Date post: | 16-Nov-2023 |
Category: |
Documents |
Upload: | independent |
View: | 0 times |
Download: | 0 times |
Ecological Applications, 17(4), 2007, pp. 1127–1141� 2007 by the Ecological Society of America
SOURCES AND DYNAMICS OF LARGE LOGSIN A TEMPERATE FLOODPLAIN RIVER
JOSHUA J. LATTERELL1
AND ROBERT J. NAIMAN
School of Aquatic and Fishery Sciences, University of Washington, Box 355020, Seattle, Washington 98195 USA
Abstract. Large logs, important agents of biophysical heterogeneity in temperatefloodplain rivers, have been virtually eliminated from modified systems. Our purpose was toquantify the sources and dynamics of large logs (�1 m diameter) in the mainstem of a nearlypristine system: the Queets River, Washington, USA. Erosion of forests by the river supplies0.40 logs�(100 m)�1�yr�1 to the channel. Most (72%) are new logs entering the river for the firsttime as the river undercuts mature fluvial terraces dominated by large conifers. Retrospectiveairphoto analyses demonstrate that, over 63 years, the Queets River recruits 95% of new logsfrom a riparian corridor extending 265 m laterally on both banks, mostly through channelmeandering. However, input rates are patchy, with 10% of the valley length supplying 38% ofthe new logs. As the river moves laterally, remnant logs are left on channel surfaces that laterdevelop riparian forests and reenter the river when those forests erode. Remnant logs lying onthe floodplain forest floor surface or buried in alluvium constitute 21% and 7% of the annualinputs from bank erosion, respectively. We estimate that 50% of logs deposited in the channelin a given year, including those underpinning logjams, are transported downriver within fiveyears. Over the next 55 years, bank erosion reclaims an additional 23%, leaving 27% of the logsstable for .60 years. Simulations indicate that recurrent transport is common, with half of thelarge conifers being deposited in �3 locations and transported �1.5 km prior to disintegrating.One in ten logs links distant reaches by occupying �7 locations spanning �12.0 km. Instreamsupplies are therefore a mixture of new and old logs from nearby and upstream forests,sustained by the recapture and transport of stockpiled remnant logs during periods when newinputs are low. We propose that patchy input rates and the periodic rearrangement of largelogs are important drivers of temporal variation in river valley habitats, adding to the spatialcomplexity created by stable logs. These findings underscore the importance of extensivemature forests and connectivity in temperate floodplain rivers.
Key words: connectivity; floodplain; flow regime; heterogeneity; large woody debris; Queets River,Washington State (USA); riparian ecosystem; spiraling.
INTRODUCTION
The ecological vitality of laterally unstable floodplain
rivers in temperate regions is underpinned, to a large
degree, by natural flow variability and the delivery of
large wood from riparian forests to the channel (Naiman
et al. 2005a, b). These processes support high levels of
biocomplexity, a widely valued trait now lacking in
many developed rivers (Poff et al. 1997, Gurnell et al.
2001, 2005). The goal of this study was to inform
conservation of temperate floodplain rivers and riparian
forests by quantifying the delivery of large logs to the
river through bank erosion (e.g., Piegay et al. 1999) and
their cycling within the channel and floodplain over time
(1939–2002) and space in the near pristine Queets River,
Washington, USA, in Olympic National Park (Fig. 1).
We focused on large logs that play an irreplaceable role
in underpinning logjams (key pieces or key members)
but have declined precipitously in most modified rivers
(Gregory et al. 2003, Gurnell et al. 2005, Naiman et al.
2005b). Our specific objectives were to (1) quantify the
rate at which trees are delivered to the channel for the
first time by the undercutting of mature forests
(becoming ‘‘new’’ logs) as well as the lateral extent of
the source areas, (2) quantify the retention and transport
dynamics of instream logs, and (3) estimate the relative
importance of inputs of remnant logs lying on the
surface of floodplain forests, or buried in the alluvium.
The natural history of large wood in floodplain rivers
has been often described (see photos in Appendix A) but
rarely quantified, hindering understanding of these
dynamic systems. River-deposited logs are most con-
spicuous in the river channel but also pervade the forests
and alluvium of the river valley floor (Fig. 2; Maser et al.
1988). Large logs entering the channel via bank erosion
function as key pieces by trapping additional wood,
promoting sediment deposition, and shaping the sur-
rounding channel (Keller and Swanson 1979, Abbe and
Montgomery 1996). Resulting logjams can initiate
lateral channel migration by increasing the elevation of
the streambed or water surface (Brummer et al. 2006).
Some logs are temporarily stranded on gravel bars and
Manuscript received 7 June 2006; revised 13 October 2006;accepted 27 October 2006. Corresponding Editor: J. A. Antos.
1 E-mail: [email protected]
1127
subsequently flushed downstream (van der Nat et al.
2003a). Other logs (hereafter, ‘‘remnant logs’’) remain
where they were deposited by the river and, over several
years, become veiled by stands of pioneering vegetation
growing in the old channels (Fetherston et al. 1995,
Abbe and Montgomery 1996). These logs remain on the
forest floor (i.e., ‘‘surface remnants’’) until they disinte-
grate or are buried under alluvium (Guyette et al. 2002,
Montgomery and Abbe 2006), hereafter referred to as
‘‘buried remnants.’’ The net result is that river-deposited
logs exist in most patches composing the forested
floodplain mosaic (Latterell et al. 2006). This is
important because, when the laterally migrating channel
eventually returns, remnant logs that persist can reenter
the channel (Nanson et al. 1995, Hyatt and Naiman
2001, Guyette et al. 2002) and potentially underpin new
logjams (Collins and Montgomery 2002).
Observations from intact low-gradient rivers suggest
the ongoing loss of wood substantially reduces bio-
complexity (Gurnell et al. 2005) and alters key
biophysical patterns in developed rivers. When present,
logs enhance instream complexity and promote flood-
plain inundation (Kellerhals et al. 1976, Abbe and
Montgomery 1996, Brooks et al. 2003). Large logs are
central to organic matter retention (Bilby and Likens
1980, Bilby 1981), to pool formation (Abbe and
Montgomery 1996, Beechie and Sibley 1997), and to
nutrient uptake (Aumen et al. 1990, Valett et al. 2002).
Remnant logs provide habitat for a variety of terrestrial
organisms (Bull 2002, Steel et al. 2003) and facilitate
conifer establishment (Fetherston et al. 1995, Fetherston
2005, Van Pelt et al. 2006). Most logs reside in
floodplain river valleys for decades, though some
fraction lasts for centuries or more (Nanson et al.
1995, Hyatt and Naiman 2001, Guyette et al. 2002,
Montgomery and Abbe 2006). Those remaining stable
over long periods may represent a sizeable carbon
reservoir (Guyette et al. 2002) and aid in replenishing
supplies of new large logs by protecting developing
forests from erosion long enough for trees to grow large
(Montgomery and Abbe 2006). In the absence of large
wood, few structures in low-gradient rivers are suitable
ecological surrogates for these functions.
Sustaining large wood in rivers is both a scientific and
a practical challenge that exemplifies the difficulty of
protecting river environments while meeting human
needs (Nilsson and Berggren 2000). Floodplain rivers
face burgeoning threats (Hughes 2003; Tockner et al., in
press) as logjams are extracted to improve navigation
and reduce bridge damage; future supplies of large logs
diminish as mature riparian forests are cleared for
agriculture and development; and recruitment processes
are cut off as channel movement and flooding are
artificially restricted to prevent loss of life and property
(Sedell and Froggatt 1984, Bilby and Bisson 1998,
Collins et al. 2002, 2003). Not surprisingly, these factors
have transformed riverine-riparian ecosystems at the
global scale (Nilsson and Berggren 2000, Nilsson et al.
2005) and are expected to intensify as regional water
scarcity is exacerbated by population and economic
growth and climate change (Vorosmarty et al. 2004).
The net result is that large logs, associated jams, and
habitats they sustain are virtually eliminated, and once
lost, are difficult to replace at meaningful scales. Our
hope is that results of this study will be useful in
informing conservation efforts aimed at maintaining or
restoring a self-sustaining supply of large wood to
floodplain rivers.
STUDY AREA
The Queets River, Washington, USA, is located at
4783201700 N, 12481805200 W NAD27 in Olympic Nation-
al Park (Fig. 1). The river drains ;1200 km2 before
reaching the Pacific Ocean. Although a wide range of
biophysical characteristics is evident in rivers of the
Pacific coastal rain forest (Naiman and Anderson 1997,
Naiman et al. 2005b), the Queets River possesses many
characteristics typical of pristine lower floodplains,
FIG. 1. Map of the mainstem of the Queets River, Washington, USA (bold), and its tributaries, indicating the extent ofOlympic National Park (shaded) and the locations of four intensively studied 2-km study reaches (open circles).
JOSHUA J. LATTERELL AND ROBERT J. NAIMAN1128 Ecological ApplicationsVol. 17, No. 4
which often meander through a wide U-shaped valley
flanked by old forests on glacial landforms (Thackray
2001). The climate is mild and wet with mean annual air
temperatures of 108C and precipitation averaging 387
cm/yr. Typical riparian vegetation consists of willow
(mostly Salix sitchensis Sanson), red alder (Alnus rubra
Bong.), Sitka spruce (Picea sitchensis [Bong.] Carr),
black cottonwood (Populus balsamifera L. ssp. tricho-
carpa [Torrey and A. Gray] Brayshaw), bigleaf maple
(Acer macrophyllum Pursh), and vine maple (Acer
circinatum Pursh) (Balian and Naiman 2005). Western
hemlock (Tsuga heterophylla [Raf.] Sarg.), Douglas-fir
FIG. 2. Aerial view of a representative reach of the upper mainstem of the Queets River, Washington, USA. This photo depictsnewly formed jams near the channel on the left, and several remnant logs in the early stages of incorporation by pioneeringvegetation on the upper right.
June 2007 1129FLOODPLAIN RIVER WOOD DYNAMICS
(Pseudotsuga menziesii [Mirbell] Franco), and western
redcedar (Thuja plicataDonn.) are also present (Naiman
et al. 2005a, Van Pelt et al. 2006). Coastal Sitka spruce
forests rarely burn, and fires have not been observed in
the Queets valley floor (Agee 1993). The fire return
interval is not well-defined, but is estimated to exceed
1100 years (Agee 1993, Greenwald and Brubaker 2001).
Terraces and hillslopes are dominated by western
hemlock (Franklin and Dyrness 1988). The channel is
gravel-dominated, low gradient (slope¼ 0.006 m/m) and
laterally active (i.e., moving 13 m/yr on average;
Latterell et al. 2006). The channel averages 128 m in
width (64 SE), but ranks in the fifth-order (based on
streams mapped at 1:24 000 scale) due to the elongate,
rectangular shape of the basin (e.g., Benda et al. 2004).
Daily flows average 124 m3/s, and annual peak flows
range from 931 to 3764 m3/s from rainstorms and rain-
on-snow events (Station 12040500, U.S. Geological
Survey 2005). Flooding is the primary disturbance
affecting the riparian forest. Human disturbances in
the valley prior to the establishment of the Park were
minor, limited to forest clearing around homesteads
established after 1890 and abandoned prior to 1953.
METHODS
We integrated airphoto analyses over the entire
mainstem river inside Olympic National Park, with
intensive studies of log dynamics and burial at four
randomly selected 2-km study reaches draining 207–565
km2 (Appendix B). We define large logs as those �1 m in
diameter at breast height (dbh; or 1.4 m from the root
mass), which approximates the observed median size of
key pieces in the Queets River mainstem (Latterell 2005).
Prior studies demonstrate that the capacity of a log to
remain stable is related, in part, to channel size
(Lienkaemper and Swanson 1987, Braudrick and Grant
2001, Abbe and Montgomery 2003). We reasoned that
many of the logs meeting or exceeding the observed
median diameter of key pieces are likely to eventually
initiate jams in some mainstem reaches even if that
function is not yet realized, and are thus potential key
pieces (e.g., Collins et al. 2002).
Inputs of new and remnant logs
We quantified delivery of new large logs by mapping
mature fluvial terraces as they existed in 1939 and
measuring the cumulative erosion of these patches by
the channel over subsequent years from georeferenced
digital airphotos (0.4–1.5 m ground resolution). Mature
fluvial terraces contain heterogeneous forest canopies
dominated by conifers from ;130 to .330 years old.
Living trees �1 m dbh are almost exclusively found in
these patches (Balian and Naiman 2005, Latterell et al.
2006, Van Pelt et al. 2006). We used ArcMap (ESRI,
Redlands, California, USA) to quantify the cumulative
erosion of mature fluvial terraces over six to seven
intervals, depending on the spatial extent of available
photos. Photos from 1939, 1950, 1954, 1968, 1973, 1993,
2000, and 2002 were used in the first 21.2 km of the
valley upstream from the Park boundary. The remaining
portion of the valley was analyzed with photos from
1939, 1968, 1976, 1981, 1993, 2000, and 2002. We
assumed relationships from multiyear comparisons were
applicable at an annual scale (as in Piegay et al. 1999,
Richter and Richter 2000). We quantified longitudinal
patterns in the delivery of new logs by subdividing the
river valley into 0.2 km long sections (n¼257), following
O’Connor et al. (2003). Transects perpendicular to the
valley axis delineated the upstream and downstream
boundaries of each section. We estimated delivery rates
(pieces�100 m�1�yr�1) by multiplying the area of each
eroded patch (from each section and interval) by an
independent estimate of stem density. Van Pelt et al.
(2006) estimated that mature fluvial terraces in the
Queets River contain 36.3 6 5.8 stems/ha (mean 6 SD)
�1 m dbh, which were mostly (91%) living trees. We
assumed this variable was normally distributed and
drew consecutive, independent estimates of stem density
from this distribution to simulate spatial variation.
Finally, we quantified mean annual delivery rates over
time for each section and overall.
The first step in quantifying the rate at which remnant
logs reentered the channel through erosion of younger
forested patches (i.e., pioneer bars to transitional fluvial
terraces) was to estimate the density of surface and
buried remnant large logs (pieces/ha) in each forested
patch type. Latterell et al. (2006) estimated the density of
surface remnant (or abandoned) large logs in each patch
type. We quantified the density of buried remnant large
logs in riparian forests by mapping key pieces (logs
underpinning jams) in airphotos while they were visible
in the active channel, tracking them through time as
forests developed around them, and determining which
were eventually buried. Hereafter, we use the term
‘‘cohort’’ in reference to a group of individual logs
fluvially deposited (or present) in the unvegetated river
channel in the same year. Cohorts of instream logs were
identified at each 2-km study reach in eight years where
instream logs were clearly visible at those locations (i.e.,
1939, 1950, 1962, 1968, 1990, 1993, 1996, and 1997).
Maps of the active channel in subsequent years indicated
which members of each cohort the river had displaced.
Logs remaining stable in 2002 were located with a
geographic positioning system (GPS) and we determined
whether the logs were buried or remained visible on or
protruding from the forest floor. This enabled us to
model the relationship between cohort ‘‘age’’ (years since
deposition) and the buried percentage (percentage of
initial cohort) with regression analysis. We had to
assume that (1) burial usually occurred after initiation
of vegetative patches, which was (2) coincident with log
deposition; (3) the initial remnant log density in forested
patches was equivalent to that observed in pioneer bars;
and that (4) remnant logs did not disintegrate complete-
ly in �63 yr, but (5) instead remained visible until the
time of observation unless the surrounding forest was
JOSHUA J. LATTERELL AND ROBERT J. NAIMAN1130 Ecological ApplicationsVol. 17, No. 4
eroded or the log was buried. If so, the density of buried
remnant logs in each forested patch type should be equal
to the product of the initial log density and buried
percentage predicted by the regression model from the
mean age of each patch type, which was adapted from
Van Pelt et al. (2006).
Estimating the density of buried remnant logs in
mature fluvial terraces required a more complex
approach because many of the surface remnant large
logs were expected to decay prior to being buried.
Simulations identified the fraction of the initial number
of large logs that would decompose prior to burial. This
entailed generating functional lifetimes (described in
following sections) for 1000 logs (conifers only, for
simplicity), which we assumed were buried at the
empirically determined burial rate, described before.
We assumed that because wood decay slows after burial
(Guyette et al. 2002), logs buried prior to reaching their
estimated minimum functional lifetime retain their
functional capacity when later exhumed. In the case
that the predicted functional lifetime was reached prior
to burial, we assumed the log was no longer a potential
key piece. Inputs to the channel were estimated from the
product of buried remnant log density and the mean
annual area of mature fluvial terrace eroded by the river.
The resulting estimates of buried remnant logs in mature
terraces must be considered preliminary because extrap-
olation of the burial model beyond the range of
observations introduces substantial uncertainty.
For all other patch types, the next step was to estimate
the area of each patch type eroded by the river each
year, thereby returning some quantity of remnant logs to
the channel. Due to ongoing disturbance and plant
succession, we assumed the relative area of each patch
type at the reach scale mirrored that at the valley scale
(reported by Latterell et al. 2006) over the long term
because the area of each patch type fluctuates among
decades at the reach scale. Patch area was scaled to that
of the channel in each study section. Latterell et al.
(2006) present negative exponential models of patch
longevity on the Queets River, which indicate the
fraction of each patch type that is eroded annually.
Thus, the quantity of remnant logs annually returned to
the channel in each section was the product of the
estimated eroded area of each patch type in that section
and the patch specific estimates of surface and buried
remnant log densities.
Lateral source areas and input mechanisms
We quantified lateral source areas and the relative
importance of various physical types of channel change
for the delivery of new large logs in the entire mainstem.
We accomplished this by measuring the lateral recruit-
ment distance, defined as the shortest distance between
the center of each eroded patch (n ¼ 1341 patches) and
the initial channel margin (as it existed in 1939). These
measurements produced a frequency histogram illus-
trating the lateral extent of large log source areas over
six decades. This approach differs from prior studies
based on the distance between the growth location of
individual trees and the present margin of relatively
stable channels (McDade et al. 1990, Van Sickle and
Gregory 1990). Contributions of new large logs from
channel meandering, cutoffs or avulsions, and island
dissection (Knighton 1998) were evaluated with air-
photo interpretation. Each eroded patch was attributed
to one of these mechanisms or else to indeterminate
erosion.
We used regression analysis to determine whether
input rates of new large logs were related to the
proximity of mature forests to the channel or channel
movement rates. Source area proximity, defined as the
lateral distance (m) from the center of the low-flow
channel to the nearest mature fluvial terrace patch in
1939, was measured for each 0.2-km section in
unconfined channel reaches. We compared the mean
annual delivery rate in each section to the mean annual
lateral movement rate of the primary low-flow channel,
as reported for each section by Latterell et al. (2006). We
deemed relationships were significant at P � 0.05.
Dynamics of large logs
We quantified the cycling of key pieces and large logs,
in general, as a series of spirals in space and time (Fig.
3). The spiraling metaphor is well established in river
ecology, and usually invoked to describe nutrient uptake
and release by organisms or transport of particulate
organic matter (e.g., Newbold et al. 1982, Newbold
1992, Fisher et al. 1998). In this application, we consider
each spiral to represent the reciprocal transfer of a log
between the channel and forest patches in the valley
floor.
The process of wood spiraling begins when a tree falls
into the channel for the first time and ends when the log
can no longer initiate jams due to fragmentation or
decay (Fig. 3). This is a simple adaptation of the lifetime
transport distance metric (n) from recent conceptual
advances in wood budgeting (Benda and Sias 2003).
Benda and Sias (2003) propose that the mean distance a
log is transported before disintegrating (n) is a function
of the mean distance between transport-impeding jams,
the ratio of piece to jam longevity, and the proportion of
the channel spanned by a jam. Their model applies to
rivers where wood transport is primarily limited by jam
spacing (e.g., Martin and Benda 2001), and is not
intended for floodplain systems where large logs are
commonly deposited between preexisting jams and
become temporarily incorporated into riparian vegeta-
tion. We modified the model form and parameters to
reflect the spiraling metaphor.
We estimated large log turnover length (Fig. 3, nLL;the total downstream distance from the initial point of
entry into the channel to loss of function) from empirical
observations of large log retention time (in situ; aLL) andannual transport distance (JLL). Large log retention time
(aLL) is the duration of log stability (years) at one
June 2007 1131FLOODPLAIN RIVER WOOD DYNAMICS
location, in contrast to ‘‘residence time,’’ which refers to
the duration over which a log resides in the river system
(e.g., Hyatt and Naiman 2001). This metric is compa-
rable to jam longevity ‘‘a’’ (Benda and Sias 2003), except
that it includes the time spent in the riparian forest. We
quantified short-term retention for three cohorts (2002,
2003, 2004) in each study reach by mapping (63 m),
measuring, tagging (n ¼ 222) instream key pieces, and
relocating the tagged logs in each reach and year of the
study. We integrated long-term retention by mapping all
visible instream key pieces (n¼ 1285) in airphotos from
11 prior years (cohorts), across the entire valley in the
Park. The number of logs visible in photos varied among
years, depending on photo clarity and extent. We
therefore determined the proportion (rather than the
absolute number) of mapped logs comprising each
cohort that were retained. Map overlays of the active
channel in subsequent years after log deposition
indicated whether logs were retained or reentered the
channel. The annual transport distance (JLL) is how far
logs move after displacement or recapture, and is
comparable to the distance between jams (J; Benda
and Sias 2003). The difference is that large log transport
was empirically derived rather than assumed to be
limited by jam-spacing. We quantified JLL by tracking
mobile logs (that previously functioned as key pieces)
between 2002 and 2005 in each study reach and over
;25 km of the mainstem; the stream distance between
the initial and new locations was measured in ArcMap.
The number of spirals (pLL) corresponds to the
number of locations a large log resides, which is
restricted by log lifetime (qLL), the duration (years)
over which a large log retains the capacity to initiate
jams. The qLL is comparable to the lifetime of a piece in
Benda and Sias (2003) but refers to a functional
threshold. Logs can spiral repeatedly until either (1)
the sum of their retention times aLL at each deposition
point meets or exceeds qLL or (2) they are transported
out of the system (e.g., past the river mouth to the
ocean). Martin and Benda (2001) speculated that wood
becomes mechanically weak and disintegrates in streams
when it is 95% decayed. We speculated that large logs
are unlikely to initiate logjams in a floodplain river when
they lose 75% of their mass, recognizing that mass or
size alone does not determine key piece functionality (see
Abbe 2000 for further details).
We simulated probability distributions for nLL, qLL,
and pLL from 1000 trials, each representing an
individual large log (adapted from Martin and Benda
2001). The process of wood decay in rivers is poorly
understood, and estimates for k are quite rare (but see
Graham and Cromack 1982 and Bilby et al. 1999). We
used species-specific decay constants (k) from ongoing
decomposition studies at the Cascade Head and H. J.
Andrews Experimental Forests in Oregon, USA (i.e.,
CHEF and HJAF; Harmon and Fasth 2005) to simulate
the lifetime (qLL) of individual logs. These sites are mild
and very wet, similar to the Queets River valley. We
assumed k was normally distributed about a mean of
0.023 for P. sitchensis and T. heterophylla, which
compose most key pieces (Latterell 2005), and ignored
potential variation from differences in log diameter. We
thought this was justified because our study is limited to
large logs, and a study in a neighboring valley concludes
that decay rates for P. sitchensis and T. heterophylla (the
dominant species of key pieces) are similar, and
differences among diameter classes are not statistically
FIG. 3. Conceptual diagram of the dynamics of large logs ina temperate floodplain river. The spiraling process begins whena log enters the active channel for the first time (e.g., fromerosion of a mature fluvial terrace). The log is deposited in thechannel, where it is retained for a period (retention time; aLL).This value corresponds to the width of a single spiral (the timeelapsed between transport events). Some become remnant logsas floodplain forests develop on old channel surfaces after thechannel migrates laterally. At a later time, the log is displacedor reenters the river through bank erosion and movesdownstream (JLL). The turnover length (nLL) is the totaldistance (the sum of all JLL) between the point of entry and thepoint where the log reaches the end of its functional lifetime(qLL, the number of years elapsed before decay inhibits theability of a log to underpin a jam). The number of spiralscompleted is pLL, which corresponds to the number of locationsin which the log resides. Values pLL are influenced by theconsecutive values of aLL between transport events. Themaximum value for pLL is restricted by the difference betweenaLL and qLL, or by the distance to the boundary of the systemof interest. If aLL exceeds qLL, the log will not be recapturedprior to disintegration (or become buried; pLL ¼ 1).
JOSHUA J. LATTERELL AND ROBERT J. NAIMAN1132 Ecological ApplicationsVol. 17, No. 4
significant (Graham and Cromack 1982). Prior studies
considering the influence of log diameter on decay ratesin upland forests have reached inconsistent conclusions(see Laiho and Prescott 2004 for a review). In some
cases, large logs appear to decay slowest, but wood-boring insects may preferentially attack them and causethem to decay faster (Edmonds and Eglitis 1989). The
standard deviation of the distribution around k (6.63 3
10�3) was derived from time series regressions of decay
for Tsuga heterophylla at HJAF. In each trial, a valuefor k was drawn from this distribution, and used in anegative exponential decay model to estimate the time
(years) required to lose 75% of the wood mass. This wasalso done for hardwoods, such as A. rubra (k¼0.055), toevaluate the influence of qLL on nLL. The fitted
nonlinear model for aLL was used to generate a discreteprobability function (i.e., annual time steps) from whichindependent samples were drawn at random until the
simulated qLL was met or exceeded for each log. Thenumber of samples was tallied to calculate pLL for eachtrial. Independent samples (n) of the natural logarithm
of observed qLL were drawn at random from a two-parameter lognormal distribution (l¼ 5.26, r¼ 2.07) sothat n¼pLL and summed to determine nLL for each trial.
RESULTS
Inputs of new and remnant logs
The average annual input from bank erosion from1939 to 2002 was 0.40 large logs�100 m�1�yr�1 (along the
valley axis) across years and sections, primarily new logsfrom eroding mature fluvial terraces (Table 1). Mature
fluvial terraces were eroded at an average rate of 7.8 3
10�3 ha�100 m�1�yr�1 across sections. We estimated thatthis generated 0.29 6 0.34 new large logs�100 m�1�yr�1(mean 6 SD) representing 72% of the annual input tothe channel, though input rates varied substantiallyamong locations (Fig. 4). In comparison, 21% were
surface remnants and 7% were buried remnants that
were recaptured by bank erosion at a combined rate of
0.11 logs�100 m�1�yr�1 across years and sections (Table
1). Most remnant log inputs originated from the erosion
of pioneer bars and developing floodplain patches
(Table 1), which are eroded frequently, relative to older
patches (Latterell et al. 2006). Most remnant logs are
conifers with the potential to function as key pieces
when recaptured by the channel, based on the simulated
distributions for the functional lifetimes of large
conifers. However, we estimate ,10% of the remnant
hardwood logs recaptured from established floodplains
and transitional fluvial terraces retain the potential to
initiate jams.
We observed a positive, linear relationship (r2¼ 0.88)
between the burial rate and estimated cohort age (Fig.
5):
y ¼ 0:88t � 4:11 ð1Þ
where y is the percentage of the remnant (stable) logs
from a cohort that are buried t years after initially being
deposited. This relationship holds for six decades.
Whether it eventually plateaus over long time periods
is unknown. If the existing linear model (Eq. 1) is
extrapolated beyond the range of observations (in the
absence of known historical data), all remaining logs are
predicted to be buried within roughly 120 yr. Simula-
tions suggested that, in transitional fluvial terraces, 14%
of large logs in the initial cohort reach the end of their
functional lifetime prior to burial. This percentage is
predicted to grow to 43% in mature fluvial terraces, and
is reflected in estimated standing stocks and inputs of
buried remnant logs (Table 1). In spite of these losses,
roughly 60% of the buried remnant logs recruited by the
channel may originate from mature fluvial terraces
because they compose 48% of the valley floor (Latterell
et al. 2006). As mentioned earlier, we caution that these
burial estimates are based on several simplifying
TABLE 1. Mean quantities and sources of large logs (�1 m diameter; new, surface remnant, and buried remnant) in forest patchesof the Queets River, Washington, USA, in Olympic National Park.
Forest patch type
Variable, by source of logs
Standing stock(no./ha)
Input rate(no. logs�100 m�1�yr�1)�
Relative importance of inputs(% of total)
Newlogs
Surfaceremnant
Buriedremnant Total
Newlogs
Surfaceremnant
Buriedremnant Total
Newlogs
Surfaceremnant
Buriedremnant Total
Pioneer bar 4 0.1 4.1 0.026 0.0006 0.027 0.0 6.4 0.1 6.6Developing floodplain 3 0.2 3.2 0.042 0.0031 0.045 0.0 10.4 0.8 11.2Established floodplain 3 0.7 3.7 0.015 0.0036 0.019 0.0 3.7 0.9 4.6Transitional fluvialterrace
1 1.5� 2.5 0.0024 0.0034� 0.0058 0.0 0.6 0.8� 1.4
Mature fluvial terrace§ 36.3 2.3� 38.6 0.29 0.018� 0.31 71.8 0.0 4.5� 76.2
Notes: The relative importance of each source is given as the percentage of the total mean annual lateral input of large logs frombank erosion. Estimates for the standing stock of new logs are from Van Pelt et al. (2006), and surface remnant logs are fromLatterell (2005).
� Measured as m/yr along valley axis.� Values have been reduced by the estimated quantity of logs reaching the end of their functional lifetime prior to reentering the
channel or to being buried, based on simulations for conifer logs.§ Downed trees from stand mortality could not be differentiated from those deposited by the river.
June 2007 1133FLOODPLAIN RIVER WOOD DYNAMICS
assumptions warranting validation and therefore repre-
sent a first attempt at quantification.
Lateral source areas and input mechanisms
Bank erosion associated with channel meandering was
the primary input mechanism for new large logs,
contributing 82% of the total, whereas erosion from
cutoffs and avulsions delivered 14%, and only 4% came
from island dissection and indeterminate erosion com-
bined. Meanders just beginning to form at the earliest
time of observation with cutbanks fringed by mature
forests often functioned as ‘‘wood hotspots’’; sites
contributing disproportionately high numbers of new
logs to the channel. Specifically, these were sections with
input rates at or above the 90th percentile (i.e., �0.7 newlarge logs�100 m�1�yr�1). Wood hotspots represented
10% of the valley length but contributed 38% of the new
large logs (Fig. 4). They were distributed up to 8.4 km
apart (along the valley axis) though one-half were
separated by �1.6 km (median). Lateral source areas
extended nearly 0.5 km from the initial (1939) channel
margin by 2002 (Fig. 6). One-half of all new large logs
recruited by bank erosion from 1939 to 2002 originated
�92 m from the 1939 channel margin, though 95% were
from �265 m.
Spatial variation in new large log input rates (Fig. 4)
was weakly related to the proximity of mature fluvial
terraces. Relationships between large log delivery to the
mainstem and the mean annual lateral movement rate of
the low-flow channel were not detected (P . 0.05),
though channel movement rates were greater at hotspots
than in the remainder of the valley (Mann-Whitney U
test; P¼ 0.01). In unconfined sections (n¼ 214), mature
fluvial terraces were closer to the channel (median ¼ 32
m) at hotspots than other sections (median ¼ 47 m)
(Mann-Whitney U test; P , 0.01). Regression analyses
indicated that input rates declined in unconfined reaches
as source areas became more distant (P¼ 0.02), but this
factor explained little of the variation (r2 ¼ 0.05). The
relationship is given by the following model:
y ¼ �1:12 3 10�3d þ 0:40 ð2Þ
where y represents the mean number of new large logs
delivered to the channel (logs�100 m�1�yr�1) and d is the
distance (m) between the center of the low flow channel
and the margin of the nearest mature fluvial terrace in
each section. Confidence intervals (95%) for the model
coefficient ranged from �1.8 3 10�3 to �4.3 3 10�4.
Dynamics of large logs
Key pieces (n ¼ 273) averaged 109 6 3.1 cm dbh
(mean 6 SE, range 28–243 cm) and 26.0 60.9 m long
(range 2–65 m). Jams averaged 10.6 6 0.6 m (range 1–67
m) across the center of mass upstream from the key
piece, containing 16 6 1.1 pieces (range 1–129 pieces) of
large woody debris �2 m long and 10 cm diameter. As in
previous studies (Abbe and Montgomery 1996), nearly
all key pieces (97%) had attached rootwads. Most (75%)
were relatively undecayed, with hard, intact sapwood
and 82% were Sitka spruce, western hemlock, or
Douglas-fir. Red alder (9%), bigleaf maple (5%), or
black cottonwood (4%) sometimes functioned as key
pieces, as well.
A negative power function (r2 ¼ 0.60) quantified
retention of 14 cohorts of large logs spanning 1939–2004
(Fig. 7):
FIG. 4. Spatial and temporal variation in annual input rates (mean 6 SE) of new large logs (�1 m dbh) from the erosion ofmature fluvial terraces in the mainstem of the Queets River, Washington, USA, from 1939 to 2002 (SE is across years), fromdownstream (left) to upstream (right). The right-hand panel illustrates the percentage (y-axis) of sections of river channel (200 m)with mean input rates equal to or lower than the corresponding value on the x-axis. Values on the line represent input rates at 10%increments, to simplify interpretation.
JOSHUA J. LATTERELL AND ROBERT J. NAIMAN1134 Ecological ApplicationsVol. 17, No. 4
y ¼ 75:4t�0:25 ð3Þ
where y is the percentage of large logs in a cohort
remaining stable t years after initially being deposited in
the channel. Approximately one-half remained stable for
at least five years, 42% for �10 yr, and 27% for �60 yr;
and those remaining stable were incorporated into
emerging forests. The river moved 81 of the key pieces
tagged in 2002–2004, and we relocated 28 (35%). The
rest were either transported outside the Park, hidden in
logjams or buried. The stability of tagged key pieces
could not be predicted solely from the size of the log or
jam (P � 0.05; logistic regression) because mean
dimensions of stable and mobilized pieces were indis-
tinguishable, including dbh (P¼ 0.91; t test), length (P¼0.66; t test), and volume (P ¼ 0.92; Mann-Whitney U
test). Jams formed by stable and mobilized logs were
also similar in size (P ¼ 0.46; Mann-Whitney U test).
Spiraling simulations suggested that conifers are likely
to be transported further and deposited in more
locations than hardwoods, which decay faster, all else
being equal (Fig. 8). The median functional lifetime
(qLL) was estimated to be 60 yr for conifers and 25 yr for
hardwoods. There was a 50% probability that coniferous
logs would spiral (pLL) �3 times (up to 16 times) before
disintegrating, and have turnover lengths (nLL) �1520m. In comparison, 50% of the hardwoods were predicted
to spiral � 2 times (up to 11 times) and be transported
�916 m, reflecting shorter qLL (Fig. 8). The interquartile
ranges of nLL were 334–5094 m for conifers and 206–
3291 m for hardwoods. Conifers had a 90% chance of
traveling �76 m and 10% chance of traveling �11 959 m
(Fig. 8). In comparison, hardwoods had a 90% chance of
traveling �49 m and a 10% chance of traveling �9746m. After their initial deposition, aLL is expected to
exceed the median qLL for 27% of conifers and 33% of
hardwoods, meaning they will disintegrate before the
channel returns unless buried or otherwise preserved.
DISCUSSION
Mature forests and longitudinal and lateral connec-
tions are vital factors shaping the abundance and
distribution of large logs in the Queets River. Instream
supplies of large logs rely on inputs from the undercut-
ting of adjacent and upstream mature forests but also
logs displaced from upstream reaches or recaptured
FIG. 6. (A) Lateral recruitment distances of new large logs(�1 m dbh) delivered to the mainstem of the Queets River,Washington, USA, through erosion of mature fluvial terracesover 63 years (i.e., 1939–2002). (B) The aerial photo depicts theQueets River in 1939 to illustrate these distances. Matureforests within 265 m (solid line) of the channel margincontributed 95% of the new large logs entering the channelover the following 63 years. Half of the new logs originatedfrom forests �92 m (dotted line) from the channel margin, as itexisted in 1939.
FIG. 5. Burial rate of remnant large logs deposited in thechannel of the Queets River, Washington, USA, over 70 years.The relationship indicates the percentage of the remnant(stable) logs in a cohort (a group of large logs) that are buriedt years after being deposited by the river. The average numberof large logs in a cohort for this analysis was 28, ranging from 6to 66.
June 2007 1135FLOODPLAIN RIVER WOOD DYNAMICS
through the erosion of young floodplain forests by the
river. Our results support the hypothesis that large log
inputs are patchy in space and time due, in part, to local
variation in channel movement patterns and the
proximity of mature forests. Large log abundance and
distribution fluctuates over time at individual reaches,
and these changes may not be synchronized. We
hypothesize that these fluctuations enhance riverine
biocomplexity and system integrity, and stabilize the
availability of wood-formed habitats in the river valley
by promoting a state of dynamic equilibrium over
decades to centuries.
The importance of mature forests
Mature forests are nonsubstitutable sources of new
large logs, and their variable proximity to channels
contributes to local heterogeneity in input rates, as
expected. Our findings demonstrate that patches of
mature forests currently hundreds of meters from the
channel can eventually become important sources of
large logs because the river can potentially erode them in
the future. For example, input rates vary widely
throughout the valley (Fig. 4), but the Queets River
recruits 95% of the large logs from a riparian corridor
extending �265 m laterally on both banks within only 63
years (Fig. 6). Stands .400 m from the channel
contribute little over six decades, but it would be
erroneous to conclude outlying areas are unimportant
over the long term. The lateral extent of mature forests
needed for a self-sustaining system is probably greater
than we observed over 63 years because, in the Queets
River, stands containing large-diameter trees are often
centuries old (Van Pelt et al. 2006).
The importance of connectivity
Lateral and longitudinal connectivity supports the
recruitment, transport, storage, and recapture of large
logs (the spiraling process) in the Queets River (see Plate
1). Widespread channel meandering delivers most new
large logs but episodic cutoffs and avulsions are locally
important. We contribute empirical evidence that a
temperate floodplain river often rearranges and trans-
ports large logs downstream, including those previously
underpinning logjams (even those exceeding 2.4 m dbh
and 60 m in length). In the Queets River, large logs often
reside in several locations hundreds of meters apart
before disintegrating, linking the structure and distur-
bance history of widely separated reaches (Fig. 8). This
finding concurs with both Marcus et al. (2002) and van
der Nat et al. (2003a), who observed high rates of wood
turnover, though total quantities remained stable.
Whereas stable logs promote spatial complexity in the
river landscape, we propose that the periodic rearrange-
ment of large logs contributes to system resilience by
driving temporal fluctuations in habitat availability that
are asynchronous among stream reaches. The river often
moves some of the large logs in the first few years after
FIG. 8. (A) Cumulative probability of turnover lengths(nLL, note log scale) for large logs in the mainstem of the QueetsRiver, Washington, USA. The two lines represent predictionsfor logs with decomposition rates typical of conifers (bold line)and of hardwoods (thin line). (B) The percentage of large logsthat would exhibit turnover lengths meeting or exceeding thedistance between the ‘‘input site’’ and the approximate locationin the river indicated by each arrow. For example, 10% of thelogs from the input site are predicted to be transported (or‘‘spiral’’) at least as far as the last arrow on the left within theirfunctional lifetime.
FIG. 7. Relationship between large log retention (percent-age remaining stable) and the years elapsed (t) since the cohortwas deposited in the active channel in the mainstem of theQueets River, Washington, USA. Each point represents adifferent cohort (n¼ 14). The average number of large logs in acohort was 108, ranging from 12 to 361 logs.
JOSHUA J. LATTERELL AND ROBERT J. NAIMAN1136 Ecological ApplicationsVol. 17, No. 4
deposition. The remaining fraction persist as remnants
in dense stands of pioneering vegetation (Fig. 6, Gurnellet al. 2000a, b). Some logs remain stable for decades ormore. The ecological importance of these stable logs is
well established (see Abbe and Montgomery 1996,2003); a small but important fraction is buried andremains for centuries, protecting associated vegetation
from erosion (Montgomery and Abbe 2006). However,many logs are eventually recaptured by ongoing lateral
channel movements, so that instream supplies consist ofa mixture of fresh, new logs as well as remnants from thedistant past (e.g., Hyatt and Naiman 2001). This
transient storage of logs in the floodplain likely ensuresthat jams are continually present in the active channel,
even during periods with low inputs of new logs.Similarly, the periodic rearrangement of large logs
may stabilize the availability of important habitats at
large spatial scales. Both aquatic and riparian habitatsare highly dynamic in floodplain rivers (Piegay et al.1999, Arscott et al. 2002, van der Nat et al. 2003b,
Gurnell et al. 2005, Latterell et al. 2006). Temporalheterogeneity in the distribution of pools (Naiman and
Latterell 2005) and pioneer bars (Latterell et al. 2006) inthe Queets River may be attributable, in part, tofluctuations in the abundance and distribution of large
logs. These habitats fluctuate over time, but the patternsof change are asynchronous among reaches, dampening
the degree of change at the valley scale. In this way, the
periodic rearrangement of large logs may contribute to
the resilience of riverine communities that utilize these
habitats for parts of their life cycle.
Our spiraling analyses suggest that floodplain storage
increases the probability that some pieces move only
short distances because retention is not solely restricted
by the longevity of instream jams. For example, Martin
and Benda (2001) estimate that smaller pieces of wood in
a 20–30 m wide low-gradient Alaskan river are
transported between existing jams and have a 90%
probability of traveling �300 m. In contrast, large logs
in the much wider Queets River have a 90% chance of
traveling only �76 m because they often come to rest
between existing jams and may enter floodplain stock-
piles. However, the river is simultaneously capable of
moving large logs substantial distances. Large logs in the
Queets River have a 10% chance of traveling �11 959 m
(Fig. 8), whereas wood in the Alaskan stream has the
same chance of traveling only �2500 m. This finding is
consistent with the notion that potential transport
distances are relatively high in rivers where channel
width exceeds the height of the tallest riparian trees (see
Gurnell 2003).
Recommended improvements
Most field studies of wood movement, including ours,
are limited by an inability to recover a large percentage
PLATE 1. Large log in the Queets River, Olympic National Park, Washington, USA. This log, shown in its original location inAppendix A, was displaced and transported by the river during the study. Photo credit: J. J. Latterell.
June 2007 1137FLOODPLAIN RIVER WOOD DYNAMICS
of the tagged pieces (e.g., Lienkaemper and Swanson
1987, Jacobsen et al. 1999). This is inevitable because
mobile logs in large rivers cannot be trapped as they exit
the reach or system of interest (e.g., like leaves, Speaker
et al. 1984). We suspect that some were lost to the ocean.
If so, JLL and nLL (Fig. 3) are actually higher than
reported and therefore wood retention is overestimated.
It seems more reasonable that most were hidden instead
in side channels and among abundant logjams (i.e.,
.600 logjams in the portion of the river that was
searched). Regardless, our simulations account for the
possibility that the river moves some large logs further
than we observed in field trials. Additional determina-
tions of how nLL and pLL vary among rivers and regions,
and whether these differences reflect system retentive-
ness and the decay resistance of tree species, will tell
much about the generic capacity of rivers to retain and
process wood.
Field studies are required to better understand the
actual transport dynamics of wood in rivers (e.g., Young
1994, Jacobsen et al. 1999, van der Nat et al. 2003a) and
to put our observations from the Queets River into a
broader context. Studies of larger rivers will be
particularly valuable because wood dynamics in small
streams are increasingly well known (e.g., Lienkaemper
and Swanson 1987, Murphy and Koski 1989, Berg et al.
1998, Martin and Benda 2001, May and Gresswell
2003). The physical basis for log stability in rivers (and
flumes) has been clearly articulated in previous studies in
the field and in experimental flumes (Abbe 2000,
Braudrick and Grant 2001, Abbe and Montgomery
2003, Wallerstein and Thorne 2004). Predicting the long-
term stability of logs in a dynamic river environment is
particularly challenging because so many covariates may
be involved, and few hold steady over time or among
locations (Appendix A).
The accuracy of the spiraling simulations in this study
would be greatly improved by a better understanding of
how large logs decay in floodplains and beneath alluvium.
The true functional lifetime of large logs is currently
unknown but is likely quite variable, depending on key
factors influencing the decay rate (i.e., wood chemistry or
substrate quality, temperature, moisture, and aeration;
Melillo et al. 1984, Laiho and Prescott 2004), as well as the
initial size of the log relative to local channel dimensions
(e.g., Abbe 2000), and colonization by wood-boring
invertebrates (J. Latterell, unpublished data). We speculate
that these factors vary substantially as logs are subjected
to changing conditions throughout vegetative succession,
recapture, or eventual burial (see Appendix A). In this
study, we use decay coefficients that fell within the upper
and lower bounds for Picea (0.023–0.028), Tsuga (0.018–
0.026), and Alnus (0.055–0.083) across experimental
forests throughout the western USA (Harmon and Fasth
2005), though none were from riparian or instream
environments. Nonetheless, we view the 75% decay
threshold as a reasonable first approximation.
Implications for river conservation and restoration
How can natural, self-sustaining processes supplyingand retaining wood in floodplain river systems in
temperate forests be restored in densely populated rivervalleys? Solutions must provide sustained delivery of
large logs, variable flows (including some floods), andsufficient room for some of the riverbed to be
abandoned long enough to grow mature forests, whichrequires considerable time. Existing mature riparian
forests warrant protection (especially in developedrivers) because they have high ecological value (Franklin
et al. 1981) and take many years to produce (e.g.,generally .100–300þ years in the Queets River valley;
Balian and Naiman 2005, Van Pelt et al. 2006). Wherelarge logs are lacking, Collins and Montgomery (2002)
proposed a ‘‘restoration succession’’ strategy, couplingartificial logjams (e.g., Brooks et al. 2004) and levee
setbacks or removals with riparian reforestation usingboth fast- and slow-growing (but durable) trees toensure wood supplies would become self-sustaining. In
regulated rivers, the provision of flood flows isimportant because high flow events play an important
role in wood recruitment and redistribution (Keller andSwanson 1979, Piegay and Bravard 1997, Marcus et al.
2002, Piegay 2003, van der Nat et al. 2003a, Pettit andNaiman 2005). Managing the channel and floodplain
forest as a unified system is also advisable (Junk et al.1989, Mac Nally et al. 2002, Angradi et al. 2004) because
when they are connected, the river may tap andreplenish stockpiles of remnant logs in the floodplain
(Piegay et al. 1999, Gurnell et al. 2001).Wood hotspots (Fig. 4) warrant special consideration
in land use planning because they supply a dispropor-tionate number of large logs to the river. Ecological
theory suggests that the loss of organisms withdisproportionately large ecological effects relative to
their abundances results in substantial, sometimesdetrimental changes to ecosystem structure or organi-
zation (Power et al. 1996). We suspect that restrictingthe river’s access to wood hotspots may have similarconsequences. In the Queets River, hotspots tend to
occur in unconfined alluvial valley segments (Montgom-ery and Buffington 1997) with mature forests in close
proximity to laterally migrating channels. We cautionthat the spatial distribution of individual hotspots
within these segments will likely shift over decades tocenturies. If so, establishing an interconnected system of
large, strategically located (see Roni et al. 2002, Berg etal. 2003, Lorang et al. 2005) riparian reserves spanning
entire stream segments would be preferable to restrictingriparian protections to short reaches presently function-
ing as hotspots.We acknowledge that significant challenges remain.
Floodplain reclamation (from development) and pres-ervation are prohibitively expensive and significant
technical, social, and political obstacles exist. Surmount-ing these challenges demands a deeper ecological
understanding of floodplain rivers but also unprece-
JOSHUA J. LATTERELL AND ROBERT J. NAIMAN1138 Ecological ApplicationsVol. 17, No. 4
dented innovation in alleviating human impacts in
developed systems. In our view, successful conservation
efforts will be guided, in part, by the principle that our
actions must evolve over time, addressing present
conditions while planning for the anticipated future,
and by embracing change as a fundamental attribute of
temperate floodplain rivers (Naiman and Latterell 2005).
ACKNOWLEDGMENTS
We thank L. Benda, R. E. Bilby, P. A. Bisson, and D.Peterson for their comments on versions of this manuscript, andL. Conquest for statistical advice. Comments from twoanonymous reviewers improved the manuscript and aregratefully acknowledged. Research support from the PacificNorthwest Research Station of the USDA Forest Service,Weyerhaeuser Company, the Andrew W. Mellon Foundation,and the National Science Foundation is gratefully acknowl-edged. Field assistance was provided by E. Meehan, N.Hurtado, A. Dotolo, J. T. Jackson, and M. G. Logsdon. Fieldresearch was conducted from 2002 to 2005 within OlympicNational Park under research permits OLYMP-0047 andOLYM-0096. We thank the U.S. Department of the InteriorNational Park Service, especially S. Acker, B. Bacchus, J.Freilich, C. Hoffman, and R. Hoffman at Olympic NationalPark. We also thank J. O’Connor and L. Fueste of the U.S.Geological Survey for sharing geospatial data and historicalairphotos, respectively.
LITERATURE CITED
Abbe, T. B. 2000. Patterns, mechanics, and geomorphic effectsof wood accumulations in a forest river system. Dissertation.University of Washington, Seattle, Washington, USA.
Abbe, T. B., and D. R. Montgomery. 1996. Large woody debrisjams, channel hydraulics, and habitat formation in largerivers. Regulated Rivers: Research and Management 12:201–221.
Abbe, T. B., and D. R. Montgomery. 2003. Patterns andprocesses of wood debris accumulation in the Queets riverbasin, Washington. Geomorphology 51:81–107.
Agee, J. K. 1993. Fire ecology of Pacific Northwest forests.Island Press, Washington, D.C., USA.
Angradi, T. R., E. W. Schweiger, D. W. Bolgrien, P. Ismert,and T. Selle. 2004. Bank stabilization, riparian land use andthe distribution of large woody debris in a regulated reach ofthe upper Missouri River, North Dakota, USA. RiverResearch and Applications 20:829–846.
Arscott, D. B., K. Tockner, D. van der Nat, and J. V. Ward.2002. Aquatic habitat dynamics along a braided alpine riverecosystem (Tagliamento River, Northeast Italy). Ecosystems5:802–814.
Aumen, N. G., C. P. Hawkins, and K. J. Gregory. 1990.Influence of woody debris on nutrient retention in cata-strophically disturbed streams. Hydrobiologia 190:183–192.
Balian, E. V., and R. J. Naiman. 2005. Abundance andproduction of riparian trees in the lowland floodplain of theQueets River, Washington. Ecosystems 8:841–861.
Beechie, T., and T. H. Sibley. 1997. Relationships betweenchannel characteristics, woody debris, and fish habitat innorthwestern Washington streams. Transactions of theAmerican Fisheries Society 126:217–229.
Benda, L. E., N. L. Poff, D. Miller, T. Dunne, G. H. Reeves,G. R. Pess, and M. M. Pollack. 2004. The network dynamicshypothesis: how channel networks structure riverine habitats.BioScience 54:413–427.
Benda, L. E., and J. C. Sias. 2003. A quantitative frameworkfor evaluating the mass balance of in-stream organic debris.Forest Ecology and Management 172:1–16.
Berg, N., A. Carlson, and D. Azuma. 1998. Function anddynamics of woody debris in stream reaches in the central
Sierra Nevada, California. Canadian Journal of Fisheries andAquatic Sciences 55:1807–1820.
Berg, D. R., A. McKee, and M. J. Maki. 2003. Restoringfloodplain forests. Pages 248–291 in D. R. Montgomery, S.Bolton, D. B. Booth, and L. Wall, editors. Restoration ofPuget Sound Rivers. University of Washington Press, Seattle,Washington, USA.
Bilby, R. E. 1981. Role of organic debris dams in regulating theexport of dissolved and particulate matter from a forestedwatershed. Ecology 62:1234–1243.
Bilby, R. E., and P. A. Bisson. 1998. Function and distributionof large woody debris. Pages 324–338 in R. J. Naiman andR. E. Bilby, editors. River ecology and management: lessonsfrom the Pacific Coastal Ecoregion. Springer-Verlag, NewYork, New York, USA.
Bilby, R. E., J. T. Heffner, B. R. Fransen, and J. W. Ward.1999. Effects of immersion in water on deterioration of woodfrom five species of trees used for habitat enhancementprojects. North American Journal of Fisheries Management19:687–695.
Bilby, R. E., and G. E. Likens. 1980. Importance of organicdebris dams in the structure and function of streamecosystems. Ecology 61:1107–1113.
Braudrick, C. A., and G. E. Grant. 2001. Transport anddeposition of large woody debris in streams: a flumeexperiment. Geomorphology 41:263–283.
Brooks, A. P., G. J. Brierley, and R. G. Millar. 2003. The long-term control of vegetation and woody debris on channel andflood-plain evolution: insights from a paired catchment studyin southeastern Australia. Geomorphology 51:7–29.
Brooks, A. P., P. C. Gehrke, J. D. Jansen, and T. B. Abbe.2004. Experimental reintroduction of woody debris on theWilliams River, NSW: geomorphic and ecological responses.River Research and Applications 20:513–536.
Brummer, C. J., T. B. Abbe, J. R. Sampson, and D. R.Montgomery. 2006. Influence of vertical channel changeassociated with wood accumulations on delineating channelmigration zones, Washington, USA. Geomorphology 80:295–309.
Bull, E. L. 2002. The value of coarse woody debris tovertebrates in the Pacific Northwest. Pages 171–178 in P. J.Shea, W. F. Laudenslayer, B. Valentine, C. P. Weatherspoon,and T. E. Lisle, editors. Symposium on the Ecology andManagement of Dead Wood in Western Forests. USDAForest Service, Pacific Southwest Research Station, Albany,California, USA.
Collins, B. D., and D. R. Montgomery. 2002. Forestdevelopment, wood jams, and restoration of floodplain riversin the Puget lowland, Washington. Restoration Ecology 10:237–247.
Collins, B. D., D. R. Montgomery, and A. D. Haas. 2002.Historical changes in the distribution and functions of largewood in Puget Lowland rivers. Canadian Journal of Fisheriesand Aquatic Sciences 59:66–76.
Collins, B. D., D. R. Montgomery, and A. J. Sheikh. 2003.Reconstructing the historical riverine landscape of the Pugetlowland. Pages 79–128 in D. R. Montgomery, S. Bolton,D. B. Booth, and L. Wall, editors. Restoration of PugetSound Rivers. University of Washington Press, Seattle,Washington, USA.
Edmonds, R. L., and A. Eglitis. 1989. The role of Douglas-firbeetle and wood borers in the decomposition of and nutrientrelease from Douglas-fir logs. Canadian Journal of ForestResearch 19:853–859.
Fetherston, K. L. 2005. Pattern and process in mountain rivervalley forests. Dissertation. University of Washington,Seattle, Washington, USA.
Fetherston, K. L., R. J. Naiman, and R. E. Bilby. 1995. Largewoody debris, physical process, and riparian forest develop-ment in montane river networks of the Pacific Northwest.Geomorphology 13:133–144.
June 2007 1139FLOODPLAIN RIVER WOOD DYNAMICS
Fisher, S. G., N. B. Grimm, E. Marti, R. M. Holmes, and J. B.Jones, Jr. 1998. Material spiraling in stream corridors: atelescoping ecosystem model. Ecosystems 1:19–34.
Franklin, J. F., K. Cromack, Jr., W. Denison, A. McKee, C.Maser, J. Sedell, F. Swanson, and G. Juday. 1981. Ecologicalcharacteristics of old-growth Douglas-fir forests. USDAForest Service General Technical Report PNW 118. PacificNorthwest Forest and Range Experiment Station.
Franklin, J. F., and C. T. Dyrness. 1988. Natural vegetation ofOregon and Washington. Oregon State University, Corvallis,Oregon, USA.
Graham, R. L., and K. Cromack, Jr. 1982. Mass, nutrientcontent, and decay rate of dead boles in rain forests ofOlympic National Park. Canadian Journal of ForestResearch 12:511–521.
Greenwald, D. N., and L. B. Brubaker. 2001. A 5000-yearrecord of disturbance and vegetation change in riparianforests of the Queets River, Washington, U.S.A. CanadianJournal of Forest Research 31:1375–1385.
Gregory, S., K. Boyer, and A. Gurnell, editors. 2003. Theecology and management of wood in world rivers. Sympo-sium 37. American Fisheries Society, Bethesda, Maryland,USA.
Gurnell, A. 2003. Wood storage and mobility. Pages 75–92 in S.Gregory, K. Boyer, and A. Gurnell, editors. The ecology andmanagement of wood in world rivers. Symposium 37.American Fisheries Society, Bethesda, Maryland, USA.
Gurnell, A. M., G. E. Petts, D. M. Hannah, P. J. Edwards, J.Kollmann, J. V. Ward, and K. Tockner. 2000b. Woodstorage within the active zone of a large European gravel-bedriver. Geomorphology 34:55–72.
Gurnell, A., G. E. Petts, D. M. Hannah, B. P. G. Smith, P. J.Edwards, J. Kollman, J. V. Ward, and K. Tockner. 2001.Riparian vegetation and island formation along the gravel-bed Fiume Tagliamento, Italy. Earth Surface Processes andLandforms 26:31–62.
Gurnell, A., G. E. Petts, N. Harris, J. V. Ward, K. Tockner,P. J. Edwards, and J. Kollman. 2000a. Large wood retentionin river channels: the case of the Fiume Tagliamento, Italy.Earth Surface Processes and Landforms 25:255–275.
Gurnell, A., K. Tockner, P. Edwards, and G. Petts. 2005.Effects of deposited wood on biocomplexity of rivercorridors. Frontiers in Ecology and the Environment 3:377–382.
Guyette, R. P., W. G. Cole, D. C. Dey, and R.-M. Muzika.2002. Perspectives on the age and distribution of large woodin riparian carbon pools. Canadian Journal of Fisheries andAquatic Sciences 59:578–585.
Harmon, M. E., and B. Fasth. 2005. Bole decomposition ratesof seventeen tree species in the western U.S.A. H.G. AndrewsExperimental Forest Report. hhttp://www.fsl.orst.edu/lter/pubs/webdocs/reports/decomp/cwd_decomp_web.htmi
Hughes, F. M. R., editor. 2003. The flooded forest: guidance forpolicy makers and river managers in Europe on therestoration of floodplain forests. FLOBAR2 (FloodplainBiodiversity and Restoration), Department of Geography,University of Cambridge, Cambridge, UK.
Hyatt, T. L., and R. J. Naiman. 2001. The residence time oflarge woody debris in the Queets River, Washington, USA.Ecological Applications 11:191–202.
Jacobsen, P. J., K. M. Jacobsen, P. L. Angermeier, and D. S.Cherry. 1999. Transport, retention, and ecological signifi-cance of woody debris within a large ephemeral river. Journalof the North American Benthological Society 18:429–444.
Junk, W. J., P. B. Bayley, and R. E. Sparks. 1989. The floodpulse concept in river-floodplain systems. Pages 110–127 inD. P. Dodge, editor. Proceedings of the International LargeRiver Symposium (LARS). Canadian Special Publication ofFisheries and Aquatic Sciences 106, Department of Fisheriesand Oceans, Ottawa, Ontario, Canada.
Keller, E. A., and F. J. Swanson. 1979. Effects of large organicmaterial on channel form and fluvial processes. Earth SurfaceProcesses 4:361–380.
Kellerhals, R., M. Church, and D. I. Bray. 1976. Classificationand analysis of river processes. Proceedings of the AmericanSociety of Civil Engineers, Journal of the HydraulicsDivision 102:813–829.
Knighton, D. 1998. Fluvial forms and processes: a newperspective. Oxford University Press, New York, New York,USA.
Laiho, R., and C. E. Prescott. 2004. Decay and nutrientdynamics of coarse woody debris in northern coniferousforests: a synthesis. Canadian Journal of Forest Research 34:763–777.
Latterell, J. J. 2005. The natural history and dynamics of largewood in the Queets River, Washington. Dissertation.University of Washington, Seattle, Washington, USA.
Latterell, J. J., J. S. Bechtold, T. C. O’Keefe, R. Van Pelt, andR. J. Naiman. 2006. Dynamic patch mosaics and channelmovement in an unconfined river valley of the OlympicMountains. Freshwater Biology 51:523–544.
Lienkaemper, G. W., and F. J. Swanson. 1987. Dynamics oflarge woody debris in streams in old-growth Douglas-firforests. Canadian Journal of Forest Research 17:150–156.
Lorang, M. S., D. C. Whited, F. R. Hauer, J. S. Kimball, andJ. A. Stanford. 2005. Using airborne multispectral imagery toevaluate geomorphic work across floodplains of gravel-bedrivers. Ecological Applications 15:1209–1222.
Mac Nally, R., A. Parkinson, G. Horrocks, and M. Young.2002. Current loads of coarse woody debris on southeasternAustralian floodplains: evaluation of change and implica-tions for restoration. Restoration Ecology 10:627–635.
Marcus, W. A., R. A. Marston, C. R. Colvard, and R. D. Gray.2002. Mapping the spatial and temporal distributions ofwoody debris in streams of the Greater YellowstoneEcosystem, USA. Geomorphology 44:323–325.
Martin, D. J., and L. E. Benda. 2001. Patterns of instreamwood recruitment and transport at the watershed scale.Transactions of the American Fisheries Society 130:940–958.
Maser, C., R. F. Tarrant, J. M. Trappe, and J. F. Franklin.1988. From forest to sea: a story of fallen trees. USDA ForestService General Technical Report PNW-GTR-229. PacificNorthwest Research Station, Portland, Oregon, USA.
May, C. L., and R. E. Gresswell. 2003. Large wood recruitmentand redistribution in headwater streams in the southernOregon Coast Range, U.S.A. Canadian Journal of ForestResearch 33:1352–1362.
McDade, M. H., F. J. Swanson, W. A. McKee, J. F. Franklin,and J. Van Sickle. 1990. Source distances for coarse woodydebris entering small streams in western Oregon andWashington. Canadian Journal of Forest Research 20:326–330.
Melillo, J. M., R. J. Naiman, J. D. Aber, and A. E. Linkins.1984. Factors controlling mass loss and nitrogen dynamics ofplant litter decaying in northern streams. Bulletin of MarineScience 35:341–356.
Montgomery, D. R., and T. Abbe. 2006. Influence of logjam-formed hard points on the formation of valley-bottomlandforms in an old-growth forest valley, Queets River,Washington, USA. Quaternary Research 65:147–155.
Montgomery, D. R., and J. M. Buffington. 1997. Channel reachmorphology in mountain drainage basins. Geological Societyof America Bulletin 109:596–611.
Murphy, M. L., and K. V. Koski. 1989. Input and depletion ofwoody debris in Alaska streams and implications forstreamside management. North American Journal of Fisher-ies Management 9:427–436.
Naiman, R. J., and E. C. Anderson. 1997. Streams and rivers:their physical and biological variability. Pages 131–148 inP. K. Schoonmaker, B. von Hagen, and E. C. Wolf, editors.
JOSHUA J. LATTERELL AND ROBERT J. NAIMAN1140 Ecological ApplicationsVol. 17, No. 4
The rain forests of home: profile of a North Americanbioregion. Island Press, Washington, D.C., USA.
Naiman, R. J., J. S. Bechtold, D. Drake, J. J. Latterell, T. C.O’Keefe, and E. V. Balian. 2005a. Origins, patterns, andimportance of heterogeneity in riparian systems. Pages 279–309 in G. Lovett, C. G. Jones, M. G. Turner, and K. C.Weathers, editors. Ecosystem function in heterogeneouslandscapes. Springer-Verlag, New York, New York, USA.
Naiman, R. J., H. Decamps, and M. E. McClain. 2005b.Riparia: ecology, conservation, and management of stream-side communities. Elsevier/Academic Press, San Diego,California, USA.
Naiman, R. J., and J. J. Latterell. 2005. Principles for linkingfish habitat to fisheries management and conservation.Journal of Fish Biology 67:166–185.
Nanson, G. C., M. Barbetti, and G. Taylor. 1995. Riverstabilisation due to changing climate and vegetation duringthe Late Quaternary in western Tasmania, Australia.Geomorphology 13:145–158.
Newbold, J. D. 1992. Cycles and spirals of nutrients. Pages379–408 in P. Calow and G. E. Petts, editors. The rivershandbook: hydrological and ecological principles. BlackwellScientific, Boston, Massachusetts, USA.
Newbold, J. D., P. J. Mulholland, J. W. Elwood, and R. V.O’Neill. 1982. Organic carbon spiralling in stream ecosys-tems. Oikos 38:266–272.
Nilsson, C., and K. Berggren. 2000. Alterations of riparianecosystems caused by river regulation. BioScience 50:783–792.
Nilsson, C., C. A. Reidy, M. Dynesius, and C. Revenga. 2005.Fragmentation and flow regulation of the world’s large riversystems. Science 308:405–408.
O’Connor, J. E., M. A. Jones, and T. L. Haluska. 2003.Floodplain and channel dynamics of the Quinault and QueetsRivers, Washington, USA. Geomorphology 51:31–59.
Pettit, N. E., and R. J. Naiman. 2005. Flood deposited wooddebris and its contribution to heterogeneity and regenerationin a semi-arid riparian landscape. Oecologia 145:434–444.
Piegay, H. 2003. Dynamics of wood in large rivers. Pages 109–134 in S. Gregory, K. Boyer, and A. Gurnell, editors. Theecology and management of wood in world rivers. Sympo-sium 37. American Fisheries Society, Bethesda, Maryland,USA.
Piegay, H., and J.-P. Bravard. 1997. Response of a Mediter-ranean riparian forest to a 1 in 400 year flood, Ouveze River,Drome-Vaucluse, France. Earth Surface Processes andLandforms 22:31–43.
Piegay, H., A. Thenevet, and A. Citterio. 1999. Input, storage,and distribution of large woody debris along a mountainriver continuum, the Drome River, France. Catena 35:19–39.
Poff, N. L., J. D. Allan, M. B. Bain, J. R. Karr, K. L.Prestegaard, B. D. Richter, R. E. Sparks, and J. C.Stromberg. 1997. The natural flow regime: a paradigm forriver conservation and restoration. BioScience 47:769–784.
Richter, B. D., and H. E. Richter. 2000. Prescribing floodregimes to sustain riparian ecosystems along meanderingrivers. Conservation Biology 14:1467–1478.
Roni, P., T. J. Beechie, R. E. Bilby, F. E. Leonetti, M. M.Pollock, and G. R. Pess. 2002. A review of stream restoration
techniques and a hierarchical strategy for prioritizingrestoration in Pacific Northwest watersheds. North AmericanJournal of Fisheries Management 22:1–20.
Sedell, J. R., and J. L. Froggatt. 1984. Importance of streamsideforests to large rivers: the isolation of the Willamette River,Oregon, USA, from its floodplain by snagging and stream-side forest removal. Verhandlungen der InternationalenVereinigung fur Theoretische und Angewandte Limnologie22:1828–1834.
Speaker, R., K. Moore, and S. Gregory. 1984. Analysis of theprocess of retention of organic matter in stream ecosystems.Verhandlungen der Internationalen Vereinigung fur Theor-etische und Angewandte Limnologie 22:1835–1841.
Steel, E. A., W. H. Richards, and K. A. Kelsey. 2003. Woodand wildlife: benefits of river wood to terrestrial and aquaticvertebrates. Pages 235–248 in S. V. Gregory, K. L. Boyer,and A. M. Gurnell, editors. The ecology and management ofwood in world rivers. American Fisheries Society, Bethesda,Maryland, USA.
Thackray, G. D. 2001. Extensive early and middle Wisconsinglaciation on the Western Olympic Peninsula, Washington,and the variability of Pacific moisture delivery to theNorthwestern United States. Quaternary Research 55:257–270.
Tockner, K., S. Bunn, C. Gordon, R. J. Naiman, G. P. Quinn,and J. A. Stanford. In press. Floodplains: critically threat-ened ecosystems. N. V. C. Polunin, editor. Aquatic ecosys-tems: trends and global prospects. Cambridge UniversityPress, Cambridge, UK.
U.S. Geological Survey. 2005. National Water InformationSystem (NWISWeb). hhttp://waterdata.usgs.gov/wa/nwis/uv/?site_no¼12040500i
Valett, H. M., C. L. Crenshaw, and P. F. Wagner. 2002. Streamnutrient uptake, forest succession, and biogeochemicaltheory. Ecology 83:2888–2901.
van der Nat, D., K. Tockner, and P. J. Edwards. 2003a. Largewood dynamics of complex alpine river floodplains. Journalof the North American Benthological Society 22:35–50.
van der Nat, D., K. Tockner, P. J. Edwards, J. V. Ward, and A.Gurnell. 2003b. Habitat change in braided floodplains(Tagliamento, NE - Italy). Freshwater Biology 48:1799–1812.
Van Pelt, R., T. C. O’Keefe, J. J. Latterell, and R. J. Naiman.2006. Riparian forest stand development along the QueetsRiver in Olympic National Park, Washington. EcologicalMonographs 76:277–298.
Van Sickle, J., and S. V. Gregory. 1990. Modeling inputs oflarge woody debris to streams from falling trees. CanadianJournal of Forest Research 20:1593–1601.
Vorosmarty, C., et al. 2004. Humans transforming the globalwater system. EOS, American Geophysical Union Transac-tions 85:509–514.
Wallerstein, N. P., and C. R. Thorne. 2004. Influence of largewoody debris on morphological evolution of incised, sand-bed channels. Geomorphology 27:53–73.
Young, M. K. 1994. Movement and characteristics of stream-borne coarse woody debris in adjacent burned and undis-turbed watersheds in Wyoming. Canadian Journal of ForestResearch 24:1933–1938.
APPENDIX A
Photographs illustrating the natural history of large logs in the Queets River, Washington, USA (Ecological Archives A017-040-A1).
APPENDIX B
Physical characteristics of study reaches used in quantifying log burial and large log dynamics in the Queets River, Washington,USA (Ecological Archives A017-040-A2).
June 2007 1141FLOODPLAIN RIVER WOOD DYNAMICS