Journal of Vegetation Science 11: 725-738, 2000

© IAVS; Opulus Press Uppsala. Printed in Sweden

- Biomass accumulation in coastal Oregon Picea-Tsuga forest - 725

Biomass accumulation over the first 150 years

in coastal Oregon Picea-Tsuga forest

Acker, Steven A.1*; Harcombe, Paul A.2; Harmon, Mark E.1 & Greene, Sarah E.3
1
National Park Service, 909 First Avenue, Seattle, WA 98104, USA;
2
Ecology and Evolutionary Biology Department MS-170, Rice University, P.O. Box 1982, Houston, TX 77251-1892, USA;
3
U.S. Forest Service, Pacific Northwest Research Station, 3200 S.W. Jefferson Way, Corvallis, OR 97331, USA;
*Author for correspondence: Fax +1 206 220 4160; E-mail [email protected]

Abstract. Production and mortality are the component proc- Introduction

esses that together determine the biomass dynamics of forests.
Due to the significant role of forests in the global carbon cycle, Forests play an important role in the global carbon
it is important to assess how these two processes affect the cycle, with natural dynamics and human management
maximum biomass attained by forests, as well as the dynamics
of forests largely determining whether the terrestrial
leading up to and following peak biomass. We address these
questions for two sets of plots in Picea sitchensis-Tsuga biosphere is a net source or sink for carbon (Dixon et al.
heterophylla forest on the northern Oregon coast that origi- 1994). Hence, to understand the role of forests in the
nated from a catastrophic wildfire in the 1840s, using new data global carbon cycle, it is necessary to explain the dy-
on dynamics of live trees and stocks of coarse woody debris namics of biomass accumulation and the patterns of
(CWD). The set of plots closest to the ocean and occupying change after maximum accumulation (Cooper 1983).
steeper, more dissected terrain with areas of thin soils has Temporal patterns of biomass change in forests and
lower biomass, lower net primary production (NPP) of bole their causes are also long-standing, fundamental issues
wood and higher tree mortality as a fraction of standing for basic and applied ecology at regional and local
biomass. The two sets of plots have similar CWD levels, most
scales, both for the aggrading phase leading to maximum
of which has accumulated in the last 25 yr. The present
disparity in biomass between the two sets of plots appears to biomass and for the following period (Peet 1981, 1992;
be the result of lower NPP on the low-biomass plots for the Cooper 1983; Pare & Bergeron 1995; Cohen et al.
entire 140+ yr history of the forest. Over the 58 yr that the 1996). In this paper we investigate the interaction be-
high-biomass plots have been measured (from stand age 85 to tween production and mortality in determining differ-
143 yr), NPP of bole wood has declined by 41%. Only ca. 6% ences in maximum biomass attained on two sets of plots
of this decline can be accounted for by an increase in mainte- in mature forest of Picea sitchensis and Tsuga hetero-
nance respiration of woody tissues. For both sets of plots phylla on the Oregon coast. We also reconstruct tempo-
relative constancy of biomass in the long term appears likely, ral patterns in the aggrading phase, and use observations
due to a short time lag in tree regeneration, asynchronous tree
of live tree dynamics and coarse woody debris stores to
mortality and little overall decline in NPP of bole wood in
recent decades. However, since tree mortality as a fraction of explore potential trends in the period following maximum
standing biomass is higher on the low-biomass plots, and NPP biomass. Long-term observations of tree growth and
of bole wood is slightly lower, the difference in biomass mortality, in association with inventories of coarse woody
between the two sets of plots should increase if current rates of debris, make it possible to assess the relative importance
production and mortality persist. of production and mortality in controlling biomass of
forests during and after the aggrading phase.
The amount of standing biomass of tree boles at any
Keywords: Coarse woody debris; Long-term biomass dynam- given time represents the integration of net primary
ics; Net primary production; Picea sitchensis; Tree mortality;
production of boles (NPPB) and tree mortality from the
Tree recruitment; Tsuga heterophylla.
initiation of the stand until the time in question. Thus,
tree mortality could limit maximum standing biomass.
Abbreviations: CHEF = Cascade Head Experimental Forest; In a previous paper (Greene et al. 1992) we reported
CWD = Coarse woody debris; DBH = Diameter at breast high local variation in biomass in 140 yr-old plots in
height; NCRNA = Neskowin Crest Research Natural Area; forest dominated by Picea sitchensis and Tsuga hetero-
NPPB = Net primary production of boles. phylla on the Oregon coast. One set of plots (CHEF);
(Harcombe 1986; Harcombe et al. 1990) had relatively
high biomass (750Mg/ha in 1988), whereas another
nearby set of plots (NCRNA); (Greene et al. 1992) had
726 Acker, S.A. et al.

considerably lower biomass (462 Mg/ha in 1989). Based ity and whether or not there are time lags in tree regen-
on long-term observations of live trees on the two sets of eration. Where tree mortality is asynchronous and tree
plots, a number of alternative scenarios could be devel- regeneration continuous, there is likely to be no decline
oped to explain the current difference in standing biom- from peak biomass. Time lags in tree regeneration lead
ass. A key, but missing, piece of evidence for discrimi- to oscillations in biomass, which are accentuated where
nating among alternative explanations was measure- tree mortality occurs in pulses.
ments of CWD (Greene et al. 1992). If the two sets of In this paper, we present new measurement data for
plots had had similar levels of production, and therefore live trees and data on CWD stocks and inputs over time.
had similar biomass at some time in the past, the differ- These data allow us to assess the relative importance of
ence in current biomass would be a result of higher loss production and mortality in causing the current difference
to mortality in the low-biomass plots, and those plots in standing biomass between stands. We also reconstruct
would have substantially higher CWD (Greene et al. NPPB, mortality, accumulation of biomass and CWD
1992). An alternative hypothesis is that if the difference accumulation over the life of these stands to gain further
in standing biomass was caused by long-term differ- information about the degree and timing of differences
ences in production, then CWD would not differ sub- between them. Finally, we use the data to distinguish
stantially between the plots. Thus, the alternative to among alternative patterns of biomass dynamics proposed
higher CWD on the low-biomass plots would be higher by Peet (1981) during the period following peak biomass.
NPPB on the high-biomass plots during the aggrading
phase. During the period that both sets of plots were
measured NPPB has declined (Harcombe et al. 1990; Study area
Greene et al. 1992). The alternatives are not mutually
exclusive, in that the low-biomass plots may have had The two sets of plots are located in close proximity
both higher mortality and lower NPPB in the past. on the north Oregon coast (45° 4' N, 123° 56' W; Fig. 1)
Implicit in both scenarios is the assumption that the rate in the Picea sitchensis-Tsuga heterophylla (Sitka spruce-
of CWD decomposition has been the same on the two western hemlock) zone of Franklin & Dyrness (1988).
sets of plots. This is reasonable, given the similarities in The existing forest cover developed after the Nestucca
climate and biota (Harmon et al. 1986).
Recently, the issue of decline in NPPB after a peak
relatively early in forest development (i.e. well before
the biomass peak) has come under scrutiny (Gower et al.
1996; Ryan et al. 1997). Until the renewed interest, the
most widely accepted explanation for the decline in
NPPB during stand maturation has been increasing res-
piration due to increasing amounts of non-photosyn-
thetic tissue (Kira & Shidei 1967; Sprugel 1985). Re-
cent work, however, suggests that respiration of non-
photosynthetic tissue may account for only a small
portion of the decline in NPPB during stand develop-
ment and that a variety of other causes are more impor-
tant including reduced leaf area, reduced assimilation
efficiency of canopies or changes in allocation from
above-ground wood production to below-ground pro-
duction (Gower et al. 1996; Ryan & Waring 1992; Ryan
et al. 1997).
With respect to long-term trends after the peak in
biomass accumulation, there are several possibilities.
Peet (1981, 1992) summarized empirical patterns and
created a theoretical framework that could account for
many different patterns of biomass variation over time.
The basic patterns include sigmoidal increase of biomass
to an asymptote, or sigmoidal increase to a peak, fol-
lowed by a decline with or without oscillations. Accord- Fig. 1. Location of Picea-Tsuga permanent plots in Cascade
ing to Peet (1981), the different temporal patterns are Head Experimental Forest (CHEF) and Neskowin Crest Re-
caused by different degrees of synchrony of tree mortal- search Natural Area (NCRNA) on the northern Oregon coast.
 - Biomass accumulation in coastal Oregon Picea-Tsuga forest - 727

Burn, a huge, catastrophic crown fire that occurred in Table 1. Summary of forest structure for CHEF and NCRNA
the late 1840s (Morris 1934; Munger 1944; Harcombe plot sets. n = 9 for CHEF (measured in 1993) and n = 44 for
1986). The Nestucca Burn is likely to have killed most NCRNA (1994). All trees ≥ 5 cm DBH were included. Values
trees within the fire area (Munger 1944). All of 157 trees are means with standard errors in parentheses.
cored on the CHEF plots established after the fire Trees per ha
(Harcombe 1986). Tree core data are not available for CHEF NCRNA
the NCRNA plots but, based on age-DBH relationships
Total 734 (102) 499 (95)
for the cored trees from the CHEF plots, it appears that Picea sitchensis 94 (24) 47 (5)
in 1979 as many as 5 trees/ha may have been survivors Tsuga heterophylla 638 (114) 446 (95)
of the Nestucca Burn. Basal area (m2/ha)
While tree densities and basal area differed between Total 104 (6) 65 (5)
P. sitchensis 52 (13) 28 (3)
the two sets of plots at the most recent measurement T. heterophylla 51 (10) 37 (8)
(1993 for CHEF, 1994 for NCRNA), other aspects of Quadratic mean diameter (cm)
forest structure and composition were similar (Table 1). P. sitchensis 83 ( 2) 85 ( 8)
T. heterophylla 35 ( 4) 40 ( 2)
Trees per ha and basal area were ca. 50% greater on the
CHEF plots. On both sets of plots, Tsuga heterophylla
accounted for ca. 90% of stems and 50% of basal area.
Mean tree sizes were similar on the two sets of plots, Methods
with P. sitchensis about twice as large as T. heterophylla.
The prevailing climate for both sets of plots is cool The two sets of plots were established at different
and wet with annual precipitation of 2496 mm, falling times, employing different designs. The nine CHEF
mostly between late fall and spring. Fog drip can add as plots, each 0.4 ha in size, were established in 1935 on
much as 800 mm precipitation during the summer months various sites throughout the Experimental Forest. Plot
(Ruth 1954). Mean annual temperature is 10.3°C for locations were selected to be representative of the post-
nearby Otis, Oregon (Franklin & Dyrness 1988). The Nestucca Burn age-class of Picea sitchensis-Tsuga
CHEF plots are mostly situated several km inland from heterophylla forest in the vicinity of Cascade Head
the Pacific Ocean on gentle, generally south or west- Experimental Forest (Brieglieb 1940; Munger 1946),
facing slopes. The NCRNA plots are directly adjacent to although there is apparently no detailed record of the
the Pacific Ocean and occur on highly dissected, moder- process of site selection. The 44 NCRNA plots were
ate to steep slopes at all aspects (Fig. 1). The soil parent established in 1978 and 1979, each nominally 0.1 ha in
material in both sets of plots is primarily tuffaceous size; area of the NCRNA plots varies due to differences
siltstones (Greene 1982; Harcombe et al. 1990), though in slope. The NCRNA plots were systematically located
basalt bedrock can cause significant local modifications 200 m apart on parallel transacts 400 m apart. Although
of the soil profile, especially in the NCRNA plots. The the two sets of plots differ with respect to number and
texture of the soil in the CHEF plots is silt loam to silty size of plots, measurements on individual trees are iden-
clay loam throughout all soil horizons (J. Norgren & tical (see below), the total area sampled is similar and
C.T. Dyrness unpubl.). Few rocks are found in any soil the frequency of measurement has been the same since
layer (< 5%). The soil is generally friable through the B2 the NCRNA plots were established. Inclusion of data
horizon, becoming much firmer in lower horizons. Struc- for the CHEF plots for the period before establishment
ture of the soil is fine granular through the A horizons, of the NCRNA plots makes direct examination of a
becoming a mix of weaker subangular particles in the B much longer interval of stand development possible. All
horizons, with more massive structure in the lowest values are reported on a slope-corrected, horizontal area
horizons. Fine, medium and coarse roots are found basis.
throughout the A horizon, fine and medium roots in the
B horizon, with virtually no roots in the C horizon. Soil Coarse woody debris stores
depths exceed 1.5 m in all the CHEF plots. Soils at the
NCRNA plots are not as well characterized (Greene The mass of CWD was estimated using the fixed
1982). Because these plots are on steeper, more dis- area plot method (Harmon et al. 1987; Harmon & Sex-
sected slopes where bedrock is often seen close to the ton 1996). Measurements were made in 1993 for the
surface we infer that the soils are not as deep and CHEF plots and in 1994 for the NCRNA plots. The
probably contain more gravel and rocks. dimensions of all CWD > 10 cm diameter and > 1 m
long were measured in both sets of plots. In addition, the
species, decay class and tag number (if found) of each
piece were recorded. For each tagged piece, the number
728 Acker, S.A. et al.

of years since death was defined as the measurement Measurements of live trees
year (i.e. 1993 or 1994) minus the middle year of the
measurement interval in which the tree was recorded as On both sets of plots, all trees > 5 cm DBH (1.37 m)
dead. Five decay classes were distinguished using a were tagged. The CHEF plots were measured in 1935,
modified version of the system of Sollins et al. (1987) 1940, 1945, 1955, 1968, 1978, 1983, 1988 and 1993 and
and Graham & Cromack (1982). Class 1 was the least the NCRNA plots in 1978/79, 1984, 1989 and 1994. At
decayed (i.e. bark and wood sound, fine twigs present), each measurement, trees that had grown to > 5 cm DBH
class 3, intermediate in decay (i.e. some bark present, since the previous measurement (ingrowth) were tagged.
heartwood decaying but still capable of supporting Dead trees were recorded at each measurement. Mortal-
weight) and class 5 was the most decayed (i.e. bark ity checks were also made during some of the years
entirely absent, wood decayed into a structureless mass). between measurements (CHEF 1950, 1961, 1979-1986;
We modified this classification system to account for NCRNA 1980-1990).
the fact that moss growth was considerably faster in To estimate volume and biomass of bole wood and
these coastal forests than in the inland forests of the bark from DBH, we used separate sets of allometric
Cascade Range where the classification was originally equations developed from optical dendrometer meas-
developed. urements for the two sets of plots. The form of the
Dimensional measurements for logs included the end equations was:
and middle diameters and the total length inside the plots. V = C * B0 * DB1 (1)
DBH, top diameter and length of snags were recorded. In
the case of broken snags, the top diameter was obtained where V = volume, B0 and B1 are regression coefficients,
by measuring the large end of the resulting log (if it could C is a factor to correct for bias introduced by back-
be located), or by making a visual estimate. Diameters to transformation from logarithmic to arithmetic units
the nearest 1 cm were measured using 100 cm calipers or (Sprugel 1983) and D = DBH. Separate equations were
diameter tapes. Log length to the nearest 0.1 m was developed for each species (Picea sitchensis, Tsuga hetero-
estimated using ultrasonic distance measuring devices phylla, Pseudotsuga menziesii at CHEF; P. sitchensis and
(Sonin Combo Pro; Sonin Inc., Scarsdale, New York, T. heterophylla at NCRNA). For the CHEF plots, equa-
USA) or tape measures. Snag height was estimated to the tions were developed for five different measurement
nearest 0.1 m using meter sticks for heights less than 3 m, years and the set of equations closest in time to each
site-specific diameter-height regressions for snags with measurement was used. A single set of equations was
intact tops and a clinometer and an ultrasonic distance developed for the NCRNA plots and was used for all
measuring device for snags > 3 m tall with broken tops. four measurements. Separate equations were necessary
The volume of each piece was calculated using Newton's for the two sets of plots because of differences in allom-
formula for logs and a cone frustum for snags (Harmon & etry: for a given DBH, trees at NCRNA are shorter and
Sexton 1996). CWD mass was calculated by multiplying have less biomass than trees on the CHEF plots (Greene
the volume of each piece by the species and decay class- et al. 1992). To convert volume to biomass, we used
specific density (Harmon & Sexton 1996). Mass and published wood density values for the three species.
volume were summed for each plot by species, decay Details of development of the equations and parameter
class and position (snag versus log). values are given in Harcombe et al. 1990 (CHEF plots) and
Three of the 44 NCRNA plots (9, 13, 32) were Greene et al. 1992 (NCRNA plots). Mortality was calcu-
unsafe to inventory because of the quantity of large, lated as the sum of the bole biomass of all trees that died
jumbled coarse wood and intermingled growth of shrubs. during an interval. Biomass at death was estimated using
The volume of CWD in these plots was estimated from DBH at the previous measurement. Net primary produc-
the live volume killed by windstorms since establish- tion of boles (NPPB) was calculated as:
ment of the plots. This is probably an underestimate of NPPB = dB + M (2)
total CWD volume on these plots. Nevertheless, the
amount added by windstorms was approximately twice where dB is the difference in standing live bole biomass
that of the other plots. Comparison with other plots with between one measurement period and the next (includ-
heavy wind damage indicated the underestimation on ing ingrowth) and M is the biomass of the trees that died
these plots to be < 30% at maximum and < 15% if the during the interval. This is equivalent to calculating the
major fraction of mortality was recent. Based on visual sum of the biomass increments for all stems that sur-
inspection of the pieces in these plots we assumed that the vived a measurement interval (Harcombe et al. 1990).
bulk of these logs were in decay class 2.
 - Biomass accumulation in coastal Oregon Picea-Tsuga forest - 729

Reconstruction of stand dynamics organisms (Harmon et al. 1986) are likely to be quite
similar in both locations.
Biomass accumulation at any point in time equals We set the lag before appreciable bole production to
the difference between production and mortality inte- 10 yr (cf. Binkley & Greene 1983) and the CWD decay
grated over time. This simple, mass-balance constraint rate to 3% per year (Sollins 1982; Spies et al. 1988). We
allowed us to explore the logical implications of our then applied the set of assumptions separately to the
records of biomass, NPPB, M and CWD for temporal CHEF and NCRNA data and fitted NPPB and M curves
patterns of production, mortality and biomass during the to the observed data. First we varied the length of the
aggrading phase of stand dynamics. The investigation time lag before appreciable bole mortality in order to
involved fitting curves of NPPB and M to the observed match the measured CWD. Then we varied the magni-
data based on the following assumptions: tude and timing of the peak of NPPB to match the first
1. There is a lag of some years after the stand- observed biomass values. We tested whether NPPB
initiating disturbance before appreciable bole produc- could have peaked at the first observed value. If this was
tion; not the case, the peak in NPPB was incrementally in-
2. Bole production rises steeply, following a para- creased and shifted back in time until the first value of
bolic curve (Peet 1981; Binkley & Greene 1983; biomass was matched.
Harcombe et al. 1990; Ryan et al. 1997) then declines We tested the sensitivity of the reconstruction ap-
linearly to the first observed value of NPPB; proach to variability in the two quantitative parameters
3. There is a lag of some years before appreciable to which we assigned assumed values: the lag before
bole mortality; bole production and the CWD decay rate. Using the data
4. Bole mortality rises linearly to match the general from the CHEF plots, we increased or decreased each
trend of observed values (i.e. extreme mortality events parameter by 10%, and compared the resulting time and
in the observed record are recognized as deviations from magnitude of peak NPPB to values from the standard
the general trend); and reconstruction (Swartzman & Kaluzny 1987).
5. The rate of CWD decay has been constant over The assumptions used to reconstruct the time trends
time and space. of NPPB and M have been assembled into a computer
The first assumption requires that all trees died in the program in the QuickBASIC language (Anon. 1985-
stand-initiating disturbance. As mentioned previously, 1988). The program, a more fully developed explana-
a small number of trees are likely to have survived the tion of the underlying ideas and test data sets are acces-
Nestucca Burn, especially on the NCRNA plots. How- sible via the worldwide web at http://www.fsl.orst.edu/
ever, given the severity of the fire (Munger 1944) and lter/pubs/res_rpts/curv_fit.htm
the sensitivity of both Picea sitchensis and Tsuga hetero-
phylla to fire (Agee 1993), it is clear that biomass and Sapwood respiration
productive capacity were almost completely removed
by the fire. The post-disturbance lag in production is Over the course of observations there has been a
also a consequence of only including trees > 5 cm DBH decrease in the rate of the bole biomass accumulation in
in biomass calculations. Tsuga heterophylla, for exam- both sets of plots (Harcombe et al. 1990; Greene et al.
ple, typically takes 4 years to reach breast height on 1992). Because we wished to investigate whether changes
productive sites (Wiley 1978) and then would require in sapwood biomass and its maintenance respiration
additional growing seasons to reach 5 cm DBH. would account for the observed changes in the rate of
The time lag before appreciable bole mortality rep- accumulation of bole biomass, we estimated the mass of
resents a period early in stand development when trees sapwood of each species using the equations in Harcombe
that have survived the seedling stage experience little et al. (1990). These equations predict the proportion of
competition (Oliver & Larson 1990; Kenkel et al. 1997). sapwood volume for individual trees from DBH. This
In addition, once started self-thinning mortality typi- proportion was multiplied by the bole biomass of each
cally affects the smallest trees in a population (Peet & tree to calculate sapwood mass.
Christensen 1987; Kenkel et al. 1997), further delaying Maintenance respiration of boles is highly corre-
a noticeable effect in terms of biomass. lated with sapwood volume in many species of conifers
The final two assumptions represent parsimonious (Ryan 1990; Ryan et al. 1995). To calculate the mainte-
use of our data or published information for the region, nance respiration of boles at each measurement period
respectively. We consider it unlikely that there is a we multiplied the mass of sapwood by 0.01 /yr, the
systematic difference between the plot sets in the rate of proportion Ryan et al. (1995) estimated for a stand of
decay of CWD, since the controlling factors of tempera- Tsuga heterophylla at low elevation on the western
ture, moisture, oxygen, substrate quality and decomposer slope of the Oregon Cascade Range (Runyon et al.
730 Acker, S.A. et al.

1994). The T. heterophylla stand in Ryan's study had an (Anon. 1989a)). Class 3 was the most common on the
annual maintenance respiration of 3.24 Mg ha−1 yr−1 CHEF plots (50% of total CWD), whereas class 2 was
(assuming a 0.5 ratio of carbon to biomass; Edwards et the most common on the NCRNA plots (46% of total
al. 1981) and a sapwood biomass of 323 Mg/ha (assum- CWD) (Fig. 2). Low CWD in classes 4 and 5 suggests
ing a sapwood density of 0.42 Mg/m3; Ryan et al. 1995). that residual CWD from the stands that burned in 1850
The estimate of maintenance respiration by Ryan et al. has essentially disappeared; this disappearance seems
(1995) was based on an annual temperature of 9.2 °C; likely, considering that, at a 3 %/yr decay rate (Sollins
this value is quite close to the temperature experienced 1982; Spies et al. 1988), only ca. 1% of the residual
at both CHEF and NCRNA plots, which are higher in mass would remain after 140 yr.
elevation and thus cooler than the Otis meteorological For the CHEF plots, we compared the time since
station (Greene 1982). As there are no published esti- death for tagged pieces in decay classes 1, 2, 3 and 4
mates of the ratio of maintenance respiration to sapwood (there were only two tagged pieces in decay class 5).
mass for Picea sitchensis or Pseudotsuga menziesii, we The percentage of tagged pieces decreased with decay
assumed that these species would have similar values to class:
those of T. heterophylla. 62% for decay class 1;
57% for decay class 2;
31% for decay class 3;
Dating mortality events
9% for decay class 4.
We took two approaches to assessing whether CWD Though the range of time since death within each
of different decay classes represents mortality in differ- decay class was broad, the bulk of observations of
ent periods in the past. First, we compared the decay decay classes 1 and 2 were distinct from those of decay
class and the time since death for tagged pieces of classes 3 and 4 (Fig. 3). Median ages for the decay
CWD. Second, we regressed the mass of trees observed classes were:
dying in various intervals in the past on CWD in differ- 1.5 yr for decay class 1;
ent decay classes. 6 yr for decay class 2;
We selected regression models to investigate based 18.5 yr for decay class 3;
27 yr for decay class 4.
on the distribution of time since death of tagged pieces
by decay class. Decay classes 1 and 2 were examined For the NCRNA plots, all tagged pieces were in decay
both separately and together due to the substantial over- classes 1, 2 and 3, with considerable overlap of classes 2
lap in ages. Due to the relatively short duration of
measurement of the NCRNA plots, only regressions of
mortality mass on CWD in decay classes 1 and 2 were
investigated. For each decay class or combination of
decay classes considered, regressions with mortality
from alternative time periods were examined. For sig-
nificant regressions, transformations of the dependent
variables were employed where necessary to satisfy
assumptions of linear regression. The three NCRNA
plots on which CWD was not measured due to unsafe
conditions were excluded from regressions.

Results and Discussion

Coarse woody debris

Mean total CWD mass was slightly higher on the

CHEF plots than the NCRNA plots (155 vs 144 Mg/ha),
not lower as predicted under the hypothesis of higher
mortality on the NCRNA plots (p = 0.80 in a one-tailed
Fig. 2. Mass of coarse woody debris by decay class for CHEF
Wilcoxon rank-sum test; Anon. 1989b). For both sets of and NCRNA plots. Inventories were conducted at stand ages
plots, most CWD was in decay classes 2 and 3 (Fig. 2). 143 yr for CHEF plots and 144 yr for NCRNA plots. Values
The two sets of plots differed with respect to the distri- are means (n = 9 for CHEF and n = 44 for NCRNA); bars
bution of CWD (p < 0.001 in a likelihood-ratio χ2 test represent standard errors.
 - Biomass accumulation in coastal Oregon Picea-Tsuga forest - 731

observed over various intervals in the past (Table 2) for

some, but not all, combinations of decay class and plot
set. For the CHEF plots, the regressions of mortality
mass 15 - 24 yr and 10 - 24 yr ago on CWD in decay class
3 were both highly significant. For the NCRNA plots,
the regressions of mortality mass at various intervals
between 0 and 14 yr ago on CWD in decay classes 1 and
2, separately and combined, were all highly significant.
The analysis of decay class of pieces with known time of
death confirms that most of the CWD present on both
sets of plots corresponds to tree mortality in the last 20
years (Fig. 3).
If the biomass difference between the two sets of
plots in 1994 had been caused by higher recent mortality
in the NCRNA plots, then CWD would be greater at
NCRNA than at CHEF (Greene et al. 1992). That CWD
was not greater at NCRNA indicates that the large
difference in biomass between the NCRNA and the
CHEF plots is not a consequence of higher recent mor-
tality at NCRNA. This leaves a long-term difference in
NPPB as the most reasonable explanation for the differ-
ence in accumulation of above-ground live biomass
over 150 yr of stand development.
Fig. 3. Time since death by decay class for tagged pieces of
coarse woody debris on CHEF and NCRNA plots. Boxes
represent the middle 50% of observations (i.e., between 25th Net primary production, mortality and live biomass of
and 75th percentiles, or lower and upper quartiles). Vertical bole wood
lines correspond to the median value. Horizontal lines extend-
ing from boxes include values up to the upper quartile plus 1.5 Net primary production of bole wood was not sig-
times the interquartile range (difference between upper and nificantly different between the two sets of plots for two
lower quartile) or the lower quartile minus 1.5 times the of the three measurement intervals. Values were most
interquartile range. Circles represent more extreme values. similar for the first interval, ending at age 133/134 yr:
CHEF, 6.2 Mg ha−1 yr−1, s.e. 0.3;
NCRNA, 5.8 Mg ha−1 yr−1, s.e. 0.6; p = 0.54; two-tailed t-test, df =
and 3 (Fig. 3). Again, the percentage of tagged pieces 50.9 by Satterthwaite correction due to unequal variance (Anon.
decreased with decay class: 1989b).
37% for decay class 1; For the second interval, ending at age 138/139 yr, pro-
21% for decay class 2; duction of bole wood was significantly greater for the
2% for decay class 3.
CHEF plots:
Median ages were
CHEF, 6.8 Mg ha−1 yr−1, s.e. 0.3;
1.5 yr for decay class 1;
NCRNA, 4.5 Mg ha −1 yr−1, s.e. 0.4; p = 0.0001; two-tailed t-test,
9 yr for decay class 2;
df =36.6
10.5 yr for decay class 3.
Ages of CWD in decay classes 1, 2 and 3 were com- In the most recent interval (up to age 143/144 yr),
pared between the two sets of plots with two-tailed production was greater on the CHEF plots, but the
Wilcoxon rank-sum tests (Snedecor & Cochran 1980; difference was not significant:
Anon. 1989b). For decay class 1, there was no signifi- CHEF, 6.7 Mg ha−1 yr−1, s.e. 0.6;
cant difference in the distribution of ages (p = 0.16). NCRNA, 5.0 Mg ha−1 yr−1, s.e. 0.4; p = 0.10
For decay classes 2 and 3, differences were significant in two-tailed t-test with df = 51 (assumption of equal variance not
(p = 0.015 and p = 0.002, respectively). However, the violated).
direction of the difference was not consistent, as CWD
in decay class 2 tended to be older on NCRNA plots Over the more than 60 years of observation, net
while the opposite was true for decay class 3 (Fig. 3). production of bole wood has declined on the CHEF
There is a significant correspondence between in- plots (Fig. 4). Although variation from one interval to
ventoried CWD of different decay classes and mortality the next has increased over time, the decrease in NPPB
732 Acker, S.A. et al.

Table 2. Regressions of mass lost to mortality on CWD mass. n = 9 for CHEF and 41 for NCRNA (see text for details).

Plot set Tree mortality CWD

during interval decay class
(years ago) P-value Regression model R2

CHEF 0-4 1 0.06 n/a n/a

CHEF 0-9 1 0.24 n/a n/a
CHEF 5-9 2 0.07 n/a n/a
CHEF 0-9 2 0.5 n/a n/a
CHEF 0-14 2 0.06 n/a n/a
CHEF 0-9 1 and 2 0.16 n/a n/a
CHEF 0-14 1 and 2 0.08 n/a n/a
CHEF 15-24 3 0.002 log(M) = 2.58 + 0.0118*CWD 0.73
CHEF 10-24 3 0.0004 log(M) = Ð0.329 + 1.07*log(CWD) 0.83
CHEF 25-37 4 0.11 n/a n/a
CHEF 15-37 4 0.86 n/a n/a
NCRNA 0-4 1 0.0001 √ M = 1.69 + 0.102*CWD 0.46
NCRNA 0-9 1 0.0003 √ M = 4.09 + 0.093*CWD 0.27
NCRNA 5-9 2 0.003 √ M = 1.84 + 0.026*CWD 0.18
NCRNA 0-9 2 0.01 M = 25.3 + 0.441*CWD 0.14
NCRNA 0-14 2 0.0001 √ M = 4.92 + 0.052*CWD 0.37
NCRNA 0-9 1 and 2 0.0001 √ M = 2.41 + 0.042*CWD 0.4
NCRNA 0-14 1 and 2 0.0001 √ M = 3.74 + 0.057*CWD 0.51

from the first interval to the last interval is highly NCRNA, 6.4 Mg ha−1 yr−1, s.e. 1.6; p = 0.42, df = 45.3
significant (mean change – 4.7 Mg ha−1 yr−1, s.e. 0.8; p and in the third interval:
= 0.0003 in two-tailed t-test, df = 9).
CHEF, 6.1 Mg ha−1 yr−1, s.e. 1.3;
Biomass lost to mortality was not significantly dif-
NCRNA, 5.5 Mg ha−1 yr−1, s.e. 1.6; p = 0.75, df = 34.7.
ferent between the plot sets for any of the three intervals
of measurement. Mortality was high in the first interval: Over the period that both sets of plots have been
CHEF, 10.0 Mg ha−1 yr−1, s.e. 1.9; observed, CHEF and NCRNA have had similar percent-
NCRNA, 11.0 Mg ha−1 yr−1, s.e. 2.4; p = 0.73, 37.5 d.f. ages of standing trees among those dying (49% and
but mortality was lower in the second interval 47%, respectively). This suggests that mortality due to
CHEF, 4.9 Mg ha−1 yr−1, s.e. 1.0; competition and other biotic factors is of similar impor-

Fig. 4. Annual net primary production of bole wood plus bark

versus stand age for CHEF plots. Each of the nine plots is Fig. 5. Annual loss of biomass due to tree mortality versus
represented by a unique symbol to facilitate recognition of stand age for CHEF plots. Each of the nine plots is represented
trends. by a unique symbol to facilitate recognition of trends.
 - Biomass accumulation in coastal Oregon Picea-Tsuga forest - 733

tance on both sets of plots. Thus, over the period of observation of both sets of
Over the 60 years of observations annual mortality plots, live biomass has consistently been higher on the
has increased on the CHEF plots (Fig. 5). The increase CHEF plots while differences in NPPB and mortality
in mortality from the first interval to the last interval is have been minimal. Since the analysis of CWD sug-
significant (mean change 3.8 Mg ha−1 yr−1, s.e. 1.1; p = gested that production, rather than mortality, explains
0.009; two-tailed t-test, df = 9). Variability between the differences in biomass between the two sets of
plots and from one interval to the next has increased plots, it appears that production on the two sets of plots
over time. must have been markedly different at some time prior
The greatest observed loss of biomass to mortality to the establishment of the NCRNA plots in 1979. The
occurred at about the same age, 133/134 yr, for both sets plausibility of this speculation can be addressed by
of plots (Fig. 6). This peak is related to two major wind reconstructing dynamics of biomass and CWD.
events that occurred in 1981 and 1983. A relatively
recent peak in biomass lost to mortality is consistent Reconstruction of stand dynamics
with the distribution of CWD by decay class. This is
further evidence that early differences in NPPB is a The reconstruction of stand dynamics supports the
more probable explanation than early differences in conclusion that NPPB has been higher on the CHEF
mortality for the current difference in above-ground
biomass between the two plots sets.
Over the period of observation, the fraction of live
biomass lost to mortality has consistently been higher
on the NCRNA plots than on the CHEF plots (Fig. 6).
For the first three decades, annual loss rate was between
0.003 and 0.005 for the CHEF plots. The peak in annual
loss rate was at the same age for both sets of plots, 133/
134 yr. Annual loss rate peaked at 0.013 on the CHEF
plots and dropped to 0.008 in the most recent interval.
The highest annual loss rate on the NCRNA plots (0.022)
was observed in the first measurement interval. Annual
loss rate on the NCRNA plots dropped to 0.012 in the
most recent interval.
The maximum observed biomass values for both the
CHEF and the NCRNA plots (774.6 Mg/ha, s.e. 39.1,
and 499.0 Mg/ha, s.e. 38.4, respectively) occurred at age
128 and 129 yr (Harcombe et al. 1990; Greene et al.
1992). Biomass changed little for both sets of plots over
the most recent measurement interval (0.3% increase
for CHEF; 0.4% decrease for NCRNA).

Fig. 7. Reconstruction of biomass dynamics for the aggrading

phase for CHEF and NCRNA plots, constrained by initial
biomass of zero and measured biomass, mortality, NPPB and
CWD. a. Annual NPPB; b. Annual loss of biomass to tree
Fig. 6. Annual loss of biomass due to tree mortality (solid line) mortality; c. Live biomass of tree bole wood plus bark (B) and
and annual fractional biomass loss (dashed line) versus stand CWD. Observed and reconstructed trends for NCRNA plots
age for CHEF and NCRNA plots. are marked with open circles.
734 Acker, S.A. et al.

plots for most of the life of the current stands (Fig. 7a). mortality, maximum NPPB and the year of maximum
NPPB on both sets of plots appears to have diverged in NPPB that will simultaneously match the observed
the first decades of growth, reaching a maximum dis- values of CWD and initial measured biomass, NPPB
crepancy at age 75 yr, shortly before NPPB peaked on and mortality. The long duration of records of tree
the CHEF plots. Since then, NPPB of the plots has mortality and growth, in conjunction with the CWD
apparently been converging to a lower value. For most inventory, imposes severe limits on plausible dynam-
of the life of the current stands, gross mortality on the ics prior to plot establishment. In particular, observed
two sets of plots has been similar (Fig. 7b). tree mortality is closely enough matched to invento-
As a consequence of production and mortality trends, ried CWD that there is very little room for CWD
the reconstruction also demonstrates that over the life of generated prior to plot establishment.
the current stands, biomass has always been higher on The reconstruction should be viewed as a parsi-
the CHEF plots (Fig. 7c). Moreover, it appears that at no monious extrapolation of the available data. A radically
time since establishment of the current stand has CWD different reconstruction would require a larger number
been higher on the NCRNA plots than on the CHEF of more complex assumptions. For example, an earlier
plots. Biomass on the CHEF plots approached and fluc- peak in NPPB would require a corresponding early
tuated around a plateau, whereas biomass on the NCRNA increase in rates of both mortality and decomposition to
plots apparently peaked later and then declined slightly make the increase in wood produced early in the life of
to a plateau. Since biomass has consistently been lower the stands disappear. However, we cannot rule out the
on the NCRNA plots while gross mortality has been possibility that tree species composition was markedly
similar, the fraction of biomass lost to mortality has different early in the history of these stands, or that
consistently been higher on the NCRNA plots. Further decomposition rates have fluctuated due to climatic
investigation of this apparent difference in disturbance variability. Either of these factors could make the his-
regime between the two sets of plots would be worth- tory of these stands more complicated than our simple
while. scenario.
Although the reconstruction depends, in part, on A difference between the two sets of plots in NPPB
assumptions concerning the shape of the time trend of is the logical explanation for differences in standing
NPPB prior to establishment of the plots, it is severely biomass. However, the difference in NPPB for the pe-
constrained by the biomass value of zero at stand riod when both were measured (1978-1994) is too small
initiation and observed biomass, mortality, NPPB and to account for this difference. Therefore, NPPB differ-
CWD values. The sensitivity analysis demonstrates ences must have been greater at some time in the past;
both the degree of this constraint, and that the recon- the reconstruction indicates that at its peak, NPPB was
struction method is robust to small changes in param- ca. 60% higher at CHEF than at NCRNA (Fig. 7a). This
eter values that decrease net accumulation of living difference may be related to better site quality at CHEF
and dead mass (Table 3). Small increases (10%) in due to gentler slopes and deeper soils. Recent mapping
either the lag before appreciable NPPB or the decay shows considerable heterogeneity of bedrock geology
rate of CWD, do not alter substantially the estimates of in the study area (Snavely et al. 1996), which could
the magnitude or timing of peak NPPB. Small de- contribute to differences in potential productivity be-
creases in these parameters result in an inability to tween the plots.
reconcile estimated trends with observations. That is, In addition to indicating the magnitude of differ-
there is no set of values of the lag before appreciable ence between the plots in maximum NPPB, recon-

Table 3. Sensitivity analysis of reconstruction of biomass dynamics.

Parameters (inputs) Estimates (outputs)

Lag before Annual rate Lag before Maximum Time of
Reconstruction appreciable of CWD decay tree mortality NPPB maximum
scenario NPPB (yr) (yr) (Mg ha−1 yr−1) NPPB (yr)

Standard 10 0.03 80 11.4 90

Test 1 11 0.03 85 11.4 90
Test 2 9 0.03 No solution possible1
Test 3 10 0.033 40 11.9 86
Test 4 10 0.027 No solution possible1

1See text for explanation.

 - Biomass accumulation in coastal Oregon Picea-Tsuga forest - 735

struction of the time trend of NPPB for the CHEF plots the CHEF plots declined from 11.4 to 6.7 Mg ha−1 yr−1,
is consistent with the observation that NPPB peaks while estimated sapwood respiration rose from 2.2 Mg
relatively early in development of spruce-hemlock ha−1 yr−1 to 2.5 Mg ha−1 yr−1, accounting for only ca.
stands (Fujimori 1971; Grier 1976). However, our 6% of the decline in NPPB. Declining NPPB with
reconstruction suggests that the peak in NPPB occurs forest age is more likely due to a combination of
at 90 yr for the CHEF plots rather than the 25 - 40 yr declining leaf area, reduced photosynthesis per unit
predicted by Harcombe et al. (1990). The NPPB peak leaf area and increased below-ground allocation
at 90 yr is two decades later than the oldest ages for (Gower et al. 1996; Ryan et al. 1997). Assessing the
peak forest growth in chronosequence studies reviewed relative magnitudes of these and other changes con-
by Ryan et al. (1997), which were reported for other tributing to declining NPPB is an important area for
species of spruce. The reconstruction also illustrates future research.
the decline in NPPB associated with even-aged stands
(Ryan & Waring 1992; Gower et al. 1996; Ryan et al. Biomass dynamics after the peak
1997).
Our observations of mortality, recruitment and NPPB
Sapwood respiration allow us to evaluate which of Peet's (1981, 1992) sce-
narios for biomass dynamics after the peak are most
In the CHEF plots, sapwood biomass increased from applicable: declining biomass, periodic oscillations or
216 Mg/ha at age 85 yr to a high of 262 Mg/ha at age 128 relative constancy with random fluctuations. Although
yr. This was followed by a decline to 253 Mg/ha at age long-term decline in biomass cannot be ruled out based
133 yr. Since then, sapwood biomass in these plots has on existing data, several observations suggest that it is
remained relatively constant at 252 Mg/ha. In the unlikely. First, NPPB seems to be levelling off in the
NCRNA plots there was a similar decline in sapwood range of 5 to 7 Mg ha−1 yr−1 in both sets of plots;
biomass from 154 Mg/ha at 129 yr and declining to 142 sequestration of resources would result in a continuing
Mg/ha at 134 yr. From 134 to 144 yr sapwood biomass long-term decline in NPPB. Second, inspection of NPPB
in the NCRNA plots was ca. 139 Mg/ha. This was 55% values for plots regenerating after blowdown indicates
of that estimated for the CHEF plots over a similar time that NPPB in post-disturbance plots can be high. This
period, a reflection of the lesser amount of bole biomass suggests that as the forest ages and the plots come to
on the NCRNA plots. The decrease in sapwood biomass have a more even distribution among various biomass
between 128/129 yr and 133/134 yr was probably caused and disturbance classes and stages of recovery, mean
by wind-related mortality, which removed large trees NPPB over all plots will not fall substantially and mean
with high sapwood biomass. biomass will not decline. In addition, the large accumu-
Changes in maintenance respiration for sapwood lation of CWD, which could lead to a decline in nutrient
paralleled those of sapwood mass for both sets of availability (Gower et al. 1996), does not appear to be
plots (Fig. 8) using our method of estimating sapwood inhibiting production. Finally, there are documented
respiration. These changes demonstrate that the de- examples of high-biomass, old-growth hemlock-spruce
cline in NPPB is not a consequence of increasing stands in Washington (483-830 Mg/ha at 205 to 266 yr;
respiratory load. Over the period of observation, NPPB on McKee et al. 1982) and Alaska (ca. 450 to 750 Mg/ha for
ages ca. 200 to 390 yr; Alaback 1982), suggesting that
live biomass may be relatively stable for intervals longer
than the current age of our stands.
Biomass oscillation in the period after the peak
could be caused by a long interval between the opening
of stands due to tree mortality and the onset of tree
regeneration (Peet 1981). This seems unlikely in the
present instance, since tree regeneration can commence
prior to opening of the canopy, and regeneration lags
that do occur are short. For example, relatively large
ingrowth numbers were recorded on several NCRNA
plots that had high initial biomass and experienced no
mortality during the first measurement interval. Sub-
stantial ingrowth (40 stems ha−1 yr−1) was observed on
Fig. 8. Mean annual sapwood maintenance respiration versus NCRNA plots with biomass values as high as 744 Mg/
stand age for CHEF and NCRNA plots. ha; only six of the 44 NCRNA plots had higher biomass
736 Acker, S.A. et al.

values at the beginning of the interval. Thus mortality tality rates, constant biomass would be ca. 6% lower
within a plot is not required to trigger recruitment. than the present value. For the NCRNA plots, constant
Concerning regeneration lags, six of the 10 NCRNA biomass would be ca. 30% lower than at present.
plots that lost at least 50% of their biomass over the Hypothetical constant values of CWD can be calcu-
course of the study did not show a lag: recruitment lated similarly. Since mortality equals production when
occurred in all three measurement intervals. The other biomass is constant, the input rate would equal the
four plots experienced most mortality in the first inter- NPPB values stated above. Assuming a 3%/yr decay
val (1979-1984) and had no ingrowth until the last rate, stable CWD values would be 220 Mg/ha for CHEF
interval (1989-1994), indicating a maximum recruit- and 170 Mg/ha for NCRNA. Thus under hypothetical
ment lag of ca. 10 yr. For CHEF plot 13, significant tree stable conditions based on current production and mor-
mortality due to wind-throw commenced at 122 yr tality rates, the sum of live bole biomass and CWD
(Harcombe et al. 1990). Ingrowth on this plot was first would be ca. 20 Mg/ha (2%) greater than at present for
recorded at 133 yr (492 stems/ha), indicating a recruit- CHEF. For the NCRNA plots, the sum of live bole
ment lag of between 6 and 11 yr. Thus, in the absence of biomass and CWD under hypothetical stable conditions
advanced regeneration or chronic recruitment due to would be ca. 110 Mg/ha (20%) lower than at present.
low stocking, there appears to be a recruitment lag of ca. Long-term biomass dynamics in the post-peak phase
10 yr. is most likely to be characterized by relative stability
Another reason that biomass oscillation is unlikely with random fluctuations. If current production and
is that the large temporal variability in windstorm dam- mortality rates prevail, the long-term mean of live,
age would seem to reduce the likelihood of synchronous above-ground biomass will be somewhat lower than at
onset of mortality (cf. Peet 1981,1992). Because plots present for both CHEF and NCRNA. While differences
do not necessarily break up in a single wind-throw event in production appear to explain current differences in
(Harcombe et al. 1990), recruitment can commence live bole biomass between the plot sets, future trends in
while there is sufficient canopy remaining to maintain mortality may be more important for long-term biomass
relatively high levels of NPPB. By the time the original dynamics, as mortality has fluctuated much more than
canopy is gone, the newly recruited cohort will have production in recent decades.
attained high NPPB. The net result is that overall NPPB
does not decline due to a recruitment time lag.
Thus, according to the theory of long-term biomass Acknowledgements. This work would not have been possible
dynamics (Peet 1981, 1992) , this system appears to lack without the dedication to the long-term measurement program
the conditions for peak/decline (synchronous mortality of many scientists and technicians, most notably Jerry Franklin
of a dominant age class, replacement of species with and Jack Booth. For critical readings of the manuscript we
thank Rick Busing, Mike Wimberly and anonymous review-
large maximum tree size by species with smaller
ers. This work was supported by grants from the National
maximum tree size, long-term declines in resource avail- Science Foundation to the H.J. Andrews Experimental Forest
ability) or periodic oscillation (synchronous mortality Long-Term Ecological Research Program (DEB-76-11978,
and/or recruitment time lags). Thus, relative constancy DEB-79-25939, DEB-80-12162, BSR-83-00370, BSR-85-
of biomass appears more likely than the alternatives. 14325, and BSR-90-11663), and cooperative agreements be-
Constant biomass over time implies a balance be- tween Oregon State University and the Pacific Northwest
tween NPPB and mortality. This condition can be for- Research Station of the United States Forest Service.
mulated as equality between NPPB and the product of
biomass and annual loss fraction (Reiners 1983;
Harcombe et al. 1998). We used this formulation to References
assess the hypothetical constant biomass implied by
current NPPB and annual loss fraction. This hypotheti- Anon. 1985-1988. QuickBASIC version 4.5. Microsoft Corpo-
cally stable state is best viewed as a long-term value, ration, Redmond, WA.
about which biomass will fluctuate. Anon. 1988. SAS(R) Procedures guide, Release 6.03 Edition.
To estimate current NPPB and mortality we used the SAS Institute Inc., Cary, NC.
mean of the last three intervals. For the CHEF plots, Anon. 1989a. SAS/STAT(R) User's Guide. Version 6, 4th ed.,
Vol. 1. SAS Institute Inc., Cary, NC.
with current NPPB of 6.6 Mg ha−1 yr−1 and annual loss
Anon. 1989b. SAS/STAT(R) User's Guide. Version 6, 4th ed.,
fraction of 0.0091, the hypothetical constant biomass Vol. 2. SAS Institute Inc., Cary, NC.
is 725 Mg/ha. For NCRNA, with current NPPB of 5.1 Agee, J.K. 1993. Fire ecology of Pacific Northwest forests.
Mg ha−1 yr−1 and annual loss fraction of 0.0158, the Island Press, Washington, DC.
hypothetical constant biomass is 323 Mg/ha. Thus, for Alaback, P.B. 1982. Forest community structural changes
the CHEF plots, based on current production and mor-
 - Biomass accumulation in coastal Oregon Picea-Tsuga forest - 737

during secondary succession in southeast Alaska. In: spruce-hemlock forest based on age structure and 50 years
Means, J.E. (ed.) Forest succession and stand develop- of mortality data. Forest Ecol. Manag. 14: 41-58.
ment research in the Northwest: Proceedings of a sympo- Harcombe, P.A., Harmon, M.E. & Greene, S.E. 1990. Changes
sium, pp. 70-79. Forest Research Laboratory, Oregon in biomass and production over 53 years in a coastal Picea
State University, Corvallis, OR. sitchensis-Tsuga heterophylla forest approaching matu-
Binkley, D. & Greene, S. 1983. Production in mixtures of rity. Can. J. For. Res. 20: 1602-1610.
conifers and red alder: the importance of site fertility and Harcombe, P.A., Allen, R.B., Wardle, J.A. & Platt, K.H. 1998.
stand age. In: Ballard, R. & Gessell, S.P. (tech. eds.) Spatial and temporal patterns in stand structure, biomass,
IUFRO symposium on forest site and continuous produc- growth, and mortality in a monospecific Nothofagus
tivity, pp. 112-117. USDA, Forest Service, Pacific North- solandri var. cliffortioides (Hook. f.) Poole forest in New
west Forest & Range Experiment Station, General Techni- Zealand. J. Sustainable Forestry 6: 313-345.
cal Report PNW-163, Portland, OR. Harmon, M.E. & Sexton, J. 1996. Guidelines for measure-
Brieglieb, P.A. 1940. Spruce-hemlock forest shows prodi- ments of woody detritus in forest ecosystems. United States
gious growth. In: PNW Research Note 31, pp. 1-2. USDA LTER Network Office, University of Washington, Seattle,
Forest Service, Portland, OR. WA, Publ. 20.
Cohen, W.B., Harmon, M.E., Wallin, D.O. & Fiorella, M. Harmon, M.E., Cromack, K. Jr. & Smith, B.G. 1987. Coarse
1996. Two decades of carbon flux from forests of the woody debris in mixed-conifer forests, Sequoia National
Pacific Northwest. BioScience 46: 836-844. Park, California. Can. J. For. Res. 17: 1265-1272.
Cooper, C.F. 1983. Carbon storage in managed forests. Can. J. Harmon, M.E. et al. 1986. Ecology of coarse woody debris in
For. Res. 13: 155-166. temperate ecosystems. Adv. Ecol. Res. 15: 133-302.
Dixon, R.K., Brown, S., Houghton, R.A., Solomon, A.M., Kenkel, N.C., Hendrie, M.L. & Bella, I.E. 1997. A long-term
Trexler, M.C. & Wisniewski, J. 1994. Carbon pools and study of Pinus banksiana population dynamics. J. Veg.
flux of global forest ecosystems. Science 263: 185-190. Sci. 8: 241-254.
Edwards, N.T., Shugart, H.H. Jr., McLaughlin, S.B., Harris, Kira, T. & Shidei, T. 1967. Primary production and turnover
W.F. & Reichle, D.E. 1981. Carbon metabolism in terres- of organic matter in different forest ecosystems of the
trial ecosystems. In: Reichle, D.E. (ed.) Dynamic proper- western Pacific. Jpn. J. Ecol. 17: 70-87.
ties of forest ecosystems, pp. 499-536. Cambridge Univer- McKee, A., LaRoi, G. & Franklin, J.F. 1982. Structure, com-
sity Press, Cambridge. position and reproductive behavior of terrace forests, South
Franklin, J.F. & Dyrness, C.T. 1988. Natural vegetation of Fork Hoh River, Olympic National Park. In: Starkey, E.
Oregon and Washington. Oregon State University Press, E., Franklin, J.F. & Matthews, J.W. (Tech. Coords.) Eco-
Corvallis, OR. logical research in National Parks of the Pacific North-
Fujimori, T. 1971. Primary productivity of a young Tsuga west, pp. 22-29. Forest Research Laboratory, Oregon State
heterophylla stand and some speculations about biomass University, Corvallis, OR.
of forest communities on the Oregon coast. USDA, Forest Morris, R.F. 1934. Forest fires in western Oregon and Wash-
Service, Pacific Northwest Forest & Range Experiment ington. Oregon Histor. Quart. 35: 313-339.
Station, Research Paper PNW-123. Munger, T.T. 1944. Out of the ashes of Nestucca. Am. For. 50:
Gower, S.T., McMurtrie, R.E. & Murty, D. 1996. Aboveground 342-345, 366-368.
net primary production decline with stand age: potential Munger, T.T. 1946. Spruce-hemlock grew 17,523 board feet
causes. Trends Ecol. Evol. 11: 378-382. (per acre) in 11 years. In: PNW Research Note 34, pp. 5-
Graham, R.L. & Cromack, K. Jr. 1982. Mass, nutrient content 6. USDA Forest Service, Portland, OR.
and decay rate of dead boles in rain forests of Olympic Oliver, C.D. & Larson, B.C. 1990. Forest stand dynamics.
National Park. Can. J. For. Res. 12: 511-521. McGraw-Hill, Inc., San Francisco, CA.
Greene, S.E. 1982. Neskowin Crest Research Natural Area. In: Pare, D. & Bergeron, Y. 1995. Above-ground biomass accu-
Franklin, J.F., Hall, F.C., Dyrness, C.T. & Maser, C. (eds.) mulation along a 230-year chronosequence in the southern
Federal Research Natural Areas in Oregon and Washing- portion of the Canadian boreal forest. J. Ecol. 83: 1001-
ton: a guidebook for scientists and educators, Supplement 1007.
No. 13. USDA, Forest Service, Pacific Northwest Forest Peet, R.K. 1981. Changes in biomass and production during
and Range Experiment Station, Portland, OR. secondary forest succession. In: Shugart, H.H., West,
Greene, S.E., Harcombe, P.A., Harmon, M.E. & Spycher, G. D.C. & Emanuel, W.R. (eds.) Forest succession: concepts
1992. Patterns of growth, mortality and biomass change in and applications, pp. 324-338. Springer-Verlag, New York,
a coastal Picea sitchensis-Tsuga heterophylla forest. J. NY.
Veg. Sci. 3: 697-706. Peet, R.K. 1992. Community structure and ecosystem func-
Grier, C.G. 1976. Biomass, productivity, and nitrogen-phos- tion. In: Glenn-Lewin, D.C., Peet, R.K. & Veblen, T.T.
phorus cycles in hemlock-spruce stands of the central (eds.) Plant succession: theory and prediction, pp. 103-
Oregon coast. In: Atkinson, W.A. & Zasoski, R.J. (eds.) 151. Chapman & Hall, London.
Western hemlock management, pp. 71-81. College of Peet, R.K. & Christensen, N.L. 1987. Competition and tree
Forest Resources, University of Washington, Seattle, death. Bioscience 37: 586-595.
WA, Institute of Forest Products Contr. 34A. Reiners, W.A. 1983. Disturbance and basic properties of eco-
Harcombe, P.A. 1986. Stand development in a 130-year-old system energetics. In: Mooney, H. A. & Godron, M. (eds.)
738 Acker, S.A. et al.

Disturbance and ecosystems, pp. 82-98. Springer-Verlag, Nestucca Bay, Hebo, and Dolph 7.5 minute quadrangles).
Berlin. Open File Report 96-0534, U.S. Dept. of Interior, U.S.
Runyon, J., Waring, R.H., Goward, S.N. & Welles, J.M. 1994. Geological Survey, Reston, VA.
Environmental limits on net primary productivity and Snedecor, G.W. & Cochran, W.G. 1980. Statistical methods.
light-use efficiency across the Oregon transect. Ecol. Appl. 7th ed. The Iowa State University Press, Ames, IA.
4: 226-237. Sollins, P. 1982. Input and decay of coarse woody debris in
Ruth, R.H. 1954. Cascade Head climatological data 1936 to coniferous stands in western Oregon and Washington.
1952. USDA, Forest Service, Pacific Northwest Forest Can. J. For. Res. 12: 18-28.
and Range Experiment Station, Portland, OR. Sollins P., Cline, S.P., Verhoeven, T., Sachs, D. & Spycher, G.
Ryan, M.G. 1990. Growth and maintenance respiration in 1987. Patterns of log decay in old-growth Douglas-fir
stems of Pinus contorta and Picea engelmannii. Can. J. forest. Can. J. For. Res. 17: 1585-1595.
For. Res. 20: 48-57. Spies, T.A., Franklin, J.F. & Thomas, T.B. 1988. Coarse
Ryan, M.G., Binkley, D. & Fownes, J.H. 1997. Age-related woody debris in Douglas-fir forests of western Oregon
decline in forest productivity: pattern and process. Adv. and Washington. Ecology 69: 1689-1702.
Ecol. Res. 27: 213-262. Sprugel, D.G. 1983. Correcting for bias in log-transformed
Ryan, M.G. & Waring, R.H. 1992. Maintenance respiration allometric equations. Ecology 64: 209-210.
and stand development in a subalpine lodgepole pine Sprugel, D.G. 1985. Natural disturbance and ecosystem
forest. Ecology 73: 2100-2108. energetics. In: Pickett, S.T.A. & White, P. (eds.) The
Ryan, M.G., Gower, S.T., Hubbard, R.M., Waring, R.H., ecology of natural disturbance and patch dynamics, pp.
Gholz, H.L., Cropper, W.P. & Running, S.W. 1995. Woody 335-355. Springer-Verlag, New York, NY.
tissue maintenance respiration of four conifers in Swartzman, G.L. & Kaluzny, S.P. 1987. Ecological simula-
constrasting climates. Oecologica 101: 133-140. tion primer. Macmillan Publ. Co., New York, NY.
Snavely, P.D. Jr., Niem, A., Wong, F.L., MacLeod, N.S. & Wiley, K.N. 1978. Site index tables for western hemlock in the
Calhoun, T.K. 1996. Geologic map of the Cascade Head Pacific Northwest. Weyerhaeuser Forestry Paper No. 17,
Area, northwestern Oregon Coast Range (Neskowin, Western Forestry Research Center, Weyerhaeuser Com-
pany, Centralia, WA.

Received 4 March 1999;

Revision received 18 November 1999;
Final revision received 23 May 2000;
Accepted 30 May 2000.
Coordinating Editor: J.S. Denslow.