Content uploaded by Thomas T Veblen
Author content
All content in this area was uploaded by Thomas T Veblen on Jan 04, 2019
Content may be subject to copyright.
678
Ecology,
78(3), 1997, pp. 678–692
q
1997 by the Ecological Society of America
BLOWDOWN HISTORY AND LANDSCAPE PATTERNS IN THE ANDES OF
TIERRA DEL FUEGO, ARGENTINA
A
LAN
J. R
EBERTUS
,
1
T
HOMAS
K
ITZBERGER
,
2,4
T
HOMAS
T. V
EBLEN
,
2
AND
L
YNN
M. R
OOVERS
3
1
School of Natural Resources, University of Missouri, Columbia, Missouri 65211 USA
2
Geography Department, University of Colorado, Boulder, Colorado 80309 USA
3
North Central Forest Experiment Station, Columbia, Missouri 65211 USA
Abstract.
The effects of periodic gales on
Nothofagus
(southern beech) stand devel-
opment and landscape dynamics were studied in a 10.4-km
2
study area in the Sierra de las
Pinturas, part of the Andes in Argentine Tierra del Fuego. We reconstructed blowdown
history (patch sizes, dates of origin, return intervals) since the late 1700s to assess how
periodic large-scale wind disturbance influenced landscape pattern in a relatively simple
system. Most previous studies have focused on single storms in more diverse forests and
in landscapes influenced by several disturbance types and human activities. Boundaries of
post-blowdown stands were digitized from aerial photographs and analyzed within a geo-
graphic information system. Ages of blowdowns and return intervals were determined from
scars, growth releases, and maximum tree ages. Discrete blowdown patches (0.1 to
.
100
ha) covered two-thirds of the study area and ranged from 19 to
ø
200 yr of age, with 20–
30 yr between major events. The meteorology of these storms was unknown, but we suspect
they were caused by intense low-pressure systems originating in Antarctica. The mean
return interval for 34 sites was 145 yr, with a range of 103–218 yr. Based on treefall size
distributions, most stands blown over in the past 100 yr were even-aged, with mean tree
dbh (diameter at breast height) between 20 and 32 cm. Aerial photographs taken in 1970
were used for before-and-after comparisons of a 1972 blowdown. Seventy-one percent of
the area blown over in 1972 was even-aged, and 35% of the boundaries from this storm
exactly coincided with previous stand boundaries. Damage patterns from storms in 1924
and 1972 suggested that forests located on valleys parallel to the wind, windward side
slopes, and possibly upper leeward slopes were most vulnerable to blowdown, but few
landscape positions escaped being hit by repeated storms. Return intervals were not sig-
nificantly related to slope, elevation, or aspect; but surprisingly, shorter return intervals
were associated with deeper soils. The landscape pattern of blowdown and recovery shifted
over time because of variation among individual storms and because a small proportion of
old-growth stands were converted to blowdowns and vice versa. Browsing by guanacos
(
Lama guanaco
), a large native camelid, has severely inhibited tree regeneration during
the past 75 yr in small blowdowns and the perimeters of larger ones, converting some
stands to open meadows and incipient alpine communities. In the relatively simple
Noth-
ofagus
forests of Tierra del Fuego, periodic gales are the main determinants of forest
structure and pattern across a range of scales from small patches to entire landscapes.
Key words: blowdowns; browsing; disturbance; guanaco; interactions, disturbance-type; land-
scape pattern;
Nothofagus pumilio;
southern beech; Tierra del Fuego; treefalls; wind.
I
NTRODUCTION
Complete characterization of a disturbance regime
requires means and variances for return intervals, sizes,
shapes, and intensities, and information on how these
factors vary across the landscape. Yet, such statistics
have not even been determined for the simplest land-
scapes (Peet 1992). Among the best described distur-
bance regimes include studies of wildfire by Heinsel-
man (1973) in Minnesota and Romme (1982, and oth-
ers) in Yellowstone National Park, and fine-scale tree-
Manuscript received 4 August 1995; revised 26 July 1996;
accepted 6 August 1996.
4
Present address: Departamento de Ecologı´a, Universidad
Nacional del Comahue, Casilla de Correo 1336, 8400 Bar-
iloche, Argentina.
fall gaps in eastern deciduous forests by Runkle (1985,
and references therein), but even these studies focus
on only a single type or scale of disturbance. To un-
derstand the role of disturbance in creating landscape
patterns, the full range of disturbances affecting an area
must be considered (e.g., Harmon et al. 1983, Veblen
et al. 1992).
Wind creates a spatial-temporal mosaic of patches
or gaps in different stages of recovery from distur-
bance. Forest turnover is dominated by fine-scale tree-
fall gaps in much of the eastern deciduous forests of
North America (Runkle 1990) and in neotropical rain
forests (Brokaw 1985), where gaps are formed at
ø
1%
of the canopy per year. Simulations by Turner et al.
(1993) suggest that such disturbance regimes allow
April 1997 679
BLOWDOWN HISTORY AND LANDSCAPE PATTERNS
F
IG
. 1. Location of study area on Tierra del Fuego (top)
and regional setting (bottom). Thin solid lines indicate ap-
proximate boundaries between major vegetation zones (see
Study area
). Grey areas are mountain ranges (elevations in
meters).
both persistence and constancy of seral types or patches
at scales of ‘‘stand’’ or larger; i.e., a ‘‘steady-state mo-
saic.’’ Coarse-scale blowdowns (
.
1000 m
2
) are much
less frequent, with return-time estimates for many tem-
perate and tropical forests generally
.
1000 yr (Lorimer
1977, Canham and Loucks 1984, Whitney 1986, Glitz-
enstein and Harcombe 1988, Schaetzl et al. 1989, Nel-
son et al. 1994). Return intervals (i.e., inverse of fre-
quency; White and Pickett 1985) for destructive storms
are much shorter in hurricane/typhoon or gale-prone
areas of the Caribbean and eastern United States (Foster
1988
a
, Lugo and Waide 1993, Boose et al. 1994),
Southeast Asia (Whitmore 1984), Australia (Webb
1958), New Zealand (Shaw 1983, Jane 1986), and the
North American Pacific Northwest (Harris 1989). How-
ever, damage patterns from these storms are very com-
plex, with partial and complete destruction occurring
at various scales.
Landscape patterns of coarse-scale blowdowns are
best known from detailed analyses of the 1938 hurri-
cane in New England and several hurricanes in the
Caribbean (Lugo et al. 1983, Foster 1988
a
, Foster and
Boose 1992, Boose et al. 1994). Blowdowns tend to
increase landscape diversity by creating a mosaic of
patches of different ages and successional status (Webb
1958, Foster and Boose 1992). Variation in blowdown
intensity also contributes to landscape heterogeneity by
enhancing beta diversity (e.g., severe, intermediate,
and light levels of disturbance fostering different com-
munities; Savage et al. 1992). Blowdowns may also
strike some aspects and slope positions more than oth-
ers, modifying vegetation patterns along environmental
gradients (Foster and Boose 1992). The landscape pat-
terns created by hurricanes are very complex, reflecting
interactions of biotic, edaphic, and historical factors
with meteorological and stochastic processes (Foster
1988
b
, Foster and Boose 1992).
Most studies of blowdowns are limited almost ex-
clusively to single events superimposed on relatively
complex landscapes with many tree species, several
disturbance types, and anthropogenic activities. In ad-
dition, these studies usually focus on recent storms and
initial stages of stand recovery. Not surprisingly, the
spatial and temporal dimensions of blowdown dynam-
ics under these circumstances are often difficult to as-
sess. In this study we describe a disturbance regime in
10.4 km
2
of
Nothofagus pumilio
(Poeppig and Endl.)
Krasser (lenga, southern beech) forest in the Andes of
Tierra del Fuego. The system is ideal for addressing
how disturbances generate landscape pattern because
there is a single dominant disturbance, catastrophic
windthrow, which reoccurs frequently and creates very
discrete patches, and there is a single modifying dis-
turbance, herbivory by guanacos (
Lama guanaco
), a
large native herbivore. Furthermore, this study area has
not been significantly affected by human activities, and
the presence of a single tree species simplifies the task
of examining tree responses to blowdown. Evidence
for blowdown persists for
ø
200 yr, providing oppor-
tunities to examine the long-term effects of disturbance
on landscape processes. The objectives of this study
were to determine spatial and temporal characteristics
of the wind-dominated disturbance regime in the south-
ern Andes, and to investigate long-term stand and land-
scape development controlled by wind disturbance and
interactions with guanacos.
S
TUDY
A
REA
The combination of westerly atmospheric circulation
and the southwest–northeast orientation of the Andes
results in a strong precipitation gradient across Tierra
del Fuego. The associated vegetation gradient com-
prises evergreen
Nothofagus betuloides
rain forest in
the southwest, deciduous
N. pumilio
‘‘summer green’’
forest in the interior, and deciduous
N. antarctica
and
steppe in the northeast (Fig. 1). With the exception of
the Andean rain shadow, the climate of Tierra delFuego
is similar to southern Alaska and western Scandinavia
with average monthly temperatures ranging from
2
1
8
C
to 10
8
C, annual precipitation of 450 mm, and prevailing
winds and major windstorms from the southwest (Ly-
ons 1983, Tuhkanen 1992).
680 Ecology, Vol. 78, No. 3
ALAN J. REBERTUS ET AL.
The 10.4-km
2
study area is located in
N. pumilio
forest in the Sierra de las Pinturas, a range of the south-
ernmost Andes running east–west along Lago Fagnano
(54
8
30
9
S, 67
8
40
9
W; Fig. 1). Cerro Yakush (elevation
950 m; relief 600–700 m) dominates the northeast part
of the study area, with several smaller foothills (relief
100–200 m) occupying the southwest half. The shallow
(
,
50 cm deep), acidic brown soils are slightly pod-
zolized, and derived from sandstone and conglomerate
parent material (Moore 1983).
Eighty-nine percent of the study area is forested and
11% is alpine tundra. There are only slight changes in
forest structure with elevation, and the transition to
alpine tundra is very abrupt. Timberline ranges from
400 to 700 m. Northeast of Cerro Yakush, the Sierra
descends into the sparsely populated steppe ecotone.
Cerro Yakush is at the edge of a vast, uninhabited,
roadless area that extends to the Pacific Ocean. Recent
logging prevented us from sampling farther down the
northeast side of the mountain, but the study area prop-
er was pristine and had little evidence of past human
activities.
Guanaco browsing plays a major role in forest re-
generation near the steppe ecotone. These animals are
large (80–100 kg), polygamous camelids native to for-
ests, pampas, and steppe from Peru to Tierra delFuego.
Roughly one-tenth of the total guanaco population re-
sides on Tierra del Fuego, where there are perhaps 10–
20 thousand animals (Raedecke 1978). Based on
ground and aerial sightings, we estimated that there
were 5–10 guanacos/km
2
in our study area during sum-
mer. In Tierra del Fuego, their diet consists mainly of
grasses and forbs, but they also browse leaves and twigs
of
Nothofagus antarctica, N. pumilio,
and several
shrubs (Raedecke 1978).
M
ETHODS
Mapping of blowdowns
An orthophoto image and digital elevation model
(DEM) were constructed from 1970 1:40 000 aerial
photographs using ERDAS (1992) DIGITAL ORTHO
module. The orthophoto image removes distortions
caused by terrain and the oblique camera angle. No
adequate topographic maps existed for the study area,
so a DEM was created from the 1970 aerial photo-
graphs. In this procedure, image matching of stereo
pairs automatically generates two corresponding grids
of
.
4000 points, representing identical ground loca-
tions. Ground coordinates, including elevation, are
computed from matched image pairs and known camera
orientation. The coordinates were scaled from external
control points taken from a 1950 1:200 000 topographic
map that overlapped part of the study area.
Blowdowns were digitized directly off the ortho-
photo image using IDRISI (Eastman 1992), a geograph-
ic information system (GIS). Blowdowns
,
200 yr old
are comprised of even-aged stands that appear as ho-
mogeneous, fine-textured patches on aerial photo-
graphs. Old growth (
.
225–250 yr) has a much coarser
texture. Several hundred oblique 35-mm color infrared
photographs were taken from
ø
2000 m altitude in
1990–1991 to aid in mapping blowdowns that occurred
in 1972. The 1972 blowdowns were transferred to the
1970 base image by carefully locating corresponding
landmarks on each photograph. The 1972 event also
provided an opportunity for detailed before-and-after
comparisons to assess (1) how much of the area blown
down in 1972 was previously even-aged vs. old growth
and (2) what percent of the 1972 blowdown boundaries
matched older blowdown boundaries visible in the
1970 photographs.
Areas where browsing by guanacos had inhibited
post-blowdown
Nothofagus pumilio
regeneration were
identified on the 1970 and 1990–1991 aerial photo-
graphs, and with field checks. We mappedpatches with-
in blowdowns where guanacos had browsed nearly all
stems
,
1.5 m in height. Guanacos typically browsed
seedlings in discrete patches wherever treefalls did not
block their access. Trees that remained unbrowsed for
.
20 yr grew out of reach of guanacos, but in areas
where animals concentrated, seedlings became severely
deformed and hedge-like after decades of persistent
browsing. Patches of complete regeneration failure
were still widespread in blowdowns that occurred in
1924.
Field procedures
Blowdown dating.
— To quantify the temporal and
spatial patterns of wind disturbance, stands of different
origin were identified on the 1970 photographs and
stand history and regeneration patterns were deter-
mined by field sampling and dendrochronology. Tree-
falls from blowdowns were visible for 150 yr after the
event, and were confirmed by the even-aged stands
(single cohort) and lack of charcoal. Several even-aged
stands between 170 and 200 yr old having pit-and-
mound topography were also assumed to be blow-
downs. Stands that reach 225–250 yr gradually become
old growth (sensu Oliver and Larson 1990), where a
second cohort develops in small treefall gaps. In this
paper, ‘‘old growth’’ is synonymous with‘‘no evidence
of previous blowdown.’’ We attempted to sample all
blowdown patches within the study area, but some data
were only available for a smaller subset of blowdowns;
an overview is presented in Table 1.
Blowdowns were dated from scars, growth releases
on remnant trees, and maximum age of post-blowdown
regeneration. At 21 blowdowns, we cut wedges from
1–4 live trees (60 total) with bole scars formed from
tree collision injury. These scars were usually several
meters long and most common on live windthrown
trees, apparently from windward trees ‘‘sliding down’’
their leeward neighbors in domino-like fashion. Sam-
ples were not cross dated, but they provided approxi-
mate dates used in conjunction with other methods. We
April 1997 681
BLOWDOWN HISTORY AND LANDSCAPE PATTERNS
T
ABLE
1. Subsets of total blowdown pool used in various analyses and selection criteria in
a study of
Nothofagus
forests on Tierra del Fuego.
Sample
n
Selection criteria
Total blowdown patches 114 All patches
.
0.1 ha on aerial photographs
Patch size distribution 90 All patches
,
100 yr old; original boundaries as-
sumed to be intact (i.e., not covered by more re-
cent event)
Treefall characteristics 61 Patches
,
100 yr old; some smaller blowdowns in
a cluster of patches were not sampled
Live basal area and dbh 41 All patches with tree (
$
4 cm dbh) regeneration;
heavily browsed blowdowns lacking trees were
excluded
Patch size and browsing 32 1972 blowdown patches; most recent major event
Return intervals 34 Opportunistic, based on availability of dendro-
chronological material and biased towards recent
events
Tree age structures 3 Arbitrarily selected
Landscape distribution All old-growth areas (no evidence of blowdowns
in
.
225 yr) compared to 1924 and 1972 blow-
downs, the two largest events with original
patch boundaries intact
cored remnant trees within and along blowdown edges
and examined the cores for abrupt increases in growth
(release) associated with disturbance (Lorimer 1985).
Release was defined as
a 2.5-fold increase in growth
when adjacent groups of five rings were compared
(Veblen et al. 1992). Because many trees have delayed
growth response to disturbance, we assumed that the
oldest release date among a group of similar dates cor-
responded closely with the event (Lorimer 1985). Ini-
tially, 10–30 cores were taken per blowdown, but after
we became familiar with the ages and appearances of
blowdowns, fewer cores were needed to confirm a sus-
pected date.
Nothofagus pumilio
can live 300–375 yr (Rebertus
and Veblen 1993), but trees
.
250 yr old are often un-
sound and it was not practical to core them for release
dates. Therefore, dates of older blowdowns (150–200
yr) were assigned in 25-yr age classes based on max-
imum stem age of five to ten trees cored at 0.3–0.4 m
above the soil surface (Heinselman 1973, Veblen et al.
1992).
N. pumilio
typically regenerates in even-aged
stands after major canopy disturbance (Veblen et al.
1996) and maximum tree ages usually correspond
closely with the time of disturbance (Veblen et al. 1992,
Rebertus and Veblen 1993).
Approximate point return intervals were determined
for 34 stands which represented a wide range of site
characteristics and blowdown events. Return intervals
were based on dates of two successive events from the
same locale (
ø
1–3 ha). A common situation was a
small blowdown patch that occurred within a much
larger, older patch, but in most cases the return intervals
only correspond to points and not whole stands. Return
intervals were sampled opportunistically, wherever
sufficient dendrochronological material was available
to date overlapping events.
Stand response.
—A rapid assessment of treefall
characteristics and stand response was made at 61
blowdowns along a randomly oriented, 60-m transect,
starting from an arbitrary point near the center. For
larger blowdowns, one or two extra transects were
used. In some cases where small patches were nearly
connected, only one blowdown within the group was
sampled ( Table 1). For blowdowns
,
100 yr old, the
first 30 treefalls intersecting the tape were measured
for dbh, direction, mode of treefall, and whether the
treefall was dead or alive.
Post-blowdown tree regeneration (
$
4 cm dbh) was
documented along the transects by the point-centered-
quarter method, with 10 m between points. Two sup-
plemental trees were sampled at the end of the transect,
for 30 trees total. Many recent blowdowns were heavily
browsed and lacked sufficient regeneration for sam-
pling (Table 1). At three blowdowns ranging from 67
to 130 yr old, all trees along the transects were cored
at 0.3–0.4 m above ground. These blowdowns were
subjectively selected to demonstrate stand response.
Site variables collected at each site included slope,
aspect, elevation, and depth to bedrock. Depth to bed-
rock was determined by pounding a 0.5 cm diameter
steel rod in as far as possible. Ten measurements were
taken from relatively undisturbed areas along each tran-
sect, avoiding pits and mounds whenever possible.
Cobble within the profile may have sometimes pre-
vented the rod from reaching bedrock, but the site av-
erage (
D
) is probably a good index of relative soil
depth.
Data analysis
Circular means, angular deviations, and mean vec-
tors were calculated for treefall direction in each blow-
down patch (Batschelet 1981). The Rayleigh test was
used to test whether treefall directions within patches
differed from a random distribution. Differences in tree
survival among blowdown age classes were analyzed
682 Ecology, Vol. 78, No. 3
ALAN J. REBERTUS ET AL.
using the Kruskal-Wallis test with Dunn’s method of
pairwise comparisons.
The areas and landscape distribution of blowdowns,
old growth (no evidence of blowdown
.
225 yr), and
browsed areas were analyzed using IDRISI. Blow-
downs up to 100 yr old were assumed to have intact
boundaries (i.e., not partially covered by more recent
blowdowns), and were used in analysis of patch size
distribution. Cole’s coefficient (Cole 1949) was used
to test for association between blowdowns and aspect,
slope, and elevation classes. This technique compares
the presence or absence of two classes on a cell-by-
cell basis (Foster and Boose 1992). In addition, we
digitized the study area into several landform catego-
ries: valleys, ridges, and side slopes. Valleyboundaries
extended from a stream channel up side slopes 20 m
and 30 m in vertical relief from first- and second-order
streams, respectively. Ridges extended downslope 15
m in vertical relief from the crest. Valleys were sub-
divided into those oriented north–south and east–west;
side slopes were subdivided into north, east, south, and
west exposure.
Stepwise multiple regression (SAS 1988) was used
to examine the relationship between return intervals
and site variables: slope, elevation, and mean soil
depth. Aspect, a circular variable, was analyzed sep-
arately using periodic regression (Batschelet 1981).
Values of the site variables were based on the few
hectares where tree-coring established the return in-
terval.
Assuming a random sample of disturbance return
intervals, the shape of their frequency distribution may
suggest how often a blowdown is likely to occur. Al-
though many functions could be used to describe such
a distribution, we chose the Weibull becauseof its flex-
ibility and traditional use in fire history studies. The
probability density form of the Weibull function is
Y
5
[(
ct
c
2
1
)/
b
]
3
exp[
2
(
t
/
b
)
c
], where
Y
is the frequency
or probability of having disturbances with intervals of
age
t
(in years),
b
is a scale parameter (in years) that
has been called disturbance ‘‘recurrence,’’ and
c
is a
dimensionless shape parameter (Johnson and VanWag-
ner 1985). When
c
.
1, the instantaneous rate of dis-
turbance increases with age. This Weibull model as-
sumes that the probability of disturbance is a power
function of the time since the last disturbance, and that
the disturbance regime is spatially and temporally con-
stant, or reasonably so, at scales relevant to the study.
R
ESULTS
Blowdown history and patch structure
One-hundred and fourteen patches
.
0.10 ha, 65%
(605 ha) of the forested part of the study area, were
identified as post-blowdown in origin (Figs. 2 and 3A).
The distribution of patch sizes for blowdowns
,
100
yr old was highly skewed right and approximated a
power curve: 62% of 90 total patches were
,
1 ha and
90% were
,
6 ha (Fig. 4). Nevertheless, patches
.
6ha
comprised
.
70% of the total blowdown area. The larg-
est single patch was
ø
150 ha, of which 101 ha was
within the study area.
Seventy-three percent of the 61 blowdowns
,
100 yr
old had mean treefall directions within 60
8
of due north
(Fig. 5). Treefalls were strongly directional within
stands: all 61 stands had treefall directions significantly
different from random (Rayleigh test, all stands
P
,
0.05), and more than half had mean vectors
.
0.8 (a
vector of 1.0 indicates all trees are oriented in the same
direction). In contrast, treefall directions in an old-
growth stand on the north edge of the study area were
not significantly different from random (mean vector
5
0.21,
n
5
40 trees,
P
.
0.05).
Although most damaging winds from these storms
were southerly, local topography clearly modified this
general pattern. For example, winds from a 1972 event
were deflected northwesterly, based on blowdowns
along the windward flank of Cerro Yakush ( Fig. 6A).
Similarly, several stands were hit by winds from the
east-northeast and northwest, apparently due to wind
funnelling up curved valleys (Fig. 6A). Winds from a
1924 storm came from both the south and east-north-
east (Fig. 6B). Hills in the southwest part of the study
area deflected winds westerly, but differences in wind
direction on opposite sides of Cerro Yakush were dif-
ficult to explain within the context of local topography.
We also could not rule out a second event.
We had variable success in dating blowdowns to the
exact year. Release and scar dates were often incon-
clusive, so we present the composite blowdown chro-
nology in 25-yr classes (Fig. 3A). Over the entire area,
major blowdowns have occurred fairly regularly,
roughly every 20–30 yr, over the past two centuries
(Fig. 3A). Nearly half of the study area consisted of
stands
,
150 yr old, and two major storms, 1972 and
1924, affected 18% and 13% of the study area, re-
spectively. Given that
Nothofagus pumilio
may reach
.
300 yr of age in old-growth stands (Rebertus and
Veblen 1993), the preponderance of stands
,
150 yr old
suggests a high disturbance rate.
There were some unavoidable problems in recon-
structing this blowdown history. First, aerial photo-
graphs and ‘‘groundtruthing’’ probably failed to reveal
some boundaries between older blowdowns of approx-
imately the same age (see Heinselman 1973). For ex-
ample, a large 150-yr-old blowdown could easily
‘‘hide’’ many smaller imbedded or adjoining patches
that were 10–20 yr younger or older. Second, both
boundaries and sizes of older blowdowns were inde-
terminate, because these blowdowns were often par-
tially or completely covered by younger events. There
is good evidence from treefall sizes and point return
intervals, however, that new blowdowns only occurred
in stands
.
100 yr old, so the sizes and shapes of blow-
downs occurring after 1890 are completely intact (Fig.
3A).
April 1997 683
BLOWDOWN HISTORY AND LANDSCAPE PATTERNS
F
IG
. 2. Maps of the study area on Tierra del Fuego showing origins of forest stands from tree blowdowns (A) and areas
where guanacos have severely inhibited regeneration of
Nothofagus
(B). Topography in 100-m contours.
At 34 stands we were able to estimate the interval
between two successive blowdowns at the same point.
The mean return interval was 145 yr and ranged from
103 to 218 yr (Fig. 3B). The distribution of return
intervals did not differ significantly from a normal dis-
tribution (Kolmogorov-Smirnov test,
P
5
0.555). The
Weibull probability function (Johnson and Van Wagner
1985) provided a good fit to the return interval fre-
quency distribution (
r
2
5
0.705), and the expected re-
turn interval,
b
5
147 yr, was close to the mean. The
Weibull shape parameter,
c
5
5.922, indicates an in-
creasing hazard of blowdown with stand age (Johnson
684 Ecology, Vol. 78, No. 3
ALAN J. REBERTUS ET AL.
F
IG
. 3. Blowdown areas in 1991 by time since disturbance
and dates of origin (A), and frequency of blowdown return
intervals with fitted Weibull function (B):
Y
5
[(5.9
3
t
4.9
)/
147]
3
exp[
2
(
t
/147)
5.9
], where
Y
is the probability of having
a disturbance with return interval
t
(
r
2
5
0.705).
F
IG
. 4. Distribution of blowdown patch sizes on Tierra
del Fuego in 1991 for events
,
100 yr old. Original patch
boundaries were assumed to be intact. Frequency data were
fit to a power function:
Y
5
0.22
1
[18.82
3
(Area)
2
1.56
](
r
2
5
0.994). Circles signify the cumulative percentage of total
blowdown area contributed by each size class for the past
100 yr.
F
IG
. 5. Distribution of mean treefall directions in 10-de-
gree classes for 61 blowdowns on Tierra del Fuego. Each
circle represents one stand; circle patterns indicateblowdown
ages.
and van Wagner 1985). Unlike most fire regimes where
these models have been applied, blowdown hazard ap-
pears to increase between 100 and 200 yr post-distur-
bance, and then possibly decrease in older stands. No
return interval was
,
100 yr, which is consistent with
the behavior of the 1972 event. Blowdowns from this
storm completely enveloped several 50- to 75-yr-old
stands without causing significant damage. The distri-
bution of return intervals probably overestimates the
disturbance rate, because the dendrochronological rec-
ord diminishes with stand age (Fox 1989). For example,
to detect a return interval
.
220 yr in a patch created
in 1924, we would have needed to find remnant trees
.
280 yr old, which is approaching the typical lifespan
of
N. pumilio.
Although return intervals were not randomly sam-
pled, and despite the dendrochronological bias, it is
reassuring that the return interval distribution is con-
sistent with the treefall size distribution of 61 stands
,
100 yr old (Fig. 7). The structural threshold of vul-
nerability occurred when trees reached 18–20 cm dbh.
Based on characteristics of live stands, post-distur-
bance cohorts reached this size between 100 and 125
yr after disturbance (Fig. 8A), which corresponds
closely with the minimum estimates of return interval
(Fig. 3B). Most of the treefall samples came from even-
aged stands. Sixty-five percent of the treefall samples
had mean dbh’s
,
32 cm (also upper quartile
,
40 cm),
which is within the range of tree diameters found in
post-blowdown stands
,
200 yr (compare Figs. 7 and
8). Thirteen percent of the treefall samples had diam-
April 1997 685
BLOWDOWN HISTORY AND LANDSCAPE PATTERNS
F
IG
. 6. Treefall directions in blowdowns from (A) 1972 and (B) 1924, illustrating the possible effects of topography on
wind direction. Ground transect data (30 measured tree azimuths per locale) were supplemented with estimates of treefall
directions from aerial photographs and ‘‘groundtruthed’’ estimates of general treefall direction (single azimuth measured per
locale). The circled ‘‘1’’ in chart (A) refers to the area of greatest local impact.
eters
.
32 cm (upper quartile
.
40 cm dbh), and were
structurally similar to the old-growth sample and other
old-growth stands on Tierra del Fuego (Rebertus and
Veblen 1993). Twenty-two percent of the samples had
characteristics intermediate between even-aged post-
blowdown stands and old growth. The transects only
represented a small portion of some blowdowns; nev-
ertheless, the treefall data further suggested that most
686 Ecology, Vol. 78, No. 3
ALAN J. REBERTUS ET AL.
T
ABLE
2. Cole’s coefficient of association between topographic features and blowdown damage in a study of forests on
Tierra del Fuego. The 1924 and 1972 events were the most extensive blowdowns in the past 100 yr. Old-growth stands
Aspect (quadrant)
North East South West
Elevation (m)
,
300 300–500
.
500
Old growth
2
0.25
1
0.12
1
0.06
2
0.16
1
0.06
1
0.03
2
0.25
(32.0) (26.0) (11.0) (31.0) (43.4) (41.1) (15.5)
1924
1
0.03
1
0.21
1
0.03
2
0.60
2
0.05
1
0.20
2
0.51
(44.2) (33.4) (7.7) (14.7) (38.2) (51.7) (10.1)
1972
2
0.29
1
0.01
1
0.04
1
0.13
1
0.19
2
0.22
2
0.14
(30.2) (16.5) (8.7) (44.6) (51.7) (30.7) (17.6)
F
IG
. 8. Mean tree dbh (A) and basal area (B) of post-
blowdown regeneration in 25-yr age classes. Individual blow-
downs are plotted to show range and variation.
F
IG
. 7. Distribution of mean treefall diameters at breast
height (dbh) from 61 blowdowns
,
100 yr old. Sample size
for each blowdown is
ø
30 trees.
stands were destroyed by blowdown before they
reached old-growth status.
Where do blowdowns occur?
Old-growth stands (no evidence of blowdown in
.
225 yr) were fairly evenly distributed with respect to
topographic positions (Table 2). Old growth was neg-
atively associated with ridges and positively associated
with easterly aspects and elevations
,
300 m, but the
relationships were relatively weak. Stands with no ev-
idence of previous blowdown were more common 2–
5 km northeast of Cerro Yakush, but we did notsample
this area because it marked the transition to steppe and
was more heavily impacted by human activities.
Damage from the 1924 blowdown was positively as-
sociated with east aspects and ridges, and negatively
associated with valleys and south and west side slopes
(Table 2). In contrast, the 1972 storm was positively
associated with valleys oriented north–south, west-fac-
ing side slopes, moderate slope steepness, and eleva-
tions
,
300 m; and negatively associated with ridges.
Sixty-four percent (108 ha) of the area destroyed in
1972 was associated with north–south valleys and east–
west side slopes, oriented parallel to the predominantly
southerly winds. Indeed, one blowdown followed a
stream valley for
.
5 km before flaring out on the south-
west flank of Cerro Yakush (Fig. 6A, solid black circle
labeled ‘‘1’’). Despite the negative association with
north and east side slopes, a line of blowdowns oc-
curred just below timberline on the northeast side of
Cerro Yakush, some clearly in leewardslope positions,
accounting for 28.6% of the damage from the 1972
storm.
We used the 1970 aerial photographs to better un-
derstand the damage patterns from the 1972 storm,
which was carefully reconstructed from 1990–1991
photographs and groundtruthing. Before the storm,
60% of the study area was even-aged and 40% was old
April 1997 687
BLOWDOWN HISTORY AND LANDSCAPE PATTERNS
had no evidence of severe damage in
.
225 yr. Cole’s index ranges from
2
1 (maximum disassociation) to
1
1 (maximum
association). The percentage of old growth or blowdown area in each category is given in parentheses.
Slope (
8
)
,
10 10–20 20–30
.
30 N–S
valley E–W
valley Ridge
Side slopes
North East South West
1
0.05
1
0.03
2
0.18
1
0.01
1
0.01
1
0.005
2
0.17
2
0.09
1
0.03
1
0.002
2
0.01
(20.4) (44.2) (31.6) (3.8) (13.6) (6.0) (5.7) (25.2) (21.7) (7.0) (20.9)
1
0.02
2
0.03 0.00
2
0.19
2
0.11
2
0.51
1
0.10
1
0.05
1
0.09
2
0.35
2
0.65
(17.7) (41.3) (38.9) (2.1) (11.4) (2.7) (15.8) (31.5) (26.9) (4.4) (7.5)
2
0.08
1
0.14
2
0.14
2
0.65
1
0.16
1
0.01
2
0.47
2
0.16
2
0.47
2
0.57
1
0.08
(14.7) (50.8) (33.5) (1.0) (26.6) (6.2) (10.3) (23.3) (10.3) (2.9) (27.2)
F
IG
. 10. Distribution of return intervals (
t
) across aspects,
fitted with a trigonometric polynomial:
t
5
147
1
[9.0
3
cos(Aspect
2
229)].
F
IG
. 9. Linear regression model of the blowdown return
interval (
t
) vs. mean soil depth (
D
) for 34 stands.
t
5
179.38
2
1.33
D
(
r
2
5
0.27); dotted lines are 95% confidence limits.
growth. However, 71% of the area blown downin 1972
was comprised of even-aged stands and 29% was old
growth. The before-and-after comparison also revealed
that 35% of the total perimeter length of the 1972 blow-
downs matched previous blowdown boundaries.
In the multiple regression model, variation in return
interval was significantly related only to mean soil
depth (
r
2
5
0.27;
F
1, 32
5
11.9;
P
5
0.004; Fig. 9).
Elevation, slope, and interaction terms were not sig-
nificant in the regression model. Surprisingly, higher
disturbance rates were associated with deeper soils,
although the relationship was highly variable. Return
intervals also tended to be slightly shorter on north and
east aspects (Fig. 10), but the correlation was weak and
not significant (
r
5
0.24,
n
5
34,
P
5
0.32). Negative
findings should be interpreted cautiously because any
error associated with dating blowdowns (see Lorimer
1985, Norton and Ogden 1990) may have been com-
pounded when determining return intervals.
Stand response
Of 1839 treefalls sampled in 61 blowdowns
,
100
yr old, 92% were uprooted, 5% were wind-snapped,
and 3% were unknown. Survival of windthrown trees
was observed in 8.1% of these 1839 treefalls
,
100 yr
old and was more prevalent in stands
,
25 yr old (Krus-
kal-Wallis test,
P
,
0.001, 3 df; Fig. 11). Surviving
trees were supported by both original and adventitious
root systems.
In the three blowdowns where we cored trees for age
structures, 67–97% (total
n
5
30 trees cored per blow-
down) of
N. pumilio
originated within a two-decade
span immediately after blowdowns (Fig. 12). The pres-
ence of occasional older, ‘‘open-grown’’ trees suggests
that they were probably saplings when the storm oc-
curred. Based on stand chronosequences, average dbh
and basal area increase asymptotically the first 200 yr
after disturbance, with basal area possibly declining
slightly in old growth (Fig. 8B).
Guanaco browsing in blowdowns
In 1991, guanaco browsing inhibited
N. pumilio
re-
generation in 130.7 ha (41.5%) of the area affected by
blowdowns since 1915 ( Fig. 2B). Persistent browsing,
with nearly all stems pruned below 1.5 m in height,
was visible in 62.4% of the area blown down in 1972,
61.5% of the area impacted in 1940–1965, and 6.9%
of the areas hit in 1924. These older browse areas have
now been partially converted to meadow. On Cerro
Yakush the percentage of blowdown area severely
browsed increased with proximity to timberline (Table
688 Ecology, Vol. 78, No. 3
ALAN J. REBERTUS ET AL.
F
IG
. 11. Percentage of treefalls alive in 1991 (mean
6
1
SD
) as a function of time since disturbance. The number of
blowdowns sampled appears above each bar.
F
IG
. 12. Post-blowdown age structures for three
Notho-
fagus pumilio
stands. Thirty trees were sampled from each
stand. Arrows indicate approximate blowdown dates.
T
ABLE
3. Percentage of blowdown area heavily browsed as
a function of travelling distance from timberline on Cerro
Yakush, Tierra del Fuego. Total blowdown area (in hec-
tares; browsed
1
unbrowsed) for each category appears in
parentheses. The foothills in the southwest third of the
study area usually lack timberline and were not included
in the analysis.
Blowdown
period
Slope distance from timberline (m)
0–199 200–399 400–599
.
600
1965–1990 85 (17) 62 (26) 59 (31) 47 (22)
1940–1965 72 (4) 55 (3) 58 (1) 18 (4)
1915–1940 3 (6) 3 (15) 1 (8) 0 (5)
All 64 (28) 43 (44) 41 (40) 38 (32)
3). Browsing was also negatively correlated with blow-
down area for the 1972 event (Pearson rank order cor-
relation,
r
52
0.64,
P
,
0.01,
n
5
32). In blowdowns
.
5 ha, severe browsing was often restricted to discrete
perimeter bands 10–50 m wide (Fig. 2B).
D
ISCUSSION
Regional dynamics of forests and wind
The distribution of patch ages, return intervals, and
treefall diameters provide three lines of evidence for
periodic blowdowns every few decades and patch turn-
over times of
ø
150 yr for most of the study area. One-
third of the study area is old growth, suggesting that
the blowdown regime is not spatially homogeneous.
Nevertheless, the disturbance rate is at least twice that
reported for
Nothofagus pumilio
and similar species in
old-growth stands dominated by fine-scale treefall gap
dynamics (Veblen 1985, Rebertus and Veblen 1993,
Veblen et al. 1996).
Because of the remoteness and limited aerial pho-
tograph coverage of the interior, we do notknow wheth-
er this blowdown regime is representative of other ar-
eas. The west end of the Sierra de las Pinturas forms
a promontory in Lago Fagnano (Fig. 1), which may
expose the range to more intense winds. Indeed, there
appear to be fewer blowdowns in the Sierra Beauvoir,
immediately to the west (A. J. Rebertus,
personal ob-
servation
). Extensive even-aged stands of probable
windstorm origin have been documented in
Nothofagus
forests throughout Patagonia (Agostini 1941, Eskuche
1973, Mutarelli and Orfila 1973, Alvarez and Gross
1979, Schmidt and Urzua 1982, Armesto et al. 1992).
We have no direct information of what type(s) of
storms were responsible for these blowdowns. How-
ever, Lyons (1983) described infrequent, intense Ant-
arctic depressions that are consistent with the level of
damage observed in this study. These storms develop
rapidly offshore of Antarctica, move northeast into the
Drake Passage, and sweep across Tierra del Fuego,
generating winds
.
60 m/s (215 km/h), equivalent to a
category 3 hurricane (Lyons 1983). The life cycle of
these storms is extremely short and their genesis is
poorly known, but they are most numerous in spring
and fall when trees are in full leaf. Given the proximity
of Tierra del Fuego to the Antarctic pack ice, it is likely
April 1997 689
BLOWDOWN HISTORY AND LANDSCAPE PATTERNS
that this disturbance regime is highly sensitive to cli-
mate change (cf. Clark 1989). Global warming studies
predict a decrease in the frequency of low pressure
systems between 30
8
and 90
8
S latitude but an increase
in their intensity (Lambert 1993).
Although wind explains the current landscape pat-
terns, charcoal from bogs in Tierra del Fuego suggests
a long history of fire on the island (Heusser 1987).
Lightning is extremely rare in Tierra del Fuego (Veblen
et al. 1996), and burning patterns would have been
subject to vagaries of the nomadic Ona, who roamed
the interior steppe and deciduous forests in search of
guanaco (Goodall 1979). Most of the indigenous peo-
ples were wiped out by measles and smallpox in the
late 1800s, but extensive logging and burning has oc-
curred north and east of the study area where estancias
(ranches) were built by European immigrants in the
early 1900s (Goodall 1979).
Stand response
Effective dispersal of
Nothofagus
seeds is often lim-
ited to distances of 50–80 m, rarely 200 m (Veblen et
al. 1996), so most blowdowns are small or narrow
enough to capture adequate seed from adjacent stands
or remnant trees. In addition,
N. pumilio
stands
.
100
yr old have a short-lived pool of seedlings, most
,
10
cm high, that survive
en masse
following a blowdown
(Rebertus and Veblen 1993; T. T. Veblen,
unpublished
data
). The development of even-aged stands after
blowdowns is similar to the response of
N. pumilio
after
fire, although regeneration failure following large, in-
tense fires or burns on steep slopes is a common pattern
throughout its range (Veblen et al. 1996). Vegetative
reproduction of any kind has not been previously doc-
umented for
N. pumilio.
Based on our data,
.
50 tree-
falls (15–40 cm dbh) per hectare will survive in a typ-
ical blowdown, and these stems may play an important
role in establishing early vegetation and minimizing
erosion and nutrient loss, particularly near timberline
or in areas heavily browsed by guanacos.
Factors involved in blowdown susceptibility
Based on our treefall data and point return intervals,
the threshold of stand vulnerability for
N. pumilio
oc-
curred between 100 and 125 yr, when trees attained a
mean dbh of
ø
20 cm, basal area of 50 m
2
/ha, and
ø
16–
19 m in height, which is very similar to results reported
from
N. solandri
var.
cliffortioides
in New Zealand
(Jane 1986). Windfirmness is strongly related to stem
taper but tends to decrease with tree size (mass, vol-
ume, and height; Fredericksen et al. 1993). As tree size
increases, the additional strain on the root–soil inter-
face may exceed the soil shear strength (Putz et al.
1983), and trees may have stem and root rots that fur-
ther compromise their strength (Foster 1988
b
, Harris
1989).
There is no consensus in the literature whether old-
growth or even-aged structural characteristics are more
susceptible to blowdowns. Of the 61 treefall samples
we took from blowdowns
,
100 yr old (Fig. 7), only
13% were unequivocally from previously old-growth
stands. In addition, even-aged stands were dispropor-
tionately affected by the 1972 storm.
Nothofagus
es-
tablish in dense, even-aged stands, where trees tend to
grow tall and thin, with poorly tapered stems—all fac-
tors that decrease windfirmness (Oliver and Larson
1990). Wardle (1984) also noted that
Nothofagus
in
New Zealand were more vulnerable to windthrow in
monospecific, even-aged stands.
The landscape distribution of individual blowdowns
indicated that valleys parallel to the storm winds, wind-
ward slopes, and some upper leeward slopes were more
vulnerable to storm damage than other landscape po-
sitions. However, few areas were completely sheltered
and storms displayed only modest levels of discrimi-
nation (Table 2). The return interval model also sug-
gests that blowdowns were likely to
reoccur
over a
wide range of elevation, slope, and aspect. Neverthe-
less, once a blowdown occurred, the odds were in favor
of future blowdowns on the same site. However, the
landscape pattern of blowdown and recovery shifted
over time because of variations among individual
storms and because a small proportion of old-growth
stands were converted to blowdowns and vice versa.
The reciprocal pattern of damage for the intense 1924
and 1972 storms can partly be explained by differences
in wind direction. In addition, some landscape positions
hit hard in 1924, such as ridges, had mostly younger
and less vulnerable stands in 1972 (Table 2). The lack
of strong discrimination of these blowdowns is in
marked contrast to hurricanes in New England and the
Caribbean (Lugo et al. 1983, Foster and Boose 1992,
Boose et al. 1994), where damage is concentrated on
windward aspects and variation can be predicted from
relatively simple wind sheltering models (Boose et al.
1994).
The location and shapes of blowdowns may indicate
the types of destructive winds accompanying these
storms. The association of the 1972 blowdown with
low elevations, moderate slope steepness, and valleys
oriented north–south is consistent with convergence of
laminar flow. Canalization in valleys is accompanied
by an increase in windspeed, particularly as winds are
compressed moving upslope (Buck 1964, Jane 1986,
Harris 1989). The elongate patches from 1972 (Fig.
6A, solid circle labeled ‘‘1’’) were apparently caused
by this phenomenon, comprising almost two-thirds of
the damage. Harris (1989) also reported that most gale
damage in southeast Alaska occurred in valleys and
side slopes parallel to the wind. Turbulent wind flow
may explain the extensive damage on the upper north-
east (leeward) slopes of Cerro Yakush (Fig. 6A). Roll
eddies form downslope of sharp ridges in mountainous
terrain, but any abrupt change in topography will cause
turbulence (Buck 1964, Jane 1986). In New Zealand,
Jane (1986) also reported that wind damage to
Noth-
690 Ecology, Vol. 78, No. 3
ALAN J. REBERTUS ET AL.
ofagus
forests was often concentrated on leeward
slopes.
Although soils were very shallow in the Sierra, we
believe the unstable sandstone-conglomerate bedrock
was a more critical predisposing factor. Root masses
were often heavily laden with rock, indicating possible
slippage. In addition roots may be susceptible to me-
chanical abrasion from wind-rocking in the shallow,
rocky soils, leaving them vulnerable to compression
breaks and/or fungal pathogens (Stone 1977). The as-
sociation of shorter return intervals with deeper soils
runs counter to the majority of the literature (Schaetzl
et al. 1989). The simplest explanation is that trees on
deeper soils grow faster and become vulnerable quicker
(Boyd and Webb 1981, Harris 1989:48), but unfortu-
nately our treefall data were inadequate to address this
hypothesis.
Landscape pattern development
In the Sierra de las Pinturas, the majority of the
landscape was composed of discrete patches of com-
plete blowdown (
.
99% mortality), creating a coarse-
scale mosaic that closely resembled some landscapes
dominated by wildfire. We suspect this reflected not
only the intensity of the storms but also their lack of
discrimination in simple, monospecific, even-aged
stands lacking height differentiation. Canopy emer-
gents and suppressed trees were relatively rare. The
more diffuse damage pattern of hurricanes partly re-
flects differences in susceptibility among species or
canopy positions
within
stands (Foster 1988
b
, Foster
and Boose 1992). Lugo et al. (1983) also noted that
structurally and compositionally ‘‘complex’’ tropical
forests were resistant to hurricane damage. Despite the
intensity of some hurricanes, the damage is best char-
acterized as a heterogenous mixture of patches, mostly
,
2 ha but some much larger,in various damage classes
(Foster and Boose 1992).
Cooper’s (1913) pioneering study of forest succes-
sion on Isle Royale, Michigan, presaged the shifting-
mosaic steady-state concept (sensu Bormann and Lik-
ens 1979) in describing the blowdown patch dynamics
of
Abies balsamea
as a ‘‘mosaic or patchwork [that]
changes continually in a manner that may almost be
called kaleidoscopic.’’ This implied landscape equilib-
rium, while appealing conceptually, has been very dif-
ficult to demonstrate empirically for any disturbance
(Pickett and White 1985, Turner et al. 1993). Turner
et al. noted that definitions and criteria for assessing
equilibrium were not applied consistently, and tem-
poral and spatial scales were often confounded. They
proposed using a spatial parameter (
S
5
ratio of dis-
turbance size to landscape size) and a temporal param-
eter (
T
5
ratio of disturbance interval to the recovery
time required to reach a mature stage). Landscape size
is arbitrary, as long as there are reasonable estimates
for disturbance size and frequency. For our study area,
S
ø
0.04–0.1 (4–100 ha/1000 ha) and
T
ø
0.6–0.7
(150 yr/225–250 yr). Based on simulations by Turner
et al. (1993), these values should permit an equilibrium
or steady state in the proportions of the 1000-ha land-
scape in various age classes. The behavior of damaging
winds in 1972, which followed along previous stand
boundaries, also suggested that blowdown patterns may
exhibit some degree of constancy with respect to patch
location. Similar dynamics are characteristic of some
fire-dominated landscapes, where the juxtaposition of
young and old stands reduces the likelihood of fire
spread and maintains patch diversity (Minnich 1983).
Interactions between guanacos, wind, and
landscape patterns
Although guanacos are known to browse trees (Rae-
decke 1978), their impact on forests has not been pre-
viously documented. Variation in browsing intensity
across the landscape influenced patch development by
creating persistent meadows in accessible and/or highly
preferred habitat. Few small blowdowns created in the
past 50 yr escaped being heavily browsed; however,
for blowdowns
.
200 m across there is a threshold of
obstacles beyond which guanacos gave up and browsed
solely the perimeter. This browsing patterncreated con-
spicuous linear meadows between adjacent stands (Fig.
2), which may persist for centuries because they are
favored by guanacos as travel corridors long after they
cease being a significant source of browse (
unpublished
data
). ‘‘Seedlings’’ survive persistent browsing for at
least 60 yr, eventually becoming hedge-like, but many
areas kept open since 1924 have gradually been con-
verted to meadow and thickets of a thorny shrub,
Ber-
beris buxifolia.
Guanacos had greatest impact near timberline, where
they concentrated in summer for browse and grazing
in the tundra above (Table 3). The near-timberline por-
tions of several blowdowns have been kept open for
.
60 yr, whereas lower parts have regenerated nor-
mally. Regeneration failure probably arose from a com-
bination of browsing pressure, climatic stresses, and
seed dispersal limitations. There was circumstantial ev-
idence of longer-term conversion of blowdowns to al-
pine throughout the Sierra, some of which may be at-
tributable to blowdowns and guanaco browsing: sharp
timberlines lacking krummholz, stranded tree islands,
wedge-shaped alpine intrusions, and dead wood frag-
ments above timberline. Timberline was typically 600–
700 m in nearby interior ranges (A. J. Rebertus,
per-
sonal observation
), but in many peaks of the Sierra de
las Pinturas, blowdowns and guanaco browsing to-
gether have probably depressed timberline
.
200 m be-
low its climatic limit. Indeed, we have found nascent
colonies of
Bolax gummifera,
the dominant alpine
cushion plant growing in old blowdowns 200–300 m
below this plant’s typical alpine limit.
C
ONCLUSIONS
There is growing realization of the importance of
coarse-scale windthrow in forest dynamics worldwide.
April 1997 691
BLOWDOWN HISTORY AND LANDSCAPE PATTERNS
In Tierra del Fuego we found a system where periodic
gales from Antarctica dominated many characteristics
in
Nothofagus pumilio
forests, from microsite to land-
scape. The distribution of patch ages, treefall sizes, and
return intervals across the landscape all indicated that
vulnerability to blowdown increased acutely when
stands reached ages of
ø
100–125 yr. Although major
valleys parallel to the prevailing storm winds, wind-
ward side slopes, and upper leeward slopes appeared
to be most vulnerable to blowdowns, we could not
detect any major differences in return intervals due to
aspect, slope, or elevation. Some areas may have been
more vulnerable to blowdowns, but the cumulative
damage from 200 yr of repeated storms had affected
most landscape positions. Guanacos also played a ma-
jor, synergistic role in the disturbance regime, influ-
encing forest regeneration and creating persistent open
patches along edges of stands and adjacent to timber-
line.
Most disturbance regimes were difficult to charac-
terize because of the multiplicity of disturbance types,
the diffuseness of their effects, and the confounding
influence of historical factors. The overwhelming dom-
inance of wind as the coarse-scale disturbance agent
and the relatively simple forests of our study site on
Tierra del Fuego, however, revealed a relatively clear
relationship between forest patterns and the disturbance
regime.
A
CKNOWLEDGMENTS
We thank Mike Hodgson and Don Cline for help in con-
structing the ortho image and DEM. Sergio DeMarco and
Julio Escobar assisted in the field. For comments and sug-
gestions on the manuscript we thank David Foster, Emery
Boose, and Chris Peterson. Funding for this research was
provided by the School of Natural Resources, University of
Missouri; Geography Department, University of Colorado;
the National Geographic Society; and the National Science
Foundation.
L
ITERATURE
C
ITED
Alvarez Guttie´rrez, S., and H. Grosse Werener. 1979. La
lenga: un recurso manejable. Trapananda 3:45–54.
Armesto, J. J., I. Casassa, and O. Dollenz. 1992. Age struc-
ture and dynamics of Patagonian beech forests in Torres
del Paine National Park, Chile. Vegetatio 98:13–22.
Batschelet, E. 1981. Circular statistics in biology. Academic
Press, London, England.
Boose, E. R., D. R. Foster, and M. Fluet. 1994. Hurricane
impacts to tropical and temperate forest landscapes. Eco-
logical Monographs 64:369–400.
Bormann, F. H., and G. E. Likens. 1979. Catastrophic dis-
turbance and the steady state in northern hardwood forests.
American Scientist 67:660–669.
Boyd, D. M., and T. H. Webb. 1981. The influence of the
soil factor on tree stability in Balmoral Forest, during the
gale of August 1975. New Zealand Journal of Forestry 26:
96–102.
Brokaw, N. V. L. 1985. Treefalls, regrowth, and community
structure in tropical forests. Pages 53–69
in
S. T. A. Pickett
and P. S. White, editors. The ecology of natural disturbance
and patch dynamics. Academic Press, Orlando, Florida,
USA.
Buck, C. C. 1964. Wind over wildlands—a guide for forest
management. U.S. Department of Agriculture Forest Ser-
vice Agriculture Handbook Number 272, Washington,
D.C., USA.
Canham, C. C., and O. L. Loucks. 1984. Catastrophic
windthrow in the presettlement forests of Wisconsin. Ecol-
ogy 65:803–809.
Clark, J. S. 1989. Ecological disturbance as a renewal pro-
cess: theory and application to fire history. Oikos 56:17–
30.
Cole, L. C. 1949. The measurement of interspecific asso-
ciation. Ecology 30:411–424.
Cooper, W. S. 1913. The climax forest of Isle Royale, Lake
Superior, and its development. Botanical Gazette 15:1–44,
115–140, 189–235.
Eastman, J. R. 1992. IDRISI. Version 4.0. User’s guide.
Clark University, Graduate School of Geography, Worces-
ter, Massachusetts, USA.
ERDAS. 1992. Digital Ortho Module. Earth Resources Data
Analysis, Atlanta, Georgia, USA.
Eskuche, U. 1973. Estudios fitosociolo´gicos en el norte de
Patagonia. I. Investigacio´n de algunos factores de ambiente
en comunidades de bosque y de chaparral. Phytocoenologia
1:164–213.
Foster, D. R. 1988
a.
Disturbance history, community orga-
nization and vegetation dynamics of the old-growth Pisgah
Forest, south-western New Hampshire, U.S.A. Journal of
Ecology 76:105–134.
. 1988
b.
Species and stand response to catastrophic
wind in central New England, U.S.A. Journal of Ecology
76:135–151.
Foster, D. R., and E. R. Boose. 1992. Patterns of forest
damage resulting from catastrophic wind in central New
England, U.S.A. Journal of Ecology 80:79–98.
Fox, J. F. 1989. Bias in estimating forest disturbance rates
and tree lifetimes. Ecology 70:1267–1272.
Fredericksen, T. S., R. L. Hedden, and S. A. Williams. 1993.
Testing loblolly pine wind firmness with simulated wind
stress. Canadian Journal of Forest Research 23:1760–1765.
Glitzenstein, J. S., and P. A. Harcombe. 1988. Effects of the
December 1983 tornado on forest vegetation of the Big
Thicket, Southeast Texas, U.S.A. Forest Ecology andMan-
agement 25:269–290.
Goodall, R. N. C. 1979. Tierra del Fuego. Ediciones Shan-
amaiim, Buenos Aires, Argentina.
Harmon, M. E., S. P. Bratton, and P. S. White. 1983. Dis-
turbance and vegetation response in relation to environ-
mental gradients in the Great Smoky Mountains. Vegetatio
55:129–139.
Harris, A. S. 1989. Wind in the forests of Southeast Alaska
and guides for reducing damage. U.S. Department of Ag-
riculture Forest Service General Technical Report PNW-
244. Pacific Northwest Forest and Range Experiment Sta-
tion, Portland, Oregon, USA.
Heinselman, M. L. 1973. Fire in the virgin forests of the
Boundary Waters Canoe Area, Minnesota. Quaternary Re-
search 3:329–382.
Heusser, C. J. 1987. Fire history of Fuego-Patagonia. Qua-
ternary of South America and the Antarctic Peninsula 5:
93–109.
Jane, G. T. 1986. Wind damage as an ecological process in
mountain beech forests of Canterbury, New Zealand. New
Zealand Journal of Ecology 9:25–39.
Johnson, E. A., and C. E. Van Wagner. 1985. The theory
and use of two fire history models. Canadian Journal of
Forest Research 15:214–220.
Lambert, S. J. 1993. Southern Hemisphere cyclone clima-
tology and the greenhouse effect. Pages 320–321
in
Fourth
International Conference on Southern Hemisphere Mete-
orology and Oceanography, 29 March–2 April 1993, Ho-
bart, Australia.
Lorimer, C. G. 1977. The presettlement forest and natural
692 Ecology, Vol. 78, No. 3
ALAN J. REBERTUS ET AL.
disturbance cycle of northeastern Maine. Ecology 58:139–
148. . 1985. Methodological considerations in the analysis
of forest disturbance history. Canadian Journal of Forest
Research 15:200–213.
Lugo, A. E., M. Applefield, D. J. Pool, and R. B. McDonald.
1983. The impact of Hurricane David on the forests of
Dominica. Canadian Journal of Forest Research 13:201–
211.
Lugo, A. E., and R. B. Waide. 1993. Catastrophic and back-
ground disturbance of tropical ecosystems at the Luquillo
Experimental Forest. Journal of Bioscience 18:475–481.
Lyons, S. K. 1983. Characteristics of intense Antarctic De-
pressions near the Drake Passage. Pages 238–240
in
First
International Conference on Southern Hemisphere Mete-
orology (Sao dos Campos, Brazil, 31 July–6 August 1983).
Minnich, R. A. 1983. Fire mosaics in southern California
and northern Baja California. Science 219:1287–1294.
Moore, D. M. 1983. Flora of Tierra del Fuego. Missouri
Botanical Garden, St. Louis, Missouri, USA.
Mutarelli, E. J., and E. N. Orfila. 1973. Algunos resultados
de las investigaciones de manejo silvicultural que se real-
izan en los bosques andino-patago´nicos de Argentina. Re-
vista Forestal Argentina 17:69–75.
Nelson, B. W., V. Kapos, J. B. Adams, W. J. Oliveira, O. P.
G. Braun, and I. L. do Amaral. 1994. Forest disturbance
by large blowdowns in the Brazilian Amazon. Ecology 75:
853–858.
Norton, D. A., and J. Ogden. 1990. Problems with the use
of tree rings in the study of forest population dynamics.
Pages 284–288
in
E. R. Cook and L. A. Kairkiukstis, ed-
itors. Methods of dendrochronology. Kluwer Academic,
Dordtrecht, Netherlands.
Oliver, C. D., and B. C. Larson. 1990. Forest stand dynamics.
McGraw-Hill, New York, New York, USA.
Peet, R. K. 1992. Community structure and ecosystem func-
tion. Pages 103–151
in
D. C. Glenn-Lewin, R. K. Peet, and
T. T. Veblen, editors. Plant succession: theory and predic-
tion. Chapman and Hall, London, England.
Pickett, S. T. A., and P. S. White. 1985. Patch dynamics: a
synthesis. Pages 371–384
in
S. T. A. Pickett and P. S. White,
editors. The ecology of natural disturbance and patch dy-
namics. Academic Press, Orlando, Florida, USA.
Putz, F. E., P. D. Coley, K. Lu, A. Montalvo, and A. Aiello.
1983. Uprooting and snapping of trees: structural deter-
minants and ecological consequences. Canadian Journal of
Forest Research 13:1011–1020.
Raedecke, K. 1978. El guanaco de Magallanes, Chile. Su
distribucio´n y biologı´a. Corporacio´n Nacional Forestal
Publicacion Te´cnica Numero Quatro, Santiago, Chile.
Rebertus, A. J., and T. T Veblen. 1993. Structure and tree-
fall gap dynamics of old-growth
Nothofagus
forests in Ti-
erra del Fuego, Argentina. Journal of Vegetation Science
4:641–654.
Romme, W. H. 1982. Fire and landscape diversity in sub-
alpine forests of Yellowstone National Park. Ecological
Monographs 52:199–221.
Runkle, J. R. 1985. Disturbance regimes in temperate forests.
Pages 17–34
in
S. T. A. Pickett and P. S. White, editors.
The ecology of natural disturbance and patch dynamics.
Academic Press, Orlando, Florida, USA.
. 1990. Gap dynamics in an Ohio
Acer-Fagus
forest
and speculation on the geography of disturbance. Canadian
Journal of Forest Research 20:632–641.
SAS. 1988. SAS/STAT user’s guide. SAS Institute, Cary,
North Carolina, USA.
Savage, M., M. Reid, and T. T. Veblen. 1992. Diversity and
disturbance in a Colorado subalpine forest. Physical Ge-
ography 13:240–249.
Schaetzl, R. J., D. L. Johnson, S. F. Burns, and T. W. Small.
1989. Tree uprooting: review of terminology, process, and
environmental implications. Canadian Journal of Forest
Research 19:1–11.
Schmidt, H., and A. Urzua. 1982. Transformacio´n y manejo
de los bosques de lenga en Magallanes. Ciencias Agrı´colas
(Universidad de Chile, Santiago) Numero 11.
Shaw, W. B. 1983. Tropical cyclones: determinants of pattern
and structure in New Zealand’s indigenous forests. Pacific
Science 37:405–414.
Stone, E. L. 1977. Abrasion of tree roots by rock during
wind stress. Forest Science 23:333–336.
Tuhkanen, S. 1992. The climate of Tierra del Fuego from a
vegetation geographical point of view and its ecoclimatic
counterparts elsewhere. Acta Botanica Fennica 145:1–64.
Turner, M. G., W. H. Romme, R. H. Gardner, R. V. O’Neill,
and T. K. Kratz. 1993. A revised concept of landscape
equilibrium: disturbance and stability on scaled landscapes.
Landscape Ecology 8:213–227.
Veblen, T. T. 1985. Forest development in tree-fall gaps in
the temperate rainforests of Chile. National Geographic
Research 1:162–183.
Veblen, T. T., C. Donoso, T. Kitzberger, and A. J. Rebertus.
1996. Ecology of southern Chilean and southern Argen-
tinian
Nothofagus
forests. Pages 293–353
in
T. T. Veblen,
R. S. Hill, and J. Read, editors. Ecology and biogeography
of
Nothofagus
forests. Yale University Press, New Haven,
Connecticut, USA.
Veblen, T. T., T. Kitzberger, and A. Lara. 1992. Disturbance
and vegetation dynamics along a transect from rain forest
to Patagonian shrublands. Journal of VegetationScience 3:
507–520.
Wardle, J. A. 1984. The New Zealand beeches: ecology,
utilisation, and management. New Zealand Forest Service,
Christchurch, New Zealand.
Webb, L. J. 1958. Cyclones as an ecological factor in tropical
lowland rain forest, North Queensland. Australian Journal
of Botany 6:220–228.
White, P. S., and S. T. A. Pickett. 1985. Natural disturbance
and patch dynamics: an introduction. Pages 3–13
in
S. T.
A. Pickett and P. S. White, editors. The ecology of natural
disturbance and patch dynamics. Academic Press, Orlando,
Florida, USA.
Whitmore, T. C. 1984. Tropical rain forests of the Far East.
Clarendon, Oxford, England.
Whitney, G. G. 1986. Relation of Michigan’s presettlement
pine forests to substrate and disturbance history. Ecology
67:1548–1559.
