International Association for Ecology

Shrub Clumps of the Chilean Matorral Vegetation: Structure and Possible Maintenance
Mechanisms
Author(s): Eduardo R. Fuentes, Ricardo D. Otaiza, M. Catalina Alliende, Alicia Hoffmann, Aldo
Poiani
Source: Oecologia, Vol. 62, No. 3 (1984), pp. 405-411
Published by: Springer in cooperation with International Association for Ecology
Stable URL: http://www.jstor.org/stable/4217336 .
Accessed: 18/06/2011 15:34

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=springer. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Springer and International Association for Ecology are collaborating with JSTOR to digitize, preserve and
extend access to Oecologia.

http://www.jstor.org
Oecologia (Berlin) (1984) 62:405-411
Shrubclumpsof the Chileanmatorralvegetation:
structureand possiblemaintenancemechanisms
EduardoR. Fuentes,Ricardo D. Otaiza, M. CatalinaAlliende,Alicia Hoffmann,and Aldo Poiani
Facultad de Ciencias Biologicas, Pontificia Universidad Catolica de Chile, Casilla 114-D, Santiago, Chile
Summary. Previous studies have claimed that the Chilean
matorral is more open than the Califonia chaparral, and
have attributed this dissimilaritylargely to the role of man
in Chile. In this paper we show that in general the Chilean
matorral has a structure better described as shrub clumps
that merge to form a continuous vegetation matrix only
in very mesic habitats, where it is comparable to the Cali-
fonia chaparral.We also presentevidence that these clumps
have been present for at least the last 26 years and that
even without human disturbancethey are likely to maintain
themselves. Evidence for the latter pertains to seed dispers-
al, seed germination and establishment, seedling survival,
and the diameter size structure of shrub clumps. Finally,
we propose that differencesbetween the California chapar-
ral and Chilean matorral are more profound than pre-
viously thought and are due not only to different degrees
of human disturbance,but also to the presenceof periodical
natural fires in California and not in Chile, and to different
shrub recruitmentpatternsand mammalianherbivoreactiv-
ity in the two areas.
Introduction
Studies comparing areas of mediterranean-typeclimate in
Chile and California have shown some remarkableconver-
gences, particularly in vegetation physiognomy and shrub
species morphology (Throwerand Bradbury1977; Mooney
1977; Cody and Mooney 1978). However, those studies
also indicated some important differencesbetween the med-
iterranean-type shrublands in the two areas. In general,
Chilean sites exhibit higher species diversity of shrubs,
lower total shrub cover, and higher herb cover than compa-
rable California localities. The overall aspect of the Califor-
nia chaparralis that of an impenetrablescrubbyvegetation,
compared to the more open and structurallydiverse aspect
of the Chilean matorral. These comparisons refer only to
"'mature" sites since, both in Californiaand in Chile, direct
human disturbance and fires are known to create open
spaces, changes in vegetation and thus greater diversity
(Aschman and Bahre 1977; Trabaud 1981; Conrad and
Oechel 1982; Armesto and Gutierrez 1978).
The differences between the mature shrub communities
in Chile and California have been attributed to the dissimi-
lar treatmentsof these vegetation types by man rather than
Offprint requests to: E. R. Fuentes
Oeclogla
? Springer-Verlag 1984
different evolutionary responses of the vegetation to nearly
identical environmental conditions (Mooney et al. 1972;
also Mooney 1977; and referencestherein).This interpreta-
tion suggests that, in the absence of direct human impact,
the Chilean matorral would become more similar to the
California chaparral.
In this article we will show that the structure of the
Chilean matorralcan be better describedas being composed
of clumps of shrubs rather than as a continuous shrub ma-
trix with some man-made gaps. Moreover, we will suggest
that independently of the natural or man-made origin of
the clumps they have the potential for maintaining them-
selves through time, even in the absence of human distur-
bance.
Materials and methods
Observationsand experimentstook place in the Andes foot-
hills (ca. 1,000 m above sea level), between the Mapocho
and Maipo rivers, east of Santiago. Matorral vegetation
in this area (see Hoffmann and Hoffmann 1982) is at ap-
proximately the same altitude and latitude as the site of
previous convergence studies (see Thrower and Bradbury
1977; Mooney 1977) and has the same main species of
shrubs (see also Mooney et al. 1972).
Clumps
Vegetation clumps were measured on 1: 5000 aerial photo-
graphs produced from a 1980 flight by the Chilean Service
of Aerophotogrammetry.The 1980 photographs were com-
pared with similar 1:5000 photographs produced from a
1956 flight by the same service. Measurements of relative
clump size were made by the line transect method (Muller-
Dumbois and Ellenberg 1974). A 50-mm ruler, equivalent
to a 250-m transect, was randomly placed 10 times on each
polar-facing slope (moist-cool), and each equatorial-facing
slope (dry-hot). In valley bottoms this procedure was re-
peated 5 times at each site. In total eight equatorial-facing
slopes, eight polar-facing slopes, and five valley bottoms
were measured. Our aim was to find out if clumps tend
to be larger on one or the other type of slope. Only areas
that had not previously been cleared were chosen for these
measurements.However, as we will mention later, in some
of these areas old trunksexhibiting signs of previouscutting
and subsequent resproutingwere found.
In addition, an area of 0.80 ha on a small disturbed
valley bottom, just east of Santiago (Los Dominicos), was
406
directly censused in the field and the species and diameter
composition of the shrub clumps sampled during January
(summer) 1982. Special care was taken during these cen-
suses to avoid counting as separate individuals what could
be small ramets of a larger plant. Usually examination of
the soil up to a depth of 4-5 cm was enough to discriminate
the ramets.
Seed and seedlingshadows
Throughout the 1981-1982 season, seed shadows for the
nine most common shrub species were evaluated by sam-
pling the relative seedfall on 40 x 40 cm plates smearedwith
Tangle-Trap. Twenty such plates were randomly placed in
the open areas between clumps and five of them were lo-
cated under five randomly chosen shrub clumps, mostly
composed of Lithraea,Quillaja,Colliguaya,or Azara. These
plates with Tangle-Trap were examined and replacedevery
month.
In addition, the seed bank in the top 1 cm of soil be-
tween clumps, as well as under clumps, was estimated in
May 1982, at the end of the seed dispersal season. Soil
samples of 40 x 40 x 1 cm were randomly collected from ar-
eas between clumps and under shrubs of Lithraea, Colli-
guaya, Quillaja, and Azara. These samples (n= 10 in each
case) were dried, sifted with a 1.5-mm2 sieve, and hand
sorted for shrub seeds.
At Los Dominicos we also tested the hypothesis
(McDonnell and Stiles 1983) that large emergent objects
coul attract birds dispersing fruit and thus positively bias
the seed bank around already establishedclumps. We mea-
sured the relative seed accumulation around 20 poles of
2.5 m height, placed in open areas between shrubs on an
equatorial-facing slope and on a polar-facing slope (n =10
in each case). The base of each of these poles was protected
from large herbivores with a 1-m2 cage of chicken-wire
mesh. We also measured seed accumulations on selected
large rocks in the same two sites, and the accumulation
on ground-levelcontrol areas of similar surface area within
I m from each rock.
Seedling shadow were evaluated in two ways. During
the 0.80 ha survey, we recorded the position and identity
of all seedlings. In addition, between October and De-
cember 1981 (springand early summer)we searchedsystem-
atically for seedlings on a much wider scale. We set up
1-m-wide radial transects centred on the periphery of 85
randomly chosen shrub clumps and carefully examined the
herb and grass cover for shrub,seedlings. These 85 clumps
were located on the equatorial-facingslope, the polar-facing
slope, and on the valley bottom between them. These tran-
sects were extended for only 5 m from the periphery of
the clumps because beyond this distance we were either
approaching another clump or the number of seedlings was
zero for several consecutive metres.
Seed germinationand seedling establishment
We designed the following experimentsto test the null hy-
pothesis that seed germination and later seedling establish-
ment does not differ between clumps and under them. In
May 1982 (fall) seeds of Quillaja and Colliguaya were
placed in five different shallow wooden boxes containing
soil from a between-clumpsarea. These boxes, covered with
a wire mesh and containing in total 100 seeds of Quillaja
and 15 seeds of Colliguayawere placed either under plants
of Lithraea(n = 5), plants of Colliguaya(n = 5), or an open
area between shrubs (n = 5). The boxes were periodically
monitored and the total number of seedlings emergingand
later establishedwas recorded.
Seedlingsurvival
Experiments were also designed to test if the death rate
of seedlings protected with a small chicken-wiremesh cage,
which excluded mammalian herbivores, would be lower
than the death rate of unprotectedseedlings.
During October (early spring) 1981, 20 small (3-5 cm)
seedlings of each of three species (Quillaja, Lithraea, and
Maytenus)were planted in each of the following four situa-
tions:
A. Open space between clumps, protected with the wire
mesh cages mentioned above. These seedlings were pro-
tected from defoliation by large herbivores, but not from
desiccation.
B. Open space between clumps, not covered with the
cages.
C. Under clumps and protectedwith the wire mesh cage,
and therefore protected from herbivores and presumably
under less hydric stress than protected seedlings between
shrubs.
D. Under clumps unprotectedand thereforepresumably
under lower relative hydric stress and perhaps "hidden"
by the clumps.
The planted seedlings, about 5 cm tall, were of the same
size as the natural seedlings that we found at that time.
All the seedlings were inspected periodically to determine
if they were still alive, showed signs of browsing, or were
dead from eitherdesiccationor herbivoreattack. Frequency
distributions of time before death were later analyzed to
determine the statistical significance of the various treat-
ments. Since these frequencydistributions,even after trans-
formations, are still significantly skewed and leptokurtic,
nonparametric statistics were used in the analyses (Siegel
1956).
Results
Clumpstructure
Both in 1956 and in 1980, clumps on the equatorial-facing
slopes were relatively smaller, on average, than clumps in
the valley bottoms and these in turn were relativelysmaller
than clumps on the polar-facingslopes (t-tests on log-trans-
formed data, after correction for multiple t-tests, give a
rejection level of 0.001. See Table 1). Our measurements
Table 1. Clumpsizes. Averageclump sizes (in metres)on nine equa-
torial-facingslopes, eight polar-facingslopes, and five alluvial fans,
for 1956 and 1980. Standarddeviations are shown in parentheses
Year Exposure
Equatorial-facing Valley bottoms Polar-facing
slopes slopes
(n=8) (n=5) (n= 8)
1956 7.80 (1.66) 12.79 (9.52) 31.93 (30.32)
64.75 (6.40)
1980 7.30 (2.40) 18.60(16.60)
 407

Table 2. Shrub age-size distributions. Number of seedlings, juveniles and adult plants of each of five species of shrubs found under
the foliage projectionof nurse plants of five species (rows)

Nursed SeedlingsC Juvenilesc Adultsc

Qs. Lc. Ac. Co. Ad. Qs. Lc. Ac. Co. Ad. Qs. Lc Ac. Co. Ad.

Q. saponaria 0 5 0 3 1 0 2 0 2 0 18 15 1 11 5
n= 59
L. caustica 1 2 0 0 1 0 0 1 1 0 9 0 1 22 0
n=21
A. caven 0 0 0 1 0 0 0 2 0 0 0 0 1 0 0
n=13
C. odorifera 0 0 0 2 0 0 0 0 0 0 0 0 0 1 0
n=10
A. dentata 0 1 0 0 0 0 0 0 0 0 0 0 0 3 2
n=4
Solitary b 0 0 0 0 0 0 1 0 0 0 20 3 21 71 5

a n is the numberof nurse plants of each species found

b "Solitary" referes to plants not associated with a clump. Notice that most plants belonged to a clump and that there were no
solitary seedlings and just one solitaryjuvenile in the whole survey
c Age-classes were arbitrarilydefined in relation to maximum trunk diameter at ground level in the following way: (1) L. caustica
and Q. saponaria(seedlings ?0.9 cm, juveniles _4 cm, adults >4.1 cm); (2) for all other species (seedlings <0.9 cm, juveniles ?3 cm,
adults >3.1 cm)
d
Qs. = Quillajasaponaria,Lc. = Lithraeacaustica,Ac. = Acacia caven,Co. = Colliguayaodorifera,Ad. = Azara dentata

suggest that this difference between exposure patterns has
been present at least since 1956. Matched comparisons of
our estimates of clump sizes on slopes with the same expo-
sure, in 1956 and 1980, did not show overall statistically
significant differences(t-test, P>0.15 in all three cases).
Clumpcomposition
Field sampling in the valley bottom of Los Dominicos re-
vealed that clumps tend to be multispecific and frequently
to have more than one class of trunk diameter (Table 2).
In fact, the plant with the largest trunk, and thus presum-
ably the oldest (Kummerow et al. 1981) in each patch,
".nurses"not only other large plants, but also younger indi-
viduals, suggesting that the clumps we see today could also
persist through the next generation. Inspection of Table 2
indicates that the number of individuals in the various age-
size classes is irregular, suggesting that it is only in some
years that establishment and survival of seedlings is possi-
ble.
Further examination of Table 2 shows: (i) that solitary
individuals are infrequent, and (ii) that the smaller size
classes of all five species are never solitary. No shrub seed-
lings except for Baccharisspp. (not shown in Table 2) were
found in the areas between clumps. Baccharisspp. are fairly
ubiquitous small "'weedy"matorralshrubs. Thus in general
in and around clumps several size classes of shrub can be
found together, but in the open spaces between clumps this
is never the case.
Analysis of all clumps in the 0.80 ha sample, including
those shown in Table 2 and others, suggests that species
associations in clumps might not be random. In fact, for
the three most common shrubs - Quillaja, Lithraea, and
Colliguaya - the relative abundance in the 0.80 ha sample
and the derived expected frequenciesin clumps of all possi-
ble species combinations, of up to six individuals, do not
match in all cases. Chi-square analyses show that Lithraea
in association with both Quillajaand Colliguayaoccurs less
frequently than expected from its overall abundance in the
0.80 ha sample (P< 0.001). On the other hand, Colliguaya
alone, and associations of Lithraeaand Quillaja,occur more
frequently than expected by chance (P<0.001). All other
associations have expected and observed frequencies that
do not differ significantly. These results suggest that the
species composition of clumps is the result not only of sto-
chastic elimination of the plant cover, but at least in part
the outcome of species interactions.
As we mentioned earlier, at Los Dominicos, as well as
in other valleys, some old trunks showed evidence of old
cuts. Most of these trunks, however, had resprouted and
belonged to one of the clumps.
Seed shadows
Table 3 shows the relativeseed accumulationon the Tangle-
Trap smeared board and in the soil seed bank. Although
seeds of Baccharis spp. were the most abundant seeds in
the open areas between clumps, seeds of at least seven other
species also reach these areas. There is a soil seed bank
for Lithraeaand Quillajain open areas. With the exception
of seeds of Schinus polygamus, seeds of any species were
found under clumps that were not also present between
clumps. Seeds of Lithraea, Quillaja, Colliguaya and Bac-
charis can be found under all clumps as well as between
clumps.
Although the small numbers do not allow quantitative
comparisons, they suggest that the seed potential exists for
three and perhaps for eight species, to be present under
clumps as well as between clumps.
Table 3 reveals a tendency for higher seed numbers to
be present under clumps than in the open spaces between
them. In fact, excluding the ubiquitous Baccharisseeds, for
408

Table 3. Seed shadows and seed bank. For each of the nine species (rows) we show the between-clumpas well as under-clump(dominated
by any one of four species of shrubs) number of seeds collected on a standardized per five samples basis. Each sample consisted
of a complete survey of 40 x 40 cm. Each entry consists of two values: the first one is the average number of seeds found on our
Tangle-Trapsmearedplates; the second value in parenthesesis the average number of seeds found in the soil

Seed species Between clumps Under clumps

Lithraea Quillaja Colliguaya Azara Average/Clump

A. dentata 0.25 (0) 0 (0) 0 (0) 0 ( 0) 360 (259.5) 90 (64.88)

Baccharisspp. 102.5 (0) 13 (0.50) 25 (0) 21 ( 0) 0 ( 1) 14.75 ( 0.38)
C. odorifera 5 (1.5) 5 (3) 17 (6.5) 31 (15) 1 ( 3) 13.5 ( 6.88)
L. caustica 1.25 (4.5) 5 (17.5) 7 (2.5) 0 ( 1) 0 ( 4.5) 3 ( 6.38)
M. hastulata 0.25 (0) 0 (0) 0 (4.5) 0 ( 0.5) 0 ( 1.5) 0 (1.5)
P. cuneifolia 1.25 (0) 0 (0) 0 (0) 13 ( 0) 0 ( 0) 0 ( 3.25)
Q. saponaria 0.25 (0) 0 (0.50) 13 (68.5) 1 ( 0) 2 ( 4.5) 4 (18.38)
S. polygamus 0 (0) 0 (0) 6 (0) 0 ( 0) 0 ( 0) 1.50 (0)
T. trinervis 0.50 (0) 0 (1.5) 2 (0.50) 0 ( 0) 0 ( 0) 0 ( 0.50)

five out of eight species in the case of the smeared boards, - 120
a) ~
and for all eight species when the seed bank was measured A
100 o --o
directly, there were more seeds under clumps than between 0~~~~~~
them. For wind-dispersed seeds (Quillaja and Proustia cu- E 80 0
0
neifolia), or for autochorous ones (Colliguaya), this higher E
concentration might be due to interception by, or reduced c: 60
wind speed in, the clumps. For bird-dispersedseeds (Lith- 0)
raea, Azara, Trevoatrinervis,and Muehlenbeckiahastulata),
the greater seed density around clumps requires a different o
explanation. It seemed likely that clumps acted as perching o
DA0 /0

sites for birds carrying shrub seeds and this produced an < 400
0
accumulation of seeds under them. We used the seed accu- 0
mulation rates around rocks and poles to test this hypothe- a) 20
sis. A cumulative count, derived from our weekly records, 100
of seeds found on ten large rocks is shown in Fig. 1a. Only -o B
seeds of Lithraea were found in all three situations. Seeds 80 -
were removed after each count. The number of seeds found
on the rocks at the next visit, i.e. immigration, first increases
and then decreases (Fig. 1b). Control areas of similar size
about 1 m from each of the "experimental" rocks did not
accumulate any seeds. Under the experimental poles the
net number of Lithraea seeds recorded on each visit also
tends to increase and then decrease again as seen around
the rocks (Fig. 1c). Control areas about 1 m away from 2 0-
the poles did not accumulate any seeds at all. The concen- 0
tration of seeds around poles was more pronounced on a) 2 c
the equatorial-facing than on the polar-facing slopes. It is 40
,n
likely that these differences are related to the greater abun- -Q
B) -

dance of other, "competing", perching sites in the more

densely vegetated polar-facing slope. o1 10-
c

Seedlingshadows Jun Jul Aug Sep Oct

Results of the radial transects of our detailed search for
seedlings are shown in Fig. 2. Seedlings of Quillaja, Lith-
raea, Kageneckia,Colliguaya, Muehlenbeckia,and Solanum
tomatillo tended to be more numerous under or around
clumps than far from them. Of this group, Kageneckia,
found only on the wet, lush, polar-facing slopes, was the
species found farthest from clumps. In contrast, seedlings
of Baccharis showed no tendency to be aggregated around
clumps in any of the three exposure situations. These re-
sults, in close agreement with our findings in the detailed
0.80-ha survey and with the predictions from the seed shad-
Time
Fig. 1A-C. Seed accumulation. A Cumulative count of Lithraea
seeds arriving between June and October 1981 of 10 large rocks.
Light and dark dots are north and south facing slopes respectively.
Seeds were removed after each count. Comparable control areas
nearby did not accumulate any seeds al all. B Net number of Lith-
raea seeds found in each visit to 10 rocks on a north-(dark) and
a south-(light) facing slope. Here seeds were not removed after
each visit as in A and the shape of histogramindicates seed arrivals
as well as seed losses. C Net number of seeds under 20 poles.
Light and dark columns indicate south- and north-facing slopes
respectively. Again, as in A and B, control areas nearby did not
accumulateany seeds at all
 409

For seedlings between shrubs, those covered with wire

60 Quillaja 60 Lithraea
-
n=112 n=105 mesh had significantly (P <0.001) longer survival times
-40-
than uncovered ones, suggesting that herbivores and desic-
4
cation are important sources of mortality in the areas be-
20 - 20 -
tween clumps. Of the 60 unprotected seedlings, 37 died at
least in part due to herbivore damage, whereas the remain-
ing 23 exhibited evidence only of desiccation.
-3-2-1 1 2 3 4 5 -3-2-1 1 2 3 4 5 Wire-mesh-protected seedlings under shrubs survived
Distance (m) Distance (m) significantly longer (P< 0.001) than protected seedlings be-
tween shrubs. This suggests that shrubs provided significant
60 - Kageneckia 60 - Baccharis protection against drought. Under clumps the survival time
-
n=116
- n=158 of seedlings did not differ significantly (P>0.18) between
g40 - 40 - protected and unprotected.
These results suggest that shrub clumps protect and en-
20 -20- hance the survival of seedlings. Both herbivory and desicca-
tion appear to be important causes of mortality for seed-
lings in the open spaces between clumps and both of these
-3 -2- 1 2 3 4 5 0-3 -2-1 1 2 3 4 5
hazards are amelioratedby the clumps. Thus, independently
Distance (m) Distance (m)
of the seed bank, seed germination, and establishmentrates,
80- seedlings under clumps have significantly higher chances
of surviving throughout the dry summer than seedlings
0 -
60 MuehLenheckia Coltiguaya SoLanum
~n=28 n=21 n4 growing in the open.
~40-
70 A 70- B
20- 50 x=9.7 50 -

0 ?30 1 -30 - x=33.3

-3-2-1 1 2 -2-1 1 2 1 2
Distance (m) Distance (m) Distance (m) 10 11o -j11t ..
Fig. 2. Seedlings and nurses. The seven histograms show the per- 10 30 50 70 90 110 10 30 50 70 90 110
cent number of seedlings of seven shrub species found at various Days Days
distances from the edge of the nurse's canopy. Arrows indicate 70- 70-
the edge and from there negative (towards the trunk) or positive c D
(away from the trunk) distances were taken. Sample size (n) is 50 - 50 - x3
also shown in each case
30 - x=23.0 30-
ow results, suggest that the next generation of shrubs is 10 10
also likely to form clumps.
10 30 50 70 90 110 10 30 50 70 90 110
Days Days
Seed germinationand seedling survival Fig. 3A-D. Seedling Survival Times. Percent seedlings surviving
to various times are shown. In each one of the four cases 60 seed-
A two-way ANOVA on previously transformeddata (arcsin lings were planted on the same day in October 1981. Results of
function) of the percentage of seeds that germinated and experiments with (C, D) and without (A, B) protection against
established on the experimental boxes showed significant herbivores as well as under a nurse plant (B, D) or in the open
differences by the clump versus no clump criterion (P's< (A, C) are shown. Survival time is on the average lowest for sun-
0.01), but no significant difference by the seedling species exposed unprotected seedlings (x = 9.7 days) and highest for seed-
criterion (Quillaja,Lithraea, Colliguaya).There was no sig- lings under a nurse (x = 33.3 and 34.8 days). Notice that seedlings
nificant interaction. Between clumps about 54% of the without a nurse on the average more than double their life expec-
seeds germinated and established whereas under shrubs tancy when protected against herbivores
only 29% did so. These results suggest that not only are
seeds dispersed to the areas between clumps, but also that Discussion and conclusions
here they have higher chances of germination and establish-
ment than under clumps. Our results show that the natural shrubby vegetation on
Since results of the experiments to test survival time the Andes east of Santiago forms clumps that vary in size
under clumps or between them did not show any among- with exposure. In aerial photographs taken in 1956 and
species differences (Kruskal-Wallis ANOVA, P>0.46), in 1980, equatorial-facing slopes tend to have smaller clump
the rest of the analysis we consider all species of seedlings sizes, valley bottoms have patches of intermediate size, and
in a single category. the polar-facing slopes tend to have the largest clumps.
A Mann-Whitney U-test on the survival times (Fig. 3) In spite of the 24-year period between the photographs,
shows a significantly (P<0.001) higher average survival not only was the relationship between slope and clump size
time for unprotected seedlings under shrub clumps com- maintained, but clumps on slopes of the same exposure
pared to those between shrubs, suggesting that vegetation tended to be of similar size in 1956 and in 1980. Our data
clumps provide seedlings with protection against herbi- do not suggest a tendency for clumps to increase or decrease
vores, desiccation, or both. in size.
410
Moreover, the photographs and our more detailed re-
sults from the Los Dominicos valley suggest that clumps
have not only been part of the structure of the shrubby
vegetation for the last 26 years but that they might continue
as part of it in the future. Twenty-six years is a long time
relative to the establishment and growth times of shrubs
in Chile and in some other areas of mediterraneanclimate
(Conrad and Oechel 1982; Godron et al. 1981). It is only
on the mesic polar-facing slopes that clumps are very large
and seem to merge into one another to form a fairly contin-
uous cover. It should be pointed out that the convergence
studies previously mentioned were made at an even more
mesic site, on the coastal ranges (Thrower and Bradbury
1977), where the shrubbyvegetation tends to form a contin-
uous matrix, unless disturbed by man. In that situation
human disturbance was, correctly, claimed to be the most
important factor producing openness in the matorral.
Our results suggest that in additon to direct human dis-
turbance (clearing areas for agricultureor opening trails),
in some areas there may be other reasons for the openness
of the matorral.Our finding that some clumps at Los Domi-
nicos have species compositions that depart significantly
from random supports the idea that clumps are not merely
the results of non-specific human disturbance in the past,
but ratherthe outcome of some definite interactionsamong
species. We still do not know how the clump patterns were
originally established. They might be natural (have an ori-
gin that is independent of human disturbance) although
perhaps later modified by man's activities. However, the
alternative hypothesis that the pattern is all the product
of human influence cannot be completely discardedat pres-
ent.
At present clumps appear capable of maintainingthem-
selves through time. The fact that the smaller size-classes
of most shrubs are strongly associated with "nursing"
clumps seems to be the result of at least two processes.
First, clumps attract birds that carry shrub seeds, and inter-
cept wind-borne seeds, which accumulate around clumps.
These effects combine to produce the observed seed and
seedling shadows around preexisting clumps. The second
process that accounts for the restricted distribution of
young plants is related to the protection given by clumps
to seedlings. The risks of death due to herbivore grazing
and to desiccation are significantly less under clumps than
between them. Elsewhere (Fuentes et al. 1983) it has been
shown that Europeanrabbits(Oryctolaguscuniculus),which
in central Chile prefer to graze in open spaces between
shrubs (Jaksicet al. 1979), browse and kill even much larger
seedlings growing in the open spaces. In contrast, native
defoliating rodents, mostly degus (Octodon degus), forage
most frequently under or around clumps and thus impose
predation risks only on seedlings growing in the dome-
shaped cavity under these large shrubs (Fuentes and Simon-
etti 1982).
Although our resultsshow that the survivaltime of seed-
lings increases among the low branches of large shrubs,
none of these seedlings survived throughout the dry season.
Perhaps it is only in unusually wet years that seedlings can
survive through this stressful period. The dry season in the
area of Chile with a mediterranean-typeclimate is highly
variable and in some years it rains I or 2 mm even in the
middle of the summer (Miller et al. 1977; Fuentes and Ha-
jek, unpublished). Irregularityin rainfall and seedling sur-
vival through the summer could explain, at least in part,
our findings regardingthe size-classstructureof individuals
in clumps.
In areas like Los Dominicos, where large shrubs were
sometimes cut by man, this affected the largest individuals
and the clump structureof the vegetation was not altered.
Most matorral shrubs, and certainly Lithraea,Quillajaand
Colliguaya,are capable of resproutingafter cutting (Monte-
negro et al. 1982). Hence, when the largest shrubs are cut
for fuel or charcoal, the perturbationis compensatedwithin
a few growing seasons and the old aspect of the clump
is regained.The structurethat we describeis thereforefairly
resilient to wood-cutting and is not expected either to be
eliminated or generatedby wood removal activities.
Goats, horses, or even cows are other large perturbating
agents in the matorral,and if kept at relativelylow densities
would probably enhance ratherthan reduce the clump pat-
tern because of their preferencefor young tissues including
seedlings of shrubs. They would thus reinforce the rabbit
effect. At high densities, however, they selectively remove
some of the shrubs of each clump and eventually shift the
vegetation towards Lithraea, Acacia caven, and Trichoce-
reus chilensisstands (Fuentes 1984).
A problem which has not yet been completely solved
relates to the determinants of clump size in the various
situations. It is likely that once a certain clump size has
been attained, further establishment and survival of seed-
lings around the peripheryof the clump is no longer possi-
ble and the clump stops growing. Preliminary evidence
(Poiani and Fuentes, unpublished)suggests that density of
0. degus and risk of seedling predation increase rapidly
with clump size, perhaps enough to account for complete
seedling mortality beyond some threshold clump size. This
hypothesis is still being investigated.
Our data and the above discussion are consistent with
the hypothesis that the general openness of the matorral
is related not only to human disturbance in some areas,
but also to the general pattern of shrub recruitment.This
pattern, in which new shrubs tend to grow under a nursing
clump, is different from that described for the California
chaparral. In California the shrub cover is periodically
burnedby naturalfiresand immediatelya secondarysucces-
sion begins (Hanes 1971; and referencestherein). The new
shrub cover arises either from ubiquitous seedlings or from
the resproutingof old burned plants (Keeley and Johnson
1977). Only a few years after a fire, shrub recruitmentbe-
comes very scarce indicatingthat shrub reproductionis dis-
continuous and fairly synchronous for all species. In Chile,
in contrast, there are neither natural fires (Mooney 1977)
nor any other large-scale natural phenomena generating
open spaces for colonization, and the shrub recruitment
pattern seems to be more continuous in time and less syn-
chronous for all species.
It seems that the pattern of shrub clumps and the pecu-
liar shrub recruitmentpattern exhibited in the Chilean sites
denote more profound differencesthan previously thought
between the Californiachaparraland the Chilean matorral.
It is likely that today the Chilean matorral is a partly an-
thropogenic alternative to the California chaparral, and
that herbivores and seed dispersershave a very important
role in shaping its vegetation structure.
We are very grateful to Professor J. Harper,
Acknowledgements.
H. Mooney, W. Schlesingerand 0. Solbrig who read the manu-
script and suggested many ways to improve it. We would also

like to acknowledge the support of P. Universidad Cat6lica de
Chile through project DIUC 75/81. This paper is a contribution
to the MAB project of UNESCO.
References
Armesto JH, GutierrezJ (1978) El efecto del fuego en la estructura
de la vegetaci6n de Chile central. Anales del Museo de Historia
Natural de Valparaiso 11:43-48
Aschmann H, Bahre C (1977) Man's impact in the wild landscape.
In: HA Mooney (ed) Convergent evolution in Chile and Cali-
fornia Mediterraneanclimate ecosystems. Dowden Hutchinson
and Ross, Pa. Stroudsbourg,Pa, USA. pp 73-84
Cody ML, Mooney HA (1978) Convergence versus non-conver-
gence in Mediterranean-climateecosystems. Ann Rev Ecol Syst
9:265-321
Conrad CE, Oechel WC (eds) (1982) Proceedings of the Sympo-
sium on Dynamics and Management of MediterraneanType
Ecosystems (1981) San Diego, California. Pacific Southwest
Forest and Range Experiment Station. Berkeley, California,
USA
Fuentes ER (1984) Human impact and ecosystem resiliencein the
Southern Andes. Mountain Research and Development 4:45-
49
Fuentes ER, Simonetti J (1982) Plant patterning in the Chilean
matorral:are the roles of native and exotic mammalsdifferent?
In: Conrad CE, Oechel WC (eds) Proceedings of the Sympo-
sium on Dynamics and Management of MediterraneanType
Ecosystems (1981) San Diego, California. Pacific Southwest
Forest and Range Experiment Station. Berkeley, California,
USA pp 227-233
Fuentes ER, Jaksic F, Simonetti J (1983) European rabbits vs na-
tive mammals: effects on shrub seedlings. Oecologia (Berlin)
58:411-414
Godron M, Guillerm JL, Poissonet J, Poissonet P, Thiault M,
Trabaud L (1981) Dynamics and management of vegetation.
In: Castri F di, Goodall D, Specht R (eds) Mediterranean-type
shrublands.Ecosystemsof the world 11. ElsevierScientificPub-
lishing Co. The Hague. Netherlandspp 317-344
Hanes TL (1971) Succession after fire in the chaparralof southern
California. Ecological Monographs 41:27-52
Hoffmann J, Hoffmann AE (1982) Altitudinal ranges of phanero-
Note addedin proof
A new experimental series started in October 1982, after one of
the wettest years of this century, has shown that in fact seedlings
planted between clumps, even if protected from rabbits, all desic-
cate before the second month. But, seedlingsplanted under clumps
have survived for 2 dry seasons with an average survival rate of
41% for those protected from mammals and 22% for the unpro-
tected ones!
411
phytes and chamaephytes in central Chile. Vegetatio
48:151-163
Jaksic F, Fuentes ER, YaniezJ (1979) Two types of adaptations
of vertebrate predators to their prey. Archivos de Biologia y
Medicina Experimental(Chile) 12: 143-152
Keeley S, Johnson AW (1977) A comparison of the pattern of
herb and shrub growths in comparable sites in Chile and Cali-
fornia. American Midland Naturalist: 97:120-132
Kummerow J, Montenegro G, Krause D (1981) Biomass, pheno-
logy and growth. In: Miller P (ed) Resource use by chaparral
and matorral.SpringerNew York pp 69-96
McDonnell MJ, Stiles EW (1983) The structural complexity of
old field vegetation and the recruitmentof bird dispersedplant
species. Oecologia (Berlin) 56:109-116
Miller PC, BradburyDE, Hajek E, La Marche V, Thrower NJW
(1977) Past and present environment. In: HA Mooney (ed)
Convergent evolution in Chile and California mediterranean
climate ecosystems. Dowden Hutchinson and Ross, Strouds-
bourg. Pa. USA pp 27-72
MontenegroG, Avila G, Schatte P (1981) The significanceof ligno-
tubers in shrubs of the Chilean matorral. III Congreso Latino-
americanode Botanica, Lima, Perui
Mooney HA (1977) Convergent evolution in Chile and California
mediterraneanclimate ecosystems. Dowden Hutchinson and
Ross, Stroudsbourg.Pa USA
Mooney HA, Dunn EL, Shropshile F, Song L (1972) Land-use
history of California and Chile as related to the structure of
the sclerophyllscrub vegetations. Madrofio 21:305-319
Miller-Dumbois D, EllenbergH (1974) Aims and methods of vege-
tation ecology. Wiley and Sons. New York NY USA
Siegel S (1956) Nonparametricstatisticsfor the behavioralsciences.
McGraw Hill. New York USA
Thrower N, Bradbury D (1977) (eds) Chile-Californiamediterra-
nean scrub atlas. Dowden, Hutchinson and Ross. Strouds-
bourg, Penn USA
TrabaudL (1981) Man and fire: Impacts on mediterraneanvegeta-
tion. In: Castri F di, Goodall DW, Specht RL (eds) Mediterra-
nean type shrublands. Ecosystems of the World 11. Elsevier
ScientificPublishing,The Hague. Netherlandpp 523-538
Received October 20, 1983