Download Ecological impacts of different harvesting scenarios for temperate

Forest Ecology and Management 252 (2007) 52–66
www.elsevier.com/locate/foreco
Ecological impacts of different harvesting scenarios for temperate
evergreen rain forest in southern Chile—A simulation experiment
Nadja Rüger a,*, Álvaro G. Gutiérrez a, W. Daniel Kissling a,
Juan J. Armesto b,c, Andreas Huth a
a
UFZ Helmholtz Centre for Environmental Research, Department of Ecological Modelling, Permoserstr. 15, 04318 Leipzig, Germany
b
IEB, Departamento de Ciencias Ecológicas, Facultad de Ciencias, Universidad de Chile, Casilla 653, Santiago, Chile
c
Centre for Advanced Studies in Ecology & Biodiversity, Pontificia Universidad Católica de Chile, P.O. Box 114-D, Santiago CP 6513677, Chile
Received 7 February 2007; received in revised form 8 June 2007; accepted 8 June 2007
Abstract
Current forestry practices in Chile largely rely on exotic tree plantations, and limited management experiences are available for the species-rich
native evergreen rain forests. Yet, conservationists and forest scientists call for sustainable management of native forests as an alternative to
plantations so as to maintain important ecosystem services. We parameterised the process-based forest growth model FORMIND for a Valdivian
coastal temperate rain forest in Chiloé Island, Chile, to assess the ecological implications of different logging practices including selective logging
and strip-cutting. We tested the model by comparing simulation results with field data from the study site and carried out an extensive sensitivity
analysis to explore the impacts of parameter values on model results. Simulated logging practices were compared in regard to expected timber
harvest and long-term impacts on forest structure and composition.
Results showed that highest harvests could be achieved when strip-cutting was applied, because it promoted the regeneration of the relatively
light-demanding and fast-growing Eucryphia cordifolia. However, forest structure and composition were severely altered by this practice. In
contrast, selective logging, although providing lower harvests, better conserved old-growth forest structure and composition. Canopy gaps created
by selective logging were not large enough to ensure regeneration of E. cordifolia, but favoured the shade-tolerant Laureliopsis philippiana.
Overall, the similarity of logged stands to undisturbed forest decreased linearly with increasing harvesting intensity. Management strategies that
rely on native species and keep an uneven-aged forest structure ensure the maintenance of native biodiversity, protect ecosystems from exotic
species invasions, and promote the conservation of biotic interactions essential for tree reproduction.
# 2007 Elsevier B.V. All rights reserved.
Keywords: Selective logging; Sustainable forest management; Forest model; FORMIND; Chile; Valdivian temperate rain forest
1. Introduction
Temperate evergreen rain forests cover more than
40,000 km2 along the western margin of southern South
America, mainly in Chile and small adjacent areas of Argentina
(Donoso, 1998; CONAF-CONAMA, 1999). The Valdivian
forest subtype has been classified among the 200 biologically
most valuable and critically endangered ecoregions of the
world (Olson and Dinerstein, 1998). Today, southern temperate
forests are severely threatened by land conversion to grassland
or monospecific plantations of Pinus radiata or Eucalyptus
* Corresponding author. Tel.: +49 341 235 2598; fax: +49 341 235 3500.
E-mail address: [email protected] (N. Rüger).
0378-1127/$ – see front matter # 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.foreco.2007.06.020
species. Exotic tree plantations have increased substantially
since 1974, when the Chilean Government began subsidising
the cost of planting and management (Lara and Veblen, 1993;
Donoso and Lara, 1999). Today conifer and eucalyptus
plantations in Chile cover nearly 21,000 km2 (CONAFCONAMA, 1999) and sustain more than 95% of timber and
wood pulp production for domestic use and export. It has been
estimated that 55% of the native forests have been substituted
by exotic plantations (Lara et al., 1999).
Before being cleared and substituted by exotic tree
plantations, native forests have often been degraded through
high-grading, i.e., selective cutting of the biggest and best trees,
leaving unhealthy, twisted and small trees behind (Neira et al.,
2002). To change this situation, Chilean foresters, scientists and
environmentalists have called for a new forestry legislation that
N. Rüger et al. / Forest Ecology and Management 252 (2007) 52–66
could foster the sustainable management of remaining native
forests as a way to maintain ecosystem services (Lara et al.,
2003). Native forests deliver a number of ecosystem goods and
services including clean water, timber, fuelwood, non-timber
forest products, cultural, religious and recreational values, and
harbour a large fraction of native and endemic biodiversity
(Armesto et al., 1999a; Lara et al., 2003; Smith-Ramı́rez et al.,
2005). Sustainable management of native forests could increase
their economic value and prevent further loss to other land uses.
While largely monospecific stands of Nothofagus pumilio
and Nothofagus betuloides at higher latitudes (45–558S) offer
promising perspectives for sustainable timber harvest (e.g.
Armesto et al., 1996), the management of Valdivian temperate
rain forest (VTRF, Veblen et al., 1983) is considered more
difficult (e.g. Arroyo et al., 1999). Doubts arise from the higher
structural complexity and tree species richness of VTRF, and its
stronger dependence on biotic interactions for pollination and
seed dispersal (Armesto et al., 1996; Smith-Ramı́rez et al.,
2005). During recent decades, pilot silvicultural experiments
have explored the recovery of evergreen rain forests from
timber harvest and strip-cutting and selective logging have been
identified as feasible management options (e.g. Donoso, 1989b;
CONAF-CONAMA, 1999; Donoso et al., 1999; Lara et al.,
2000). However, the design, execution and monitoring of large
silvicultural experiments remains costly and operationally
difficult (Armesto et al., 1999c). Thus, modelling approaches
which are complementary to experimental studies are needed to
assess the long-term consequences of different management
options and to provide guidelines for forest managers aiming at
reconciling conservation and production objectives (Lindenmayer and Franklin, 2002).
Here, we apply the process-based forest growth model
FORMIND (e.g. Köhler and Huth, 1998; Köhler et al., 2001) for
the first time to Chilean evergreen rain forests to explore
potential management strategies. FORMIND calculates the
carbon balance for each individual tree on the basis of the light
53
environment in the forest. Hence, the model allows for the
detailed incorporation of different logging treatments. The
simulated logging strategies are either practiced by small forest
owners or companies (selective logging) or were suggested to
promote the regeneration of light-demanding species after
harvesting (strip-cutting, Donoso, 1989b).
The objectives of this study are (1) to parameterize
FORMIND for VTRF in Chile to evaluate the ability of the
model to reproduce observed forest characteristics, and to
explore the impact of model parameters on simulation results
by carrying out an extensive sensitivity analysis, and (2) to
compare different logging strategies in regard to expected
timber harvest and long-term consequences for the structure
and species composition of the managed forests. Our
simulation results highlight opportunities for the sustainable
use of native forest resources and for the conservation of
biodiversity outside protected areas (Franklin, 1993; Armesto
et al., 1998).
2. Materials and methods
2.1. Study area
The study site used as a reference for the simulations was a
large remnant (200 ha) of VTRF located in Guabún, Chiloé
Island, Chile (418500 S), about 30 km northwest of Ancud
(Fig. 1). The prevailing climate is wet-temperate with a strong
oceanic influence (Di Castri and Hajek, 1976). Rainfall occurs
throughout the year. The nearest meteorological station in
Punta Corona (418470 S, 738520 W) has an annual average of
2444 mm of rainfall and a mean annual temperature of 10.7 8C.
Mean maximum and minimum monthly temperatures are
13.8 8C (January) and 8.3 8C (July).
Floristically, this forest type is dominated by Eucryphia
cordifolia (Eucryphiaceae), Aextoxicon punctatum (only member of the endemic Aextoxicaceae), Laureliopsis philippiana
Fig. 1. Location of the study area Guabún in northern Chiloé Island, Chile. Old-growth forests are shown in dark grey.
54
N. Rüger et al. / Forest Ecology and Management 252 (2007) 52–66
Table 1
Density, basal area, and importance value ((% of density + % of basal area)/200%) of tree species in the Valdivian temperate rain forest of Guabún, northern Chiloé
Island, Chile
Species
Family
Aextoxicon punctatum
Eucryphia cordifolia
Laureliopsis philippiana
Myrtaceaea
Other speciesb
Aextoxicaceae
Eucryphiaceae
Atherospermataceae
Myrtaceae
Total
a
b
Density (N ha1)
Basal area (m2 ha1)
116
72
432
976
319
11.1
40.7
13.7
36.9
2.6
1915
104.9
Importance value
0.08
0.21
0.18
0.43
0.10
1
Includes Amomyrtus luma, A. meli, Luma apiculata, Myrceugenia ovata, M. planipes.
Includes Caldcluvia paniculata, Dasyphyllum diacanthoides, Drimys winteri, Ovidia pillo-pillo, Rhaphitamnus spinosus.
(Atherospermataceae), and several myrtaceous tree species
(Table 1). Dendrochronological data of oldest cohorts show no
evidence of catastrophic disturbance, such as fire or stand-scale
logging for at least 400 years (Gutiérrez et al., unpublished
manuscript). Stand structure and floristic composition are similar
to other coastal old-growth VTRF stands in Chiloé Island and
mainland sites in the Chilean Lake District (Donoso et al., 1984,
1985; Veblen, 1985; Smith-Ramı́rez et al., 2005).
2.2. Model tree species
The model focused on three main canopy-dominant tree
species and one multi-specific group of sub-canopy species. E.
cordifolia is a canopy-emergent tree species (up to 40 m height
and 2 m diameter at breast height, dbh). It is considered lightdemanding and requires medium to large-scale disturbances for
establishment (Veblen, 1985; Donoso et al., 1985). A.
punctatum and L. philippiana are both shade-tolerant species
occurring in the main canopy of the forest (e.g. Donoso et al.,
1999). They reach heights of 30 m and dbh of up to 1 m. Finally,
five tree species in the Myrtaceae family (Amomyrtus luma, A.
meli, Luma apiculata, Myrceugenia ovata, M. planipes) were
combined into one species group because of their similar
ecological characteristics. They are shade-tolerant species with
maximum heights of 15–20 m, which often dominate the lower
canopy and understorey (e.g. Donoso et al., 1999). A few other
tree species (e.g. Drimys winteri) also occur at the study site,
but they are less common in the forest and were not included in
our simulations.
2.3. The process-based forest growth model FORMIND
The forest growth model FORMIND simulates the spatial
and temporal dynamics of uneven-aged mixed species forest
stands (e.g. Köhler and Huth, 1998; Kammesheidt et al., 2001;
Köhler et al., 2001, 2003). The model simulates a forest (in
annual time steps) of up to a few hundred hectares as a mosaic
of interacting grid cells of 20 m 20 m, which is the
approximate crown size of a large mature tree. The model is
individual-oriented, i.e., small trees are treated as cohorts of
identical individuals while larger trees are modelled individually. Light availability is considered the main driver of
individual tree growth and forest succession as rainfall in the
study area is high (e.g. Pacala et al., 1994). Nutrient availability
is considered to be homogeneous at the stand-scale. Within
each grid cell all trees compete for light and space following the
gap model approach (Shugart, 1984). For the explicit modelling
of the competition for light, each grid cell is divided into
horizontal layers. In each height layer the leaf area is summed
up and the light climate in the forest interior is calculated via an
extinction law. The carbon balance of each individual tree is
modelled explicitly, including the main physiological processes
(photosynthesis, respiration) and litter fall. Growth process
equations are modified from the models FORMIX3 and
FORMIX3-Q (Ditzer et al., 2000; Huth and Ditzer, 2000,
2001).
Allometric functions and geometrical relations are used to
calculate above-ground biomass, tree height, crown diameter
and stem volume from the dbh of the tree. Tree mortality can
occur via self-thinning in densely populated grid cells,
senescence, gap formation by large falling trees, or medium-scale windthrows (800–1600 m2), which are observed in
the study area (Veblen, 1985; Gutiérrez, personal observation).
Gap formation links neighbouring grid cells. Tree regeneration
rates are maximum rates of recruitment of small trees at a dbh
threshold of 1 cm, with seed losses through predation and
seedling mortality being incorporated implicitly. These
maximum recruitment rates are reduced by shading. There
is no inter-annual variability of climatic conditions in the
model. A detailed model description is given in Supplementary
data.
2.4. Model parameterisation
The model was parameterised on the basis of values given in
the literature, data from the study site and expert estimations.
All parameters, their description, values, and sources are
summarised in Table 2.
Parameters, for which no direct measurements or literature
values were available, were estimated based on different but
related field observations. The form factor was adjusted such
that simulated stem volume corresponded to empirical volume
functions (Emanuelli and Pancel, 1999; Salas, 2002; see
Sections 2.5 and 3.1.2). A genetic algorithm was used to fit
parameters of maintenance respiration parameters such that
simulated maximum diameter growth for the different species
(i.e., under full sunlight) generated an upper envelope for
measured diameter increment (see Sections 2.5 and 3.1.1).
N. Rüger et al. / Forest Ecology and Management 252 (2007) 52–66
55
Table 2
Parameters of FORMIND2.3 for Valdivian temperate evergreen rain forest in northern Chiloé Island, Chile
Parameter
Description
AP
EC
LP
Environmental parameters
Average irradiance above canopy
I0
[mmol(photons) m2 s1]
k
Light extinction coefficient
[m2 ground m2 leaf]
Tree geometry parameters
Maximum diameter [m]
Dmax
Hmax
Maximum height [m]
h0
Parameter of diameter–height
relationship [cm m1]
h1
Parameter of diameter–height
relationship [m]
cd
Parameter of diameter–crown
diameter relationship [m cm1]
f
Form factor
c
L
Crown depth fraction
Leaf area index per tree
[m2 leaf m2 ground]
Fraction of stem wood
biomass to total biomass
sw
Biomass production parameters
r
Wood density [t m3]
pmax
a
rg
r0
r1
r2
m
codm
Maximum rate of photosynthesis
[mmol(CO2) m2 s1]
Light-use efficiency [mmol(CO2)
mmol(photons)1]
Parameter of growth respiration
Parameter of maintenance respiration
Parameter of maintenance respiration
Parameter of maintenance respiration
1
30
Mortality parameters
Basic mortality [year1]
mb
mmax
Maximum mortality of small
trees [year1]
Diameter up to which mortality
Dmort
is increased [m]
Dfall
Minimum diameter of
falling trees [m]
pfall
Probability of dying
trees to fall [%]
Technical parameters
Diameter of ingrowing trees [m]
Ds
a
Patch size [m2]
Dh
Step width of vertical
discretization [m]
Lovengreen (unpublished data)
0.5
Pierce and Running (1988) and
Brown and Parker (1994)
1
30
0.7
20
40.1
27.7
0.4
0.35
1.2
41.6
Reference
700
2
40
48.7
Brun (1969), Echeverrı́a
(unpublished data)
and Gutiérrez
(unpublished data)
0.12
0.4
0.4
0.25
4
Estimated based on Emanuelli
and Pancel (1999) and
Salas (2002)
Estimated
Saldaña and Lusk (2003)
0.7
Köhler (2000)
0.59
0.72
0.55
1.15
5.6
10
6.4
7
Pérez-Galaz (1983) and
Diaz-vaz et al. (2002)
Lusk et al. (2003)
0.25
0.2
0.2
0.35
Estimated
0.1
0.0
0.0001
0.11
0.0
0.0
0.13
0.0008
0.0
0.1
0.003
0.0
Ryan (1991)
Estimated based on diameter
increment data (Gutiérrez et al.,
in preparation)
0.2
Transmission coefficient of leaves
Parameter for conversion in organic
dry matter [t mmol(CO2)1]
Recruitment parameters
Minimum light intensity for
Imin
establishment [% of I0]
Imax
Maximum light intensity for
establishment [% of I0]
Maximum recruitment rates of
Nmax
small trees [ha1 year1]
MY
0.1
0.6344e12
Larcher (2001)
Larcher (2001)
1
70
3
1
90
100
90
100
50
100
150
250
0.01
0.006
0.006
0.004
0.12
0.1
Estimated based on
inventory data (Gutiérrez,
unpublished data)
Estimated based on Lusk (2002),
Lusk and del Pozo (2002),
Coomes et al. (2003), and
inventory data (Gutiérrez,
unpublished data)
0.45
30
0.01
400
0.5
AP, Aextoxicon punctatum; EC, Eucryphia cordifolia; LP, Laureliopsis philippiana; MY, myrtaceous species.
–
–
–
56
N. Rüger et al. / Forest Ecology and Management 252 (2007) 52–66
Recruitment and mortality parameters were adjusted such
that the simulated forest structure and composition 400 years
after a large-scale disturbance resembled the inventory data
(see Sections 2.5 and 3.1.3). Using available data on average
light environments occupied by juveniles (e.g. Lusk, 2002;
Lusk and Kelly, 2003; Lusk et al., 2006), it was not possible to
reproduce the observed forest structure and composition of the
inventory data. While the data confirm that A. punctatum, L.
philippiana and the myrtaceous species can establish and
survive at low light levels, E. cordifolia had to be assigned a
much higher minimum light intensity for establishment (70%)
to produce episodic recruitment as observed in the inventory
data.
To our knowledge, there are no data available on adult tree
mortality for this forest type and mortality parameters had to be
estimated in accordance with inventory data. We assumed a
constant mortality rate for trees >15 cm dbh. Smaller trees
experience an additional mortality that declines linearly with
dbh. This pattern of tree mortality has been found in a global
data set as well as in New Zealand forests (Coomes et al., 2003).
Experiments with artificial light environments showed that
seedling mortality of M. planipes and A. luma was low
compared to the other species (Lusk and del Pozo, 2002).
Annual juvenile (<100 cm tall) mortality under field conditions
was reported for E. cordifolia (3.7%) but sample sizes were low
and the other species showed no mortality (Lusk, 2002).
2.5. Model evaluation
To test the ability of the model to reproduce observed forest
characteristics, we compared simulation results with field data
at different levels. At the level of individual trees, we compared
simulated and measured diameter increments and stem
volumes. At the species-level, we compared inventory data
for the old-growth forest stand in Chiloé with the simulated
forest structure and species composition 400 years after a largescale disturbance.
Inventory data consisted of six transects (2 transects of
20 m 50 m and 4 transects of 5 m 100 m, together 0.4 ha)
within a forest stand that has not been affected by large-scale
disturbances for at least 400 years (Gutiérrez et al., unpublished
manuscript). All trees >5 cm dbh were measured and their
species identified. To estimate radial growth, 87 randomly
located trees (38 for A. punctatum, 35 for E. cordifolia, 14 for L.
philippiana) were cored and cores were analysed with standard
dendrochronological techniques (Gutiérrez et al., 2004;
Gutiérrez, unpublished manuscript). Annual diameter increment was approximated by multiplying tree ring width by two.
To simulate an upper envelope for actual diameter growth for
each species in the model, we simulated the growth (from
seedling to maximum size) of a single tree of each species under
full sunlight conditions.
To assess the reliability of stem volume values of single trees
calculated by FORMIND using geometrical relationships, we
compared them to empirical volume functions for the three
main tree species and the Myrtaceous species group (Emanuelli
and Pancel, 1999; Salas, 2002).
To study forest succession and the dependence of the current
forest composition on natural medium-scale canopy disturbances, we started the simulation from a treeless area that is
suitable for the establishment of all species. We simulated
forest dynamics during 1500 years with and without stochastic
medium-scale canopy disturbances. Ten simulations were
carried out for an area of 1 ha, and the means (over the 10
simulations) and standard deviations were calculated for stem
numbers and basal area of the three tree species and one species
group.
We carried out an extensive sensitivity analysis to explore
the impact of model parameters on model results. We used the
software package SimLab2.2 (Simulation environment for
uncertainty and sensitivity analysis, Saltelli et al., 2004). We
applied the extended Fourier Amplitude Sensitivity Test
method (FAST) to compute first order sensitivity indices
(Saltelli et al., 2000). This method is a variance-based method
that avoids making the assumption of a nearly linear relation
between model parameters and output. If Y is the model output
and Xi the parameter of interest, then the first order sensitivity
index Si = (V(E(Y/Xi))/V(Y)) indicates the amount of variance
that would be removed from the total output variance if the
parameter’s true value was known, and hence the relative
importance of a given model parameter for a given model
prediction (Tarantola et al., unpublished manuscript).
We simultaneously varied all model parameters by 20% of
their standard value (Table 2). Simulations were run for 119770
parameter combinations for 1 ha and 1000 years. The model
was initialised with inventory data from the study site
(extrapolated to 1 ha). Selected model predictions are
aggregated forest characteristics such as stem volume, biomass,
basal area, stem number, and leaf area index of the entire forest
stand, as well as basal area, stem numbers, maximum diameter
increment and maximum age for each species. Model
predictions for old-growth conditions were assessed at 10
points in time, every 50 years from simulation time 550–1000
to allow the model to reach dynamic equilibrium, and then
averaged.
2.6. Logging scenarios
We simulated three logging treatments (selective logging
with and without retention of large old trees and strip-cutting).
Large-scale clear-cutting is not an acceptable silvicultural
treatment for VTRF due to problems with soil erosion, nutrient
losses, and tree regeneration requirements, and was therefore
not considered in the simulations (Donoso, 1989b; Armesto
et al., 1999b,c). Within each logging treatment, we varied both
the extracted wood volume and logging cycle (in the case of
selective logging) and the logging cycle only (in the case of
strip-cutting). The model was initialised with inventory data
from the study site. Inventory data from 0.4 ha were
extrapolated to an area of 9 ha. As tree locations were not
measured in the field, we assumed a random distribution of
small individuals. Additionally, we avoided that more than one
individual >70 cm dbh was assigned to any 20 m 20 m plot.
Logging cycles were repeated over a period of 400 years.
N. Rüger et al. / Forest Ecology and Management 252 (2007) 52–66
2.6.1. Selective logging
Here, selective logging refers to the extraction of trees with a
dbh of 50–100 cm. The two simulated selective logging
strategies differed in the way large, old and probably senescent
trees were treated. In the first case (with retention of large
trees), trees >1 m dbh were left standing, as they often exhibit
heart rot and do not provide valuable timber, but they provide
habitat and resources for birds and other species (e.g. Dı́az
et al., 2005). In the second case of selective logging (without
retention of large trees), all trees > 1 m dbh were felled before
the harvesting took place to enhance the growth of potential
future crop trees by reducing shading. These trees were left in
the forest and not regarded as harvested timber. We varied the
time between two sequential harvests (logging cycle) from 10
to 50 years. For each logging cycle we also varied the annual
volume of harvested wood (harvest aim) from 1 to 10 m3 ha1.
For a 10-year logging cycle this corresponds to harvesting 10–
100 m3 ha1 and for a 50-year logging cycle to 50–
500 m3 ha1. When, at any time, the harvestable volume was
lower than the harvest aim, the logging operation was skipped.
Within the dbh range of 50–100 cm, the largest trees were
always logged first. Logging damages were simulated as direct
damage by the falling tree and additional damage due to
skidding. We assumed reduced-impact logging where falling
trees are directed to existing gaps if possible. No damage
occurred to trees >50 cm dbh. Skidding damages were
assumed to increase with increasing volume of extracted wood
(Donoso et al., 1999; Fig. 2). Donoso et al. (1999) report that
16% of trees >10 cm dbh were severely damaged by harvesting
(cutting and skidding) in evergreen temperate rain forest in
Chile, and another 29% were slightly damaged. The slightly
damaged trees might be more susceptible to pathogens and
experience elevated mortality in the future. Thus, we assumed
rather high skidding damages.
2.6.2. Strip-cutting
Strip-cutting has been suggested to promote the regeneration
of light-demanding tree species after harvesting VTRF
(Donoso, 1989b). Strip-cutting was simulated by clear-cutting
all trees in a 20 m wide strip in each ha. Thus, 20% of the forest
area was harvested in each logging operation and after five
logging operations the first strip was logged for a second time.
The return time to each strip was varied from 50 to 150 years.
Skidding damage was assumed to be only 10%, because
logging strips can be used to extract trees from the stand. No
damage occurred to trees >50 cm dbh.
2.6.3. Assessment of logging scenarios
To evaluate the economic and ecological consequences of
logging scenarios, we used four response variables, namely
mean annual harvest (in m3 stem volume), an index of structural
change (ISC), an index of compositional change (ICC), and the
stand leaf area index (LAI). Mean annual harvest is an
economic indicator of profit from timber harvest, ISC and ICC
are ecological indicators of changes in forest structure and
species composition, and LAI is an environmental indicator of
erosion risk in this region with high rainfall.
57
Fig. 2. Proportion of damaged trees due to skidding operations assumed in
model simulations.
Mean annual harvest was calculated for a period of 400 years
under each simulated logging scenario.
ISC was calculated as
ISC ¼
3
1X
jx̄si x̄ci j
;
3 i¼1
x̄ci
i.e., the difference in average stem numbers (over the last 100
years of the simulation, i.e., time steps 300–400) in three
different diameter classes (x̄si , s1: 5–50 cm, s2: 50–100 cm,
and s3: 100–200 cm dbh) in logged and control forest. In the
control forest no logging is applied (x̄ci , i = 1–3).
ICC indicates the change in relative importance of the three
tree species and one species group (i) in logged forest compared
to the control forest based on importance values (IV),
calculated as
1
bai
ni
IVi ¼
þ
;
2 batotal ntotal
i.e., the sum of relative basal area (ba, m2 ha1) and relative
density (n, trees ha1) of the focal species in relation to all
species (total). ICC was calculated as
ICC ¼
4
1X
jIVsi IVci j
;
4 i¼1
IVci
i.e., summing the differences between average IV (over the last
100 years of the simulation, i.e., time steps 300–400) of species
i in the logging scenario (IVsi ) and the unlogged control forest
(IVci ) relative to its mean IV in the control forest (‘species’ are
A. punctatum, E. cordifolia, L. philippiana, and the Myrtaceae
species group).
LAI values were directly determined from model output and
averaged over time steps 300–400.
To contrast economic benefit and ecological impact of a
logging scenario, we calculated an ecological index (EI), which
measures the similarity of a logged forest to undisturbed oldgrowth forest. EI takes into account the ecological variables
ISC, ICC, and LAI, and the number of old trees (>1 m dbh,
OLD) which were divided by the maximum value obtained
58
N. Rüger et al. / Forest Ecology and Management 252 (2007) 52–66
Fig. 3. Simulated (lines) and measured (dots) annual diameter increment. Simulations were carried out under full light conditions (700 mmol(photons) m2 s1) and
represent maximum potential growth. Field data are derived from radial growth measurements (Gutiérrez et al., unpublished manuscript).
from all logging scenarios (ISCmax, ICCmax, LAImax, OLDmax),
respectively, and summed up as follows:
EI ¼
1
4
1
ISC
ISCmax
ICC
LAI
OLD
þ 1
þ
þ
ICCmax
LAImax OLDmax
For reference purposes, we also calculated EI for a simulated
undisturbed old-growth forest, bare ground, and a fictitious
Eucalyptus plantation. For bare ground, LAI and OLD are 0,
whereas ISC and ICC are 1. For the Eucalyptus plantation we
assumed a mean annual volume increment of 22 m3 ha1, as an
average between reported average (11 m3 ha1) and maximum
(32 m3 ha1) annual volume increments of Eucalyptus globulus
plantations in the study region (Geldres and Schlatter, 2004).
Average density of the plantations was 1356 stems ha1 with
all stems being in the smallest diameter class (5–50 cm). The
resulting ISC was 0.73. ICC cannot be calculated with our
formula, because Eucalyptus is not a natural component of the
native forest, and ICC would theoretically be infinite if another
term was added to the index. To be able to calculate EI for the
Eucalyptus plantation we assigned ICC an arbitrary value of 2.
LAI of the plantation is assumed to be 4.5 (Battaglia et al.,
1998).
3. Results
3.1. Model evaluation
3.1.1. Diameter increment
Simulated maximum annual diameter increments (SMDI)
and empirical diameter increments (Gutiérrez, unpublished
data) for the three main canopy species and the Myrtaceous
species group are shown in Fig. 3. SMDI should correspond to
the mean diameter increment of trees growing under full
sunlight in the field. Outliers in field measurements are
attributed to absent tree rings or measurement errors. SMDI of
A. punctatum corresponded well with the highest values
recorded in field data. For E. cordifolia, SMDI matched well the
maximum values measured for small diameters. For intermediate diameters, SMDI was lower than highest values from
field data. For diameters >140 cm, no field data were available.
For L. philippiana, SMDI corresponded well with highest
observed values. No field data were available for myrtaceous
species at the study site. However, SMDI of 6 mm year1
compared well with measured maximum diameter increment of
6.2 mm year1 (maximum radial increment was 3.1 mm
year1) for myrtaceous species from Puyehue National Park
in the Andean Range, Chile (Pollmann and Veblen, unpublished
data). In general, we recognize that SMDI values are optimistic
because SMDI lies consistently at the upper limit of field data
and that, as a consequence, forest growth might be overestimated by the model.
3.1.2. Stem volume
For the three main canopy species and the Myrtaceous
species group, stem volumes calculated by FORMIND
corresponded well with empirical volume functions (Emanuelli
and Pancel, 1999; Salas, 2002; Fig. 4).
3.1.3. Simulated long-term forest dynamics
The simulation of long-term forest dynamics over 1500
years in the absence of medium and large-scale disturbances
(e.g. windthrow of multiple trees) is shown in Fig. 5(A and B).
Total tree density reached a dynamic equilibrium within the
first 100 years of simulation (Fig. 5A). E. cordifolia tended to
disappear from the forest after approximately 800 years. A.
N. Rüger et al. / Forest Ecology and Management 252 (2007) 52–66
59
Fig. 4. Stem volume of single trees calculated with FORMIND and empirical volume functions from Emanuelli and Pancel (1999) and Salas (2002).
punctatum stabilised at relatively low density, L. philippiana at
intermediate density, and the myrtaceous species reached a high
density. In terms of basal area, the first 400 years of forest
succession were dominated by E. cordifolia, which was then
replaced by the shade-tolerant species (Fig. 5B). The
myrtaceous species group accounted for the highest basal area
in old-growth forest, followed by L. philippiana and A.
punctatum.
According to field data, the forest at the study site has not
been affected by disturbances other than single tree falls for
about 400 years (Gutiérrez et al., unpublished manuscript).
Assuming that the forest at the study site is 400 years old, the
Fig. 5. Simulation of forest dynamics following large-scale disturbance (e.g. clear-cut or stand-devastating event) without (A and B) and with (C and D) stochastic
windthrow events. Mean and standard deviation of 10 simulations for stem numbers (A and C) and basal area (B and D) for all individuals 5 cm dbh. Simulations
were run for 1 ha and 1500 years. Inventory data from the study site (estimated age: 400 years) are shown in the grey bars on the right side of each chart.
60
N. Rüger et al. / Forest Ecology and Management 252 (2007) 52–66
simulated forest can be compared to the inventory data 400
years after the beginning of the succession (Fig. 5(A and B)).
The symbols in the grey bars on the right side of each chart
(Fig. 5) represent inventory data. Simulated stem numbers and
basal areas after 400 years (left grey bars) correspond well with
inventory data for all species, suggesting that the main trends of
forest dynamics are captured in our model.
Incorporating medium-scale disturbances (e.g. multiple tree
falls) in the model changes long-term forest dynamics (Fig. 5(C
and D)). This is to the same simulating forest dynamics on a
larger spatial scale, where forest patches representing different
successional stages co-occur. Again, total stem numbers and
total basal area reached a dynamic equilibrium after
approximately 100–200 years of simulation (Fig. 5(C and
D)). Stem numbers levelled off at about 1950 trees ha1 and
total basal area reached 95 m2 ha1. At the beginning of
succession, the forest was dominated by myrtaceous species in
terms of stem numbers and by E. cordifolia in terms of basal
area. Stem numbers of the main canopy species and the
Myrtaceous species group rapidly reached a dynamic
equilibrium, whereas basal areas continued to change for
about 1000 years. Again, there was a trend towards
displacement of E. cordifolia by shade-tolerant species. But
in contrast to forest dynamics without medium-scale disturbances, E. cordifolia now persisted over the entire simulation
period, with a few large E. cordifolia trees accounting for a
large proportion of the stand’s basal area. As a consequence of
the higher spatial and temporal variability of the forest due to
the medium-scale disturbances, the standard deviation of stem
numbers and basal area was larger than in the simulations
without disturbances.
3.1.4. Sensitivity analysis
Aggregated forest characteristics, namely total stem volume
(SV total), biomass (BM total), basal area (BA total), stem
number (N total), and leaf area index (LAI) were most strongly
affected by the parameter of the linear relationship between
stem diameter and crown diameter (cd), the light extinction
coefficient (k), the proportion of stem wood biomass to total
biomass (sw), and the LAI of a single tree (L) (Fig. 6). Average
irradiance above the canopy (I0) was a parameter of
intermediate importance. Total stem number was additionally
influenced by the parameters of increased mortality of small
trees (mmax, Dmort), and total basal area by several characteristics of the myrtaceous species (form factor f, maximum
photosynthetic capacity pmax, light-use efficiency a, wood
density r, maintenance respiration parameter r0). Basal area of
the different species (BAAP, BAEC, BALP, BAMY) was most
strongly affected by production parameters ( pmax, a, r).
Additionally, f and, to a lower extent, r0, had an effect. Stem
numbers of the different species (NAP, NEC, NLP, NMY) were
most strongly affected by production and morphological
parameters (a, r, f, cd). The number of individuals of
myrtaceous species (NMY) was also significantly influenced by
the size-dependent component of mortality (mmax, Dmort). Only
recruitment rate of A. punctatum (Nmax AP) had an impact on
stem numbers. Stem numbers of E. cordifolia depended on
minimum light intensity required for its establishment (Imin EC).
For all species, maximum diameter increment was most
strongly affected by cd. A parameter of the diameter–height
relation (h1), f, L, sw, pmax, r, and r0 had intermediate effects on
maximum diameter increment. For maximum age, again the
parameter with the strongest influence was cd. Mortality rates
(mb) and sw were of minor importance.
3.2. Logging scenarios
3.2.1. Mean annual harvest
We simulated wood extraction for three logging treatments
(selective logging with and without retention of large old trees
and strip-cutting). For the selective logging scenarios, we
varied the logging cycle from 10 to 50 years and the harvest aim
from 1 to 10 m3 ha1 on an annual basis (Fig. 7). Up to about
4 m3 ha1 year1, the harvest aim could be met by all
scenarios. Selective logging scenarios with large tree retention
reached a limit of sustained wood extraction at
6.5 m3 ha1 year1 when the harvest aim was 180 m3 ha1
and the logging cycle was 20 years (in this case six harvesting
operations had to be omitted because the harvest aim could not
be reached). When large trees were removed prior to the
simulation of logging scenarios, up to 8 m3 ha1 year1 could
be harvested when the harvest aim was 160 m3 ha1 and the
logging cycle 20 years. Strip-cutting achieved a higher annual
harvest that ranged from 6 m3 ha1 year1 for a logging cycle
of 150 years to 13.4 m3 ha1 year1 for a logging cycle of 60
years.
3.2.2. Forest composition
The impacts of the logging scenarios on forest composition
were measured by importance values (IV = relative stem
numbers plus relative basal area) for the main canopy species
and the Myrtaceous species group. Logging scenarios had
strong effects on IVs of E. cordifolia and L. philippiana (Fig. 8).
IVs of E. cordifolia were more than twice as high under stripcutting as under selective logging. This increase occurred at the
expense of L. philippiana, for which IVs in the strip-cutting
scenarios halved compared to selective logging. The inverse
pattern was observed within the selective logging scenarios for
increasing levels of wood extraction. While E. cordifolia’s IVs
decreased, IVs of L. philippiana increased. The IVs of A.
punctatum remained relatively constant under the different
logging scenarios. The myrtaceous species showed the same
trends as L. philippiana, but to a lower extent.
3.2.3. Forest structure
To assess the impact of logging on forest structure, we
distinguish three diameter classes (5–50, 50–100, 100–
200 cm dbh). The number of smaller trees (5–50 cm dbh)
increased for increasing levels of wood extraction (Fig. 9A).
The number of larger trees (50–100 cm dbh) remained
relatively constant for low levels of wood extraction (up to
5 m3 ha1 year1), but decreased sharply for higher levels of
wood extraction (Fig. 9B). For strip-cutting, the decrease of the
number of large trees occurred at higher levels of wood
N. Rüger et al. / Forest Ecology and Management 252 (2007) 52–66
61
Fig. 7. Mean annual harvest for three logging strategies. For selective logging,
logging cycle varied from 10 to 50 years, harvest aim (i.e., amount of extracted
wood aimed at by the logging scenario) varied from 10 to 500 m3 ha1,
depending on the logging cycle. Converted to an annual basis, harvest aim
ranged between 1 and 10 m3 ha1 year1. For strip-cutting the logging cycle
was varied from 50 to 150 years.
harvest (i.e., long logging cycles) had a lower impact on forest
structure than selective logging scenarios with similar mean
annual harvests. Stand leaf area index (LAI) decreased from
about 6 for low levels of wood extraction to about 4 for
intermediate levels of wood extraction (Fig. 10C). LAI further
decreased to about 2–3 for strip-cutting with highest levels of
wood extraction.
Fig. 6. First order sensitivity indices for model parameters and selected model
predictions (SV, BM, BA, N, LAI), species composition (BA, N for the different
species), maximum diameter increment (Dinc), and maximum age (Amax).
Indices were computed with the extended FAST method (Saltelli et al.,
2000). SV, stem volume; BM, biomass; BA, basal area; N, stem number;
LAI, leaf area index of forest stand; AP, A. punctatum; EC, E. cordifolia;
LP, L. philippiana; MY, Myrtaceae. For a description of model parameters see
Table 2.
extraction (8–14 m3 ha1 year1). The number of old trees
(>1 m dbh) decreased linearly up to a mean annual harvest of
8 m3 ha1 (Fig. 9C). Beyond that threshold, no old trees
remained in the forests in the long-term, because large trees
were harvested before they attained a dbh of 1 m.
3.2.4. Ecological indices
The index of compositional change (ICC) increased with
increasing levels of wood extraction for both selective logging
scenarios (Fig. 10A). ICC was very high (0.8) for all stripcutting scenarios but remained relatively constant regardless of
the level of wood extraction. The index of structural change
(ISC) increased almost linearly with increasing levels of wood
extraction (Fig. 10B). Only selective logging scenarios with the
highest logging intensities altered forest structure more than
proportionally. Strip-cutting with comparably low mean annual
3.2.5. Ecological integrity versus harvest
With increasing harvesting intensity, the ecological index
(EI), which describes the overall similarity of the logged forest
to the undisturbed old-growth forest, decreased almost linearly
for both selective logging scenarios (Fig. 11). Hence, every
increment in the amount of harvested wood was accompanied
by a proportional increase of ecological impact. For stripcutting scenarios, EI remained relatively constant at a low level.
Compared to bare soil and a pure Eucalyptus plantation,
selective logging scenarios are still relatively benign in terms of
ecological impact. This is mainly due to the conservation of
native species composition and a higher LAI.
4. Discussion
We parameterised the forest growth model FORMIND for
the first time for a temperate evergreen rain forest in southern
Chile to assess the ecological impacts of potential harvesting
methods and to augment the scientific support for forest
management aiming at sustainable use of the species-rich
native forests. We used available knowledge about the
evergreen forest type to parameterise the model. Where data
were lacking, we applied the pattern-oriented modelling
approach to estimate model parameters in a way that field
data were reproduced (Grimm et al., 2005). Here, the
individual-based approach proved to be valuable as it allowed
for a comparison of model results with field observations on
different hierarchical levels, i.e., individual trees, populations
and the forest stand.
62
N. Rüger et al. / Forest Ecology and Management 252 (2007) 52–66
Fig. 8. Impact of logging intensity on importance values ((relative abundance + relative basal area)/2) for three species and one species group and three logging
strategies. Values of simulated unlogged old-growth forest are displayed for comparison ().
4.1. Natural forest dynamics
The results of our simulation are in agreement with
empirical observations that medium to large-size canopy
openings (e.g. due to multiple tree falls during storm events) are
necessary to maintain the characteristic tree species composition of VTRF (Veblen et al., 1981; Armesto et al., 1999b)
because E. cordifolia, a relatively light-demanding tree species,
depends on large canopy openings for regeneration (Veblen
et al., 1981; Donoso et al., 1984, 1985; Veblen, 1985). Without
such disturbances which also affect much of the advance
regeneration of shade-tolerant species, the abundance of E.
cordifolia gradually declines over successional time. Only few
large individuals remain in the forest because of E. cordifolia’s
long life span, which has been estimated to be at least 400 years
(Lusk and del Pozo, 2002; Gutiérrez et al., unpublished
manuscript). The relative importance of shade-tolerant species
increased steadily for 1000 years and in the absence of medium
to large-size disturbances they eventually dominated the forest
almost completely. Such a successional trend has also been
suggested by other field studies (e.g. Donoso et al., 1984, 1985;
Veblen, 1985). The fact that the relative basal areas of the main
canopy species and the Myrtaceous species group continued to
change for 1000 years indicates that VTRF canopy has a slow
turnover, which is greatly influenced by the long life spans of its
dominant tree species. The long life spans, in turn, are an
evidence for the low frequency of large-scale disturbances (e.g.
Veblen et al., 1980; Armesto and Figueroa, 1987; Lusk and del
Pozo, 2002).
Inventory data provide relevant information about forest
dynamics, structure and composition. However, at the scale of
inventory data available for this and other stands in the region
(0.4 ha or less), forest structure and composition are highly
heterogeneous. Available inventory data (Gutiérrez et al.,
unpublished manuscript) represent an old-growth forest with
large, old E. cordifolia trees in the canopy, but lack of E.
cordifolia regeneration. We argue here that because of their
limited scale, inventory data may not fully represent the spatial
heterogeneity at larger spatial scales, which is associated with
different disturbance histories. Therefore, field data from larger
sampling areas encompassing different successional stages and
forest gaps or inventory data from secondary forests, would
provide a better basis for model evaluation.
4.2. Ecological impacts of logging
Selective logging and strip-cutting are two largely contrasting harvesting strategies considering their potential for
continuous timber yield and ecological impacts. Strip-cutting
provided highest harvests of up to 13 m3 ha1 year1.
Maximum sustainable harvests under selective logging were
limited to 7.5 m3 ha1 year1 without large tree retention and
6 m3 ha1 year1 when large trees remained in the forest.
However, these results should be compared in relative terms
rather than as absolute values, because annual diameter
increments are rather optimistic. Strip-cutting altered forest
composition and structure to a greater extent than did selective
logging. The stands regrowing after strip-cutting were
dominated by E. cordifolia with an understorey of shadetolerant species. Selective logging, in turn, favoured L.
philippiana and the myrtaceous species, which benefited from
the small canopy gaps created by logging.
Under the strip-cutting scenarios that resulted in the highest
yields, no large, old trees (>1 m dbh) remained in the forest.
The forest stand was converted into a secondary forest with a
more homogeneous structure and larger number of small trees.
For the selective logging scenarios, the number of large, old
trees also decreased with increasing harvest intensity, but the
N. Rüger et al. / Forest Ecology and Management 252 (2007) 52–66
Fig. 9. Impact of logging intensity on forest structure (i.e., stem numbers in
three diameter classes) for three logging strategies. Values of simulated
unlogged old-growth forest are displayed for comparison ().
structural complexity of the forest was maintained better.
Mature and senescent trees are known to play important roles as
habitat for many animal and plant species, such as woodpeckers
or vascular and non-vascular epiphytes (e.g. Franklin and
Armesto, 1996; Galloway, 1999; Lindenmayer and Franklin,
2002; Muñoz et al., 2003; Dı́az et al., 2005). Therefore, the
retention of some large, old, and dead trees has been
recommended to conserve components of biodiversity that
depend on them as well as to increase the structural complexity
of logged forests (Armesto et al., 1999c).
Due to the low levels of atmospheric nutrient inputs and high
rainfall in the study area, it is essential that sustainable forest
management ensures the retention of significant canopy cover
to prevent soil erosion and maintain soil processes such as
nutrient retention and recycling (Hedin et al., 1995; Galloway
et al., 1996; Pérez, 1999). In the model, LAI of single trees was
63
Fig. 10. Impact of logging intensity on (A) the index of structural change (ISC),
(B) the index of compositional change (ICC), and (C) leaf area index for three
logging strategies. Values of simulated forest without logging are displayed for
comparison (). See Section 2 for a description of indices.
Fig. 11. Impact of logging intensity on the ecological index (EI) for three
logging strategies. Values of simulated unlogged old-growth forest, bare
ground, and a fictitious Eucalyptus plantation are displayed for comparison.
64
N. Rüger et al. / Forest Ecology and Management 252 (2007) 52–66
4. This means that as the LAI of the entire stand falls below 4,
there will be areas of the soil that are unprotected by canopy
cover and exposed to rain. Therefore, logging methods should
maintain LAIs of at least 4 for the entire forest stand to ensure a
sufficient canopy cover to prevent erosion. The three stripcutting scenarios that provided the highest timber yields did not
satisfy this minimum requirement.
One fundamental assumption of ecosystem management is
that silvicultural treatments should mimic natural disturbance
regimes to minimise negative effects on biodiversity and
ecological processes (e.g. Perry, 1994; Armesto et al., 1996;
Lindenmayer and Franklin, 2002). In the study area, the natural
disturbance regime comprises frequent single and less frequent
multiple tree falls that create canopy gaps usually smaller than
400 m2 (Armesto and Fuentes, 1988; Armesto et al., 1999b), as
well as rare windthrow events that may affect larger areas. The
simulated low-intensity selective logging mimics natural gap
creation by single tree falls, and strip-cutting can be regarded as
a form of simulating medium to large-size disturbances.
Smaller canopy gaps favour advance regeneration of shadetolerant species already present in the understorey, whereas
regeneration of E. cordifolia is enhanced only by larger gaps.
To maintain spatial heterogeneity, the creation of gaps of
different sizes could be incorporated into future forest
management planning. At the same time, this would allow
for aggregated retention of original forest structural elements,
such as old trees (Armesto et al., 1999c; Lindenmayer and
Franklin, 2002).
4.3. Sensitivity analysis
The sensitivity analysis revealed that aggregated forest
characteristics such as total stem volume, biomass, basal area,
stem numbers and LAI were most strongly influenced by model
parameters directly related to an important process of the forest
model—competition for light due to canopy shading. The
parameter of the linear relationship between stem diameter and
crown diameter (cd) and LAI per tree (L) determine the stand’s
leaf area, whereas the light extinction coefficient (k) controls
how incoming light is absorbed through the canopy.
Maximum diameter increment of the different species as
well as the species composition of the forest in terms of basal
area and stem numbers were – additionally to cd and L –
strongly affected by parameters related to biomass allocation
(e.g. wood density r, form factor f, a parameter of the diameter–
height relationship h1, stem wood fraction sw) and physiology
(maximum rate of photosynthesis pmax, respiration parameter
r0). This indicates that growth characteristics (biomass
production and allocation) of the different tree species largely
determine species composition of the forest (cf. Lusk and
Matus, 2000). Recruitment and most mortality parameters only
had minor impacts on model predictions.
A previous sensitivity analysis of FORMIND parameterised
for a tropical lowland rain forest in Venezuela yielded different
results (Kammesheidt et al., 2001). In their study, model
parameters related to recruitment and mortality most strongly
affected species composition. This difference may be due to
different types of sensitivity analyses applied. Kammesheidt
et al. (2001) varied only one model parameter at the time. For
high dimensional non-linear models, however, global sensitivity analyses such as the one applied here are more
appropriate (Saltelli et al., 2000).
First order sensitivity indices reveal which model parameters provide opportunities to significantly reduce uncertainty
of simulation results. According to our results, more detailed
information on morphological parameters, as well as on
physiological processes such as photosynthesis and respiration
are needed for Chilean trees, in order to improve the data basis
for parameter estimation of process-based forest models.
Although model results were not particularly sensitive to
mortality parameters, the credibility of model results and
understanding of forest dynamics would be substantially
improved if stand-level mortality data became available.
4.4. Limitations of the modelling approach
One potential shortcoming of the developed model
parameterisation for VTRF is the omission of the understorey
bamboo species Chusquea quila, which is known to be an
aggressive coloniser of canopy gaps. Chusquea has been
reported to inhibit tree regeneration (e.g. Donoso, 1989b;
González et al., 2002; Donoso and Nyland, 2005). Likewise, the
pioneer tree species D. winteri was not included in this study.
Both species are absent or rare at the study site but can be
expected to respond positively to large-scale logging and to
have a considerable impact on forest dynamics (e.g. Veblen,
1982; Donoso, 1989a). Future simulation studies on forest
management should incorporate these species. Future studies
should also consider disaggregating the myrtaceous species
group, because their light requirements seem to differ
substantially despite their relatively high shade tolerance
(e.g. Lusk et al., 2006).
Furthermore, the model has been parameterised for one study
site in Chiloé Island. As growth, mortality and regeneration rates
may differ at other sites due to differences in climatic variables,
disturbance regime, nutrient availability and other site conditions, simulation results have to be regarded as a reference, rather
than as predictions that hold true over large areas. An important
assumption underlying our simulations is that nutrient limitation
can be neglected. For the studied ecosystem, the nutritional
balance strongly depends on nutrient retention and recycling by
the vegetation, and therefore massive biomass extraction should
be avoided (e.g. Pérez, 1999). Thus, model results have to be
examined keeping this restriction in mind.
Simulations of forest dynamics were run for 1500 years
without taking into account changing climatic conditions. This
may be justified here because we were interested in the
theoretical implications of our model parameterisation for
long-term forest dynamics. The same argument applies to
simulations of harvesting scenarios in that they provide
qualitative comparisons based on current conditions. Another
study is currently underway, which explicitly deals with climate
change effects on the dynamics of Chilean temperate rain
forests.
N. Rüger et al. / Forest Ecology and Management 252 (2007) 52–66
5. Conclusions
We simulated a wide range of possible harvesting scenarios
in species-rich Valdivian temperate rain forest in southern
Chile. Each of the scenarios achieves a different balance
between timber yield and conservation of old-growth forest
structure and tree species composition. Overall, the similarity
of the logged forest to an undisturbed old-growth forest
decreases linearly with increasing harvesting intensity. For
areas where the least possible ecological impact is desired (e.g.
buffer zones of protected areas) selective logging combined
with large tree retention should be the recommended management approach. In gently sloping terrain with reduced erosion
risk, however, strip-cutting could achieve higher yields.
Management strategies that rely on native species and keep
an uneven-aged forest structure ensure the maintenance of
native biodiversity, protect ecosystems from exotic species
invasions, and promote the conservation of biotic interactions
essential for tree reproduction.
Acknowledgements
We wish to thank C. Echeverrı́a, C. Lovengreen, C. Lusk, A.
Ortega, and W. Pollmann for provision of data. A. Altamirano
and C. Zamorano helped with literature search, and L.
Kammesheidt provided a literature review. NR is grateful to
C. Donoso, A. Lara, and I. Dı́az for discussions on forest
dynamics, V. Grimm and several anonymous reviewers for
helpful comments on the manuscript, J. Groeneveld for
important discussions on modelling issues, and J. Ávila for a
joyous walk through the Chilean woods. The work was funded
by the European Commission; project BIOCORES [ICA4-CT2001-10095]. Sampling of inventory data was partially funded
by Fondecyt-Fondap [1501-0001], and by Millennium Scientific Initiative, Chile. This is a contribution to the research
programme of Senda Darwin Biological Station, Chiloé, Chile.
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at doi:10.1016/j.foreco.2007.06.020.
References
Armesto, J.J., Figueroa, J., 1987. Stand structure and dynamics in the rain forest
of Chiloé Archipelago, Chile. J. Biogeogr. 14, 367–376.
Armesto, J.J., Fuentes, E.R., 1988. Tree species regeneration in a mid-elevation,
temperate rain forest in Isla de Chiloé, Chile. Vegetatio 74, 151–159.
Armesto, J.J., Arroyo, M.T.K., Peñaloza, A., 1996. Condiciones para la
sustentabilidad ecológica del manejo de bosques y el proyecto ‘Rı́o Condor’. An. Inst. Patagon., Ser. Cs. Nats. 24, 29–39.
Armesto, J.J., Rozzi, R., Smith-Ramı́rez, C., Arroyo, M.T.K., 1998. Conservation targets in South American temperate forests. Science 282, 1271–1272.
Armesto, J.J., Lobos, P.L., Arroyo, M.K., 1999a. Los bosques templados del sur
de Chile y Argentina: una isla biogeográfica. In: Armesto, J.J., Villagrán,
C., Arroyo, M.T.K. (Eds.), Ecologı́a de los bosques nativos de Chile. third
ed. Editorial Universitaria, Santiago, pp. 23–28.
Armesto, J.J., Aravena, J.C., Villagrán, C., Pérez, C., Parker, G.G., 1999b.
Bosques templados de la Cordillera de la Costa. In: Armesto, J.J., Villagrán,
65
C., Arroyo, M.T.K. (Eds.), Ecologı́a de los bosques nativos de Chile. third
ed. Editorial Universitaria, Santiago, pp. 199–213.
Armesto, J.J., Franklin, J.F., Arroyo, M.T.K., Smith-Ramı́rez, C., 1999c. El
sistema de cosecha con ‘‘retención variable’’: una alternativa de manejo
para conciliar los objetivos de conservación y producción en los bosques
nativos chilenos. In: Donoso, C., Lara, A. (Eds.), Silvicultura de los bosques
nativos de Chile. Editorial Universitaria, Santiago, pp. 69–94.
Arroyo, M.T.K., Armesto, J.J., Rozzi, R., Peñaloza, A., 1999. Bases de la
sustentabilidad ecológica y sus implicaciones para el manejo y conservación
del bosque nativo en Chile. In: Donoso, C., Lara, A. (Eds.), Silvicultura de
los bosques nativos de Chile. Editorial Universitaria, Santiago, pp. 35–68.
Battaglia, M., Cherry, M.L., Beadle, C.L., Sands, P.J., Hingston, A., 1998.
Prediction of leaf area index in eucalypt plantations: effects of water stress
and temperature. Tree Physiol. 18, 521–528.
Brown, M.J., Parker, G.G., 1994. Canopy light transmittance in a chronosequence of mixed-species deciduous forests. Can. J. For. Res. 24, 1694–1703.
Brun, R., 1969. Strukturstudien im gemäßigten Regenwald Südchiles als
Grundlage für Zustandserhebungen und Forstbetriebsplanung. Ph.D. Thesis. Albert-Ludwig-Universität, Freiburg i. Br., Germany.
CONAF-CONAMA, 1999. Catastro y Evaluación de Recursos Vegetacionales
Nativos de Chile. Informe Nacional con Variables Ambientales.
PROYECTO CONAF-CONAMA-BIRF, Santiago, Chile.
Coomes, D.A., Duncan, R.P., Allen, R.B., Truscott, J., 2003. Disturbances
prevent stem size-density distributions in natural forests from following
scaling relationships. Ecol. Lett. 6, 980–989.
Dı́az, I., Armesto, J.J., Reid, S., Sieving, K.E., Willson, M.F., 2005. Linking
forest structure and composition: avian diversity in successional forests of
Chiloe Island, Chile. Biol. Conserv. 123, 91–101.
Diaz-vaz, J.E., Poblete, H., Juacida, R., Devlieger, F., 2002. Maderas Comerciales de Chile, third ed. Ed. Marisa Cuneo, Valdivia.
Ditzer, T., Glauner, R., Förster, M., Köhler, P., Huth, A., 2000. The process-based
stand growth model FORMIX3-Q applied in a GIS environment for growth
and yield analysis in a tropical rain forest. Tree Physiol. 20, 367–381.
Di Castri, F., Hajek, E., 1976. Bioclimatologı́a de Chile. Universidad Católica
de Chile, Santiago.
Donoso, C., 1989a. Antecedentes básicos para la silvicultura del tipo forestal
siempreverde. Bosque 10, 37–53.
Donoso, C., 1989b. Regeneración y crecimiento en el tipo forestal siempreverde
costero y andino tras distintos tratamientos silviculturales. Bosque 10, 69–83.
Donoso, C., 1998. Bosques templados de Chile y Argentina, fourth ed. Editorial
Universitaria, Santiago.
Donoso, C., Lara, A., 1999. Introducción. In: Donoso, C., Lara, A. (Eds.),
Silvicultura de los bosques nativos de Chile. Editorial Universitaria,
Santiago, pp. 25–34.
Donoso, C., Grez, R., Escobar, B., Real, P., 1984. Estructura y dinámica de
bosques del tipo forestal siempreverde en un sector de Chiloé insular.
Bosque 5, 82–104.
Donoso, C., Escobar, B., Urrutia, J., 1985. Estructura y estrategias regenerativas
de un bosque virgen de ulmo (Eucryphia cordifolia Cav.)-tepa (Laurelia
philippiana Phil.) Looser en Chiloé, Chile. Rev. Chil. Hist. Nat. 58, 171–186.
Donoso, C., Donoso, P., González, M., Sandoval, V., 1999. Los bosques
siempreverdes. In: Donoso, C., Lara, A. (Eds.), Silvicultura de los bosques
nativos de Chile. Editorial Universitaria, Santiago.
Donoso, P.J., Nyland, R.D., 2005. Seedling density according to structure,
dominance and understorey cover in old-growth forest stands of the
evergreen forest type in the coastal range of Chile. Rev. Chil. Hist.
Nat. 78, 51–63.
Emanuelli, P., Pancel, L., 1999. Funciones de volumen para la Reserva Nacional
Valdivia. In: Documento de trabajo, Proyecto Manejo Sustentable del
Bosque Nativo, CONAF-GTZ.
Franklin, J.F., 1993. Preserving biodiversity: species, ecosystems, or
landscapes? Ecol. Appl. 3, 202–205.
Franklin, J.F., Armesto, J.J., 1996. La retención de elementos estructurales del
bosque durante la cosecha: una alternativa para el manejo de bosques
chilenos. Ambient. Desarro. 12, 69–79.
Galloway, J.N., 1999. Los lı́quenes del bosque templado de Chile. In: Armesto,
J.J., Villagrán, C., Arroyo, M.T.K. (Eds.), Ecologı́a de los bosques nativos
de Chile. third ed. Editorial Universitaria, Santiago, pp. 101–111.
66
N. Rüger et al. / Forest Ecology and Management 252 (2007) 52–66
Galloway, J.N., Keene, W.C., Likens, G.E., 1996. Processes controlling the
composition of precipitation at a remote southern hemispheric location:
Torres del Paine National Park, Chile. J. Geophys. Res. 101, 6883–6897.
Geldres, E., Schlatter, J.E., 2004. Crecimiento de las plantaciones de Eucalyptus
globulus sobre suelos rojo arcillosos de la provincia de Osorno, Décima
Región. Bosque 25, 95–101.
González, M., Veblen, T.T., Donoso, C., Valeria, L., 2002. Tree regeneration
responses in a lowland Nothofagus-dominated forest after bamboo dieback
in South-Central Chile. Plant Ecol. 161, 59–73.
Grimm, V., Revilla, E., Berger, U., Jeltsch, F., Mooij, W.M., Railsback, S.F.,
Thulke, H.-H., Weiner, J., Wiegand, T., DeAngelis, D.L., 2005. Patternoriented modeling of agent-based complex systems: lessons from ecology.
Science 310, 987–991.
Gutiérrez, Á.G., Armesto, J.J., Aravena, J.C., 2004. Disturbance and regeneration dynamics of an old-growth North Patagonian rain forest in Chiloé
Island, Chile. J. Ecol. 92, 598–608.
Hedin, L.O., Armesto, J.J., Johnson, A.H., 1995. Patterns of nutrient loss from
unpolluted, old-growth temperate forests: Evaluation of biogeochemical
theory. Ecology 76, 493–509.
Huth, A., Ditzer, T., 2000. Simulation of the growth of a lowland Dipterocarp
rain forest with FORMIX3. Ecol. Model. 134, 1–25.
Huth, A., Ditzer, T., 2001. Long-term impacts of logging in a tropical rain
forest—a simulation study. For. Ecol. Manage. 142, 33–51.
Kammesheidt, L., Köhler, P., Huth, A., 2001. Sustainable timber harvesting in
Venezuela: a modelling approach. J. Appl. Ecol. 38, 756–770.
Köhler, P., 2000. Modelling anthropogenic impacts on the growth of tropical
rain forests. Ph.D. Thesis. University of Kassel, Kassel, Germany. Der
Andere Verlag, Osnabrück.
Köhler, P., Huth, A., 1998. The effect of tree species grouping in tropical rain
forest modelling—simulation with the individual based model FORMIND.
Ecol. Model. 109, 301–321.
Köhler, P., Ditzer, T., Ong, R.C., Huth, A., 2001. Comparison of measured and
modelled growth on permanent plots in Sabahs rain forests. For. Ecol.
Manage. 144, 101–111.
Köhler, P., Chave, J., Riera, B., Huth, A., 2003. Simulating long-term response
of tropical wet forests to fragmentation. Ecosystems 6, 114–128.
Lara, A., Veblen, T.T., 1993. Forest plantations in Chile: a successful model?
In: Mather, A. (Ed.), Afforestation: Policies, Planning and Progress. Belhaven Press, London, pp. 118–139.
Lara, A., Echeverrı́a, C., Donoso, C., 2000. Guı́a de ensayos silviculturales
permanentes en los bosques nativos de Chile. Instituto de Silvicultura de la
Universidad Austral de Chile, Valdivia.
Lara, A., Solari, M.E., Rutherford, P., Thiers, O., Trecaman, R., Molina, R.,
Prieto, A.R., Montory, C., 1999. Cobertura de la vegetación original de la
Ecoregión de los bosques valdivianos en Chile hacia 1550. WWF-Universidad Austral de Chile, Valdivia.
Lara, A., Soto, D., Armesto, J., Donoso, P., Wernli, C., Nahuelhual, L., Squeo,
F. (Eds.), 2003. Componentes Cientı́ficos Clave para una Polı́tica Nacional
Sobre Usos, Servicios y Conservación de los Bosques Nativos Chilenos.
Universidad Austral de Chile. Iniciativa Cientı́fica Milenio de Mideplan.
Larcher, W., 2001. Ökophysiologie der Pflanzen, sixth ed. Verlag Eugen
Ullmer, Stuttgart.
Lindenmayer, D.B., Franklin, J.F., 2002. Conserving Forest Biodiversity: A
Comprehensive Multiscaled Approach. Island Press, Washington, DC.
Lusk, C.H., 2002. Leaf area accumulation helps juvenile evergreen trees tolerate
shade in a temperate rainforest. Oecologia 132, 188–196.
Lusk, C.H., Matus, F., 2000. Juvenile tree growth rates and species sorting on
fine-scale soil fertility gradients in a Chilean temperate rain forest. J.
Biogeogr. 27, 1011–1020.
Lusk, C.H., del Pozo, A., 2002. Survival and growth of seedlings of 12 Chilean
rainforest trees in two light environments: Gas exchange and biomass
distribution correlates. Aust. Ecol. 27, 173–182.
Lusk, C.H., Kelly, C.K., 2003. Interspecific variation in seed size and safe sites
in a temperate rain forest. New Phytol. 158, 535–542.
Lusk, C.H., Wright, I., Reich, P.B., 2003. Photosynthetic differences contribute
to competitive advantage of evergreen angiosperm trees over evergreen
conifers in productive habitats. New Phytol. 160, 329–336.
Lusk, C.H., Chazdon, R.L., Hofmann, G., 2006. A bounded null model explains
juvenile tree community structure along light availability gradients in a
temperate rain forest. Oikos 112, 131–137.
Muñoz, A., Chacón, P., Pérez, F., Barner, E., Armesto, J.J., 2003. Diversity and
host tree preferences of vascular epiphytes and vines in a temperate rain
forest in southern Chile. Aust. J. Bot. 51, 381–391.
Neira, E., Verscheure, H., Revenga, C., 2002. Chile’s Frontier Forests: Conserving a Global Treasure. WRI, CODEFF, UACH, Washington, DC.
Olson, D.M., Dinerstein, E., 1998. The global 200: a representation approach to
conserving the earth’s most biologically valuable ecoregions. Conserv. Biol.
12, 502–515.
Pacala, S.W., Canham, C.D., Silander, J.A., Kobe, R.K., 1994. Sapling growth
as a function of resources in a north temperate forest. Can. J. For. Res. 24,
2172–2183.
Pérez, C., 1999. Los procesos de descomposición de la materia orgánica de
bosques templados costeros: interacción entre suelo, clima y vegetación.
In: Armesto, J.J., Villagrán, C., Arroyo, M.T.K. (Eds.), Ecologı́a de
los bosques nativos de Chile. third ed. Editorial Universitaria, Santiago,
pp. 301–315.
Pérez-Galaz, V.A., 1983. Manual de propiedades fı́sicas y mecánicas de
maderas chilenas. Documento de trabajo No 47. Investigación y Desarrollo
Forestal CONAF/FAO, Santiago de Chile.
Perry, D.A., 1994. Forest Ecosystems. John Hopkins University Press, Baltimore.
Pierce, L.L., Running, S.W., 1988. Rapid estimation of coniferous forest leaf
area index using a portable integrating radiometer. Ecology 69, 1762–1767.
Ryan, M.G., 1991. Effects of climate change on plant respiration. Ecol. Appl. 1,
157–167.
Salas, C., 2002. Ajuste y validación de ecuaciones de volumen para un relicto
del bosque de Roble-Laurel-Lingue. Bosque 23, 81–92.
Saldaña, A., Lusk, C.H., 2003. Influencia de las especies del dosel en la
disponibilidad de recursos y regeneración avanzada en un bosque templado
lluvioso del sur de Chile. Rev. Chil. Hist. Nat. 76, 639–650.
Saltelli, A., Chan, K., Scott, E.M., 2000. Sensitivity Analysis. Wiley Series in
Probability and Statistics—Probability and Statistics Section. John Wiley &
Sons, Chichester.
Saltelli, A., Tarantola, S., Campolongo, F., Ratto, M., 2004. Sensitivity Analysis
in Practice: A Guide to Assessing Scientific Models. John Wiley & Sons,
Chichester.
Shugart, H.H., 1984. A Theory of Forest Dynamics. The Ecological Implications of Forest Succession Models. Springer, New York.
Smith-Ramı́rez, C., Armesto, J., Valdovinos, C. (Eds.), 2005. Historia, biodiversidad y ecologı́a de los bosques costeros de Chile. Editorial Universitaria, Santiago.
Veblen, T.T., 1982. Growth patterns of Chusquea bamboos in the understorey of
Chilean Nothofagus forests and their influences in forest dynamics. Bull.
Torr. Bot. Club 109, 474–487.
Veblen, T.T., 1985. Forest development in tree-fall gaps in the temperate rain
forests of Chile. Natl. Geogr. Res. 1, 161–184.
Veblen, T.T., Schlegel, F.M., Escobar, R., 1980. Structure and dynamics of oldgrowth Nothofagus forests in the Valdivian Andes. Chile. J. Ecol. 68, 1–31.
Veblen, T.T., Donoso, C., Schlegel, F.M., Escobar, R., 1981. Forest dynamics in
south-central Chile. J. Biogeogr. 8, 211–247.
Veblen, T.T., Schlegel, F.M., Oltremari, J.V., 1983. Temperate broad-leaved
evergreen forests of South America. In: Ovington, J.D. (Ed.), Temperate
Broad-Leaved Evergreen Forests, vol. 10. Elsevier Science Publishers,
Amsterdam, pp. 5–31.