channels (Allan et. al 2003). Riparian
vegetation also helps to stabilize stream banks (Gregory et al. 1991) and regulate light input and
temperature fluctuations through shading (Feller 1981, Johnson and Jones 2000), along with
providing the majority of LW to streams in Southeast Alaska (Murphy and Koski 1989, Martin
and Grotefendt 2007).
Multiple factors within streams interact to create complex habitats including variation in
flow, substrate, and LW. Salmonids rely on complex stream habitats, such as deep pools with
LW (Bisson et al., 1982, Quinn and Peterson 1996), to provide cover from predators and refuge
during extremes in flow conditions (Dolloff et al. 2003). Spawning salmonids also need clean
gravel-size substrate that is free of fine sediments. Fine sediments can inhibit embryo survival
by filling interstitial spaces and reducing oxygen supply (Chapman 1988). The input of LW to
streams is critical to salmonids for multiple reasons and can have profound influences on the
availability of complex habitat and stream substrates. Large wood in stream channels serves as
habitat in the form of cover. This helps to support several life stages of salmon. Young fishes
will hide from predators within the structure provided by LW, and feeding fish of many ages use
cover to search for drifting invertebrates (Rosenfeld et al. 2000, Roni and Quinn 2001, Bryant
and Woodsmith 2009). Large wood can create low velocity habitats utilized by adult salmonids
for resting purposes, as they may travel thousands of miles before spawning. Aside from directly
acting as physical cover, LW also creates cover by influencing channel morphology and forming
pools (Bisson et al. 1987, Martin 2001). Scour pools are common and formed when water is
10

forced over and/or under wood structures, scouring the streambed and creating deep, low
velocity habitats. Pools can also form when a log or accumulation of logs act to hold water back,
creating habitats similar to those formed behind beaver dams. In terms of sediment, LW retains
coarse substrates important for spawning, and it also stabilizes banks and reduces erosion
(Montgomery et al. 1996, Gomi et al. 2001).
Geomorphic and riparian factors are known to have a major influence in observed
variation in habitat factors of streams in Southeast Alaska and the Pacific Northwest. Large
wood input, for example has been shown to be greater in alluvial channels than confined
channels (Murphy and Koski 1989, Martin 2001). Additionally the size, abundance, and ability
for LW to function in a stream is associated with the density and type of riparian vegetation
(Bilby and Wasserman 1989, Gomi et al. 2006). Stream size also influences the density, size,
and residence time of log jams in Southeast Alaska (Martin and Benda 2001). Also, pool size
and density are known to vary by stream size (Buffington et al. 2002), while other catchment
factors such as depth to bedrock or frequency of debris flows may limit the potential of pool
formation (Benda 1990).
The earliest best management practices applied by the timber industry in the region
initially failed to provide protection to streams, despite the importance of complex habitat,
substrates, and LW to salmonids of Southeast Alaska. Alaska’s Forest Resources and Practices
Act (FRPA) was established in 1978 and dictates how access, harvest, and reforestation can
occur on state, private, and municipal land. The Act is intended to allow commercial harvest of
timber and fisheries while still protecting fish habitat, water quality, and productive forests into
the future. Prior to 1990, the FRPA recommended that low-value timber be left along streams,
but the Act did not require the retention of any trees directly on the stream bank. Under these
11

initial best management practices (BMP’s), it was typically common to harvest anything of
significant economic value directly up to stream banks (Douglas Martin, personal
communication). Amendments to the 1990 FRPA acknowledged the importance of forested
stream banks by requiring that 20 m wide riparian buffers be left along anadromous streams on
private timberlands in Southeast Alaska.
Timber harvest in the riparian zone before 1990 may have reduced wood inputs and
degraded habitat conditions in the region’s anadromous streams. Research in Alaska and the
Pacific Northwest have documented reduced production of stream fishes in watersheds
extensively harvested for timber prior to statewide implementation of ecosystem management
programs. Reductions in fish production can be attributed to negative changes to physical
habitat that can result from timber harvest including sedimentation (Scrivener and Brownlee
1982), loss of large wood (Bisson and Sedell 1984), and water temperature increases (Moring
and Lantz 1975). While effectiveness monitoring of intact stream buffers has occurred since the
early 1990’s, little research has occurred to empirically describe current condition of streams in
the region, and much is assumed about current habitat in streams harvested under the BMP’s in
place in Southeast Alaska from 1978-1990.
The goal of this study is to characterize geomorphic, riparian, and timber-related
influences on aquatic habitat in anadromous streams of Southeast Alaska with riparian zones
harvested between 1980 and 1990. To address this goal, my objectives were: a) to document
channel geomorphic characteristics, riparian stand composition, and aquatic habitat conditions
including habitat heterogeneity, substrates, and LW in streams with harvested riparian zones, b)
to characterize associations between geomorphic factors and riparian stand composition with
specific aquatic habitat factors, and c) to compare aquatic habitat factors in harvested streams to
12

similar streams in the Tongass National Forest having no riparian harvest to explore potential
differences that may be associated with differences in riparian management. By understanding
influences of geomorphic, riparian, and timber-related management actions on aquatic habitat,
we can better understand how historical timber harvest practices in the riparian zone may have
affected aquatic habitat in the region. Addressing these knowledge gaps is crucial as decision
makers and resource managers will be tasked with prioritizing restoration efforts or transitioning
to second growth timber harvest in Southeast Alaska, and results of this work can be used to
support decision making in the region.
Methods
Study region
Southeast Alaska is part of the Alexander Archipelago which contains approximately
1,100 islands, ranging from islands too small to support stands of harvestable trees to those large
enough to support multiple communities (e.g., Chichagof Island). Some of the largest islands in
North America occur in this region, including the 4th largest in the United States; Prince of
Wales Island. The region is considered part of the largest temperate rainforest in the world, as
defined by the World Wildlife Fund’s Ecoregion classification (Olson et al. 2001). The North
Pacific Coastal Forests contain over 25% of the world’s coastal temperate rainforests, many of
which have been generally unaltered by human activity. The region also encompasses the
Tongass National Forest which is the largest national forest in the United States.
Annual precipitation often exceeds 2000 mm, and a maritime climate produces mild
temperatures in the summer and winter. Many of the larger islands of the region are
characterized by mountainous terrain with steep slopes and glacially carved U-shaped valleys
13

(Harris and Farr 1974). Western hemlock (Tsuga heterophylla), Sitka spruce (Picea sichensis),
and western red-cedar (Thuja plicata) primarily make up the old growth conifer forests of the
region, and deciduous alder (Alnus spp.) commonly establishes after disturbances. The variable
topography and long periods of constant precipitation lead to occasional natural disturbance like
mass wasting and large windthrow events. Heterogeneous landscapes and aquatic habitats result
in part from such disturbances and support diverse populations of anadromous salmonids.
Native salmonids to Southeast Alaska include chinook (Oncorhynchus tshawytscha), coho
(Oncorhynchus kisutch), sockeye (Oncorhynchus nerka), dog (Oncorhynchus keta), pink
(Oncorhynchus gorbuscha), steelhead (Oncorhynchu mykiss), dolly varden (Salvelinus malma),
and cutthroat trout (Oncorhynchus clarki), all of which can or specifically do exhibit an
anadromous lifecycle.
Site selection
This study evaluated streams on Prince of Wales Island and Chichagof Island on Sealaska
Corporation timberlands in Southeast Alaska (Figure 1.1). While nine different stream types
occur in the region due to broad differences in geology, topology and hydrology (Paustian et al.
1992), we primarily evaluated streams classed as two types for this study: floodplain (FP) and
moderate gradient-mixed control (MM). These stream types were selected in part due to their
potential sensitivity to disturbances resulting from timber harvest (Paustian et al. 1992,
Woodsmith et al. 2005). FP streams are those with low gradients (i.e., < 2%) and are usually
sediment-storing stream segments located in low valley bottoms toward the mouths of
watersheds. Large wood is abundant and important in creating pools in the region, with a
predictable pool-riffle sequence commonly observed in FP channels. MM channels have

14

moderate gradients (i.e., 2-6%) with plane bed and LW forced pool-riffle sequences that can
contribute to sediment retention.
Three different stream coverages for the region were used to identify potential sites for
sampling. They are managed by 1) Alaska Department of Fish and Game (ADF&G), 2)
Sealaska Timber Corporation (STC), and 3) the United States Forest Service (USFS). Although
all three datasets are represented at the same scale of 1:63,360, each stream coverage provides a
unique set of attributes that were important in identifying potential study reaches. Streams in the
ADF&G and STC coverages are classified by their ability to support anadromous fishes, while
streams in the USFS coverage are classified by fish use where a classification of I indicates
“Streams with anadromous or adfluvial fish or fish habitat; or high quality resident fish waters or
habitat above fish migration barriers known to provide reasonable enhancement opportunities for
anadromous fish.” Furthermore, the USFS incorporates Paustian’s stream channel type
classification referenced above (1992) which was used to assist in locating unconfined and
moderately confined streams having low to moderate gradients. To be considered for site
selection, reaches had to be classified as anadromous by either ADF&G or STC or had to be
classed as Type I by USFS. Once all anadromous waters of Sealaska lands were identified,
reaches were further assigned a channel type from the USFS coverage.
Data on historical timber management practices adjacent to anadromous streams on
Sealaska lands were also used to further identify potential study sites. Data on history of the
timberlands were obtained from managers at Sealaska Timber Corporation, and possible sample
sites were further narrowed using a geographic information system and ortho-photography to
identify only reaches in which at least 100 m of at least one of the stream banks had riparian
timber harvest prior to the 1990 streamside buffer amendments to Alaska’s FRPA (1978). We
15

only considered anadromous streams that experienced conventional timber harvest (i.e., clearcut) in the riparian zone adjacent to the stream between 1980 and 1990. This process led us to
identify 326 potential study sites (i.e., target population) that qualified under the anadromous,
channel type, and timber management criteria described above. These sites were entered into a
generalized random-tessellation stratified design (GRTS, Stevens and Olsen 2004) to randomly
select a sample of sites from the target population that were spatially distributed across Sealaska
timberlands.
A global positioning system (GPS) and paper maps were used to navigate to potential
study sites that were identified by the random sample. The GRTS process resulted in a random
sequence of sites which were examined in order. Final selection occurred in the field and after a
site was located it would either be sampled or rejected, and exclusion occurred for various
reasons including the stream having: a palustrine channel, a confined channel, a dry channel, or
no channel present. Other rational for excluding a potential site includes that it was not
wadeable or not feasible to sample based on access. In addition, visual confirmation of historic
riparian timber harvest was assessed and confirmed prior to selecting a site for sampling. In
total, 44 sites were visited and after field verification, 28 sites were selected for sampling that
satisfied the selection criteria. The sample size of 28 sites was determined by field time, labor,
and funding. Study reaches were identified within sites, and the beginning and end points of all
reaches were at least 20 m away from any bridge crossing or major tributary. Average channel
width was estimated before each survey, and 20 x the average channel width was used as a
minimum reach length (Harrelson et al. 1994), while 40 x the estimated channel width was used
as a maximum length.
Field data collection
16

Geomorphic and riparian factors
Channel morphology
Measures of stream channel morphology were taken at three cross sections along each
study reach including bankfull width and depth and channel bed width. While these
measurements were taken at the stream reach scale, we evaluated relationships between these
factors and stream habitat in subsequent analyses for a number of reasons. First, channel
dimensions are known to be strongly influenced by drainage area of river systems (e.g., Schumm
1977, Ralph et al. 1994, Knighton 1998), and they have been treated as surrogates for stream size
in previous studies within the region (e.g., Bilby and Ward 1989). Second, using bankfull
dimensions as a surrogate for stream size was preferred by managers of the region to provide
consistent comparisons with other previously collected data. As bankfull width and drainage
area were highly correlated in our sample sites with an r-value of 0.92 we felt that it would
adequately represent stream size for subsequent analyses. Morphology measurements were
typically taken in straight riffle units that were free of obstructions. When physical obstructions
were present (e.g., boulder, woody debris) in or near cross sections, the type of obstruction and
degree of influence was recorded prior to making measurements. Bankfull was indicated
visually as the point at which flow would begin to enter the depositional floodplain adjacent to
low-gradient study reaches. In confined channels, bankfull width was identified by evaluating
vegetation changes, substrate changes, slope breaks along the bank, undercuts, stain lines, and
the height of depositional features (Harrelson et al. 1994). A visual ranking of stream
confinement was performed at all channel cross sections. Unconfined channels were ranked a 1,
while totally confined channels were ranked a 3. Channels that were moderately confined or had

17

mixed confinement (e.g., one bank confined, one unconfined) were ranked a 2. The average was
taken across all channel cross sections for a given study reach.
Riparian condition
On each bank and beginning at the edge of bankfull, data collected from three 30 x 3 m
rectangular plots that were oriented perpendicular to the stream were used to characterize timber
stand and landform characteristics. The upstream valley direction was estimated using a
compass, and 90 degrees was added to orient all transects perpendicular to the stream valley.
Landform slope in percent was recorded every 5 m along each transect. Additionally, the
vegetation type (deciduous vs. conifer), number, and size (dbh) of each tree along each transect
were observed and recorded. Size (dbh) for riparian trees was measured and assigned to one of
four size categories: small (10-30 cm), medium (30-60 cm), large (60-90 cm) and very large (>
90 cm). The proportion of type and size of trees were calculated to broadly characterize the LW
potential available along each study reach.
Aquatic habitat factors
Habitat complexity
Descriptions by Bisson et al. (1982) were used to define geomorphic habitat units (i.e.,
pool, riffle, glide). Length and width were measured to the nearest half meter for each unit
encountered in the channel throughout the study reach. Additional data were collected for pools
including the tail crest and maximum depths to the nearest 1 cm. The residual depth of each pool
was then calculated as the difference between the maximum and pool crest depths (Lisle 1987).
Identification of pools was confirmed by those that with a residual depth equal to or greater than
the value determined by the equation: minimum residual depth = (0.01 * mean bed width (m)) +
18

0.15 m with a length or width at least 10% of the mean bed width (U.S. Forest Service 2001).
Pool units were further classed as to whether they occurred in the main channel or in regions of
the stream adjacent to the main channel. Main channel units were identified if the thalweg
flowed through them at the time of sampling, while associated units were located adjacent to a
main channel unit, including those located within the floodplain as long as it had a surface
connection to the main channel. A hip chain was used to record the distance from the beginning
point of the reach survey to the beginning of each habitat unit. Length and one width
measurement was taken to the nearest 0.5 m for units that were approximately 1 m or less in
length or had a consistent width. Two width measurements were taken for units that were longer
than 1 m or were proportionally heterogeneous. Unit area was estimated as the product of unit
length and mean unit width. Relative pool area was determined as the percentage of pool area
relative to the total wetted area of the study reach.
Substrate
Estimates of dominant and sub-dominant substrate types were performed visually in each
geomorphic unit. Substrate types were classified as sand and fines (smooth to gritty to ladybug
size), gravel (pea to baseball size), cobble (baseball to basketball size), boulder (basketball to car
size), and bedrock (solid rock or bigger than car size) (Krumbein 1934, U.S. Environmental
Protection Agency 1992). At all locations where channel morphology measurements were taken,
a pebble count (Wolman 1954) was also performed to characterize bed material. A total of 100
substrate samples were taken along five channel-spanning cross-sections which were centered at
the channel morphology cross section.
Large wood

19

For this study, LW is defined as any log at least 10 cm in diameter and 2 m in length
within the bankfull channel following Robison and Beschta (1990) and Martin (2001). All LW
pieces were counted and measured within each 10 m long segment in smaller streams (< 10 m
average width) and every 20 m long segment in larger streams (> 10 m average width). The
diameter of the center of each log was measured and assigned to one of four size categories:
small (10-30 cm), medium (30-60 cm), large (60-90 cm) and very large (> 90 cm). The length of
each log within the bankfull channel was measured to the nearest 3 m.
All logs in which the source could be identified would be considered a LW recruit and
classified by its recruitment process. The woody debris recruitment processes include: bank
erosion, windthrow, broke stem (i.e., assumed windthrow), mortality, landslide, cut, beaver, or
unknown. Information on whether a LW recruit was in or above the active channel, and whether
it was a functioning recruit was also recorded. To be considered functioning, a recruit must do
any or all of the following: cause channel scour or flow deflection, create a sediment retention
zone, dam water, create an eddy, or catch and hold other woody debris in place.
Large wood recruits were also placed into a visual decay class, modified from Martin and
Benda (2001). Classes were determined for the part of the log that was least disturbed by flow
(e.g., the portion on the bank). Five decay classes were defined as: green leaves or needles
present (green), twigs present (twig), secondary and/or primary branches present (branch), no
branches and some nubs present (nub), and moss and saplings present (old).
Data analyses
Data reduction

20

Distributional properties (i.e., mean, range, standard deviation) of over 200 habitat
variables were calculated and used to select those that would adequately describe a range of
factors across the five categories of aquatic and riparian habitat variables listed above. For
example, variables with many zero values were not selected. Normality and linearity of
variables was also assessed by use of histograms and p-p plots. Non-linear variables were
transformed as follows: continuous data were transformed using natural log, count data were
transformed using square root, and proportions were transformed using arcsin square root.
Bankfull width, bankfull depth, and the proportion of fine substrates (i.e., from pebble counts)
were transformed to meet the statistical assumptions described above. Next, a Pearson’s
correlation analysis (Table 1.1) was performed on variables within geomorphic, riparian, and
aquatic habitat categories to further limit the number and reduce redundancy in variables used to
describe the current condition of stream reaches sampled. If a pair of variables were highly
correlated within categories (i.e., r = absolute value > 0.8), one would be retained based on its
ecological interpretability. This process led to the selection of twelve aquatic habitat variables
including four measures of habitat complexity, two measures of substrate, and six measures of
LW. Four variables representing channel morphology and two riparian stand variables were also
selected to be included in further analyses (Table 1.2).
Principal components analysis
Principal components analysis (PCA) was performed on the correlation matrix of the 18
chosen aquatic habitat, morphological, and riparian stand variables as an initial step to
investigate broad relationships by identifying which geomorphic, riparian, and aquatic habitat
factors contribute to explaining the most variation across the sampled study reaches. PCA can be
used to simplify large multivariate datasets into a smaller, more interpretable set of variables. It
21

assigns weights to individual variables to characterize axes of variance within multidimensional
space. Only principal components with eigenvalues greater than 1 were retained (Kaiser 1960),
and variables weighting individual axes with an absolute value of 0.55 or greater were used to
interpret axes. To simplify interpretation of resulting factors, Varimax rotation was applied to
the resulting axes.
Understanding influences of geomorphic and riparian factors on aquatic habitat
To relate patterns in aquatic habitat with geomorphic factors and riparian stand
composition, multiple regression models were conducted using geomorphic and riparian
variables. Eleven aquatic habitat variables were selected in the data reduction process and those
investigated include D50, proportion of fine substrates, pool density, residual pool depth, relative
pool area, the number of pools formed by wood, large wood density, the mean length of
functioning recruits, the percent of functioning recruits, the proportion of recruits entering the
stream by erosion, and the proportion of recruits entering the stream by wind. A stepwise
multiple regression approach was used to examine associations between geomorphic and riparian
factors on each aquatic habitat variable with the significance of the F statistic (P = 0.05) used as
the model entry criterion. If no significant models were produced for a particular response
variable a P = 0.10 would be used as the entry criterion. After independent variables were
selected, collinearity was assessed based on a variance inflation factor threshold of three (VIF <
3). If variables exceeded this threshold they were removed, and the stepwise method would be
rerun. I only considered models containing four or fewer independent variables to avoid overfitting and selected the model with the lowest Aikaike Information Criterion (AIC) as the best
model from the group. The proportion of variance explained is reported as R² for models with
one variable and as R²adj for models with more than one variable.
22

Understanding influences of timber harvest on aquatic habitat
Data characterizing 25 stream reaches within unharvested watersheds in the Tongass
National Forest were obtained from the U.S. Forest Service to determine whether differences in
stream habitat factors could also be attributed to differing management types (U.S. Forest
Service 2001). The dataset represents streams of similar size and channel types to the sampled
streams on Sealaska timberlands, and aquatic habitat variables characterizing pools and stream
substrates that were collected using comparable methods were pooled from both datasets for
further analyses. Pooled data were tested for linearity and normality using steps described
previously, and Pearson’s correlation analysis was performed on pooled variables. If two
variables were moderately correlated with an absolute value > 0.45, one would be retained for
further analysis based on ecologic interpretability and frequency of use in the literature (Table
1.3). This led to the selection of four key aquatic habitat variables that were used to consider the
influence of timber harvest on aquatic habitat including: median particle size, mean residual pool
depth, relative pool area, and pool density.
Discriminant function analysis (DFA) was performed on the selected variables to test
whether streams with harvested riparian zones are different from streams from unharvested
watersheds, using the four aquatic habitat variables chosen above. Finally, an analysis of
covariance (ANCOVA) was performed on each of the key response variables to test for
relationships between channel type and harvest type on the four key response variables while
controlling for stream size (i.e., bankfull width) as a covariate.
Results
Geomorphic, riparian, and timber harvest characteristics
23

Regional trends in timber harvest
Sealaska Corporation owns approximately 1180 km² of land with just over 835 km²
(70.76%) either planned for harvest, or having already been harvested for timber using
conventional (i.e., clearcut) or helicopter methods (Table 1.4). Of this, over 58% has or is
planned to occur after 1990, resulting in less than 13% of their total land area having been
harvested without being required to leave a streamside buffer. The majority of sampled reaches
(i.e., 20/28) experienced harvest by Sealaska Corporation during or after 1985 (Figure 1.2).
Riparian harvest occurred on both banks more often than one bank when considering all reaches
(54 vs. 46%, respectively; Table 1.5).
Geomorphic factors
Geomorphic characteristics varied considerably across all reaches ranging from 1 m to
over 25 m in bankfull width (Table 1.5). Gradient also varied from 0.5% to about 6% but
averaged just over 3% across all reaches. Confinement was less variable and on average, all
streams were identified somewhere between unconfined and moderately confined based on
visual assessment. Noticeable differences across the two channel types were also observed.
Low gradient channels were on average wider (10.8 vs. 3.8 m), which resulted in longer sampled
reach lengths for these streams (345 vs. 152 m). The moderate gradient sites were relatively
steeper compared to low gradient channels (3.8 vs. 1.5%).
Riparian factors
Data on riparian stand composition were collected for 876 trees across all study reaches.
Results suggest that riparian zones consisted of mostly small and medium sized trees as the
percentage of large and extra-large trees ranged from 0 to 30%, but on average, less than 10% of
24

all trees were classed as large or extra-large (Figure 1.3). Little difference in riparian tree size
was observed when comparing low gradient to moderate gradient streams (Figure 1.4), but a
slightly higher percentage of medium sized trees were found along streams having two banks
harvested compared to having just one (Figure 1.5). Riparian tree type was also variable as the
percentage of recorded trees that were deciduous ranged from 0 to over 70%, although on
average most riparian zones were comprised of coniferous trees (Figure 1.3). When comparing
across different stream types there was a slightly higher percentage of deciduous trees along low
gradient streams (Figure 1.6), although results suggest that streams having both banks harvested
also had relatively more deciduous trees when compared to those only having one side harvested
(Figure 1.7).
Aquatic habitat characteristics
Habitat complexity, substrate, and large wood
Substantial variation in aquatic habitat was observed across the study reaches (Table 1.2).
Pool density ranged from under 10 to over 125 pools per km. Relative pool area in percent was
highly variable with a minimum value of 13.2 and a max of 95.1%. Residual pool depth
averages also varied, ranging from 20 to 90 cm. Across all sites, wood was the dominant pool
forming element as most pools were formed by wood (i.e., 62%). Rock also accounted for a
substantial amount of pool formation (i.e., 14%). Active beaver ponds were observed in 7% (i.e.,
2/28) of the reaches, in which they were the dominant pool forming element for 12% and 90% of
all pools at these locations. Substrate particle measurements ranged from sand to bedrock (<4
mm - >512 mm), with the majority of measured substrates from pebble counts consisting of
gravel (i.e., 71%).

25

A total of 2237 logs were measured, with 53.2% classed in the small category (i.e.,
diameter 0.1 – 0.3 cm) and 30.5% classed as medium (i.e., diameter 0.3-0.6 cm). Large logs
were less common as 14.8% and just 1.5% of all logs were classed as large and extra-large
respectively (i.e., diameter 0.6 – 0.9 and > 0.9 cm). In channel length of logs varied in a similar
fashion as 57.3% of logs were between 2 and 5 m long, 34.8% fell between 5 and 11 m in length,
while the final 7.9% of LW pieces were greater than 11 m long (Figure 1.8). Large wood density
varied from less than 110 to just under 890 pieces per km. Large wood density was relatively
higher in low gradient streams compared to moderate gradient streams as an average of 381
pieces/km of wood were measured in 9 low gradient streams, while an average of 356 pieces/km
were measured in 19 moderate gradient streams. Little noticeable difference was observed
across the two types of channels when comparing size distributions (Figure 1.9). Overall a
higher density of wood was found in the 15 streams having two sides harvested (i.e., 378
pieces/km) when compared to the 13 streams only experiencing harvest on one bank (i.e., 348
pieces/km), and in general the wood was relatively larger (Figure 1.10).
The average in stream length of functioning recruits ranged from 3 m to just over 11 m
with an average of approximately 6.8 m. On average, recent LW recruits were present with less
than two green and twig recruits per 100 m across all reaches (Table 1.2). Only 4 out of 28
reaches had more than three recent recruits per 100 m, two of which exceeded eleven recent
recruits in each reach. The decay class (see Methods) of small and medium-size recruits was
between the branch and nub stage, while large and extra-large recruits were on average older and
fell between the nub and old decay class (Figure 1.11). Recruits were functioning about half the
time when averaged across all reaches (i.e., 48%), although this varied widely from 13% to 88%
at the individual reach level. New recruits in these reaches were shown to have the capability to
26

function, although relative to nub and old recruits, green and twig recruits functioned much less
often (i.e., 56% vs. 24%; Figure 1.12). A similar pattern was observed for recruits based on their
size where larger recruits were less common but functioned more often (Figure 1.13). The
recruitment process bringing LW into the stream was dominated by erosion and wind with 59%
and 30% respectively, of all recruits entering the stream by these mechanisms. A small
percentage (i.e., 4%) of recruits showed signs of entering the stream by harvest, while the
remaining 7% were recruited through natural mortality, beaver, or unknown processes.
Understanding influences of natural factors on aquatic habitat
Principal components analysis
The principal components analysis identified five axes that explained a majority of
variation across sampled reaches (77.57%) (Figure 1.14). The first component, explaining
27.01% of the variation, was weighted by measures related to stream size, riparian trees, and
functioning recruits. Variables positively weighting this axis included bankfull width, residual
pool depth, the average length of functioning recruits, bankfull depth, and the proportion of
deciduous riparian trees, while confinement was negatively weighted on this axis. The second
axis, explaining 16.86% of the variation in aquatic and riparian habitat across reaches, was
weighted by measures of stream power having negative weights for both the proportion of fine
substrates and the number of pools formed by wood, while this axis had positive weights on D50
grain size and gradient. The third axis was weighted by measures of LW recruitment by both
wind and erosion, explaining 13.21% of the variation. The fourth axis explained 12.23% of the
variation and was positively weighted on measures of LW and pool density. The fifth and final

27

axis explained 8.27% of the variation and included both large riparian trees and the proportion of
functioning recruits.
Stepwise multiple regressions
By performing stepwise multiple linear regression models, measures of aquatic habitat
were related to riparian and geomorphic factors including bankfull width, gradient, confinement,
and the proportion of deciduous riparian trees (Table 1.6). Most notably, bankfull width was
shown to have a significant relationship with seven out of eleven aquatic habitat variables.
Bankfull width was the only significant predictor of six variables including pool density (R² =
0.21), residual pool depth (R² = 0.68), the length of functioning recruits (R² = 0.41), the
proportion of functioning recruits (R² = 0.26), LW recruited by erosion (R² = 0.36), and LW
recruited by wind (R² = 0.14). For D50 (R²adj = 0.52), both bankfull width and gradient were
significant predictors. Additionally gradient was shown to have a significant relationship with
the proportion of fine substrates (R² = 0.28) and relative pool area (R² = 0.36) and was the only
significant predictor included in those models. The proportion of wood formed pools (R²adj =
0.30) was significantly related to both confinement and the proportion of deciduous riparian
trees. Large wood density was not significantly related to any riparian or geomorphic factors at
P < 0.05, but did show a negative relationship with the proportion of large and extra-large trees
(R² = 0.11) when the stepwise entry criterion was less conservative at P < 0.10.

Understanding influences of timber harvest on aquatic habitat
Cumulative frequencies
28

A sequence of cumulative frequency distributions comparing logged and unlogged
streams showed the potential for differences in the four variables (Figure 1.15). Frequency plots
were created for a) D50, b) residual pool depth, c) pool density, and d) relative pool area. Little
difference is apparent between harvested and unharvested streams for residual pool depth, pool
density, or relative pool area while D50 grain size may be lower in harvested streams.
Discriminant function analysis
The results from the DFA showed moderate success in discriminating between harvested
and unharvested stream reaches when developing a linear combination of D50, RPD, RPA, and
POOLKM to summarize between-management class variation. The resulting original
classification had a 76.0% and 78.6% correct classification for unharvested and harvested
streams respectively (Table 1.7). Overall 77.4% of these original grouped cases were correctly
classified. Successful classification dropped slightly when cross-validating grouped cases as
76.0% and 71.4% of harvested and unharvested streams were correctly classed with 73.6% of all
cases correctly classified in the cross-validation procedure. While this analysis does support
some influence of management on the aquatic habitat factors tested, it does not suggest that there
are widespread differences in aquatic habitat between harvest and unharvested streams.
Analysis of Covariance (ANCOVA)
All four key response habitat variables were evaluated by an ANCOVA to test for
associations with channel type and timber management as factors. The results from the multiple
regressions show the importance of stream size in influencing aquatic habitat characteristics, and
BFWD was entered into each of the models as a single covariate to control for this factor.

29

The mean RPD was lower in harvested vs. unharvested streams and in moderate vs. low
gradient streams respectively, but neither difference was statistically significant (Table 1.8, Table
1.9). RPA was lower in unharvested vs. harvested streams and higher in FP vs. MM streams,
although only significant for the channel type. The POOLKM was lower in harvested than
unharvested streams and higher in FP than MM streams, although no statistical differences were
observed. D50 showed statistical differences by both management type and channel type with
harvested and FP streams having lower values than unharvested and MM streams respectively.
No significant interactions were observed between management type (i.e., harvested vs.
unharvested) or channel type (i.e., low vs. moderate gradient) for any of the aquatic habitat
variables tested.
Discussion
This study describes the current condition of stream geomorphology, riparian stand
composition, and aquatic habitat in and along anadromous streams of Southeast Alaska that had
experienced riparian timber harvest between 1980 and 1990. In support of our first objective, we
described geomorphic, riparian stand, and aquatic habitat conditions in streams draining second
growth forests on Sealaska-owned land, and addressed our second objective by comparing key
habitat factors to these landscape-scale influences. Geomorphic controls were strongly
associated with multiple measures of aquatic habitat such as pool characteristics, substrates,
functioning LW, and LW recruitment, while riparian stand composition was related to measures
of wood-formed pools and LW density. To understand influence of management on aquatic
habitat measures, pool and substrate characteristics of study streams were compared to a set of
streams having unharvested watersheds. Differences in size of substrates were noted, with
harvested streams having smaller median particle size. These findings suggest that riparian and
30

geomorphic factors influence aquatic habitat in streams and reinforces the importance of
understanding natural controls on aquatic habitat while managing for these characteristics in
southeastern Alaska.
Geomorphic influences on aquatic habitat
Pool characteristics
Channel geomorphology was related to multiple measures of aquatic habitat, including
measures characterizing pools. Bankfull width was the only significant predictor of pool density,
functioning recruit length, proportion of functioning recruits, and residual pool depth, which was
the variable most strongly predicted by bankfull width. This relationship follows similar results
from coastal Alaska streams; Martin (2001) found pool depths to generally increase with
increasing stream size. The importance of pools to the region’s salmonids is well established
(Heifetz et al. 1986), especially during low flow periods when drying can occur (May and Lee
2004). Besides being a sig predictor of pool dimensions, Pool density was also significantly
predicted by bankfull width with fewer and longer pools in larger streams. Similar results from
Montgomery et al. (1995) showed a strong influence of stream size on pool spacing (i.e.,
distance between pools) in forested channels of Southeast Alaska and Washington having similar
loads of LW. The proportion of wood-formed pools was significantly related to channel
confinement, with a lower proportion of pools formed by wood in streams that were more
confined, while relative pool area was strongly associated with stream gradient as fewer pools
were observed in steeper streams. Additionally, results from the ANCOVA confirmed a
statistical difference in relative pool area between FP and MM channels. More pools were
formed by wood in less confined channels, and more pool area was observed in FP channels.

31

These patterns concur with other observations in the region, as Martin (2001) found higher
proportions of pools and more wood formed pools in FP streams of coastal Alaska when
compared to MM and contained channels.
Substrate characteristics
Measures of stream substrate were additionally influenced by geomorphic factors.
Gradient was the only significant predictor included in the model for the proportion of fine
substrates, while both gradient and bankfull width were strongly associated with D50 grain size.
The ANCOVA confirmed this relationship as D50 was statistically lower in FP than MM
channels. More powerful streams are able to transport larger substrates, resulting in larger
median particle sizes. However, as channels become wider, gradients tend to be less steep, so
although a higher volume of water is present, the ability to move coarse substrates diminishes
(Buffington et al. 2002). Other natural factors can additionally influence substrate characteristics
including catchment geology, landslide frequency, and proximity to tributaries (Montgomery
1999, Benda et al. 2004), as well as elements that can increase bed roughness (e.g., LW) which
can limit transport of sediment in channels, effectively reducing particle sizes in streams
(Buffington and Montgomery 1999). These landscape factors were not accounted for in this
study potentially leading to relatively low amounts of variation in substrate measurements
explained by the regression models.
Large wood characteristics
Numerous large wood characteristics varied with bankfull width including longer and
proportionately more functioning recruits occurring in larger vs. smaller channels. When a
recruit falls toward a wider stream, it is less likely to span the entire channel. This allows the
32

new recruit to function sooner than if it were spanning the riparian zones on both banks. These
results concur with Bilby and Ward (1989) who found the strongest relationship between LW
volume and sediment storage in the widest streams of Western Washington. The proportion of
large wood recruited by erosion increased with increasing channel width, while recruitment by
wind was more prevalent in smaller channels. These findings are consistent with Martin and
Benda (2001) who found more recruitment through erosion in streams having larger drainages,
and more recruitment by mortality (including windthrow) in smaller drainages. While previous
studies have related LW densities to stream size in unharvested streams of the Pacific Northwest
(Bilby and Ward 1989, Robison and Beschta 1990), this relationship was not found in this study.
My results do agree with Ralph et al. (1994) who similarly did not see this relationship and
suggested that the high variation observed in LW requires sampling longer stretches of water
(i.e., 1-2 km) to achieve proper confidence in the measured frequencies rather than extrapolating
a LW density based off shorter surveys.
Riparian influences on aquatic habitat
Riparian stand composition was significantly related to aquatic habitat based on results of
this study. Relative composition of sampled riparian stands that were deciduous was inversely
related to the proportion of pools formed by wood. The results suggest that a lower proportion of
pools are formed by wood in streams having a higher proportion of deciduous trees in the
riparian zone. This result may partially be an artifact of relationships between pool density and
stream size as the relative amount of deciduous riparian trees was significantly correlated with
stream size (Table 1.1). Nevertheless, deciduous species do not have the same size potential as
conifers in old growth forests, leading to a lower volume of LW to streams with deciduous zones
dominated by deciduous vs. conifer species, and our results suggest that larger wood (i.e.,
33

typically conifers) more readily influences channel morphology. In addition, deciduous trees are
shorter-lived and decay faster than conifers (Deal et al. 2004), but may influence size potential of
surrounding conifers in the same stand. In fact, previous research found slightly larger (i.e., dbh)
dead conifers as the percentage of red alder increased in 13 second-growth headwater riparian
stands of Southeast Alaska (Orlikowska et al. 2004). Large wood density was additionally
negatively related to the proportion of large and extra-large riparian trees. A greater proportion
of large trees in the riparian zone may be the result of other smaller trees being outcompeted for
space and resources over time. Large wood in streams is dictated partially by what is available
in the riparian zone and in streams draining old-growth forests of western Washington, LW
diameter was equal or larger than the average riparian tree diameter while younger forests
showed less association between diameter of riparian trees and in-stream LW (Rot et al. 2000).
In northeastern Oregon little difference was observed in amounts of LW or pools formed by
wood when comparing 11 streams having undisturbed riparian forests with 5 streams having
riparian zones harvested for timber up to 17 years prior (Carlson et al. 1990).
Timber harvest influences on aquatic habitat
Previous studies on the influence of timber harvest on stream habitats in Southeast
Alaska and the Pacific Northwest have documented numerous effects including habitat
simplification (Ralph et al. 1993, McIntosh et al. 2000), increased erosion and sedimentation
(Grant and Wolff 1991), and a long term reduction in pools and LW over time (Murphy et al.
1986, Murphy and Koski 1989). While many of these influences were not evident in this study,
there is evidence to support a shift to a smaller median particle size as a result of historical
timber management. The results from the DFA suggest moderate differences in aquatic habitat
between harvested and unharvested sites given our available habitat data, although similar
34

research comparing 13 pristine streams and 10 streams harvested in Southeast Alaska before
1980 had substantially more success discriminating streams based on timber management as
92% and 90% of pristine and disturbed sites were classified correctly in their study (Woodsmith
and Buffington 1996). The ANCOVA showed no significant differences in residual pool depth,
pool density, or relative pool area in regards to timber harvest, which concurs with Carlson et al.
(1990), but not with Heifetz et al. (1986) who found lower amounts of woody debris and pool
area in clearcut streams when compared to unlogged and streams with riparian buffers. My
results do suggest that median particle size was significantly lower in streams having had
riparian harvest compared to unharvested streams after controlling for stream size, a result that
contradicts Carlson et al. (1990) who did not find this relationship. This association is plausible,
and common effects documented from the absence of riparian vegetation, loss of LW, the
presence of roads, and cumulative watershed disturbances include increased erosion and
decreased bank stability leading to increased sedimentation (Murphy 1995), although in
Southeast Alaska the majority of non-federal, closed and decommissioned logging roads comply
with BMP’s issued by the State and are likely not a significant source of sediment to streams
(ADNR 2010). Additionally as LW was abundant in many of the sampled reaches, and LW for
example increases bed roughness which can reduce particle sizes in streams by limiting transport
of suspended sediments down channels (Buffington and Montgomery 1999).
The effects of timber harvest on LW was untested by this study, and timber harvest
related influences on LW dynamics may be occurring in streams having riparian zones harvested
for timber between 1980 and 1990. These factors were not tested due to inconsistencies in
defining LW between the harvested and unharvested aquatic habitat datasets, although general
conclusions can be inferred based on our data and previously published research on streams
35

draining old growth forests in the region. Most LW in Southeast Alaska is recruited within the
first 20-30 m of the stream bank (Murphy and Koski 1989, Martin and Grotefendt 2007), making
the riparian zone the main source of wood for streams in the region. Our results suggest that the
majority of riparian trees were of the smallest size class which has implications for new wood
recruited to streams, most notably that input of LW to such streams will be small pieces. It has
been estimated that on productive valley floors of Southeast Alaska, second growth trees will
reach a dbh of 60 cm (i.e., smallest size of large LW) after 75 years, and the quantity of trees this
size doesn’t equal old-growth forests until 130 years (Taylor 1934). Additionally, new
recruitment of LW is generally low in the region (Martin and Benda 2001), a result supported by
our findings as nearly 30% of the sampled sites had no green or twig recruits. While recruitment
of wood is low in the region, 20-30 years have passed since harvest and LW loads are not
substantially different when compared to streams having unharvested riparian stands (Table
1.10).
Perhaps more notably are the implications for the size distribution of LW in the future as
opposed to LW density. As the largest pieces of wood decay or are transported, the harvested
riparian zones will not be able to provide replacement large logs for potentially hundreds of
years (Murphy and Koski 1989), skewing future distributions towards smaller pieces of wood.
Our results also suggest that those largest pieces have more ability to function in streams
compared to smaller recruits, while the functioning capability also increases with log age
(Figures 1.12 and 1.13). Remnant wood was still prevalent throughout the study reaches, a
characteristic that may mitigate the loss of future recruitment of the largest LW pieces. Results
suggest that recruitment of LW since harvest has been mostly small and medium sized LW,
while some medium sized wood was additionally decayed enough to be recruited before harvest,
36

and the largest pieces of wood are almost all decayed enough to be recruited prior to harvest
(Figures 1.11 and 1.12). This follows similar research as Martin and Benda (2001) found that
logs in the later stages of decay were typically at least 18 years old and recruited before harvest.
Historical effects of timber harvest on streams often focus on catchments experiencing
more anthropogenic disturbance than timber harvest alone so teasing out mechanisms can be
difficult (McIntosh et al. 2000). Moreover, the earliest logging in Alaska and the Pacific
Northwest occurring near streams had little to no regulations requiring that aquatic resources be
taken into account while harvesting timber (Bryant 1983), so the legacy of historic riparian
management practices may still be affecting streams and may continue to do into the future
(Benda et al. 2002).
Management implications
Stream size and channel type constraints
The results of this study suggest that geomorphic factors largely control many aquatic
habitat factors in streams of Southeast Alaska, and they must be accounted for when assessing
the potential of a stream to exhibit desired pool, substrate, and LW characteristics. We found
relatively more wood in low gradient streams when compared to moderate gradient streams.
Higher gradients have more power to transport wood out of those reaches, but low gradient
floodplain streams often meander which allows more wood to be recruited to these streams
through bank erosion. More bank erosion also encourages relatively more deciduous vegetation
to colonize because of increased light reaching the forest floor as established conifers are
removed, and we did find more deciduous vegetation in low gradient streams. These natural
controls are consistent with previous findings in forested catchments of Wisconsin and Michigan
37

where natural controls explained more variation in aquatic habitat than anthropogenic landscape
factors (Wang et al. 2003). This study concurs with previous research in streams of Southeast
Alaska, suggesting that size and channel type will dictate how aquatic habitat will be distributed
throughout a given reach (Paustian et al. 1992). Larger streams will have deeper pools and
fewer, but longer pools than smaller streams. Wood will be more capable of forming pools in
less confined reaches (e.g., FP channels), and a lower proportion of pools will be found in
steeper reaches (e.g., MM channels). Efforts aimed at increasing pool habitat by adding LW
should focus on streams most suited for LW formed pools. Additionally, wood placement
projects must consider that the functioning capability of wood in Southeast Alaska streams
depends on the size of the stream in which it resides as larger streams will allow for longer, and
proportionately more pieces of LW to function.
Legacy of actions on the landscape
Riparian stand composition
While riparian stand composition can vary by landscape factors (Villarin et al. 2009),
harvesting timber influences both the type and size structure of riparian vegetation. The findings
of this study support this, and relatively few riparian trees measured were larger than 60 cm dbh.
We also found that streams having two banks harvested also had relatively more deciduous
riparian trees. This influence has implications for aquatic habitat as opened canopies and
abandoned logging roads provide patches of light following timber harvest and encourage thick
stands of deciduous trees to colonize and out-compete young-growth conifers (Deal et al. 2004).
Large wood of deciduous origin has a smaller size potential and decays faster than conifers, so
mangers tasked with conserving or creating pool habitat in streams having second growth

38

riparian zones will have to consider the ability and time frame in which potentially new LW may
function in streams. In contrast, Deal et al. (2004) states that deciduous trees can enhance
productivity through changes in light penetration and the input of leaf litter. Moreover, red alder
and other deciduous tree species can also influence nutrient levels by fixing nitrogen in soils, and
hyporheic activity can transport these nutrients into nearby streams. Based on the different types
of benefits associated with differing riparian stand composition, decision makers directing stream
enhancement projects need to have explicitly defined goals and objectives as management efforts
are being planned.
Stream substrate composition
The results of this study show a potential decrease in median particle size influenced by
timber management in the riparian zone. These conclusions are limited as other factors affecting
sediment were not accounted for, including geology and landslide frequency in the catchment,
and while an abundance of gravel was observed in the study reaches sampled (only 2/28 sites had
less than 50% gravel from pebble count summaries (Table 1.2) embeddedness by fine substrates
can drastically reduce the quality of gravel and even LW for stream biota (Richards and Host
1994). Managers must also consider natural controls on sediment by knowing floodplain streams
are likely to include smaller substrates, and future research and stream enhancement projects
should take into account geology and historic landslide information as mass wasting events can
input large volumes of not only fine, but coarse sediment.
Silviculture as a tool to manage for future large wood
Silviculture, and more specifically thinning has historically been used by foresters to
increase productivity of second growth forests with the goal of having more timber resources in a
39

shorter time frame than would occur naturally, although this can also be employed to enhance
future sources of LW to streams. Research has found that thinning is effective in dense young
stands, and it can accelerate the production of large trees by up to 20 years (Spies et al. 2013).
While tree growth may be enhanced, conventional thinning could actually result in less dead
trees over the lifetime of a stand (Garman et al. 2003, Dodson et al. 2012), potentially limiting
input of LW to streams. Non-conventional methods may supplement this loss if thinning efforts
directly added trees to streams at the time of thinning. Additionally thinning will be most
effective near streams as most LW input comes from sources closest to the stream. Stream type
and location also matter as small headwater streams may not be capable of delivering wood to
larger streams that support anadromous fishes without debris flows, which only occur in
mountainous regions, so thinning may not be beneficial in these locations.

40

APPENDIX

41

Table 1.1: Pearson's correlations among geomorphic, riparian, and aquatic habitat variables used
in analyses. An asterisk next to correlation coefficient indicates significance at a 0.05 level
while two asterisks indicate significance at a 0.01 level. Bold variable codes indicate a
transformed variable (See Methods for specific transformations and Table 1.2 for variable
codes).

Aquatic habitat

Riparian Geomorphic

BFWD
BFWD
BFDP
GRAD
CONF

1.00
0.76 **
-0.50 **
-0.55 **

DCRT

0.62 **

PLXL
LWKM
FNRL
PFNR
EROSLW
WINDLW
POOLKM
RPD
RPA
WOOD
D50
FINE

0.24
0.11
0.64
0.51
0.60
-0.38
-0.45
0.82
0.42
0.07
0.25
-0.01

**
**
**
*
*
**
*

Geomorphic
BFDP
GRAD
1.00
-0.24
-0.23

1.00
0.65 **

CONF

Riparian
DCRT PLXL

1.00

0.57 **

-0.10

-0.39 *

1.00

0.20
0.10
0.42 *
0.27
0.52 **
-0.27
-0.34
0.65 **
0.27
-0.16
0.20
0.02

-0.25
-0.15
-0.57
-0.33
-0.23
0.34
0.13
-0.56
-0.60
-0.26
0.48
-0.53

-0.13
-0.17
-0.48 **
-0.35
-0.32
0.38 *
0.22
-0.37
-0.30
-0.40 *
0.16
-0.29

0.21
-0.18
0.31
0.21
0.48 *
-0.24
-0.38 *
0.36
0.03
-0.25
0.42 *
-0.24

**

**
**
**
**

42

1.00
-0.34
0.20
0.33
0.29
-0.15
-0.19
0.20
0.03
-0.12
-0.05
0.09

Table 1.1: (cont’d).

Aquatic habitat

Riparian Geomorphic

LWKM

FNRL

1.00
0.03
0.00
-0.08
-0.14
0.43 *
0.15
0.60 **
-0.04
-0.11
0.21

1.00
0.31
0.22
0.02
-0.34
0.70 **
0.46 *
0.37
-0.01
0.15

Aquatic habitat
PFNR
EROS

WIND POOLKM

1.00
0.12
-0.17
0.07
0.37
0.39 *
0.18
-0.05
0.02

1.00
0.22
-0.35
-0.34
0.25
0.02
-0.08

BFWD
BFDP
GRAD
CONF
DCRT
PLXL
LWKM
FNRL
PFNR
EROSLW
WINDLW
POOLKM
RPD
RPA
WOOD
D50
FINE

43

1.00
-0.78 **
-0.42 *
0.48 **
0.16
-0.19
0.22
-0.08

1.00
-0.42 *
0.23
-0.10
-0.23
0.17

Table 1.1: (cont’d).

Aquatic habitat

Riparian Geomorphic

RPD

RPA

Aquatic habitat
WOOD
D50

FINE

BFWD
BFDP
GRAD
CONF
DCRT
PLXL
LWKM
FNRL
PFNR
EROSLW
WINDLW
POOLKM
RPD
RPA
WOOD
D50
FINE

1.00
0.61 **
0.01
0.07
0.05

1.00
-0.07
-0.32
0.33

1.00
-0.37
0.33

1.00
-0.75 **

44

1.00

Table 1.2: Mean, minimum, maximum, and standard deviation (SD) of calculated variables used
to describe the current aquatic and riparian habitat condition of study sites.
Category Variable name

Code

Mean

Min

Max

SD

Channel morphology
Mean gradient (%)
GRAD
3.03 0.50 6.17 1.52
Visual confinement*
CONF
1.61 1.00 2.00 0.44
Bankfull depth (m)
BFDP
0.19 0.06 0.43 0.10
Bankfull width (m)
BFWD
6.06 1.00 25.13 5.95
Riparian
Deciduous riparian trees (proportion)
DCRT
0.27 0.00 0.73 0.24
Large and extra-large trees (proportion) PLXL
0.06 0.00 0.30 0.06
Habitat complexity
Relative pool area (%)**
RPA
45.89 13.18 95.05 21.54
Pool density (#/km)**
POOLKM 61.26 9.86 125.87 24.11
Mean residual pool depth (m)**
RPD
0.44 0.20 0.90 0.20
Pools formed by wood (proportion)
WOOD
0.61 0.06 1.00 0.25
Substrate
Fine (proportion ≤ 4 mm )
FINE
0.11 0.01 0.46 0.11
D50**
D50
28.80 4.76 76.11 17.13
Large wood (LW)
LW pieces/km (#)
LWKM
363.73 109.89 887.85 187.58
Green & twig recruits (#/100 m)
GTREC
1.94 0.00 11.75 2.91
Mean length of functioning recruits (m) FNRL
6.72 3.00 11.19 2.52
Proportion functioning recruits
PFNR
0.48 0.13 0.88 0.17
LW recruitment by erosion (%)
EROSLW 58.48 12.50 100.00 23.29
LW recruitment by wind (%)
WINDLW 29.98 0.00 87.50 21.84
* 1 = unconfined, 2 = moderately confined, 3 = totally confined
** Used in comparison to 25 unharvested stream reaches

45

Table 1.3: Pearson's correlations between key response variables used in comparison between
harvested and unharvested streams. An asterisk next to a correlation coefficient indicates
significance at the 0.05 level and two asterisks indicates significance at 0.01 level. Width to
depth ratio (WD) is the ratio between bankfull width and bankfull depth, average pool area
(APA) was in m², pool spacing (SPACE) is measured in # channel widths per pool, and pool
length (RPL) is the proportion of reach length classified as pool. All other variable codes can be
found in Table 1.2.
RPD
RPD
1.00
RPA
0.40
POOLKM -0.41
D50
0.19
APA
0.83
SPACE
-0.48
RPL
0.49
WD
0.61
D84
0.12

RPA
**
**
**
**
**
**

1.00
0.21
-0.41
0.37
-0.39
0.82
0.22
-0.49

POOLKM

**
**
**
**
**

D50

APA

1.00
-0.24
1.00
-0.50 ** 0.03
1.00
-0.23
-0.19
-0.31 *
0.27 * -0.16
0.34 *
-0.41 ** 0.01
0.59 **
-0.26
0.92 ** -0.01

46

SPACE

RPL

WD

D84

1.00
-0.58 ** 1.00
-0.33 *
0.14
1.00
-0.13
-0.29 * 0.00 1.00

Table 1.4: Total area harvested and percent of total land area harvested for all of Sealaska owned
land organized for three time periods. Time periods correspond to changes to Alaska's forest
resource regulations.
Area harvested (km²)

Percent of total land area (%)

Harvest
method and Conventional Helicopter Total Conventional Helicopter Total
time period
Pre 1980
1980-1990
Post 1990
Total

53.21
93.80
278.61
425.63

0.04
0.31
411.44
411.79

53.25
94.11
690.06
837.42

4.50
7.93
23.54
35.97

47

0.00
0.03
34.77
34.80

4.50
7.95
58.31
70.76

Table 1.5: Mean and standard deviation (SD) of study reaches draining land owned by Sealaska
Corporation. Sites are grouped by gradient as either low (< 2%) or moderate gradient (2-6 %).
Low
Moderate
All reaches
(n=9)
(n=19)
(N=28)
Variable
Mean
SD
Mean SD Mean
SD
Average reach length (m)
345.11 129.62 152.08 92.83 214.13 138.39
Average bankfull width (m)
10.78
7.44
3.83 3.49
6.06
5.95
Gradient (%)
1.46
0.61
3.77 1.23
3.03
1.52
Visual confinement *
1.29
0.42
1.75 0.38
1.61
0.45
One bank harvest (% of sites)
14.29
32.14
46.43
Two bank harvest (% of sites) 17.86
35.71
53.57
* 1 = unconfined, 2 = moderately confined, 3 = totally confined

48

Table 1.6: Results of multiple linear regressions predicting each aquatic habitat variable for 28
streams on Sealaska timberlands. Independent variables are listed in order of their influence on
the model, based on standardized coefficients, which are located in parentheses. Sig. indicates
the p-value for each hypothesis test on the coefficient of the corresponding term in the linear
model. Proportion of variance explained is indicated by R² for one variable models and R²adj for
two variable models. A variance inflation factor value of 4 (VIF< 4) was used to ensure little
multicollinearity.

Aquatic habitat variable Independent variable(s) Model F
D50
FINE*
POOLKM
RPD
RPA
WOOD
LWKM**
FNRL
PFNR
EROSLW
WINDLW

Sig.

R² (R²adj) VIF
0.52
1.34

GRAD (0.81)
BFWD* (0.66)

15.86

< 0.0001
< 0.001

GRAD (-0.53)

10.04

0.00

0.28

1.00

BFWD* (-0.45)

6.70

0.02

0.21

1.00

BFWD* (0.82)

54.15

< 0.0000001 0.68

1.00

GRAD (-0.60)

14.29

< 0.001

0.36

1.00

CONF (-0.59)
DCRT (-0.48)

6.82

0.00
0.01

0.30

1.18

PLXL (-0.34)

3.28

0.08

0.11

1.00

BFWD* (0.64)

17.69

< 0.001

0.41

1.00

BFWD* (0.51)

9.00

0.01

0.26

1.00

BFWD* (0.60)

14.63

< 0.001

0.36

1.00

BFWD* (-0.38)

4.33

0.05

0.14

1.00

* Indicates transformed variables
** Significance of the F statistic (P = 0.10) used as entry criterion

49

Table 1.7: Classification results organized by count and percent correctly classified from the
discriminant function analysis for both known (Original) and unknown membership (Crossvalidated).
Original
Cross-validated
Classified membership
Predicted membership
Unharvested Harvested Total
Unharvested Harvested
Unharvested (#)
19.00
6.00 25.00 Unharvested (#)
19.00
6.00
Harvested (#)
6.00
22.00 28.00 Harvested (#)
8.00
20.00
Unharvested (%)
76.00
24.00 100.00 Unharvested (%)
76.00
24.00
Harvested (%)
21.43
78.57 100.00 Harvested (%)
28.57
71.43
* 77.4% of original grouped cases correctly classified
** 73.6% of cross-validated grouped cases correctly classified

50

Total
25.00
28.00
100.00
100.00

Table 1.8: ANCOVA adjusted means and standard errors of variables reported for both low and
moderate gradient reaches. Differences are significant (P < 0.05) when the 95% confidence
intervals do not include 0 and an asterisk denotes variables with significant differences.

Low
gradient
channels
Variable Mean SE
RPD
0.46 0.03
RPA*
0.60 0.06
POOLKM 67.51 5.80
D50*
25.15 3.94

Moderate
gradient
channels
Mean SE Difference 95% CI
0.41 0.02
0.05 -0.01 - 0.12
0.37 0.04
0.23 0.09 - 0.37
66.61 3.83
0.91 -13.74-15.56
41.95 2.60
-16.80 -26.77−-6.84

51

Table 1.9: ANCOVA adjusted means and standard errors of variables reported for both
unharvested and harvested reaches. Differences are significant (P < 0.05) when the 95%
confidence intervals do not include 0 and an asterisk denotes variables with significant
differences.

Unharvested
Variable Mean SE
RPD
0.45 0.02
RPA
0.46 0.05
POOLKM 73.09 4.72
D50*
42.03 3.21

Harvested
Mean SE Difference 95% CI
0.43 0.02 0.01
-0.04-0.07
0.50 0.04 -0.05
-0.17-0.08
61.03 4.56 12.06
-1.08-25.19
25.06 3.10 16.97
8.04-25.90

52

Table 1.10: Large wood density (pieces/km) in streams of Southeast Alaska organized by
channel type.

Channel type
FP
MM

Previous research

Forest condition

Murphy and Koski 1989
Robison and Beschta 1990
Bryant et al. 2004
Martin 2001
Ross 2013

Old growth
399
311
Undisturbed
337
330
Nonharvested
350-450 350-450
Some harvest
390
355
At least 1 bank harvested
381
356

53

Table 1.11: Data characterizing aquatic and riparian habitat in and along 28 study sites on Sealaska-owned land sampled in 2011.

Site code
9
34
52
68
69
74
77
83
102
Creek name
Perry Unnamed Unnamed Unnamed Unnamed Unnamed Unnamed Unnamed Unnamed
Latitude
-132.67 -133.11 -132.93 -133.18 -133.11 -133.03 -133.02 -132.94 -133.22
Longitude
55.28
55.62
55.61
55.62
55.64
55.64
55.63
55.66
55.6
Year
2011
2011
2011
2011
2011
2011
2011
2011
2011
Channel type
MM
MM
MM
FP
MM
MM
MM
MM
MM
Variable name
Bankfull width (m)
3.53
1.57
1.70
2.17
2.53
6.23
1.93
4.92
2.60
Bankfull depth (m)
0.17
0.12
0.14
0.17
0.14
0.22
0.06
0.34
0.11
Mean gradient (%)
3.17
3.33
2.17
2.00
4.83
2.50
2.50
4.17
2.50
Visual confinement*
2.00
2.00
2.00
2.00
2.00
1.00
1.00
2.00
2.00
Relative pool area (%)***
57.58
45.23
37.26
95.05
34.17
49.68
60.57
57.21
38.98
Pool density (#/km)***
94.53 125.87
74.77
68.32
78.26
60.24
98.77
73.47
47.30
Mean residual pool depth (m)***
0.50
0.26
0.29
0.57
0.28
0.46
0.25
0.61
0.35
Pools formed by wood (proportion)
0.45
0.60
0.29
0.00
0.56
0.79
1.00
0.67
0.73
Fine (proportion ≤ 4 mm )
0.02
0.31
0.11
0.24
0.02
0.04
0.35
0.01
0.09
D50***
39.40
9.19
13.00
13.93
43.71
43.71
5.46
27.86
20.39
LW pieces/km (#)
452.74 319.44 292.45 676.83 380.53 572.29 790.12
608.16
317.57
Mean length of functioning recruits (m)
10.13
5.50
3.00
5.63
4.00
9.28
4.36
7.00
5.67
Proportion functioning recruits
0.46
0.46
0.21
0.33
0.63
0.48
0.34
0.50
0.39
LW recruitment by erosion (proportion)
0.43
0.46
0.29
0.50
0.37
0.73
0.31
0.33
0.48
LW recruitment by wind (proportion)
0.57
0.46
0.50
0.00
0.63
0.20
0.25
0.50
0.30
Deciduous riparian trees (proportion)
0.20
0.00
0.06
0.00
0.03
0.11
0.00
0.19
0.00
Large and extra-large trees (proportion)
0.00
0.10
0.09
0.00
0.06
0.05
0.00
0.00
0.03

54

Table 1.11: (cont’d).

Site code
112
123
126
174
176
183
187
189
374
Creek name
Unnamed
Lime Unnamed Unnamed Dog Salmon Unnamed Steelhead Unnamed Unnamed
Latitude
-133.1 -133.02 -133.01 -132.97
-132.99 -133.09 -132.94 -132.94 -132.7
Longitude
55.62
55.63
55.64
55.62
55.62
55.62
55.65
55.65
55.24
Year
2011
2011
2011
2011
2011
2011
2011
2011
2011
Channel type
FP
FP
FP
MM
MM
MM
FP
MM
MM
Variable name
Bankfull width (m)
3.30
10.67
5.30
4.33
6.73
2.20
25.13
16.40
2.50
Bankfull depth (m)
0.08
0.13
0.09
0.28
0.23
0.12
0.36
0.31
0.09
Mean gradient (%)
1.83
1.17
1.50
6.17
2.17
4.67
1.83
3.33
3.00
Visual confinement*
1.00
1.00
2.00
2.00
1.00
2.00
1.33
1.67
1.33
Relative pool area (%)***
32.23
70.94
77.11
24.81
74.89
17.69
22.87
75.54
36.27
Pool density (#/km)***
58.82
59.88
63.43
45.11
75.76
53.76
9.86
42.55
54.95
Mean residual pool depth (m)***
0.38
0.59
0.60
0.32
0.61
0.26
0.84
0.90
0.26
Pools formed by wood (proportion)
0.68
0.81
0.71
0.20
0.38
0.62
0.60
0.54
0.45
Fine (proportion ≤ 4 mm )
0.17
0.10
0.12
0.03
0.02
0.08
0.04
0.04
0.03
D50***
17.75
17.75
16.56
43.71
27.86
19.03
59.71
43.71
29.86
LW pieces/km (#)
310.16 447.11 392.86 417.91
376.74 302.08 239.60 330.32 197.67
Mean length of functioning recruits (m)
6.69
10.17
8.90
3.00
6.60
4.80
7.29
9.25
5.25
Proportion functioning recruits
0.52
0.72
0.58
0.33
0.45
0.29
0.44
0.73
0.50
LW recruitment by erosion (proportion)
0.62
0.48
0.64
1.00
0.91
0.53
0.88
0.94
0.63
LW recruitment by wind (proportion)
0.23
0.48
0.26
0.00
0.00
0.41
0.12
0.03
0.00
Deciduous riparian trees (proportion)
0.07
0.58
0.00
0.49
0.61
0.00
0.51
0.46
0.41
Large and extra-large trees (proportion)
0.02
0.06
0.08
0.04
0.01
0.04
0.02
0.08
0.03

55

Table 1.11: (cont’d).

Site code
Creek name
Latitude
Longitude
Year
Channel type
Variable name
Bankfull width (m)
Bankfull depth (m)
Mean gradient (%)
Visual confinement*
Relative pool area (%)***
Pool density (#/km)***
Mean residual pool depth (m)***
Pools formed by wood (proportion)
Fine (proportion ≤ 4 mm )
D50***
LW pieces/km (#)
Mean length of functioning recruits (m)
Proportion functioning recruits
LW recruitment by erosion (proportion)
LW recruitment by wind (proportion)
Deciduous riparian trees (proportion)
Large and extra-large trees (proportion)

376
Unnamed
-132.71
55.24
2011
MM
2.07
0.10
5.17
1.67
25.72
52.63
0.20
0.82
0.07
22.63
265.96
5.14
0.58
0.17
0.58
0.11
0.00

379
414
507
511
519
599
601
Deer Unnamed Unnamed Unnamed Unnamed Unnamed Unnamed
-132.71 -135.56 -135.67 -135.67 -135.56 -132.77 -132.83
55.24
58.17
58.11
58.13
58.15
55.29
55.27
2011
2011
2011
2011
2011
2011
2011
FP
MM
MM
MM
MM
FP
MM
8.33
0.21
2.00
1.00
43.30
46.43
0.48
0.50
0.03
46.85
421.99
9.20
0.41
0.76
0.19
0.73
0.10

2.50
0.12
3.00
1.67
35.85
87.72
0.26
0.58
0.04
27.86
201.75
4.29
0.88
0.33
0.22
0.23
0.10

56

2.43
0.17
4.00
2.00
20.29
58.82
0.22
0.33
0.06
25.99
245.10
3.00
0.43
0.71
0.29
0.58
0.09

1.00
0.10
4.33
2.00
16.04
50.00
0.26
0.80
0.10
23.43
125.00
6.00
0.25
0.75
0.25
0.09
0.09

1.63
0.11
5.17
2.00
13.18
32.97
0.31
0.67
0.05
48.50
114.94
9.00
0.13
0.13
0.88
0.44
0.00

10.40
0.24
1.83
1.33
49.71
36.90
0.49
0.73
0.10
28.84
270.37
9.00
0.52
0.88
0.12
0.26
0.16

5.90
0.20
5.50
2.00
40.83
87.25
0.29
0.05
0.01
76.11
457.52
4.67
0.53
0.59
0.35
0.59
0.11

Table 1.11: (cont’d).

Site code
Creek name
Latitude
Longitude
Year
Channel type
Variable name
Bankfull width (m)
Bankfull depth (m)
Mean gradient (%)
Visual confinement*
Relative pool area (%)***
Pool density (#/km)***
Mean residual pool depth (m)***
Pools formed by wood (proportion)
Fine (proportion ≤ 4 mm )
D50***
LW pieces/km (#)
Mean length of functioning recruits (m)
Proportion functioning recruits
LW recruitment by erosion (proportion)
LW recruitment by wind (proportion)
Deciduous riparian trees (proportion)
Large and extra-large trees (proportion)

605
608
Natzuhini Natzuhini
-132.79 -132.82
55.29
55.29
2011
2011
FP
FP
12.77
0.39
0.50
1.00
63.76
26.19
0.82
0.55
0.24
8.88
240.20
11.20
0.71
0.71
0.24
0.50
0.30

18.93
0.81
0.50
1.00
68.11
58.69
0.60
0.88
0.46
4.76
431.46
10.50
0.65
0.71
0.25
0.59
0.00

57

Table 1.12: Data provided by John McDonell, 2013, U.S. Forest Service, characterizing streams in the Tongass National Forest having
unharvested riparian zones.

Site code
CHAN
CORN
DRINK
DBAYL
DBAYU
Creek name
Chanterelle Corner Bay Drinkingwater Dry Bay Lower Dry Bay Upper
Latitude
-132.79
-135.11
-132.25
-133.6
-133.58
Longitude
55.65
57.68
55.26
57.63
57.63
Year
2008
2006
2008
2006
2006
Channel type
MM
MM
MM
FP
FP
Variable name
Bankfull width (m)
4.85
2.70
4.80
5.50
5.40
Bankfull depth (m)
0.26
0.21
0.47
0.43
0.36
Relative pool area (%)
28.79
38.26
29.12
62.68
38.39
Pool density (#/km)
40.98
95.63
52.63
84.81
77.52
Mean residual pool depth (m)
0.41
0.30
0.38
0.36
0.34
D50
49.39
40.11
60.21
34.68
21.23

58

EMER
Emerald
-132.01
55.88
2008
MM
4.00
0.45
69.79
104.76
0.36
41.93

FARR FRESH
Farragut Freshwater
-133.27
-135.3
57.19
57.92
2008
2008
FP
MM
5.25
0.22
60.01
79.44
0.56
30.00

3.85
0.40
11.72
81.97
0.36
24.35

Table 1.12: (cont’d).

Site code
GUNS
GYPM
Creek name
Gunsight Gypsy Main
Latitude
-131.37
-132.19
Longitude
55.49
56.45
Year
2008
2008
Channel type
MM
MM
Variable name
Bankfull width (m)
9.65
11.60
Bankfull depth (m)
0.27
0.39
Relative pool area (%)
75.19
29.68
Pool density (#/km)
64.52
49.65
Mean residual pool depth (m)
0.64
0.69
D50
33.76
81.73

HOYA JENK
KEG
MIRK MONT OHMER OXBOW PACK
Hoya Jenkins
Keg Mirkwood Montana Ohmer Oxbow Packer
-131.71 -132.18 -133.13
NA -131.18 -132.71 -132.59 -131.37
56.18
56.43
55.3
NA
55.35
56.58
55.6
55.43
2006
2008
2008
2008
2008
2008
2008
2008
FP
FP
FP
MM
MM
FP
MM
MM
7.60
0.50
89.75
48.19
0.48
27.00

5.00
0.47
82.13
74.38
0.51
38.36

59

15.05
0.52
27.83
70.42
0.67
48.29

3.30
0.21
17.19
82.73
0.24
37.42

3.18
0.36
28.34
56.00
0.24
55.91

8.80
0.47
42.70
59.60
0.56
67.80

2.50
0.25
33.32
86.67
0.28
45.30

7.40
0.32
23.01
31.91
0.41
65.77

Table 1.12: (cont’d).

Site code
PERK SALT SFMONT TUNE VIAL
Creek name
Perkins
Salty SF Montana Tunehean
Vial
Latitude
-132.17 -131.56
NA -133.46 -132.8
Longitude
54.95
55.53
NA
56.62 56.28
Year
2008
2008
2009
2008 2008
Channel type
MM
MM
MM
MM
MM
Variable name
Bankfull width (m)
1.30
2.20
4.00
5.80 11.25
Bankfull depth (m)
0.28
0.16
0.41
0.22
0.23
Relative pool area (%)
68.32
46.21
5.43
18.41 18.94
Pool density (#/km)
135.80
51.28
82.53
70.51 26.14
Mean residual pool depth (m)
0.29
0.35
0.36
0.46
0.72
D50
18.65
25.24
55.42
51.57 71.20

60

Southeast
Alaska

(A) Chichagof
Island

(B) Prince of Wales
Island

C

Legend
Native land land
Native owned(in tax
Coast
Coast
!

Meters
0

blocks)

Study sites
Study Sites

20,000

Figure 1.1: Maps of (A) Chichagof Island and (B) Prince of Wales Island showing locations of
stream reaches sampled (n=28) on timberlands owned by Sealaska Corporation during the
summer of 2011.

61

100
90

Cumulative percent

80
70
60
50

One

40

Two

30
20
10
0
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990
Harvest year

Figure 1.2: Cumulative frequencies of earliest Sealaska Corporation harvest year organized by
number of banks that experienced riparian timber harvest.

62

700
600

Count

500

400
300
200
100
0
Small

Medium
Large
Tree size class

X-Large

Figure 1.3: Frequency of 876 riparian trees organized by size class and vegetation type. Black
bars represent coniferous trees, light gray bars represent deciduous trees, and dark gray bars
represent standing snags.

63

Percent (%)

100
90
80
70
60
50
40
30
20
10
0
Low

Moderate

Figure 1.4: Relative amount (%) of riparian trees organized by low vs. moderate gradient
streams. Black bars represent small (10-30 cm dbh), light gray bars represent medium (30-60 cm
dbh), and dark gray bars represent large and extra-large (60-90 & > 90 cm dbh) riparian trees.

64

Percent (%)

100
90
80
70
60
50
40
30
20
10
0
One

Two

Figure 1.5: Relative amount (%) of riparian trees organized by number of banks harvested.
Black bars represent small (10-30 cm dbh), light gray bars represent medium (30-60 cm dbh),
and dark gray bars represent large and extra-large (60-90 & > 90 cm dbh) riparian trees.

65

Percent (%)

100
90
80
70
60
50
40
30
20
10
0
Low

Moderate

Figure 1.6: Relative amount (%) of riparian trees organized by low vs. moderate gradient. Black
bars represent conifers, light gray bars represent deciduous, and dark gray bars represent snags.

66

Percent (%)

100
90
80
70
60
50
40
30
20
10
0
One

Two

Figure 1.7: Relative amount (%) of riparian trees organized by number of banks harvested.
Black bars represent conifers, light gray bars represent deciduous, and dark gray bars represent
snags.

67

(a)

(b)

Figure 1.8: Percent distribution frequencies of (a) diameter and (b) length of 2237 LW pieces
measured across all 28 Sealaska stream reaches (minimum diameter ≥ 10 cm and ≥ 2 m length).
X-axis values equal to minimum length requirement for 3 m length classes.

68

Percent (%)

100
90
80
70
60
50
40
30
20
10
0
Low

Moderate

Figure 1.9: Relative amount (%) of large wood pieces organized by low vs. moderate gradient
streams. Black bars represent small (10-30 cm dbh), light gray bars represent medium (30-60 cm
dbh), and dark gray bars represent large and extra-large (60-90 & > 90 cm dbh) pieces of wood.

69

Percent (%)

100
90
80
70
60
50
40
30
20
10
0
One

Two

Figure 1.10: Relative amount (%) of large wood pieces organized by number of banks harvested.
Black bars represent small (10-30 cm dbh), light gray bars represent medium (30-60 cm dbh),
and dark gray bars represent large and extra-large (60-90 & > 90 cm dbh) pieces of wood.

70

Figure 1.11: Box plots showing decay class of LW recruits organized by size class. See Methods
for decay class and size class descriptions.

71

300
250

Count

200
150
100
50
0
Green

Twig

Branch
Decay class

Nub

Old

Figure 1.12: Frequency of LW recruits organized by decay class. Darker bars represent recruits
that functioned while lighter bars represent recruits that did not function.

72

300
250

Count

200
150
100
50
0
Small

Medium
Large
LW size class

X-Large

Figure 1.13: Frequency of LW recruits organized by size class. Darker bars represent recruits
that function while lighter bars represent recruits that did not function.

73

Variable weightings

Axis 1 -- Stream size

Variable weightings

Axis 2 -- Stream power

Figure 1.14: Results of the PCA of 17 habitat measures. The four components
explained 72.93% of the variation in the habitat data. Axis 1 explained 30.48%, Axis
2, 16.60%, Axis 3, 13.44%, and Axis 4, 12.42%. The y-axes show the weights on each
variable for the individual components. Black bars indicate variables with weights
with an absolute value greater than 0.55. Variable descriptions are included in Table
1.2.

74

Figure 1.14 (cont’d):

Variable weightings

Axis 3 -- LW recruitment

Variable weightings

Axis 4 -- LW and pool density

75

76

PLXL

PFNR

BFWD*

DCRT

RPA

BFDP*

POOLKM

RPD

FNRL

EROSLW

CONF

FINE*

WINDLW

D50

GRAD

Axis 5 -- Large riparian trees and functioning recruits

WOOD

1.00
0.80
0.60
0.40
0.20
0.00
-0.20
-0.40
-0.60
-0.80
-1.00

LWKM

Variable weightings

Figure 1.14 (cont’d)

Figure 1.15: Cumulative percent distributions of (a) D50, (b) RPD, (c) POOLKM, and (d) RPA.
Black triangles represent unharvested streams and white boxes represent harvested streams.

77

Figure 1.15 (cont’d):

78

LITERATURE CITED

79

LITERATURE CITED

Alaska Department of Natural Resources. 2010. ADNR road condition/fish passage monitoring
year 5, Pages 1-4 in Alaska sustainable salmon fund project completion report.
AFRPA. 1978. Alaska’s Forest Resources and Practices Act AS 41.17.
Allan, J. D., M. S. Wipfli, J. P. Caouette, A. Prussian, and J. Rogers. 2003. Influence of
streamside vegetation on inputs of terrestrial invertebrates to salmonid food webs.
Canadian Journal of Fisheries and Aquatic Sciences 60:309-320.
Allan, J. D. 2004. Landscapes and riverscapes: the influence of land use on stream ecosystems.
Annual Review of Ecology Evolution and Systematics 35:257-284.
Benda, L. E. 1990. The influence of debris flows on channels and valley floors in the Oregon
Coast Range, USA. Earth Surface Processes and Landforms 15:457-466.
Benda, L. E., P. Bigelow, and T. M. Worsley. 2002. Recruitment of wood to streams in oldgrowth and second-growth redwood forests, northern California, USA. Canadian Journal
of Forest Research 32:1460-1477
Benda, L. E., N. L. Poff, D. Miller, T. Dunne, G. Reeves, G. Pess, and M. Pollock. 2004. The
network dynamics hypothesis: how channel networks structure riverine habitats.
Bioscience 54:413-427.
Bilby, R. E., and J. W. Ward. 1989. Changes in characteristics and functions of woody debris
with increasing size of streams in Western Washington. Transactions of the American
Fisheries Society 118:368-378
Bilby, R. E., and L. J. Wasserman. 1989. Forest practices and riparian management in
Washington state: data based regulation development. Pages 87-94 in R. E. Gresswell, B.
A. Barton, and J. L. Kershner, editors. Practical approaches to riparian resource
management. U.S. Bureau of Land Management, Billings, Montana.
Bisson, P. A., J. L. Nielsen, 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. Pages 62-73 in N. B. Armantrout, editor. Acquisition and utilization of
aquatic habitat inventory information. Proceedings of a symposium held 28-30 October,
1981, Portland Oregon. Western Division, American Fisheries Society, Bethesda,
Maryland.

80

Bisson, P. A., and J. R. Sedell. 1984. Salmonid populations in streams in clear-cut vs. old-growth
forests of western Washington. Pages 121-129 in W. R. Meehan, T. R. Merrell, Jr., and
T.A. Hanley editors Proceedings – Symposium on Fish and Wildlife Relationships in Old
Growth Forests. American Institute of Fishery Research Biologists, Asheville, N.C.

Bisson, P. A., R. E. Bilby, M. D. Bryant, C. A. Dolloff, 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. Pages 143-191 in E. 0. Salo and T. W.
Cundy editors Streamside management: forestry and fishery interactions. Contribution
No. 57, University of Washington, Institute of Forest Resources, Seattle.
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.
Boyer, E. W., C. L. Goodale, N. A. Jaworsk, and R. W. Howarth. 2002. Anthropogenic nitrogen
sources and relationships to riverine nitrogen export in the northeastern USA.
Biogeochemistry 57:137-169.
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.
Bryant, M. D. and R. D. Woodsmith. 2009. The response of salmon populations to geomorphic
measurements at three scales. North American Journal of Fisheries Management 29:549559.
Buffington, J. M., T. E. Lisle, R. D. Woodsmith, and S. Hilton. 2002. Controls on the size and
occurrence of pools in coarse-grained forest rivers. River Research and Applications
18:507-531.
Buffington, J. M., and D. R. Montgomery. 1999. Effects of hydraulic roughness on surface
textures of gravel-bed rivers. Water Resources Research 35(11):3507-3521.
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, U.S.A. Canadian Journal of Fisheries and Aquatic Sciences
47(6):1103-1111.
Chapman, D. W. 1988. Critical review of variables used to define effects of fines in redds of
large salmonids. Transactions of the American Fisheries Society 117:1-21.
Cohen M. J. and M. T. Brown. 2007. A model examining hierarchical wetland networks for
watershed stormwater management. Ecological Modeling 201:179-193.

81

Deal, R. L., P. E. Hennon, E. H. Orlikowska, and D. V. D’Amore. 2004. Stand dynamics of
mixed red alder-conifer forests of southeast Alaska. Canadian Journal of Forest Research
34:969-980.
Dodson, E. K., a. Ares, and K. J. Puettmann. 2012. Early responses to thinning treatments
designed to accelerate late successional forest structure in young coniferous stands of
western Oregon, USA. Canadian Journal of Forest Research 42:345-355.
Dolloff, C. A. Warren, M. L. Jr. 2003. Fish relationships with large wood in small streams
American Fisheries Society Symposium 37:179-193.
Feller, M. C. 1981. Effects of clearcutting and slashburning on stream temperature in
southwestern British Columbia. Journal of the American Water Resources Association
17:863-867.
Frissell, C.A.,W. J. Liss, C.E. Warren and M.D. Hurley. 1986. A hierarchical framework for
stream habitat classification: viewing streams in a watershed context. Environmental
Management, 10: 199–214.
Garman, S. L., J. H. Cissel, and J. H. Mayo. 2003. Accelerating development of late-successional
conditions in young managed Douglas-fir stands: a simulation study. Gen. Tech. Rep.
PNW-GTR-557. Portland, Oregon: U.S. Department of Agriculture, Forest Service,
Pacific Northwest Research Station. 57 p.
Gomi, T., A. C. Johnson, R. L. Deal, P. E. Hennon, E. H. Orlikowska, and M. S. Wipfli. 2006.
Factors affecting distribution of wood, detritus, and sediment in headwater streams draining
managed young growth red alder-conifer forests in southeast Alaska. Canadian Journal of
Forest Research 36:725-737.
Gomi, T., R. C. Sidle, M. D. Bryant, and R. D. Woodsmith. 2001. The characteristics of woody
debris and sediment distribution in headwater streams, southeastern Alaska. Canadian
Journal of Forest Research 31:1386-1399.
Grant, G. E., and A. L. Wolff. 1991. Long term patterns of sediment transport after timber
harvest, western Cascade Mountains, Oregon, USA. Pages 31-40 in E. N. Peters, E. D.
Walling, editors. Sediment and stream water quality in a changing environment: trends
and explanation. IAHS symposium, Vienna
Gregory, S. V., F. J. Swanson, W. A. McKee, and K. W. Cummins. 1991. An ecosystem
perspective of riparian zones: focus on links between land and water. Bioscience 41:540551.
Harrelson, C. C., Rawlins, C. L., Potyondy, J. P. 1994. Stream channel reference sites: an
illustrated guide to field technique. General Technical Report, RM-245, U.S. Department
of Agriculture, Forest Service, Fort Collins, Colorado.

82

Harris, A. S., and W. A. Farr. 1974. The forest ecosystem of southeast Alaska. USDA Forest
Service General Technical Report PNW-25.
Heifetz, J., M. L. Murphy, and K. V. Koski. 1986. Effects of Logging on Winter Habitat of
Juvenile Salmonids in Alaskan Streams. North American Journal of Fisheries
Management 6(1):52-58.
Hetrick, N. J., M. A. Brusven, W. R. Meehan, and T. C. Bjornn. 1998. Changes in solar input,
water temperature, periphyton accumulation, and allochthonous input and storage after
canopy removal along two small salmon streams in southeast Alaska. Transactions of the
American Fisheries Society 127:859-875.
Hicks, B. J., R. L. Beschta, and R. D. Harr. 1991. Long-term changes in streamflow following
logging in western Oregon and associated fisheries implications. Water Resources
Bulletin American Water Resources Association 27:217-226.
Hynes, H. B. N. 1975. The stream and its valley. Verhandlungen Internationale Vereinigung für
Theoretische und Angewandte Limnologie 19:1–15.
Imholt, C., C. Soulsby, I. A., Malcolm, and C. N. Gibbins. 2013. Influence of contrasting
riparian forest cover on stream temperature dynamics in salmonid spawning and nursery
streams. Ecohydrology 3:380-392.
Johnson, S. L. and J. A. Jones. 2000. Stream temperature responses to forest harvest and debris
flows in western Cascades, Oregon. Canadian Journal of Fisheries and Aquatic Sciences
57:30-39.
Jones, J. A., and G. E. Grant. 1996. Peak flow responses to clear-cutting and roads in small and
large basins, western Cascades, Oregon. Water Resources Research 32:959-974.
Kaiser, H. F. 1960. The application of electronic computers to factor analysis. Educational and
Psychological Measurement 20:141-151.
Knighton, D. 1998. Fluvial forms and processes: a new perspective. John Wiley and Sons, Inc.,
New York, New York.
Krumbein, W. C. 1934. Size frequency distributions of sediments. Journal of Sediment Petrology
4:65-77.
Lisle, T. E. 1987. Using residual depths to monitor pool depths independently of discharge.
Research Note PSW-394. Berkely, CA: Pacific Southwest Forest and Range Experiment
Station, U.S. Forest Service, Department of Agriculture.
Martin, D.J. 2001. The influence of geomorphic factors and geographic region on large woody
debris loading and fish habitat in Alaska coastal streams. North American Journal of
Fisheries Management 21:429-440
83

Martin, D. J. and L. E. Benda. 2001. Patterns of instream wood recruitment and transport at
the watershed scale. Transactions of the American Fisheries Society 130:940-958.
Martin, D. J., and R. A. Grotefendt. 2007. Stand mortality in buffer strips and the supply of
woody debris to streams in southeast Alaska. Canadian Journal of Forest Research
37:36-49.
May, C. L., and D. C. Lee. 2004. The relationships among in-channel sediment storage, pool
depth, and summer survival of juvenile salmonids in Oregon Coast Range streams. North
American Journal of Fisheries Management 24:761-774.
McIntosh, B. A., J. R. Sedell, R. F. Thurow, S. E. Clarke, and G. L. Chandler. 2000. Historical
changes in pool habitats in the Columbia River Basin. Ecological Applications 10:14781496.
Montgomery, D. R. 1999. Process domains and the river continuum. Journal of American Water
Resources Association 35:397-410.
Montgomery, D. R., T. B. Abbe, J. M. Buffington, N. P. Peterson, K. M. Schmidt, and J. D.
Stock. 1996. Distribution of bedrock and alluvial channels in forested mountain drainage
basins. Nature 381:587-589.
Montgomery, D. R., J. M. Buffington, R. D. Smith, K. M. Schmidt, and G. Pess. 1995. Pool
spacing in forest channels. Water Resources Research 31:1097-1105.
Moring, J. R., and R. L. Lantz. 1975. The Alsea Watershed Study: effects of logging on the
aquatic resources of three headwater streams of the Alsea River, Oregon. Oregon
Department of Fish and Wildlife. Fisheries Resource Report 9:66.
Murphy, M. L. 1995. Forestry impacts on freshwater habitat of anadromous salmonids in the
Pacific Northwest and Alaska—Requirements for protection and restoration. NOAA
Coastal Ocean Program Decision Analysis Series No. 7. NOAA Coastal Ocean Office,
Silver Spring, MD. 156 pp.
Murphy, M. L., and K. V. Koski. 1989. Input and depletion of woody debris in Alaska streams
and implications for streamside management. North American Journal of Fisheries
Management 9:427-436
Murphy, M. L., J. Heifetz, S. W. Johnson, K. V. Koski, and J. F. Thedinga. 1986. Effects of
clear-cut logging with and without buffer strips on juvenile salmonids in Alaskan
streams. Canadian Journal of Fisheries and Aquatic Sciences 43:1521-1533.
Olson, D.M., Dinerstein, E., Wikramanayake, E.D., Burgess, N.D., Powell, G.V.N., Underwood,
E.C., D'Amico, J.A., Itoua, I., Strand, H.E., Morrison, J.C., Loucks, C.J., Allnutt, T.F.,
Ricketts, T.H., Kura, Y., Lamoreux, J.F., Wettengel, W.W., Hedao, P. & Kassem, K.R.
84

2001. Terrestrial ecoregions of the world: a new map of life on Earth. BioScience
51:933-938.
Orlikowska, E. H., R. L. Deal, P. E. Hennon, and M. S. Wipfli. 2004. The role of red alder in
riparian forest structure along headwater streams in Southeastern Alaska. Northwest
Science 78(2):111-123.
Paustian, S. J., K. Anderson, D. Blanchet, S. Brady, M. Cropley, J. Edgington, J. Fryxell,
G. Johnejack, D. Kelliher, M. Kuehn, S. Maki, R. Olsen, J. Seesz, and M. Wolanek.
1992. A channel type user’s guide for the Tongass National Forest, southeast Alaska.
Technical Paper 26, U.S. Forest Service, Juneau, Alaska.
Quinn, T. P., and P. N. Peterson. 1996. The influence of habitat complexity and fish size on
over-winter survival and growth of individually marked juvenile coho salmon
(Oncorhynchus kisutch) in Big Beef Creek, Washington. Canadian Journal of Fisheries
and Aquatic Sciences 53:1555-1564.
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, C., and G. Host. 1994. Examining land use influences on stream habitats and
macroinvertebrates: a gis approach. Water Resources Bulletin American Water Resources
Association 30:729-738.
Roni, P., and T. P. Quinn. 2001. Distribution and size of juvenile salmonids in response to
placement of large woody debris in western Oregon and Washington streams. Canadian
Journal of Fisheries and Aquatic Sciences 58:282-292.
Robison, E. G., and R. L. Beschta. 1990. Identifying trees in riparian areas that can provide
coarse woody debris to streams. Forest Science 36:790-801.
Rosenfeld, J., M. Porter, and E. Parkinson. 2000. Habitat factors affecting the abundance and
distribution of juvenile cutthroat trout (Oncorhynchus clarki) and coho salmon
(Oncorhynchus kisutch). Canadian Journal of Fisheries and Aquatic Sciences 57:766-774.
Rot, B. W., R. J. Naiman, and R. E. Bilby. 2000. Stream channel configuration, landform, and
riparian forest structure in the Cascade Mountains, Washington. Canadian Journal
Fisheries Aquatic Sciences 57:699-707.
Scrivener, J. C., and M. J. Brownlee. 1982. An analysis of Carnation Creek gravel-quality data,
1973-1981, pages 154-176 in G. F. Hartman editor Proceedings of the Carnation Creek
Workshop: a ten-year review, Feb 24-26, 1982. Pacific Biological Station, Nanaimo,
British Columbia.
Schumm, S. A. 1977. The fluvial system. John Wiley and Sons, New York, New York.
85

Spies, T., M. Pollock, G. Reeves, and T. Beechie. 2013. Effects of riparian thinning on wood
recruitment: a scientific synthesis. Science Review Team Wood Recruitment Subgroup
46 p.
Stevens, D.L., Olsen, A.R. 2004. Spatially balanced sampling of natural resources. Journal of the
American Statistical Association. 99:262-278.
Sutherland, A. B., J. L. Meyer, and E. P. Gardiner. 2002. Effects of land cover on sediment
regime and fish assemblage structure in four southern Appalachian streams. Freshwater
Biology 47:1791-1805.
Swanson, D. N., and D. A. Marion. 1991. Landslide response to timber harvest in Southeast
Alaska. Pages 10-49 to 10-56 in Shou-Shan Fan and Yung Huang Kuo, editors.
Proceedings of the Fifth Federal Interagency Sedimentation Conference, Las Vegas,
Nevada, March 18-21, 1991, Volume 2.
Taylor, R. F. 1934. Yield of second-growth western hemlock-Sitka spruce stands in southeastern
Alaska. Tech. Bull. 412. Washington DC: U.S. Department of Agriculture. 34 p.
Townsend, C. R., S. Doledec, R. H. Norris, K. Peacock, and C. J. Arbuckle. 2003. The influence
of scale and geography on relationships between stream community composition and
landscape variables: description and prediction. Freshwater Biology 48:768-785.
U.S. Environmental Protection Agency. 1992. Streamwalk manual. U.S. EPA Region 10, Water
Management Division, Seattle, WA.
U.S. Forest Service, 2001. Aquatic Habitat Management Handbook, FSH-2090, U.S.
Department of Agriculture. Forest Service, Region 10, Juneau Alaska
Villarin, L.A., D.M. Chapin, and J.E. Jones. 2009. Riparian forest structure and succession in
second growth stands of the central Cascade Mountains, Washington, USA. Forest
Ecology and Management 257:1375-1385.
Wang, L., J. Lyons, P. Rasmussen, P. Kanehl, P. Seelbach, T. Simon, M. Wiley, E. Baker, S.
Niemela, and M. Stewart. 2003. Influences of landscape and reach scale habitat on stream
fish communities in the Northern Lakes and Forest ecoregion. Canadian Journal of
Fisheries and Aquatic Sciences 60:491-505.
Wang, L., P. W. Seelbach, and R. M. Hughes. 2006. Introduction to landscape influences on
stream habitats and biological assemblages. Pages 1-23 in R.M. Hughes, L. Wang, and P.
W. Seelbach, editors. Landscape influences on stream habitats and biological
assemblages. American Fisheries Society, Symposium 48, Bethesda, Maryland.
Wiley, M. J., S. L. Kohler, and P. W. Seelbach. 1997. Reconciling landscape and site based
views of aquatic stream communities. Freshwater Biology 37:133-148.

86

Wolman, M. G. 1954. A method of sampling coarse river bed material. Transactions of the
American Geophysical Union 35:951-956.
Woodsmith, R. D., and J. M. Buffington. 1996. Multivariate geomorphic analysis of forest
streams: implications for assessment of land use impacts on channel condition. Earth
Surface Processes and Landforms 21:377-393.
Woodsmith, R. D., J. R. Noel, and M. L. Dilger. 2005. An approach to effectiveness monitoring
of floodplain channel aquatic habitat: channel condition assessment. Landscape and
Urban Planning 72:177-204.
Wu, T. H., W. P. McKinnell III, and D. N. Swanson. 1978. Strength of tree roots and landslides
on Prince of Wales Island, Alaska. Canadian Geotechnical Journal 16:19-33.

87

CHAPTER 2
EFFECTS OF LANDSCAPE FACTORS ON LARGE WOOD IN STREAMS DRAINING
SECOND GROWTH WATERSHEDS OF SOUTHEAST ALASKA: COMPARING
NATURAL AND TIMBER HARVEST-RELATED INFLUENCES

Abstract
Large wood is an important resource to fishes and streams in Southeast Alaska, and
currently, much is known about influences of landscape factors in riparian zones, including
timber harvest activities, on large wood dynamics. However, comparatively less is known about
influences of landscape factors within catchments on streams of Alaska, due in part to a
historical lack of consistent catchment-scale data available for the region. This study addresses
these limitations by characterizing catchment-scale influences on large wood in streams of
Southeast Alaska. We first summarized multiple measures of landscape factors including natural
and anthropogenic variables in catchments of streams throughout the region. Next, we predicted
multiple measures of large wood habitat factors from a subset of streams using both natural and
timber harvest-related landscape factors summarized within catchments to identify those that
may be most influential to large wood. Natural factors including catchment area, deciduous
forests, wetlands, and stream slope all had significant influences on measures of large wood, as
did various measures of current and historical timber harvest in the region. Finally, we
compared relative amounts of variation in a group of large wood habitat factors explained by
natural vs. harvest-related landscape factors as a way to broadly compare types of influences on
the region’s streams. Results indicate that in our study streams, timber harvest explained more
variation in large wood factors, underscoring its importance to stream habitat and the importance
of considering catchment landscape factors in efforts to manage streams of Southeast Alaska.

88

Introduction
Large wood (LW) is acknowledged as an important component of habitat in stream
systems. Large wood can directly act as habitat for aquatic organisms, and it can influence
numerous other physical and chemical factors important for supporting aquatic communities
such as channel morphology and habitat heterogeneity; size and distribution of sediment; and
nutrient cycling (Bisson et al. 1987, Gomi et al. 2001, Brookshire and Dwire 2003). For
example, LW in stream channels can serve as cover for both fishes and aquatic
macroinvertebrates (McMahon and Hartman 1989, Bryant and Woodsmith 2009), and
macroinvertebrates are also known to rely on LW as a food source (Sweeney 1993). LW can
slow water velocities, and young fishes will hide from predators within the structure provided by
LW. Feeding adult and juvenile fishes use cover to search for drifting invertebrates (Rosenfeld
et al. 2000, Roni and Quinn 2001), and large wood can also create cover by influencing channel
morphology and forming pools (Fausch and Northcote 1992).
In streams of Southeast Alaska, LW is acknowledged as especially important for
supporting salmonids, a group of native fishes that are critical to the economy and a major focus
of management actions in the region. Large wood in streams of Southeast Alaska benefits
salmonids in a number of ways, including promoting channel complexity, forming pools, and
retaining coarse substrate. Robison and Beschta (1990) found LW loadings to be positively
related to heterogeneous channel characteristics in five low-gradient streams of Southeast
Alaska. Pools are acknowledged as being critical habitat for resident trout and juvenile salmon
(Bryant et al. 2005, Bryant et al. 2007). Scour pools are most commonly associated with LW in
low and moderate gradient streams of Southeast Alaska (Martin 2001) and form when water is
forced over and/or under wood structures, scouring the streambed and creating deep, low
89

velocity habitats which can be utilized by adult salmonids for resting purposes during migration,
as well as providing refuge for young and resident fishes during periods of flooding and droughts
(Lisle 1986). Pools can also form when a log or accumulation of logs acts to hold water back,
creating habitats similar to those formed behind beaver dams. Deposition and transport of
organic matter and fine sediment are influenced by the presence of LW (Gomi et al. 2001).
Large wood also retains coarse substrates important for lithophilic spawners (including
salmonids), and it stabilizes banks and reduces erosion (Montgomery et al. 1996, Gomi et al.
2001).
Just as large wood influences stream habitats and organisms, the amount, type, and
movement of LW to and through stream channels is influenced by properties of the landscapes
drained by streams (e.g., Hynes 1975, Frissell et al. 1986, Wiley et al. 1997, Allan 2004).
Further, the influence of landscape factors on physical characteristics of streams have been
shown to be differentially affected by landscape factors operating over different spatial scales
including riparian zones vs. watersheds (e.g., Frissell et al. 1986, Townsend et al. 2003, Cohen
and Brown 2007). That differential influence results from specific mechanisms by which
landscape factors influence stream conditions (Allan 2004), and teasing apart the scale at which
landscape factors affect different components of stream habitat, including LW, is critical for
identifying restoration opportunities when habitat factors are found to be limiting.
Currently, much is known about influences of landscape factors in riparian zones on LW
in streams of Southeast Alaska. For example, forested riparian zones provide a majority of LW
to streams in the region (Murphy and Koski 1989, Martin and Grotefendt 2007). Bank erosion,
driven by climatic factors, geology, flooding frequency, and position within river networks
(Martin and Benda 2001), is acknowledged as an important LW input process. Landslides and
90

mass wasting events can deliver large quantities of LW to streams locally, and river valley
characteristics, such as slope, can also influence the potential for material to reach stream
channels (Benda 1990, Paustian et al. 1992, May and Gresswell 2004). Riparian tree
composition, including type and size of tress, also influences LW habitat factors such as the
relative amount of pools formed by wood and the size of LW functioning within stream channels
(Ross Chapter 1 this volume).
Compared to influences of landscape factors in riparian zones, less is known about
influences of landscape factors within watersheds on LW throughout Southeast Alaska. One
study, however, found catchment area to be related to aspects of LW storage and transport in
streams of the region. Longevity of debris jams decreased with stream size, and large streams
were found to move logs farther and more frequently than smaller streams (Martin and Benda
2001). This relationship likely results at least in part from hydrologic controls within a
catchment, as streams having greater catchment areas will deliver more water to and through
stream channels (Knighton 1998). In a similar manner, other natural and anthropogenic
landscape factors that influence catchment water routing are also likely influential in controlling
LW in streams of Southeast Alaska including amount of forest and wetland cover within
catchments (e.g., Roth et al. 1996, Cohen and Brown 2007). Further, anthropogenic disturbances
that alter these natural catchment influences may also have detectable effects on LW in stream
channels. In Southeast Alaska, large-scale timber harvest represents the greatest potential
anthropogenic landscape disturbance to streams, yet we are aware of no studies that relate
logging efforts throughout stream catchments to amount and type of LW in stream channels of
the region. Such understanding could provide critical information for managing the region’s

91

streams and ultimately provide insights for improvement of salmonid habitats in previously and
currently-logged watersheds.
To meet those needs, the goal of this study is to characterize influences of catchment
landscape factors, including both current and historical timber harvest activities, on LW in
streams of Southeast Alaska. One factor potentially contributing to the absence of research
characterizing catchment landscape influences on streams of the region may be the absence of
comparable landscape data summarized in stream catchments throughout Southeast Alaska.
Consequently, our first objective will be to summarize landscape factors for catchments
throughout Southeast Alaska using a geographic information system (GIS) and region-wide data
characterizing various catchment features. This summarization will provide insights into the
range of natural and anthropogenic landscape conditions across stream catchments of the region,
and it will inform our second objective, which will be to predict multiple measures of LW habitat
factors in a subset of the region’s streams from natural and timber harvest-related landscape
factors summarized within catchments to identify those that may be most influential to LW.
Finally, we will compare relative amounts of variation in a group of LW habitat factors
explained by natural vs. harvest-related landscape factors as a way to broadly compare types of
influences on the region’s streams. By addressing these objectives, we expect to gain a broader
understanding of influences on LW habitat in streams of Southeast Alaska. Our results can
ultimately be used to support the region’s fisheries by providing a framework capable of
summarizing landscape factors that are biologically and socio-culturally important as decision
makers are tasked with administering actions in a holistic manner, considering a myriad of
factors required for management.
Methods
92

Study area, catchment creation, and landscape data
Landscape data were summarized for this study from the entire region of Southeast
Alaska (Figure 1) in two spatial extents, local and network stream catchments. Streams
throughout this area were identified from the 1:63,360 scale National Hydrography Dataset
(NHD) with reaches defined as confluence to confluence stretches of water. For all stream
reaches of the region, local catchment boundaries were delineated from a 60 m digital elevation
model (DEM) to encompass the land area draining to the downstream end of a particular reach.
Additionally, network catchments were also delineated and defined as the entire upstream area
draining to the downstream end of any given reach; these steps were performed using a
geographic information system (GIS).
Various natural landscape variables were calculated and attributed to the local catchment
of each study reach. Data for network catchments were then created by aggregating data from all
upstream local catchments for each particular study reach. Natural landscape variables
associated with each reach include catchment area and catchment slope, derived from the 60 m
DEM and used to create catchment boundaries. The center of each cell in the DEM was used to
derive slope by taking the maximum rate of change in value from a single cell with all
neighboring cells. Topologically Integrated Geographic Encoding and Referencing (TIGER)
roads data were obtained from the U.S. Census Bureau and summarized as density within local
and network catchments. The National Wetlands Inventory was acquired for Alaska and
summarized in local and network catchments for the entire region. The National Land Cover
Database for Alaska (NLCD 2001) was also acquired, and variables used from this database
include measures of urban land use and forest type which were summarized as percentages of
total land cover in local and network catchments (Table 2.1).
93

Study site description
Measurements were taken in 2011 at twenty-eight sample reaches that drain second
growth watersheds for multiple LW characteristics on Prince of Wales and Chichagof Islands
following methods described in Ross (Chapter 1 this volume). Sampled reaches ranged from 72
to 507 m in length. All reaches sampled were identified as supporting anadromous fisheries
based on multiple datasets for the region (Paustian et al. 1992, Alaska Department of Fish and
Game 2011). Sampled reaches were additionally classified as either floodplain or moderate
gradient, mixed control stream channels following Paustian et al. (1992). These stream types
were selected in part due to their ability to retain abundant and functioning LW, as well as their
potential sensitivity to disturbances resulting from timber harvest (Paustian et al. 1992,
Woodsmith et al. 2005).
Timber harvest data
Timber harvest and logging road data were obtained from managers at Sealaska Timber
Corporation and from the U.S. Forest Service (USFS). All data were in a GIS format and
polygons represented harvested areas for timberlands. Network catchments of study streams
occasionally extended beyond Sealaska-owned land, so additional comparable timber harvest
and logging road data were obtained from the USFS for the Tongass National Forest to ensure
that all harvest-related activities could be accounted for. Due to changes in regulations regarding
timber harvest in Southeast Alaska over time, harvest data were summarized into three time
periods. The first time period includes harvest that occurred before 1980. The initial Forest
Resources and Practices Act (FRPA) for Alaska was adopted in 1978, and prior to the FRPA,
there were no standards regarding how timber could be harvested, felled, and transported in

94

proximity to anadromous streams in the region. While the initial FRPA required more stringent
guidelines for how timber activities could occur, there were no standards regarding timber
harvest within the riparian zone, and trees would regularly be harvested directly up to the stream
bank if valuable timber was present. In 1990, amendments to the FRPA required that 20 m uncut
riparian buffers be left along all streams on private timberlands that support anadromous fishes.
To match these regulations, the second period for which harvest data were summarized occurred
between 1980 and 1990, while the third included any harvest occurring after 1990. Additionally
there are two types of harvest summarized which include conventional clear-cut harvest and
selective helicopter harvest. Harvest data were summarized for the three time periods for both
local and network catchments in all 28 sampled reaches (Table 2.2). Logging road density
(m/km²) and logging road/stream crossing density (#/km²) were also calculated and summarized
for local and network catchments.
Large wood data
For this study, large wood (LW) is defined as any log at least 10 cm in diameter and 2 m
in length within the bankfull channel following Robison and Beschta (1990) and Martin (2001).
In the 28 study streams, all LW pieces were counted and measured within each 10 m long
segment in smaller streams (< 10 m average width) and every 20 m long segment in larger
streams (> 10 m average width). The diameter of each log was measured and assigned to one of
four size categories: small (10-30 cm), medium (30-60 cm), large (60-90 cm) and very large (>
90 cm). The length of each log was measured to the nearest 3 m. The number of wood-formed
pools was additionally counted to calculate the density (#/km) of wood formed pools for each
study reach.

95

All logs in which the source could be identified would be considered a LW recruit and
classified by its recruitment process. Woody debris recruitment processes include: bank erosion,
windthrow, broke stem (i.e., assumed windthrow), mortality, landslide, cut, beaver, or unknown.
Information on whether a LW recruit was in or above the active channel and whether it was a
functioning recruit was also recorded. Recruits in the channel were often interacting with the
active channel bed, while recruits above the channel were often at or near bankfull, spanning
across the channel but not typically associated with physical stream characteristics at lower
flows. To be considered functioning, a recruit must do any or all of the following: cause channel
scour or flow deflection, create a sediment retention zone, dam water, create an eddy, or catch
and hold other woody debris in place (Martin 2001).
Statistical methods
Data preparation
Distributional properties including the mean, range, and standard deviation of all LW
habitat, natural landscape, and harvest-related landscape variables were calculated and assessed
to remove those variables having very few records or that failed to characterize a range in
conditions across study sites. Normality and linearity of variables was also assessed by use of
histograms and p-p plots to identify those that would be suitable for analyses to identifying
relationships between LW habitat factors and landscape factors. All variables were transformed
to maximize linearity, and transformations occurred as follows: continuous data were
transformed using natural log, count data were transformed using square root, and proportions
were transformed using arcsin square root.

96

Selection of LW variables for analyses occurred based on their ecological relevance and
data distribution across study sites. Selected variables included average LW length in reaches,
large and extra-large sized LW density, medium sized LW density, and the ratio of the number of
recruits found in channel to those found above channel. The length of LW is important as longer
vs. smaller pieces have a greater ability to function within the stream channel. Similarly, the size
(i.e., dbh) of LW is important, as bigger pieces proportionately function more vs. small pieces, so
density of the largest pieces was also selected. Additionally, medium sized LW density was
selected for a number of reasons. It is functionally important to streams in a manner similar to
larger wood (Ross Chapter 1 this volume), the medium size class encompasses the average size
that deciduous alder tends to achieve (Worthington et al. 1962), and the maximum diameter at
breast height for this class (i.e., 60 cm) is about the largest that Sitka spruce will grow in 80 years
in Oregon where trees are on average larger than in Southeast Alaska (Smith et al. 1984). These
facts make it a relevant size to evaluate landscape influences in the second growth watersheds
targeted for this study. Finally, the ratio of LW recruits that are located within the stream
channel as opposed to positioned above the channel was selected as a measure of how often
recruits are actually functioning along a sampled reach in relation to the number of recruits
identified in the reach.
Landscape influences on LW habitat factors used in analysis included natural and timber
harvest-related landscape factors summarized in both local and network catchments. Natural
factors included: network catchment area (km²), network catchment forested wetlands (%), local
catchment max slope (%), and local catchment deciduous forest (%). Timber-harvest related
landscape variables included local catchment conventional harvest between 1980 and 1990, local
catchment other harvest post-1990, network catchment logging road/stream crossing density
97

(#/km²), and network catchment conventional harvest post-1990. These landscape variables
were selected based on 1) variable distributions throughout the study streams and 2) theorized
influences on LW habitat factors. Pearson’s correlation analysis on pairs of transformed
landscape variables was performed to ensure that measures were minimally redundant (R <
absolute value of 0.50, Table 2.3).
Best subsets multiple regression
Two sets of regression models were developed to relate LW habitat factors to timber
harvest and natural landscape factors. Due to catchment area’s known influence on LW (Martin
and Benda 2001, Ross this volume), each LW variable was initially regressed with only
catchment area to better compare the performance of subsequent models with additional
variables (Burnett et al. 2006). Models having an AIC near or lower than the model with
catchment area alone indicate that they perform as well or better than that model. Next, I
performed a best subsets regression (R leaps package) on each LW variable with the suite of
landscape and timber harvest factors to identify the five best models, characterized by those
having the largest R²adj value. I considered only significant models that contained no more than
four explanatory variables to avoid model overfitting (Johnston et al. 1990) and only those
within 2 AIC units of the “best” model. Collinearity was assessed based on a variance inflation
factor threshold of three (VIF < 3).
Constrained ordination analysis
A constrained ordination analysis was run in the program CANOCO to assess the relative
importance of timber harvest vs. natural landscape factors on LW in the study reaches. To
determine whether a canonical correspondence analysis (CCA) or redundancy analysis (RDA)
98

was more appropriate given the data distribution, a detrended correspondence analysis (DCA)
was first performed to assess the gradient lengths in the response dataset. This analysis
suggested that the data follow a linear distribution, making RDA a more appropriate technique
(ter Braak and Prentice 1988). RDA is a direct gradient analysis and can be used to quantify
variation in a set of response variables (in this case, LW variables) by constraining their
ordination to a linear combination of independent predictors (i.e., landscape variables) (ter Braak
and Prentice 1988). An equal number of harvest-related and other landscape variables were
chosen so as not to over or underestimate variance explained due to difference in numbers of
predictor variables.
Results
Regional trends of natural landscape factors in Southeast Alaska watersheds
Conditions of network catchments in Southeast Alaska showed considerable variation in
natural factors occurring across the landscape (Table 2.4). Catchment area ranged from less than
1 km² up to as large as 2168 km² across the entire region. Deciduous forests average 2.5%, but
ranged from 0-100% across all catchments. Evergreen forests were more prominent in the
region and network catchments averaged about 50%, while mixed forests were on average about
13% of catchment area. Percent of forested wetlands in network catchments ranged from 0100% but averaged 17% across all catchments in the region. Very little human influence is
present in Southeast Alaska catchments as low, medium, and high intensity urban land use all
averaged less than 0.25% in network catchments across the region with no agricultural land use
present in any catchments according to the 2001 NLCD. Timber harvest data were only

99

available for the Tongass National Forest so we were unable to summarize these factors for the
entire Southeast Alaska region.
Trends in natural landscape factors of study streams
Considerable variation was observed in natural landscape factors of catchments in the
sampled study streams. The maximum slope ranged from 1.21% to 7.13% and averaged 3.52%
locally (Table 2.1). Deciduous forests comprised on average about 7% of local catchment area.
Network catchment area averaged 4.24 km², and ranged from less than 1 km² to just below 20
km². The percent of network catchment area comprised of forested wetlands ranged from none
to a maximum of over 90%, but averaged slightly more than 25% across all study catchments.
Non-harvest related human influences were not evident in the study catchments as low-intensity
urban land use occurred in network catchments of only 5 sites and averaged 0.25% across all
sites, while only 4 sites had roads in their network catchments.
Trends in timber harvest-related factors of study streams
Conventional timber harvest occurring from 1980-1990 in local catchments was the most
pervasive timber harvest practice across study sites; it averaged over 40% across study site
catchments and exceeded 80% in one site’s local catchment (Table 2.2). Other harvest after
1990 averaged 10% of local catchment area and ranged from 0 to slightly more than 30% of
local catchments for sampled reaches. Conventional harvest occurring in network catchments
after 1990 averaged about 14% across sites with a max of 49%. Logging roads were present in
all network catchments, and logging road/stream crossing density averaged 3/km² across all sites
with a maximum of about 9/km². Conventional harvest before 1980 was relatively uncommon

100

within network catchments and ranged from 0 to more than 6% but averaged less than 1% across
all study sites.
Explaining variation in large wood habitat factors using natural and timber harvest-related
landscape factors
A substantial amount of variation (Adj. r² = 0.16-0.59) was explained by using a best
subsets regression approach that derived multiple significant linear models predicting average
LW length, the ratio of recruits that are located within the channel vs. recruits positioned above
the channel, and medium-sized LW density by using both natural and harvest-related landscape
variables. The procedure only derived one significant model predicting large and extra-large LW
density, and this included only timber harvest related variables.
Average large wood length
The average length of LW was significantly related to natural and harvest-related
landscape factors in 5 linear regression models. A substantial amount of variation (Adj. r²
ranging from 0.57 to 0.59, Table 2.5) was explained by models, and all 5 included network
catchment area, local catchment deciduous forest, and network catchment conventional harvest
occurring after 1990, with the “best” model containing only these three variables. In all models,
average length of LW in the study reaches was positively associated with these landscape
factors. Subsequent models included network catchment logging road density, network
catchment forested wetlands, local catchment conventional harvest from 1980-1990, and local
catchment other harvest after 1990. Network catchment area had the greatest contribution in all
models predicting LW length, followed by local deciduous forest, and network catchment
conventional harvest after 1990 based on standardized beta coefficients.
101

Large wood recruits located within the active channel vs. above the channel
The ratio of recruits that are in vs. above the active channel was best predicted by a
model incorporating three natural and one harvest-related landscape factors; it explained a
moderate amount of variation in this wood habitat factor (Adj. r² = 0.34, Table 2.6).
Conventional harvest occurring from 1980-1990 in the local catchment explained the most
variation and was positively associated with this habitat factor in all models. Local deciduous
forest, local max slope, and finally network catchment area were additionally included in this
model, with max slope having the only negative relationship with the habitat factor. Four
subsequent models that included both local catchment conventional harvest occurring from
1980-1990 and local max slope, also included local deciduous forest as having a positive
association in two of the remaining four models. Local other harvest after 1990 and network
catchment density of logging road/stream crossings were negatively related to the LW habitat
factor, and included in two of the five models.
Medium-sized large wood density
A moderate amount of variation (Adj. r² = 0.37, Table 2.7) was explained in mediumsized LW by two measures of timber harvest and two natural landscape factors. Variables
included in this model – network catchment conventional harvest after 1990, local deciduous
forest, network catchment forested wetlands, and local conventional harvest occurring from
1980-1990 – all had positive relationships with medium-sized large wood density. Three
subsequent models significantly explained variation (Adj. r² = 0.32 – 0.33) in this LW habitat
factor, and all of these included network catchment conventional harvest after 1990 and network
catchment forested wetlands, while local catchment deciduous and local catchment other harvest

102

after 1990 were included in two of the remaining three models. Local catchment other harvest
after 1990 had the only negative association with medium-sized LW density in any models.
Network catchment area was related to medium-size LW density in only one of the four
regression models and was the weakest contributing landscape factor included for that model.
Large and extra-large-sized large wood density
Large and extra-large-sized LW density was significantly related to only timber harvest
landscape factors. The amount of variation explained for this model was relatively low (Adj. r² =
0.16, Table 2.8). Harvest-related landscape variables explaining this LW habitat factor include
local catchment conventional harvest occurring from 1980-1990, which had a positive
association, while a negative association existed between this habitat factor and local other
harvest after 1990.
Assessing the relative importance of natural vs. harvest-related landscape factors in explaining
variation in large wood habitat factors
CCA variance partitions indicated that the four natural and four harvest-related landscape
factors evaluated in this study explained 44% of the variation in the set of LW habitat factors
tested (Figure 2.2). Of the explained variation, timber harvest-related factors explained about
54% and natural factors explained about 28%, while about 18% was explained by the
interactions among harvest-related and natural landscape factors.
Plots of landscape and habitat factors based on CCA axes scored supported some findings
shown by best subsets regression models (Figure 2.3). Catchment area, local deciduous forest,
and local catchment conventional harvest occurring from 1980-1990 were positively associated
with all four habitat factors along Axis 1, while local other harvest after 1990 was negatively
103

associated with these factors. Along Axis 2, conventional harvest occurring after 1990 was
positively associated with length of wood and medium-sized LW density. While interpretation
of CCA axes plots is in part subjective, corroboration of our findings using multiple techniques
lends confidence to our results.
Discussion
Results from best subsets regressions predicting LW habitat factors from landscape
variables indicate that LW is significantly associated with multiple factors summarized in
catchments of study streams, including both natural and timber harvest-related landscape
variables. CCA results further support association of specific LW habitat variables with major
landscape drivers and also show that unique amounts of variation in a set of four LW habitat
factors can be attributed to both natural and anthropogenic landscape factors. Timber harvest
activities occurring within current and historical time periods were found to explain the most
variation in the set of LW habitat variables investigated in this study, underscoring the
complexities associated with understanding factors that influence stream conditions in the region
given the natural and human context of the landscapes they drain. The legacy of timber harvest
practices may have influenced changes associated with LW abundance and distribution in the
region’s streams, increased proportions of deciduous trees following harvest for example, may
increase the average length of large wood as deciduous trees have rapid growth early after
colonization, providing more woody debris to streams than early-successional conifer stands. To
our knowledge, this is the first study comparing relationships between multiple catchment factors
and LW in Southeast Alaska. It establishes the importance of considering watershed-level
landscape summaries to support decision making in the region and may ultimately contribute to
more effective management of the region’s streams.
104

Natural landscape factors
Catchment area
Catchment area was included in all five models produced for average LW length, and
over half of the variance in this variable was explained. An increase in LW length with
increasing catchment area follows previous research from the Pacific Northwest. Pieces of wood
are on average longer in wider channels as LW is generally stored when wood length is longer
than the stream width, with wider channels holding longer pieces (Martin and Benda 2001).
Therefore, larger streams have on average, longer pieces of wood. This relationship occurs up to
a point as wood is more easily transported when the channel becomes wider than the tallest tree
(Benda et al. 2002). My results also concur with Bilby and Ward (1989) who explained over
75% of variance in LW diameter, length, and volume through 22 sampled reaches of Western
Washington using stream size alone. Catchment area was also included as a predictor in two
other LW factors (i.e., the ratio of recruits located in the channel vs. positioned above the
channel and medium LW density), but only did so in a single model each, suggesting that area
has relatively less influence on these factors.
Catchment area predicted medium-sized LW density in the third best model for that
variable, and its influence was positive. While its inclusion into only a single model suggests
that catchment area is not the strongest landscape variable influencing LW density, the positive
relationship does not follow Bilby and Ward (1989) who found decreasing wood density with
increasing stream size. This may be due to the fact that they did not analyze LW by separate size
classes, and they also sampled streams having a wider range of stream slopes (i.e., up to 18%),
with slope acknowledged as an important factor influencing stream power and transporting

105

capability of streams (Knighton 1998), potentially leading to less wood in streams with higher
slopes. As bank erosion generally accounts for a majority of identifiable wood in streams of the
Pacific Northwest and Alaska (Martin and Benda 2001, Benda et al. 2002), steep confined
headwater channels are less susceptible to bank erosion than low to moderate gradient streams
having active floodplains, additionally accounting for contradicting results between the two
studies regarding LW density. Furthermore, regional differences in geology and climate, not
accounted for in this study, as well as regional land use differences, have the potential to affect
relationships with LW density and stream size. While Bilby and Ward (1989) sampled streams
draining old-growth forests, all reaches sampled for this study drained mixed old-growth and
second-growth catchments.
Catchment slope
Maximum slope in the local catchment was included in the prediction of one LW habitat
factor, the ratio of recruits located within vs. positioned above the active channel. It was
included in all five models predicting this habitat factor, although it was never the most
influential landscape factor as indicated by standardized coefficients. Results show that with
increasing gradient there are fewer recruits located within the channel than spanning above the
channel. Stream power, and consequently a stream’s capability to move LW and other materials,
increases with increasing slope, so wood that falls into a stream channel with a relatively higher
slope may not remain in the channel as long due to increased power associated with increasing
gradient (Knighton 1998). Steep headwater streams are typically narrower and more confined
than low to moderate gradient streams, and this characteristic may restrict the capacity for LW to
intersect stream channels (Liquori et al. 2008). May and Greswell (2004) found 33 % of locally
recruited wood pieces were spanning the full width or suspended over the channel of headwater
106

streams in a coastal Oregon watershed. Furthermore, steep channels are less prone to bank
erosion, resulting in overall low LW recruitment in the absence of wood delivering landslides.
When tree-fall occurs due to windthrow or mortality, for example, tree roots often remain
positioned on the bank, whereas recruitment by erosion generally occurs as undercut banks begin
to slump over time, and rootwads have a better probability of slipping down the bank with at
least one end in the channel.
Deciduous forests
Deciduous forest was associated with average LW length in all five models predicting
this factor, although its influence was always second to that of catchment area. These models
indicate that with more deciduous forests in local catchments, streams will contain longer pieces
of LW. Floodplain streams are typically lower in the stream network having larger catchments
and wider channels which allows for longer pieces of wood to be located in the channel.
Floodplain streams also have migrating channels, providing more disturbances for deciduous
trees to colonize (Villarin et al. 2009), although my data showed almost no correlation between
catchment area and proportion of local deciduous forest in catchments (Table 3). Another
potential factor may be due to growth rates as red alder in Southeast Alaska grow rapidly during
early stages of succession (Deal et al. 2004). Higher densities and death rates occur for alder
than conifers in similar young growth stands of the Pacific Northwest (Minore and Weatherly
1994, Hibbs and Giordano 1996), initially providing more pieces of deciduous wood than conifer
stands due to quickly-growing deciduous juvenile trees (Andrus et al. 1988, Gomi et al. 2006).
Hanley and Hoel (1996) also found that 40 years after timber harvest in Southeast Alaska,
riparian stands became dominated by red alder trees large enough to be considered LW (i.e., 1525 cm dbh). They also found old growth floodplains of a 4th-order stream to produce higher
107

densities of large spruce trees, while old growth upland stands were characterized by higher
frequencies of smaller conifers. The red alder dominated riparian zones were additionally on a
trajectory towards conifer dominated stands as the youngest trees were spruce. This may change
size distributions and functioning capability of LW in streams over time as deciduous trees are
smaller and decompose faster than conifers in streams (Bilby et al. 1999). While some evidence
exists that high deciduous productivity early after disturbances may provide short term
mitigation in LW production, Gomi et al. (2006), studying woody debris in headwater streams of
Southeast Alaska, found little contribution of second growth trees larger than 10 cm dbh from
second growth stands. However, they did find more deciduous pieces of wood in streams having
greater amounts of deciduous forests. In addition to length, a positive relationship was also
suggested between deciduous forest and medium-sized LW in streams having more deciduous
forest, a finding also supported by previous research that found alder in second growth stands of
Southeast Alaska grow to sizes near the medium class (i.e., 30 cm dbh) used in my study (Hanley
and Hoel 1996, Deal et al. 2004).
Forested wetlands
Wetlands have a significant influence on the hydrologic cycle, and while there are many
examples of wetlands reducing stream flooding by storing large amounts of precipitation (e.g.,
Maltby 1991, Hey and Philippi 1995), wetlands can also function to increase floods and high
flows (Bullock and Acreman 2003). Wetlands can store large quantities of water, potentially
restricting LW transport out of stream reaches draining catchments with wetlands having very
high water storing capacity, and holding smaller pieces of wood due to lack of transport. In
contrast, when wetlands are full and become saturated, they can act like an impervious surface,
routing essentially any new hydrologic input in catchments to stream channels, contributing to
108

high flows, potentially increasing fluvial transport of LW previously stored in upstream
channels. The amount of forested wetlands located within the network catchment was positively
related to medium-sized LW density in all four regression models produced in this study. As
described above, wetlands are a critical aspect regulating water routing within watersheds, and in
a small headwater temperate rainforest catchment of British Columbia, 30-99% of stream
discharge was regulated by a headwater wetland, depending on preceding conditions (Fitzgerald
et al. 2003), so substantial amounts of wetland in a catchment combined with large, or longlasting or intense precipitation events may increase transport of wood from upstream channels.
Poor stand productivity in wetlands may contribute to this result if medium-sized wood is the
largest these stands can produce and recruit to streams.
Forested wetlands of Southeast Alaska also have the potential to produce trees large
enough to be considered LW, and models estimate that forested wetlands can produce wood
volumes nearly twice the U.S. Forest Service minimum standard to be considered productive
forestland (Julin and D’Amore 2003). Wetlands are known to influence stream physical,
chemical, and biological processes in Alaska (King et al. 2012, Walker et al. 2012), but more
research is needed to determine specific mechanisms in which wetland land cover influences LW
dynamics in streams. Only forested wetlands were included in this analysis, and the National
Wetlands Inventory defined forested wetlands as being those dominated by woody vegetation at
least 20 feet tall, but differing wetland types in conjunction with spatial positioning throughout
watersheds will likely influence LW in streams differentially.
Timber harvest-related landscape factors
Historic timber harvest (1980-1990)

109

Timber harvest efforts occurring from 1980-1990 in Southeast Alaska were not required
to leave riparian buffers along anadromous streams, but rules in place during this time regulated
heavy equipment use near streams and removal of LW from streams. Timber harvest occurring
as a land use during this time period was summarized in local catchments of the study streams,
and it was related to all four key LW habitat factors tested, though it was most strongly
associated with LW recruit position. This suggests that with more conventional harvest from
1980-1990 in local catchments of study streams, more LW recruits may have been made
available to stream channels in relation to being positioned above the channel. While this may
be due to LW being left within streams after harvest (Gomi et al. 2001, Benda et al. 2002), only
4% of LW in which the source could be identified was attributed to timber harvest in my study.
As LW ages and decays over time, it breaks down and will be located within the channel more
often than more recently-recruited LW (Powell et al. 2009, Ross Chapter 1 this volume).
Harvest from 1980-1990 near streams removed the main source of LW from riparian zones.
Because of this, a more plausible explanation may be that the higher proportion of LW recruits
positioned within these streams are remnant and were recruited before timber harvest occurred,
while the lower proportion of LW located above the channel in these streams was recruited more
recently and may originate from second growth stands. Consequently, stream catchments having
more of this type of harvest may consist of relatively more LW recruits located within the stream
channel as the loss of riparian trees would eliminate new LW recruitment for a period of time
skewing the ratio toward relatively more in-stream recruits as remnant pieces decay and are
repositioned from above the channel into the channel (Powell et al. 2009).
Harvest occurring from 1980-1990 was also positively, associated with the density of
medium-sized LW. In Southeast Alaska, local disturbances to riparian zones such as timber
110

harvest, landslides, and bank erosion can remove patches of old-growth conifers which invite
quickly-growing deciduous species like red-alder to colonize. As deciduous tree types grow
more rapidly than conifers (Deal et al. 2004), they also die more quickly and potentially enter the
stream as LW, and the medium size-class is the largest these trees tend to grow (Worthington et
al. 1962). For this study, only LW recruits were determined to be deciduous or conifer, so I
lacked sufficient data to determine the type of all logs measured making the above explanation
plausible. Deciduous recruits were as high as 56% at one reach, although across all reaches they
averaged 12%, and 46% of reaches had no deciduous recruits identified. The association
contradicts Gomi et al. (2006) who found little contribution of trees large enough to be
considered LW (i.e., 10 cm dbh) from second growth stands of previously logged headwater
streams in Southeast Alaska. Differences may be due to their focus on headwater streams having
relatively narrow bankfull widths and some stream slopes > 10%. Also, their catchments
included logging that occurred as early as the 1950’s.
More medium sized large wood with increasing harvest from 1980-1990 may also be a
result of altered hydrology in extensively harvested catchments. Increased timber activity within
a watershed removes vegetation that intercepts precipitation. Heavy equipment can compact
soils and decrease infiltration, essentially increasing soil water moisture and the rate of water
movement from upland areas into stream channels (Chamberlin et al. 1991), although evidence
from a small California watershed with more than 15% of the watershed compacted by roads
suggest that only the smallest storm volumes were affected by logging road construction while
no differences were observed in water volumes following large storms (Wright et al. 1990).
Clearcuts can acquire more snow resulting in higher peaks in flows occurring in the spring (Harr
1986) which may have increased transporting capacity of catchments having more harvest during
111

this time period. Keppeler and Ziemer (1990) studying harvest effects on stream flow in
northwestern California suggest an average 15% increase in annual water yield should be
expected following intensive logging based on their results. Cumulative percent of area logged
and percent area covered by roads, landings, and skid trails were the most significant variables
associated with six flow-related dependent variables, although both watersheds were second
growth forests previously logged in the 1800’s. Harr et al. (1979) suggests similar effects of
harvest with increased peak and summer low flows, although they studied small headwater
basins which may be more responsive to flow alterations following harvest. Reduced evapotranspiration in heavily logged watersheds can lead to higher soil moisture content at the
beginning of the wet season, providing more precipitation to streams as less water is needed to
recharge soils (Douglas and Swank 1975, Ziemer 1981).
While increases in flow following the removal of watershed vegetation have been
suggested for quite some time (Hibbert 1967, Ponce and Meiman 1983, Surfleet and Skaugset
2013), this effect is strongest immediately following harvest (Keppeler and Ziemer 1990), and
diminishes over time as rapidly growing understories and second growth stands take advantage
of increased light and have higher evapotranspiration rates than old growth forests which may
ultimately reduce stream flow as succession proceeds (Harr 1983, Jarvis 1985, Greenwood et al.
1985). Few studies have documented these hydrologic effects following harvest in the temperate
rainforests of Southeast Alaska, although greater flow increases are suggested to occur in wetter
years (Harr et al. 1979, Ponce and Meiman 1983) and wetter climates (Hewlett 1982). In-stream
processing of wood such as transport from upstream sources, for example, is not well understood
and has historically been ignored or assumed to be equal to the output of a stream reach (Murphy

112

and Koski 1989, Van Sickle and Gregory 1990, Liquori et al. 2008), and transport is typically
limited in low order stream reaches.
Another plausible influence of harvest on increased LW density could be due to altered
rates of landslides and slope failure following harvest. Logging roads and loss of hillside
vegetation from harvest often increase the rate of landslides and mass wasting events
(Amaranthus et al. 1985, Wu and Sidle 1995, Tang et al 1997). These soil movements have the
potential to deliver large amounts of wood and other materials from upland areas into stream
channels, and debris flows down channels have been documented to contribute large amounts of
wood from upstream sources into high-order streams (Benda et al. 2002, Reeves et al. 1993),
although attributing LW transport by debris flows is difficult because of complexities in aging
these deposits (Liquori et al. 2008). Data on the occurrence of landslides and mass wasting
events within catchments sampled for this study were not acquired and would need to be
investigated to further support such timber harvest influences.
Current timber harvest (post-1990)
Two types of timber harvest were summarized post-1990 including conventional in the
network catchment, and other types, including selective and helicopter harvest in the local
catchment. Harvest after 1990 required that forested streamside buffers be left along
anadromous streams, while selective harvest within riparian buffers could occur under certain
circumstances. Conventional harvest after 1990 was positively related to average LW length in
all models predicting this factor, although it was typically less influential compared to natural
factors in the models (i.e., catchment area and deciduous forest). Nevertheless riparian buffers
were required to be left after 1990, so these catchments should have more intact riparian zones

113

compared to catchments harvested previously, providing taller trees as longer LW to streams. A
similar relationship is present between this landscape factor and medium-sized LW density.
More medium-sized LW was observed in stream catchments with greater amounts of
conventional harvest after 1990. Timber harvest can increase LW input in Southeast Alaska, not
only from harvest related debris being left over (Gomi et al. 2001), but also from increased
windthrow of standing riparian buffers (Martin and Grotefendt 2007), and medium-sized trees
may be more vulnerable to this effect due to potentially shallower rooting depths than larger
trees. In contrast, other types of harvest after 1990 in local catchments are negatively related to
the density of large and extra-large LW. There was only one model significantly predicting the
largest wood in streams, and it included two measures of harvest. While it was a comparatively
weaker model and explained little variation, and these largest pieces of wood are typically rare
even in streams having old growth riparian stands, the negative relationship between other
harvest after 1990 and the density of the largest pieces of LW may be a product of selective
harvest methods targeting very large trees from the riparian zone.
Logging road and stream crossings
The density of logging road/stream crossings within network catchments was present in
three models predicting the ratio of recruits located within vs. above the channel (although it was
less influential than other predictors). The recruit ratio decreased as logging road/stream
crossings increased within catchments. This result may be due to the fact that logging
road/stream crossings typically occur at locations where the channel is confined and has less
opportunity to meander. The lack of channel migration and bank erosion in confined channels
additionally results in low overall LW recruitment and narrow, confined channels restrict wood
from intersecting the stream (Liquori et al. 2008), although we lacked data to fully evaluate this
114

relationship. Furthermore, confined channels are more powerful and may additionally be
capable of moving LW out of those reaches. An alternate explanation may simply be a reflection
of streamside buffering as streams harvested after 1990 required riparian buffers be retained and
increased windthrow in riparian buffers shortly after harvest can increase wood loading (Martin
2001).
Comparing landscape influences on large wood habitat factors
CCA results indicate that a substantial amount of variation (i.e., 44%) in average LW
length, LW recruit position in relation to the active channel, medium LW density, and large and
extra-large LW density can be explained using natural and timber harvest-related landscape
factors. While managers and scientists have typically assessed and documented effects of timber
management on the region’s streams at local scales, these findings suggest that landscape scale
associations are also important in understanding LW dynamics in streams.
The findings of this study additionally suggest that timber harvest factors at a landscape
scale have significant influences on LW habitat factors in streams of Southeast Alaska, and its
interaction with natural landscape factors accounts for a comparatively low amount of variation
in LW habitat factors which is not surprising as little covariation exists within the set of
landscape variables selected. More compelling is that timber harvest factors across the landscape
account for more variation than natural landscape factors in the second-growth study catchments.
Our results contradict similar approaches investigating human impacts on streams. Wang et al.
(2003) found substantial amounts of variation can be explained using natural landscape factors,
and in relatively undisturbed regions these influences were strongest locally. As my study did
not investigate the influence of scale, local and network factors characterizing natural land cover

115

and timber harvest activities were successful in explaining variation in LW habitat factors, and
while links between the riparian zone and in-stream LW have been well studied, the findings of
this research enforce the idea that influences on LW characteristics occur at scales larger than the
riparian zone alone. Windthrow within streamside buffers, for example, may increase as large
areas outside the riparian zone are harvested. Peak flows can increase after extensive timber
harvest, especially in small catchments during wet years and wet climates, and while this effect
is typically not long-lived, future research may investigate if short term increases effectively
increase transport of wood from channels that would otherwise be characterized as LW storage
locations. While source distances of LW recruitment from landscapes are typically confined to
riparian zones, altered hillsides can increase landslide rates and essentially increase this source
distance in steep catchments. The most important scales of influence and specific mechanisms
affecting LW in streams is currently unknown and likely varies regionally. Johnson et al. (2006)
found that both natural and anthropogenic factors occurring in stream catchments influenced LW
habitat factors differentially across two Midwestern states. This study concurs with their
findings and demonstrates that the legacy of actions on the landscape affect LW in Southeast
Alaska and are important considerations when managing for natural resources in the region.
Management implications
Landscape considerations in managing the region’s streams
The results of this study suggest that landscape factors are important for describing
characteristics of LW abundance and distribution in streams of Southeast Alaska. Natural
conditions, current and historic land use, and LW habitat factors differ throughout catchments of
Southeast Alaska, so understanding these factors and how they are distributed across multiple

116

scales is important in characterizing the current condition and developing management
prescriptions to restore and protect streams of the region. Forested wetlands in watersheds may
influence water routing and LW production, although small streams generally do not transport
LW in the absence of debris flows so a better understanding of land cover and disturbance
history of watersheds is needed provide insight into the potential for a stream to store and
transport LW. Increased timber harvest in catchments might affect peak flows in the short term
(Harr et al. 1979) and the spatial location of harvested areas influences the risk of slope failures
(Tang et al. 1997) that can have major effects on LW recruitment and transport throughout
watersheds, so a thorough investigation of landscapes in relation to streams is important when
identifying factors limiting LW production. The landscape approach can also assist in locating
those streams most vulnerable to the effects of timber harvest, or most likely to benefit from
restoration efforts.
Future research
Numerous landscape factors known to affect habitat factors within streams were untested
in this study including underlying geology and silviculture efforts like thinning and pruning, for
example, which can accelerate the production of trees and increase the rate of succession from
early to old forests. Various geologic differences exist across Southeast Alaska and geology is
known to influence LW availability. Confined bedrock channels, for example, have low rates of
bank erosion and more power to transport materials resulting in low wood storage. Steep slopes
in headwater catchments may additionally be prone to landslides and debris flows capable of
delivering LW into higher order channels, and timber harvest in these catchments can increase
the rate of hillslope failures. Much of the region contains previously glaciated landscapes, and
coarse deposits of surficial materials are known to contribute to groundwater delivery in streams
117

systems (Wiley et al. 1997). Also, streams draining karst landscapes (found, for example, on
Prince of Wales Island) have higher alkalinities, and higher fish production than non-karst
streams (Bryant et al. 1998), while differences in salmonid densities between northern and
southern Southeast Alaska watersheds have additionally been attributed to differences in geology
(Bryant and Woodsmith 2009). The potential of landscapes to produce large trees is also mainly
controlled by catchment topology, local landforms, as well as soil moisture and structure (Hanley
and Hoel 1996, Villarin et al. 2009). Silviculture practices like thinning and pruning are
intended to maximize forest productivity and can be employed to enhance the quality of potential
LW available to streams which affects numerous aquatic habitat factors important for fish (e.g.,
pool formation). Thinning dense stands can increase primary production in streams during the
summer through increased light and deciduous benefits, while still maintaining inputs of LW
(Sedell and Swanston 1984). Thinning dense second growth vegetation in streams of Wisconsin
has been shown to enhance trout production compared to similar, unmanaged second growth
streams (Hunt 1979). Inclusion of these factors may improve the predicting capability in future
efforts to relate landscape factors to LW habitat factors in streams of Southeast Alaska.

118

APPENDIX

119

Table 2.1: Mean, min, max, and standard deviation (SD) of natural landscape variables for 28
sampled stream reaches draining second growth watersheds.
Scale
Variable name
Code
Mean
Min
Max
SD
Local
Area (km²)
T1areasqkm
0.71
0.02
2.83
0.75
Ave slope (%)
T1slpAVE
1.16
0.43
2.95
0.62
Max slope (%)*
T1slpMAX
3.52
1.21
7.13
1.97
Open water (%)
T1open
NA
NA
NA
NA
Deciduous forest (%)*
T1decid
6.59
0.00
37.93
9.02
Evergreen forest (%)
T1evgrn
36.64
0.00
65.62
16.91
Mixed forest (%)
T1mixd
27.40
2.74
57.14
15.29
All forest (%)
T1allFOR
70.62 23.59
99.99
19.54
Dwarf scrub (%)
T1dwfscrb
0.02
0.00
0.35
0.08
Scrub/shrub (%)
T1shrbscrb
25.91
0.00
76.40
20.67
Forested wetlands (%)
T1WETfor
33.17
0.00
100.00
30.77
Developed-low intensity (%) T1dvlLOW
0.76
0.00
9.37
2.01
Road density (m/km²)
T1rDens
58.80
0.00
680.81 181.91
Network
Area (km²)*
T2areasqkm 4.24
0.12
19.75
5.63
Slope (%)
T2slpAVE
1.38
0.50
2.48
0.50
Max slope (%)
T2slpMAX
3.82
1.47
6.54
1.38
No Harvest/No data (%)
T2noHVST
40.76
0.00
87.05
27.71
Open water (%)
T2open
0.02
0.00
0.24
0.05
Deciduous forest (%)
T2decid
3.95
0.00
37.93
7.01
Evergreen forest (%)
T2evgrn
42.03
0.00
66.98
19.77
Mixed forest (%)
T2mixd
19.17
2.75
44.63
12.65
All forest (%)
T2allFOR
65.16 23.59
100.00
19.92
Dwarf scrub (%)
T2dwfscrb
0.09
0.00
0.68
0.20
Scrub/shrub (%)
T2shrbscrb
32.47
0.00
76.41
21.53
Forested wetlands (%)*
T2WETfor
25.69
0.00
93.16
22.65
Developed-low intensity (%) T2dvlLOW
0.25
0.00
2.97
0.67
Road density (m/km²)
T2rDens
1425.75
0.00 24972.66 4978.11

120

Table 2.2: Mean, min, max, and standard deviation (SD) of timber harvest-related variables
for 28 sampled stream reaches draining second growth watersheds.
Scale
Variable name
Code
Mean Min Max
SD
Local
Conventional pre 1980 (%)
T1CONpre80 1.76 0.00
27.27
6.49
Conventional 1980-1990 (%)*
T1CON8090 41.58 5.78
80.23 21.03
Conventional post 1990 (%)
T1CONpo90 10.30 0.00
37.81 11.88
Other pre 1980 (%)
T1OTpre80
0.00 0.00
0.00
0.00
Other 1980-1990 (%)
T1OT8090
0.25 0.00
3.46
0.79
Other post 1990 (%)*
T1OTpo90
10.34 0.00
31.61
9.81
Logging rd density
T1LogXDens 2.79 0.00
6.49
1.67
Logging rd/stream crossing density
(#/km²)
T1LogRDens 5.58 0.00
75.73 14.41
Network
Conventional pre 1980 (%)
T2CONpre80 0.46 0.00
6.43
1.44
Conventional 1980-1990 (%)
T2CON8090 30.29 3.64
77.55 22.43
Conventional post 1990 (%)*
T2CONpo90 14.24 0.00
49.10 12.69
Other pre 1980 (%)
T2OTpre80
0.00 0.00
0.00
0.00
Other 1980-1990 (%)
T2OT8090
0.18 0.00
2.12
0.55
Other post 1990 (%)
T2OTpo90
14.07 0.00
41.91 11.93
Logging rd density (m/km²)
T2LogXDens 37.88 0.27 244.29 62.18
Logging rd/stream crossing density
(#/km²)*
T2LogRDens 3.00 0.00
8.92
2.53

121

Table 2.3: Pearson’s correlations between timber harvest-related and other landscape factors used in analyses. An asterisk indicates
significance at a 0.05 level.
T2areasqkm T2WETfor
T2areasqkm
1.00
T2WETfor
-0.24
1.00
T1slpMAX
0.01
-0.27
T1decid
-0.07
-0.24
T1OTpo90
-0.21
-0.03
T2LogXDens
-0.13
0.10
T2CONpo90
0.13
-0.11
T1CON8090
0.14
-0.26

T1slpMAX

1.00
-0.09
0.15
-0.42 *
0.06
0.01

T1decid T1OTpo90 T2LogXDens T2CONpo90 T1CON8090

1.00
-0.20
-0.17
-0.03
0.26

122

1.00
0.14
-0.07
-0.08

1.00
-0.06
0.20

1.00
-0.21

1.00

Table 2.4: Landscape factors for streams in Southeast Alaska summarized for local and network
catchments.
Scale
Mean
Min
Max
Area (km²)
Local
0.47 0.00
1833.10
Ave slope (%)
1.57 0.00
10.78
Max slope (%)*
3.95 0.00
43.55
Open water (%)
0.01
40.00
Deciduous forest (%)*
3.58 0.00
100.00
Evergreen forest (%)
53.17 0.00
100.00
Mixed forest (%)
14.04 0.00
100.00
Dwarf scrub (%)
1.19 0.00
100.00
Scrub/shrub (%)
17.94 0.00
100.00
Forested wetlands (%)
0.00 0.00
100.00
Developed-low intensity (%)
0.22 0.00
100.00
0.01 0.00
100.00
Developed-medium intensity (%)
Developed-high intensity (%)
0.00 0.00
37.50
Network
Area (km²)*
Ave slope (%)
Max slope (%)
Open water (%)
Deciduous forest (%)
Evergreen forest (%)
Mixed forest (%)
Dwarf scrub (%)
Scrub/shrub (%)
Forested wetlands (%)*
Developed-low intensity (%)
Developed-medium intensity (%)
Developed-high intensity (%)

0.51
1.70
4.82
0.79
2.55
48.18
12.87
1.82
20.60
0.00
0.13
0.00
0.00

123

0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00

2168.90
10.68
43.55
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
23.08
32.24

Table 2.5: Results from best subsets regression for LW length. Models are in order of most
variance explained, predictor variables are ordered by their influence on the model based on
standardized beta coefficients. Variance inflation factor value of four (VIF < 4) was used to
ensure low multicollinearity between landscape variables, and ∆ AIC represents the difference
when compared to catchment area alone.
Adj r²
0.59

Standard beta
0.66
0.32
0.28

F
13.70

Model sig
0.000021

VIF
1.02
1.01
1.02

∆ AIC
-6.89

2 T2areasqkm
T1decid
T2CONpo90
T2LogXDens

0.58

0.68
0.34
0.28
0.11

10.36

0.00006

1.04
1.04
1.02
1.05

-5.79

3 T2areasqkm
T1decid
T2CONpo90
T2WETfor

0.57

0.68
0.34
0.28
0.07

10.03

0.000075

1.09
1.08
1.03
1.15

-5.22

4 T2areasqkm
T1decid
T2CONpo90
T1CON8090

0.57

0.65
0.31
0.28
0.03

9.89

0.000083

1.07
1.09
1.08
1.17

-4.96

5 T2areasqkm
T1decid
T2CONpo90
T1OTpo90

0.57

0.66
0.32
0.28
-0.02

9.86

0.000085

1.07
1.05
1.02
1.10

-4.92

Model Landscape variables
1 T2areasqkm
T1decid
T2CONpo90

124

Table 2.6: Results from best subsets regression for the ratio of LW recruits that are in:above the
channel. Models are in order of most variance explained, predictor variables are ordered by
their influence on the model based on standardized beta coefficients. Variance inflation factor
value of four (VIF < 4) was used to ensure low multicollinearity between landscape variables,
and ∆ AIC represents the difference when compared to catchment area alone.
Adj r²
0.34

Standard Beta
0.40
0.28
-0.26
0.23

F
4.50

Model sig
0.008

VIF
1.11
1.10
1.01
1.03

∆ AIC
-8.20

2 T1CON8090
T1slpMAX
T1decid
T1OTpo90

0.33

0.43
-0.24
0.21
-0.21

4.38

0.009

1.08
1.03
1.12
1.06

-7.88

3 T1CON8090
T1slpMAX
T2LogXDens
T1OTpo90

0.33

0.53
-0.36
-0.23
-0.20

4.33

0.009

1.07
1.31
1.37
1.10

-7.74

4 T1CON8090
T1slpMAX
T2LogXDens
T1decid

0.32

0.50
-0.36
-0.22
0.19

4.23

0.01

1.18
1.30
1.42
1.20

-7.47

0.56
-0.41
-0.28

5.22

0.006

1.05
1.23
1.28

-8.08

Model Landscape variables
1 T1CON8090
T1decid
T1slpMAX
T2areasqkm

5 T1CON8090
T1slpMAX
T2LogXDens

0.32

125

Table 2.7: Results from best subsets regression for medium sized LW density. Models are in
order of most variance explained, predictor variables are ordered by their influence on the
model based on standardized beta coefficients. Variance inflation factor value of four (VIF < 4)
was used to ensure low multicollinearity between landscape variables, and ∆ AIC represents the
difference when compared to catchment area alone.
Adj r²
0.37

Standard beta
0.50
0.39
0.37
0.34

F
4.96

Model sig
0.005

VIF
1.08
1.11
1.15
1.20

∆ AIC
-10.37

2 T1decid
T2CONpo90
T1OTpo90
T2WETfor

0.33

0.40
0.40
-0.26
0.26

4.37

0.009

1.11
1.02
1.05
1.08

-8.79

3 T1decid
T2CONpo90
T2WETfor
T2areasqkm

0.32

0.49
0.39
0.35
0.25

4.24

0.010

1.08
1.03
1.15
1.09

-8.44

4 T2CONpo90
T2CON8090
T1OTpo90
T2WETfor

0.32

0.47
0.39
-0.31
0.28

4.15

0.011

1.09
1.16
1.02
1.11

-8.19

Model Landscape variables
1 T2CONpo90
T1decid
T2WETfor
T1CON8090

126

Table 2.8: Results from best subsets regression for large and extra-large sized LW density.
Models are in order of most variance explained, predictor variables are ordered by their
influence on the model based on standardized beta coefficients. Variance inflation factor value
of four (VIF < 4) was used to ensure low multicollinearity between landscape variables, and ∆
AIC represents the difference when compared to catchment area alone.
Model Landscape variables
1 T1CON8090
T1OTpo90

Adj r²
0.16

Standard beta
0.36
-0.28

127

F
3.50

Model sig
0.046

VIF
1.01

∆ AIC
-4.11

No data
Open water
Ice / snow
Open space
Low urban
Medium urban
High urban
Baren
Deciduous forest
Evergreen forest
Mixed forest
Dwarf scrub
Shrub / scrub
Grassland
Sedge
Wooded wetlands
Emergent wetlands

N
W
N

E
S

0
N

300
Kilometers

N
Figure 2.1: National Land Cover Database (2001) for Southeast Alaska. For interpretation of the
references to color in this and all other figures, the reader is referred to the electronic version of
this thesis.

128

18%
Timber harvest
54%

Natural
Interaction

28%

Figure 2.2: Results of variance partitioning from the CCA. A total of 44% of variation in LW
habitat factors is partitioned into timber harvest-related factors in gray, natural factors in black,
and the interaction in white.

129

1.0

1.0

T2CONpo90

T2WETfor
T1slpMAX
MEDLWKM
T1OTpo90
LWLENGTH

CCA II
LXLWKM
T2LogXDens

T1decid
T2areasqkm

T1CON8090

-1.0

INABOVE

-1.0
-1.0
-1.0

1.0

1.0

CCA I
Figure 2.3: Graph of CCA axis one vs. CCA axis 2. Black arrows are natural landscape factors,
red arrows are timber harvest-related factors, and black squares are LW habitat factors.
LWLENGTH is average LW length, MEDLWKM and LXLWKM are densities of medium sized
and large and extra-large sized LW, while INABOVE is the ratio of LW recruits that are located
within vs. positioned above the active channel.

130

LITERATURE CITED

131

LITERATURE CITED

Alaska Department of Fish and Game. 2011. The catalog of waters important for spawning,
rearing, or migration of anadromous fishes.
Allan, J. D. 2004. Landscapes and riverscapes: The influence of land use on stream ecosystems.
Annual Review of Ecology Evolution and Systematics 35:257-284.
Amaranthus, M. P., R. M. Rice, N. R. Barr, and R. R. Ziemer. 1985. Logging and forest roads
related to increased debris slides in southwestern Oregon. Journal of Forestry 83:229233.
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.
Benda, L. E. 1990. The influence of debris flows on channels and valley floors in the Oregon
Coast Range, USA. Earth Surface Processes and Landforms 15:457-466.
Benda, L. E., P. Bigelow, and T. M. Worsley. 2002. Recruitment of wood to streams in oldgrowth and second-growth redwood forests, northern California, USA. Canadian Journal
of Forest Research 32:1460-1477
Bilby, R. E., and J. W. Ward. 1989. Changes in characteristics and functions of woody debris
with increasing size of streams in Western Washington. Transactions of the American
Fisheries Society 118:368-378
Bilby, R. E., J. T. Heffner, B. R. Fransen, J. W. Ward, and P. A. Bisson. 1999. Effects of
immersion in water on deterioration of wood from five species of trees used for habitat
enhancement projects. North American Journal of Fisheries Management 19:687-695.
Bisson, P. A., R. E. Bilby, M. D. Bryant, C. A. Dolloff, 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. Pages 143-191 in E. 0. Salo and T. W.
Cundy editors Streamside management: forestry and fishery interactions. Contribution
No. 57, University of Washington, Institute of Forest Resources, Seattle.
ter Braak, C. J. F., and I. C. Prentice. 1988. A theory of gradient analysis. Advanced Ecological
Research 18:271-313
Brookshire, E. N. J., and K. A. Dwire. 2003. Controls on patterns of coarse particle retention in
headwater streams. Journal of the North American Benthological Society 22:17-34.

132

Bryant, M. D., R. T. Edwards, and R. D. Woodsmith. 2005. An approach to effectiveness
monitoring of floodplain channel aquatic habitat: salmonid relationships. Landscape and
Urban Planning 72:157-176.
Bryant, M. D., T. Gomi, and J. J. Piccolo. 2007. Structures linking physical and biological
processes in headwater streams of the Maybeso watershed, southeast Alaska. Forest
Science 53:371-383.
Bryant, M. D., D. N. Swanson, R. C. Wissmar, and B. E. Wright. 1998. Coho salmon
populations in the karst landscape of north Prince of Wales Island, Southeast Alaska.
Transactions of the American Fisheries Society 127:425-433.
Bryant, M. D. and R. D. Woodsmith. 2009. The response of salmon populations to geomorphic
measurements at three scales. North American Journal of Fisheries Management 29:549559.
Bullock, A., and M. Acreman. 2003. The role of wetlands in the hydrologic cycle. Hydrology
and Earth Systems 7:358-389.
Burnett, K. M., G. H. Reeves, S. E. Clarke, and K. R. Christiansen. 2006. Comparing riparian
and catchment influences on stream habitat in a forested, montane landscape. Pages 175197 in R.M. Hughes, L. Wang, and P. W. Seelbach, editors. Landscape influences on
stream habitats and biological assemblages. American Fisheries Society, Symposium
48, Bethesda, Maryland.
Chamberlin, T. W., R. D. Harr, and F. H. Everest. 1991. Timber harvesting, silviculture, and
watershed processes. Pages 181-205 in W. R. Meehan (editor) Influences of forest and
rangeland management on salmonid fishes and their habitats. Special Publication 19.
American Fisheries Society, Bethesda, Maryland.
Cohen M. J. and M. T. Brown. 2007. A model examining hierarchical wetland networks for
watershed stormwater management. Ecological Modeling 201:179-193.
Deal, R. L., P. E. Hennon, E. H. Orlikowska, and D. V. D’Amore. 2004. Stand dynamics of
mixed red alder-conifer forests of southeast Alaska. Canadian Journal of Forest Research
34:969-980.
Douglas, J. E., and W. T. Swank. 1975. Effects of management practices on water quality and
quantity: Coweeta Hydrologic Laboratory, North Carolina. In Municipal Watershed
Management Symp. Proc. 1975:1-13. U.S. Department of Agriculture Forest Service
Gen. Tech. Rep. NE-13. Northeast For. Exp. Stn.,Upper Darby, PA.
Fausche, 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.

133

Fitzgerald, D. F., J. S. Price, and J. J. Gibson. Hillslope-swamp interactions and flow pathways
in a hypermaritime rainforest, British Columbia. 2003. Hydrologic Processes 17:30053022.
Frissell, C.A.,W.J. Liss, C.E. Warren and M.D. Hurley. 1986. A hierarchical framework for
stream habitat classification: viewing streams in a watershed context. Environmental
Management, 10: 199–214.
Gomi, T., R. C. Sidle, M. D. Bryant, and R. D. Woodsmith. 2001. The characteristics of woody
debris and sediment distribution in headwater streams, southeastern Alaska. Canadian
Journal of Forest Research 31:1386-1399.
Gomi, T., A. C. Johnson, R. L. Deal, P. E. Hennon, E. H. Orlikowska, and M. S. Wipfli. 2006.
Factors affecting distribution of wood, detritus, and sediment in headwater streams draining
managed young growth red alder-conifer forests in southeast Alaska. Canadian Journal of
Forest Research 36:725-737.
Greenwood, E. A., L. Klein, J. D. Beresford, G. D. Watson, and K. D. Wright. 1985. Evaporation
from the understory in the Jarrah Forest, southwestern Australia. Journal of Hydrology
80:337-349.
Hanley, T. A., and T. Hoel. 1996. Species composition of old-growth and riparian Sitka sprucewestern hemlock forests in southeastern Alaska. Canadian Journal of Forest Research
26:1703-1708.
Harr, D. R., R. L. Fredriksen, and J. S. Rothacher. 1979. Changes in stream flow following
timber harvest in southwestern Oregon. Research Paper PNW-249. U.S.D.A. Forest Service,
Pacific Northwest Forest and Range Experiment Station, Portland, Oregon.
Harr, D. R. 1986. Effects of clearcutting on rain-on-snow runoff in western Oregon: a new look
at old studies. Water Resources Research 22:1095-1100.
Hey, D. L., and N. S. Philippi. 1995. Flood reduction through wetland restoration: the Upper
Mississippi River Basin as a case history. Restoration Ecology 3:4-17.
Hunt, R. L. 1979. Removal of woody streambank vegetation to improve trout habitat. Wisconsin
Department of Natural Resources Technical Bulletin 115 36 p.
Hynes, H. B. N. 1975. The stream and its valley. Verhandlungen Internationale Vereinigung für
Theoretische und Angewandte Limnologie 19:1–15.
Jarvis, P. G. 1985. Increasing forest productivity and value of temperate coniferous forest by
manipulating site water balance. Pages 39-74 in R. Ballard, P. Farnum, G. A. Ritchie, and
J. K. Winjum, editors. Symposium: forest potentials, productivity and values.
Weyerhaeuser Company, Tacoma, WA.

134

Johnson, L. B., G. E. Host, and J. H. Olker. 2006. Landscape and reach scale predictors of large
wood abundance in low gradient streams. Pages 151-173 in R.M. Hughes, L. Wang, and
P. W. Seelbach, editors. Landscape influences on stream habitats and biological
assemblages. American Fisheries Society, Symposium 48, Bethesda, Maryland.
Johnston, C. A., N. E. Detenbeck, and G. J. Niemi. 1990. The cumulative effect of wetlands on
stream water quality and quantity: a landscape approach. Biogeochemistry 10:105-141.
Julin, K. R., and D. V. D’Amore. 2003. Tree growth on forested wetlands of southeastern Alaska
following clearcutting. Western Journal of Applied Forestry 18:30-34.
Hanley, T. A., and T. Hoel. 1996. Species composition of old-growth and riparian Sitka sprucewestern hemlock forests in southeastern Alaska. Canadian Journal of Forest Research
26:1703-1708.
Harr, R. D. 1983. Potential for augmenting water yield through forest practices in western
Washington and western Oregon. Water Resources Bulletin 19:383-392.
Hewlett, J. D. 1982. Forests and floods in the light of recent investigation. Pages 545-560 in
Proc. Canadian Hydrology Symposium Natural Resources Council of Canada, Ottawa.
Hibbert, A. R., 1967. Forest treatment effects on water yield. In Proceedings of the International
Symposium on Forest Hydrology, W. E. Sopper and H. W. Lull, editors. p.527-543,
Pergamon, New York.
Hibbs, D. E., and P. A. Giordano. 1996. Vegetation characteristics of alder-dominated riparian
buffer strips in Oregon Coast Range streams. Northwest Science 70:213-222.
Keppeler, E. T., and R. R. Ziemer. 1990. Logging effects on streamflow: water yield and
summer low flows at Caspar Creek in northwestern California. Water Resources
Research 26:1669-1679.
King, R. S., C. M. Walker, D. F. Whigham, S. J. Baird, and J. A. Back. 2012. Catchment
topography and wetland geomorphology drive macroinvertebrate community structure
and juvenile salmonid distributions in south-central Alaska headwater streams.
Freshwater Science 31:341-364.
Knighton, D. 1998. Fluvial forms and processes: A new perspective. John Wiley and Sons, Inc.,
New York, New York.
Leak, W. B. 1985. Relationships of tree age to diameter in old growth northern hardwoods and
spruce-fir. Research Note NE-329 USDA Forest Service, Northeastern Forest Experiment
Station.
Lisle, T. E. 1986. Effects of woody debris on anadromous salmonid habitat, Prince of Wales
Island, southeast Alaska. North American Journal of Fisheries Management 6:538-550.
135

Lisle, T. E. 1987. Using residual depths to monitor pool depths independently of discharge.
Research Note PSW-394. Berkely, CA: Pacific Southwest Forest and Range Experiment
Station, U.S. Forest Service, Department of Agriculture.
Liquori, M. D. Martin, L. Benda, R. Coats, and D. Ganz. 2008. Scientific literature review of
forest management effects on riparian functions for anadromous salmonids. Report of
Sound Watershed Consulting to the California State Board of Forestry and Fire
Protection, Sacramento, CA.
Maltby, E. 1991. Wetland management goals: wise use and conservation. Landscape Urban
Planning 20:9-18.
Martin, D.J. 2001. The influence of geomorphic factors and geographic region on large woody
debris loading and fish habitat in Alaska coastal streams. North American Journal of
Fisheries Management 21:429-440
Martin, D. J. and L. E. Benda. 2001. Patterns of instream wood recruitment and transport at
the watershed scale. Transactions of the American Fisheries Society 130:940-958.
Martin, D. J., and R. A. Grotefendt. 2007. Stand mortality in buffer strips and the supply of
woody debris to streams in southeast Alaska. Canadian Journal of Forest Research
37:36-49.
May, C. L., and R. E. Gresswell. 2004. Spatial and temporal patterns of debris-flow deposition in
the Oregon Coast Range, USA. Geomorphology 57:135-149.
McMahon, T. E., and G. F. Hartman. 1989. Influence of cover complexity and current velocity
on winter habitat use by juvenile coho salmon (Oncorhynchus kisutch). Canadian Journal
of Fish and Aquatic Sciences 46:1551-1557.
Minore, D., and H. G. Weatherly. 1994. Riparian trees, shrubs, and forest regeneration in the
coastal mountains of Oregon. New Forests 8:249-263.
Montgomery, D. R., T. B. Abbe, J. M. Buffington, N. P. Peterson, K. M. Schmidt, and J. D.
Stock. 1996. Distribution of bedrock and alluvial channels in forested mountain drainage
basins. Nature 381:587-589.
Murphy, M. L., and K. V. Koski. 1989. Input and depletion of woody debris in Alaska streams
and implications for streamside management. North American Journal of Fisheries
Management 9:427-436
Paustian, S. J., K. Anderson, D. Blanchet, S. Brady, M. Cropley, J. Edgington, J. Fryxell,
G. Johnejack, D. Kelliher, M. Kuehn, S. Maki, R. Olsen, J. Seesz, and M. Wolanek.
1992. A channel type user’s guide for the Tongass National Forest, southeast Alaska.
Technical Paper 26, U.S. Forest Service, Juneau, Alaska.
136

Ponce, S. L., and J. R. Meiman. 1983. Water yield augmentation through forest and range
management-issues for the future. Water Resources Bulletin 19:415-419.
Powell, S. R., L. D. Daniels, and T. A. Jones. 2009. Temporal dynamics of large woody debris in
small streams of the Alberta foothills, Canada. Canadian Journal of Forest Research
39:1159-1170.
Robison, E. G., and R. L. Beschta. 1990. Identifying trees in riparian areas that can provide
coarse woody debris to streams. Forest Science 36:790-801.
Roni, P., and T. P. Quinn. 2001. Distribution and size of juvenile salmonids in response to
placement of large woody debris in western Oregon and Washington streams. Canadian
Journal of Fisheries and Aquatic Sciences 58:282-292.
Rosenfeld, J., M. Porter, and E. Parkinson. 2000. Habitat factors affecting the abundance and
distribution of juvenile cutthroat trout (Oncorhynchus clarki) and coho salmon
(Oncorhynchus kisutch). Canadian Journal of Fisheries and Aquatic Sciences 57:766-774.
Ross, J. A. 2013. Influences of timber management and natural landscape factors on anadromous
streams of southeastern Alaska: relating local and catchment factors to aquatic habitat.
M.S. Thesis. Michigan State University, East Lansing, MI.
Roth, N. E., J. D. Allan, and D. L. Erickson. 1996. Landscape influences on stream biotic
integrity assessed at multiple spatial scales. Landscape Ecology 11:141-156.
Sedell, J.R. and F.J. Swanston. 1984. Ecological characteristics of streams in old growth forests
of the pacific northwest. Pages 9-16 in W.R. Meehan, T.R. Merrell Jr, and T.A. Hanley,
editors, Fish and wildlife relationships in old growth forests symposium. American
Institute of Fishery Research Biologists, Juneau, Alaska.
Smith, S. H., J. F. Bell, F. R. Herman, and T. See. 1984. Growth and yield of sitka spruce and
western hemlock at Cascade Head Experimental Forest, Oregon. U.S.D.A. Forest Service
Research Paper PNW-325 30 p.
Surfleet, C. G., and A. E. Skaugset. 2013. The effect of timber harvest on summer low flows,
Hinkle Creek, Oregon. Western Journal of Applied Forestry. 28:13-21.
Sweeney, B. W. 1993. Effects of streamside vegetation on macroinvertebrate communities of
White Clay Creek in eastern North America. Proceedings of the Academy of Natural
Sciences of Philadelphia 144:291-340.
Tang, S. M., J. F. Franklin, and D. R. Montgomery. 1997. Forest harvest patterns and landscape
disturbance processes. Landscape Ecology 12:349-363.

137

Townsend, C. R., S. Doledec, R. H. Norris, K. Peacock, and C. J. Arbuckle. 2003. The influence
of scale and geography on relationships between stream community composition and
landscape variables: description and prediction. Freshwater Biology 48:768-785.
Van Sickle, J. and S. V. Gregory. 1990. Modeling inputs of large woody debris to streams from
falling trees. Canadian Journal of Forest Research 20:1593-1601.
Villarin, L.A., D.M. Chapin, and J.E. Jones. 2009. Riparian forest structure and succession in
second growth stands of the central Cascade Mountains, Washington, USA. Forest
Ecology and Management 257:1375-1385.
Walker, C. M., R. S. King, D. F. Whigham, and S. J. Baird. Landscape and wetland influences
on headwater stream chemistry in the Kenai Lowlands, Alaska. Wetlands 32:301-310.
Wang, L., J. Lyons, P. Rasmussen, P. Kanehl, P. Seelbach, T. Simon, M. Wiley, E. Baker, S.
Niemela, and M. Stewart. 2003. Influences of landscape and reach scale habitat on stream
fish communities in the Northern Lakes and Forest ecoregion. Canadian Journal of
Fisheries and Aquatic Sciences 60:491-505.
Wiley, M. J., S. L. Kohler, and P. W. Seelbach. 1997. Reconciling landscape and site based
views of aquatic stream communities. Freshwater Biology 37:133-148.
Woodsmith, R. D., J. R. Noel, and M. L. Dilger. 2005. An approach to effectiveness monitoring
of floodplain channel aquatic habitat: channel condition assessment. Landscape and
Urban Planning 72:177-204.
Worthington, N. P., R. H. Ruth, and E. E. Matson. 1962. Red alder, its management and
utilization. U.S.D.A. Forest Service Miscellaneous Publication 881 44 p.
Wright, K. A., K. H. Sendek, R. M. Rice, and R. B. Thomas. 1990. Logging effects on
streamflow: storm runoff at Caspar Creek in northwestern California. Water Resources
Research 26:1657-1667.
Wu, W., and R. C. Sidle. 1995. A distributed slope stability model for steep forested basins.
Water Resources Research 31:2097-2110.
Ziemer, R. R. 1981. Stormflow response to roadbuilding and partial cutting in small streams of
northern California. Water Resources Research 17:907-917.

138

CHAPTER 3
CONCLUSIONS: TIMBER HARVEST, NATURAL LANDSCAPE FACTORS, AND
RELATIONSHIPS WITH FLUVIAL HABITAT; FROM REACHES TO WATERSHEDS OF
SOUTHEASTERN ALASKA – SUMMARY OF FINDINGS, MANAGEMENT
IMPLICATIONS, AND STUDY LIMITATIONS
This chapter provides a brief overview of the main findings of my MS thesis. I will
present a suite of ideas supported by the results and provide managers of timber and aquatic
resources in Southeast Alaska and the Pacific Northwest with the implications that have been
suggested by this work. These include management considerations related to natural landscape
factors such as stream size or the amount and type of forest within a streams catchment. Timber
harvest related influences on stream habitat were also considered and included harvest type and
intensity at reach and watershed scales. Principal findings include that a potential reduction in
median particle size has been detected in streams with riparian zones harvested during initial best
management practices (i.e., 1978-1990) for the timber industry in the region. Associations
between landscape factors and large wood were additionally made, including timber harvest
efforts occasionally showing stronger relationships with woody debris than natural landscape
factors. The importance of natural factors in controlling aquatic habitat in streams is also shown,
and associations supported in the literature from multiple regions have also been detected in my
study area. Study limitations will also be presented to guide and improve future research on
topics related to stream ecology and management in regions with streams that depend on woody
debris and/or support anadromous fishes.
Principal findings
Chapter 1: Aquatic habitat and influences from reach scale geomorphic, riparian, and timber
harvest factors of Chichagof, and Prince of Wales Islands, Alaska, U.S.A.

139

While studies have documented the importance of riparian zones to stream habitats in
Southeast Alaska and across the world (e.g., Murphy and Koski 1989, Allan et al. 2003, Pusey
and Arthington 2003), this study builds from previously-conducted work to explore linkages
occurring at the transition zone between streams and the landscapes they drain. Additionally,
timber harvest in Southeast Alaska is a major human land use, and regulations in the state did not
initially restrict harvesting trees from the riparian zone, potentially altering processes occurring
between aquatic and terrestrial habitats. Streams and riparian zones are complex ecosystems,
and their connections are not fully understood, so it is challenging to identify the most sensitive
and meaningful descriptors that still adequately describe variation observed across a particular
set of study streams. For this study, stream habitat was characterized at the reach-level using
both quantitative and visual measures of channel morphology, substrate, and large wood (LW)
characteristics, while reach-level landform and tree measurements characterized riparian zones,
resulting in over 200 variables being calculated. Redundancy in aquatic and riparian habitat
variables was reduced initially by investigating correlation values between related variables.
Next, a principal components analysis (PCA) was run on 17 minimally-redundant measures of
stream and riparian habitat to identify main gradients in habitat across the 28 study sites. Five
axes explaining 77.6% of the variation were interpreted as gradients in stream size, power, LW
recruitment, density of LW and pools, and large-sized riparian trees and functioning LW recruits.
Further exploration of reach-level relationships between channel morphology and riparian zones
with aquatic habitat characteristics in the study streams was based on these PCA results.
To further associate geomorphic and riparian factors with aquatic habitat in streams,
eleven variables characterizing channel substrate, LW, and pool habitat were selected based on
the initial data reduction. Geomorphic variables deemed important included bankfull width,
140

gradient, and channel confinement. Riparian variables selected include the proportion of riparian
trees that are deciduous and the proportion of riparian trees that are large and extra-large.
Multiple regression models were run using the independent (i.e., geomorphology and riparian)
variables to predict the suite of aquatic habitat variables selected. To determine how timber
harvest may have influenced aquatic habitat in streams, unpublished data from the U.S. Forest
Service were obtained, and select measures were compared to data from the study sites by using
analysis of covariance (ANCOVA) and also a discriminant function analysis (DFA), while
controlling for known natural controls on aquatic habitat. The results of these analyses suggest
that stream size (i.e., bankfull width) was associated with multiple aquatic habitat measures
including substrate, LW, and complex habitat, while stream power (represented by gradient and
confinement) was related to measures of substrate and wood-formed pools. Riparian factors
(e.g., tree type and size) were less influential, but still associated with wood formed pools and
density of LW in stream reaches. ANCOVA results showed little difference in aquatic habitat
when considering habitat complexity, but a significantly lower median particle size was observed
in streams having experienced riparian zone harvest compared to unharvested streams.
Classification results from the DFA had moderate success discriminating between harvested and
unharvested streams and correctly classified 73.6% of grouped cases after cross-validation.
These results emphasize the importance of considering natural controls on stream habitat before
attributing differences to human land use. While timber harvest may have the potential to reduce
median particle sizes in streams similar to those sampled for this study, the abundance of wood
and presence of beavers will also influence this habitat factor.
Chapter 2: Large wood and influences from the landscape – natural and timber harvest factors

141

Large wood influences stream habitats, and likewise, landscapes influence LW. For
example, LW is increasingly recruited to streams through bank erosion as catchment size
increases, and most LW recruitment comes from the immediate riparian zone. As much is
known about landscape influences in riparian zones on LW, less is known about larger
catchment influences on wood in streams of Southeast Alaska, partially due to limited
availability of landscape data for the region. This study was intended to fill this gap in
knowledge by associating landscape factors at catchment scales, both natural and timber harvest
related to LW in stream channels. Catchment scale data are limited in much of Alaska, so the
first step was to delineate catchment boundaries for stream reaches of the southeastern region of
the state and attribute landscape data to those spatial units. Natural factors calculated for
catchments include watershed area and slope, forest type, and wetland cover. Timber harvest
data were obtained and additionally attributed to study catchments including percent of area
harvested. Logging type, as well as when logging occurred, were summarized and grouped
separately due to potential differences in methods used to harvest trees as regulations over time
became more restrictive in how operations could occur near streams. Logging road and logging
road/stream crossing densities were additionally calculated and summarized for stream
catchments. Study catchments showed considerable variation in landscape factors as catchment
area ranged from 1 to about 20 km² and maximum slope ranged from 1.2 to 7.1%. The average
amount of deciduous forest in local catchments was about 7% while average amount of forested
wetlands in catchments was slightly more than 25%. Timber harvest was also highly variable,
and conventional logging occurring from 1980-1990 was the most pervasive type of practice
across study catchments, averaging over 40% in local catchments, while conventional harvest
after 1990 averaged 14% in study catchments.

142

To determine how influential natural vs. timber harvest related landscape factors were in
explaining variation in measures of LW, two analyses were performed; a best subsets regression,
and a canonical correspondence analysis (CCA). Response variables included LW length, LW
size (i.e., two categories; medium and large and extra-large), as well as LW position in relation
to the active stream channel. Results indicate that LW length was most strongly associated with
catchment size and deciduous forests. Large wood position showed a moderately strong
relationship with conventional harvest from 1980-1990 as well as with slope, while moderate
associations were made between medium-sized LW density and conventional harvest after 1990,
along with deciduous forests. Large and extra-large sized LW density was poorly predicted, but
had a significant relationship with conventional harvest from 1980-1990 as well as selective
harvest after 1990. Variance partitioning results from the CCA suggest that at a catchment scale,
timber harvest factors were more influential in explaining variation in the four LW response
variables as 44% of the total variance was explained; timber harvest accounted for 54% of this
while natural factors accounted for 28%, and their interaction was responsible for the remaining
18%.
Together the results of these two chapters suggest that influences of natural and timber
harvest-related factors on stream habitat vary depending on the scale at which they are
investigated. At a reach-level, natural factors (i.e., channel morphology and riparian tree size
and type) were more important than timber harvest in influencing aquatic habitat measures, while
at a catchment scale, timber harvest explained more variation in LW measures than the natural
factors investigated. These results support the notion that physical characteristics of streams are
differentially affected by landscape factors operating over different spatial scales (Wiens 2002),

143

and these influences result from specific mechanisms by which landscape factors influence
stream conditions.
Management implications
Natural landscape considerations
Natural landscape factors have long been known to influence fluvial systems, and
classifications based on natural landscape conditions are useful to more appropriately manage for
streams that behave in similar ways (e.g., Strahler 1957, Paustian et al. 1984, Montgomery and
Buffington 1997). The results from both chapters of this study concur with previous research
and reinforce the notion that natural factors such as stream size, channel type, and surrounding
vegetative cover can largely dictate how aquatic habitat will be distributed throughout a given
stream reach (Paustian et al. 1992).
This study found catchment size to be greatly important in driving physical stream
characteristics in the study streams, and this is a known control on physical and biological
characteristics of streams around the world (Frissell et al. 1986, Bilby and Ward 1989, Hughes et
al. 1986, Robison and Beschta 1990). While watersheds of the region are generally quite small,
results from multiple regressions in Chapter 1 and Chapter 2 indicate the importance of
catchment size in determining the potential of a stream, whether it be substrate size and
distribution, LW, or habitat complexity. Larger catchments will have more water routed to and
through the channel while slope tends to decrease with increasing basin area (Schumm 1977,
Leopold 1994, Knighton 1998). Longer pieces of wood will be found in larger catchments, and
these streams will have more ability to transport sediment and wood more efficiently down
channels (Benda et al. 2002). Bilby and Ward (1989) also found stream size to be a strong
144

predictor for wood size in Washington. Additionally recruitment of wood is affected by
catchment size and more erosion is found along streams that are larger (Martin and Benda 2001,
Ross Chapter 1 this volume). Managers tasked with prioritizing restoration efforts must
acknowledge how stream size will affect those efforts. Streams that are adequately wide, and
powerful enough to transport large amounts of wood might not benefit as much from wood
placement projects compared to moderately-sized streams that transport less large pieces of
wood. Gravel-poor streams that are moderate in size may benefit from wood placement projects
to increase retention of coarse substrates important for lithophilic spawners, and to act as pool
forming elements. Channels having larger catchments will also be more dynamic in terms of
meandering and bank erosion so they may not need supplemental wood placement, given the
riparian zones have adequate sources and sizes of trees to be recruited in the future.
Deciduous and forested wetlands were additionally important in the density of mediumsized LW, both having a positive relationship. Channel migration can promote colonization of
deciduous trees which grow and die faster than conifers (Minore and Weatherly 1994, Hibbs and
Giordano 1996), and more deciduous pieces have been found in streams with riparian zones that
consist of deciduous vegetation (Andrus et al. 1988, Gomi et al. 2006). Second growth forests
have also been documented to produce LW sized deciduous trees, and increasing proportions of
deciduous trees may promote larger conifers in the long term (Hanley and Hoel 1996,
Orlikowska et al. 2004). Forested wetlands additionally can regulate water routing and increase
transport capacity of small headwater streams after long periods of precipitation that fill wetlands
(Fitzgerald et al. 2003), and this may play a role in moving wood out of fishless headwater
streams into larger fish bearing streams, although LW loads 25 years after harvest were similar to
unlogged streams in the region. Further, forested wetlands of Southeast Alaska have the
145

potential to produce trees large enough to be considered LW, and models estimate that forested
wetlands can produce wood volumes nearly twice the U.S. Forest Service minimum standard to
be considered productive forestland (Julin and D’Amore 2003), while forested wetlands may
also limit the size of LW available to streams. Managers concerned with improving wood
dynamics in streams must consider forest type in the catchment and how it may play a role in
producing, or transporting LW to and through channels.
We also found channel slope to be influential to LW position in relation to the active
stream channel. Steep headwater streams are typically narrower and more confined than low to
moderate gradient streams and this characteristic restricts the capacity for LW to intersect stream
channels (Liquori et al. 2008). May and Greswell (2003) found 33 % of locally recruited wood
pieces were spanning the full width or suspended over the channel of headwater streams in a
coastal Oregon watershed. Furthermore, steep channels are less prone to bank erosion, resulting
in overall low LW recruitment in the absence of wood delivering landslides. Management
efforts regarding wood in these channels will need to balance whether a stream may be too
narrow for LW to function with whether or not sufficient power exists to transport wood out of
the channel; a natural characteristic that can negate efforts intended to increase more woody
debris. Streams having moderate gradients that lack sufficient woody debris may also transport
gravel and other spawning substrates out of these reaches, so increased LW can increase bed
roughness which has been shown to lower particle size distributions by holding substrate and
reducing channel slopes locally (Buffington and Montgomery 1999).
Timber harvest considerations

146

Results of this study suggest the importance of scale when considering effects of timber
harvest on streams in Southeast Alaska. Chapter 1 suggests a potential reduction in median
particle size when considering streams that have had riparian zones harvested for timber,
although the effect may be minimal as most substrates measured were still gravel; the ideal
spawning substrate for salmonids, and the presence of beavers (a factor untested in this study)
can also influence stream substrate sizes. While ample gravel still exists in the study streams,
fine sediment can fill intersticial spaces and reduce spawning success while remaining
undetected by pebble counts until larger substrates are mostly embedded, as pebble counts are
known to be biased toward sampling larger particles. Increased erosion during harvest and
increased landslide rates post-harvest may have contributed to a reduction in particle size by
increasing sedimentation (Murphy 1995). While roads in catchments can also increase
sedimentation, pilot data suggest most non-federal, decommissioned logging roads in Southeast
Alaska are in good shape and likely do not contribute much sediment to streams (ADNR 2010).
Managers tasked with increasing spawning habitat will want to investigate history of timber
harvest near streams, and additionally explore methods used and the time period in which trees
were harvested. A gradient of protection exists through time in regards to timber harvest near
streams and while little difference was observed in aquatic habitat characteristics when timber
harvest occurred at the reach-scale between 1980 and 1990, harvest efforts occurring prior to
1980 will potentially have more pronounced effects. These include less large wood and reduced
particle sizes due to LW removal and use of heavy equipment in and near stream channels, while
timber harvested post-1990 were required to leave riparian buffers and may not be different than
their old-growth counterparts when considering aquatic habitat characteristics.

147

Results from Chapter 2 indicate that at a catchment scale, patterns in timber harvest
explained more variation in measures of LW than natural factors (e.g., catchment area, deciduous
forest). In general, study streams with more harvest from 1980-1990 also had more mediumsized pieces of LW. Some pieces measured had signs that they entered the stream through
timber harvest (e.g., one end cut), and streams having both riparian zones harvested also had
relatively more medium and large sized pieces of LW when compared to only one side
harvested. Additionally timber harvest encourages deciduous vegetation to colonize, and they
can reach a size large enough to be considered a medium piece of LW. Streams having more
deciduous forest in the riparian zone will also have more deciduous pieces of LW in the channel.
Chapter one results indicate that streams having both banks harvested for timber typically had
more deciduous trees in the riparian zone as well. While deciduous trees provide a lower quality
of wood to streams in terms of physical functionality to create pools or armor banks for example,
they do provide a better food source for macroinvertebrates, are nitrogen fixers for streams as
well as other riparian vegetation (e.g., conifers), and this may enhance growth of very large
conifers in the future. Balancing how much deciduous vegetation is beneficial to streams will
depend on clearly defined goals and objectives. Enhancing productivity in streams may require
more light penetration and more nutritious leaf litter that comes from deciduous trees, while
wood-poor streams may benefit more from the long-term channel shaping and substrate holding
characteristics that come with input of coniferous trees. Selectively thinning out either conifers
or deciduous trees can assist in reaching these goals although little research empirically tests for
theoretical benefits provided to streams that can be provided by thinning.
Study limitations

148

Aquatic landscape ecology is an increasingly useful focus of study that investigates
relationships between landscapes and stream conditions, reinforcing the notion that the valley
rules the stream (Hynes 1975). Landscapes influence stream conditions, and humans alter
landscapes, resulting in potential for altered stream conditions, and while research continues to
find associations between landscapes and streams, mechanistic drivers behind these relationships
continue to be unclear. Mechanisms driving linkages between landscapes and streams remain
unknown and hard to test, a point that Allan (2004) acknowledges. This paper also identifies
four main challenges to riverscape-based studies including: covariation of natural and
anthropogenic gradients, multiple interacting stressors, scale of measurement and influence, and
possible legacy effects from historical disturbance.
This study shares all these limitations and more. Multiple factors known to be important
to fishes were unavailable or untested in this study including: geology, landslide history and
frequency, as well as silviculture efforts (e.g., selective thinning) in second-growth riparian
stands, although the spatial location and human influence present in the study region limits some
of the challenges identified by Allan (2004) in riverscape research. Some covariation between
natural and anthropogenic landscape factors do exist in the study catchments. For example more
harvest surely occurred in areas having the most productive soils and largest trees, but landscape
variables used in Chapter 2 analyses showed very little covariation or collinearity with each
other, as indicated by results from the multiple regressions which had low variance inflation
factor values. Additionally the CCA results showed the interaction between natural and timber
harvest related factors explained the least amount of variation in measures of woody debris.
This study did not specifically test for the effect of scale, or the most appropriate scale to
test the effects of timber harvest on streams, but it did investigate multiple scales from reach to
149

watershed-level and found varying associations between stream habitat and timber harvest
factors, depending on the scale investigated. This research still lacks the ability to elucidate
mechanisms that underlie the relationships presented or how scale affects these patterns, which
continues to be a major challenge in ecological studies (Levin 1992). It is important that future
research considers scales that are relevant to the questions being asked, or the resource being
managed.
Southeast Alaska has very few urban areas and virtually no agriculture, so timber harvest
is the most pervasive human land use, making the problem of multiple interacting stressors rare
for watersheds in the region. Legacy effects from historical disturbances are mostly limited to
activities occurring after 1950 because most large-scale timber exploitation occurred much later
in Alaska than in the conterminous U.S., but there are localized areas that have historic and
current mining operations. Because of the remoteness and lack of historical human influence, in
addition to the sociocultural and economic importance of salmonids in Alaska, it may be possible
to identify the closest thing to “true” reference conditions. This would allow us to link
mechanisms to associations as legacy effects are currently limited and there is not a suite of
stressors acting in conjunction to influence aquatic habitat in the state. This certainly differs
from most watersheds in North America as much of the continent has been altered from its “precolonial” condition. It is important for future researchers and natural resource managers and
decision makers to remember to use the term reference condition, even in Alaska or other remote
regions, in an appropriate context as likely no watershed on earth is void of human influence.
The earliest North Americans quite possibly entered the continent through Alaska so some of the
earliest exploitation of resources occurred here. Native Alaskans established relatively large,
permanent communities and cultures that revolved around wood resources, so timber harvest has
150

occurred here for thousands of years, and even the most remote watersheds on earth have
certainly not escaped exposure to deposition of atmospheric pollutants, so terms like minimally
disturbed, or least disturbed (Stoddard et al. 2006) may be more appropriate as physical and
biological assessments of stream condition continues into the future.

151

LITERATURE CITED

152

LITERATURE CITED

Alaska Department of Natural Resources. 2010. ADNR road condition/fish passage monitoring
year 5, Pages 1-4 in Alaska sustainable salmon fund project completion report.
Allan, J. D., M. S. Wipfli, J. P. Caouette, A. Prussian, and J. Rogers. 2003. Influence of
streamside vegetation on inputs of terrestrial invertebrates to salmonid food webs.
Canadian Journal of Fisheries and Aquatic Sciences 60:309-320.
Allan, J. D. 2004. Landscapes and riverscapes: The influence of land use on stream ecosystems.
Annual Review of Ecology Evolution and Systematics 35:257-284.
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.
Benda, L. E., P. Bigelow, and T. M. Worsley. 2002. Recruitment of wood to streams in oldgrowth and second-growth redwood forests, northern California, USA. Canadian Journal
of Forest Research 32:1460-1477
Bilby, R. E., and J. W. Ward. 1989. Changes in characteristics and functions of woody debris
with increasing size of streams in Western Washington. Transactions of the American
Fisheries Society 118:368-378
Buffington, J. M., and D. R. Montgomery. 1999. Effects of hydraulic roughness on surface
textures of gravel-bed rivers. Water Resources Research 35(11):3507-3521.
Fitzgerald, D. F., J. S. Price, and J. J. Gibson. Hillslope-swamp interactions and flow pathways
in a hypermaritime rainforest, British Columbia. 2003. Hydrologic Processes 17:30053022.
Frissell, C.A.,W.J. Liss, C.E. Warren and M.D. Hurley. 1986. A hierarchical framework for
stream habitat classification: viewing streams in a watershed context. Environmental
Management, 10: 199–214.
Gomi, T., A. C. Johnson, R. L. Deal, P. E. Hennon, E. H. Orlikowska, and M. S. Wipfli. 2006.
Factors affecting distribution of wood, detritus, and sediment in headwater streams draining
managed young growth red alder-conifer forests in southeast Alaska. Canadian Journal of
Forest Research 36:725-737.
Hanley, T. A., and T. Hoel. 1996. Species composition of old growth and riparian Sitka sprucewestern hemlock forests in southeastern Alaska. Canadian Journal of Forest Research
26:1703-1708

153

Hibbs, D. E., and P. A. Giordano. 1996. Vegetation characteristics of alder-dominated riparian
buffer strips in Oregon Coast Range streams. Northwest Science 70:213-222.
Hughes, R. M., D. P. Larsen, and J. M. Omernik. 1986. Regional reference sites: a method for
assessing stream potentials. Environmental Management 10:629-635.
Hynes, H. B. N. 1975. The stream and its valley. Verhandlungen Internationale Vereinigung für
Theoretische und Angewandte Limnologie 19:1–15.
Julin, K. R., and D. V. D’Amore. 2003. Tree growth on forested wetlands of southeastern Alaska
following clearcutting. Western Journal of Applied Forestry 18:30-34.
Knighton, D. 1998. Fluvial forms and processes: A new perspective. John Wiley and Sons, Inc.,
New York, New York.
Leopold, L. B. 1994. A view of the river. Harvard University Press, Cambridge, Massachusetts.
Levin, S. A. 1992. The problem with pattern and scale in ecology: The Robert H. MacArthur
Award Lecture. Ecology 73:1943-1967.
Liquori, M. D. Martin, L. Benda, R. Coats, and D. Ganz. 2008. Scientific literature review of
forest management effects on riparian functions for anadromous salmonids. Report of
Sound Watershed Consulting to the California State Board of Forestry and Fire
Protection, Sacramento, CA.
Martin, D. J. and L. E. Benda. 2001. Patterns of instream wood recruitment and transport at
the watershed scale. Transactions of the American Fisheries Society 130:940-958.
May, C. L., and R. E. Gresswell. 2003. Large wood recruitment and redistribution in headwater
streams in the southern Oregon Coast Range, U.S.A. Canadian Journal of Forest
Research 33:1352-1362.
Minore, D., and H. G. Weatherly. 1994. Riparian trees, shrubs, and forest regeneration in the
coastal mountains of Oregon. New Forests 8:249-263.
Montgomery, D. R., and J. M. Buffington. 1997. Channel reach morphology in mountain
drainage basins. Geological Society of America Bulletin 109:596-611.

Murphy, M. L. 1995. Forestry impacts on freshwater habitat of anadromous salmonids in the
Pacific Northwest and Alaska—Requirements for protection and restoration. NOAA
Coastal Ocean Program Decision Analysis Series No. 7. NOAA Coastal Ocean Office,
Silver Spring, MD. 156 pp.

154

Murphy, M. L., and K. V. Koski. 1989. Input and depletion of woody debris in Alaska streams
and implications for streamside management. North American Journal of Fisheries
Management 9:427-436
Orlikowska, E. H., R. L. Deal, P. E. Hennon, and M. S. Wipfli. 2004. The role of red alder in
riparian forest structure along headwater streams in Southeastern Alaska. Northwest
Science 78(2):111-123.
Paustian, S. J., D. A. Marion, and D. F. Kelliher. 1984. Stream channel classification using large
scale aerial photography for southeast Alaska watershed management. Pages 670-677 in
Renewable resources management: applications of remote sensing. American Society of
Photogrammetry, Bethesda, Maryland.
Paustian, S. J., K. Anderson, D. Blanchet, S. Brady, M. Cropley, J. Edgington, J. Fryxell,
G. Johnejack, D. Kelliher, M. Kuehn, S. Maki, R. Olsen, J. Seesz, and M. Wolanek.
1992. A channel type user’s guide for the Tongass National Forest, southeast Alaska.
Technical Paper 26, U.S. Forest Service, Juneau, Alaska.
Pusey, B. J., and A. H. Arthington. 2003. Importance of the riparian zone to the conservation and
management of freshwater fish: a review. Marine and Freshwater Research 54:1-16.
Robison, E. G., and R. L. Beschta. 1990. Identifying trees in riparian areas that can provide
coarse woody debris to streams. Forest Science 36:790-801.
Schumm, S. A. 1977. The fluvial system. John Wiley and Sons, New York, New York.
Stoddard, J. L., D. P. Larsen, C. P. Hawkins, R. K. Johnson, and R. H. Norris. 2006. Setting
expectations for the ecological condition of streams: the concept of reference condition.
Ecological Applications 16:1267-1276.
Strahler, A. N., 1957. Quantitative analysis of watershed geomorphology. Transactions of the
American Geophysical Union 38:913-920.
Wiens, J. A. 2002. Riverine landscapes: taking landscape ecology into the water. Freshwater
Biology 47:501-515.

155