1435

Conservation Biology, Pages 1435–1440
Volume 14, No. 5, October 2000
Scale-Dependent Effects of Habitat Disturbance on
Species Richness in Tropical Forests

K. C. HAMER* AND J. K. HILL

Centre for Tropical Ecology, Department of Biological Sciences, University of Durham, South Road, Durham
DH1 3LE, United Kingdom

Abstract:

Despite growing concern, no consensus has emerged over the effects of habitat modification on
species diversity in tropical forests. Even for comparatively well-studied taxa such as Lepidoptera, disturbance
has been reported to increase and decrease diversity with approximately equal frequency. Species diversity
within landscapes depends on the spatial scale at which communities are sampled, and the effects of distur-
bance in tropical forests have been studied at a wide range of spatial scales. Yet the question of how distur-
bance affects diversity at different spatial scales has not been addressed. We reanalyzed data from previous
studies to examine the relationship between spatial scale and effects of disturbance on tropical-forest Lepi-
doptera. Disturbance had opposite effects on diversity at large and small scales: as scale decreased, the proba-
bility of a positive effect of disturbance on diversity increased. We also explicitly examined the relationship
between spatial scale and the diversity of butterflies in selectively logged and unlogged forest in Maluku Prov-
ince, Indonesia. Species richness increased with spatial scale in both logged and unlogged forest, but at a sig-
nificantly faster rate in unlogged forest, whereas species evenness increased with scale in unlogged forest but
did not increase with scale in logged forest. These data indicate that the effects of habitat modification on spe-
cies diversity are heavily scale-dependent. As a result, recorded effects of disturbance were strongly influenced
by the spatial scale at which species assemblages were sampled. Future studies need to account for this by ex-
plicitly examining the effects of disturbance at a number of different spatial scales. A further problem arises
because the relationship between scale and diversity is likely to differ among taxa in relation to mobility.
This may explain to some extent why the measured effects of disturbance have differed between relatively
mobile and immobile taxa.

Efectos Escala-Dependientes de la Perturbación del Hábitat Sobre la Riqueza de Especies en Bosques Tropicales

Resumen:

A pesar de una creciente preocupación, aún no ha emergido un consenso sobre los efectos de la
modificación del hábitat sobre la diversidad de especies en bosques tropicales. Aún para taxones comparati-
vamente bien estudiados, como Lepidoptera, se ha reportado con aproximadamente la misma frecuencia
que la perturbación incrementa y disminuye la diversidad. La diversidad de especies dentro de paisajes de-
pende de la escala espacial a la cual las comunidades son muestreadas y los efectos de la perturbación en
bosques tropicales han sido estudiados en un amplio rango de escalas espaciales. No obstante, la pregunta
sobre cómo la perturbación afecta la diversidad a diferentes escalas espaciales no ha sido abordada aún.
Volvimos a analizar los datos de estudios previos para examinar la relación entre la escala espacial y los
efectos de la perturbación en Lepidópteros de bosques tropicales. La perturbación tiene efectos opuestos en la
diversidad a escalas grandes y pequeñas; si la escala disminuye, la probabilidad de un efecto positivo de la
perturbación sobre la diversidad incrementa. También analizamos explícitamente la relación entre la escala
espacial y la diversidad de mariposas en bosques talados selectivamente y en bosques no talados de la pro-
vincia de Maluku en Indonesia. La riqueza de especies incrementó con la escala espacial tanto en bosques ta-
lados como en los no talados, pero a una escala significativamente más rápida en los bosques sin talar,
mientras que la uniformidad de las especies incrementó con la escala en bosques sin tala pero no incrementó
con la escala en bosques talados. Estos datos indican que los efectos de la modificación del hábitat sobre la
diversidad de especies son altamente dependientes de la escala. Como resultado, los resultados registrados de

*

email k.c.hamer@durham.ac.uk
Paper submitted September 7, 1999; revised manuscript accepted January 12, 2000.

1436

Scale-Dependent Effects of Habitat Disturbance Hamer & Hill

Conservation Biology
Volume 14, No. 5, October 2000

Introduction

Over half of global diversity in terms of numbers of spe-
cies is represented by insects (Stork 1991), so the manner
in which these taxa respond to changes in their habitat is
a major concern. The responses of some insect groups to
severe forms of forest disturbance such as clearfelling and
conversion to grassland or farm fallow have been rela-
tively well documented (Holloway et al. 1992; Daily &
Ehrlich 1995). Responses to less severe forms of distur-
bance are poorly understood, however, even for compar-
atively well-studied taxa such as Lepidoptera. Some stud-
ies have recorded a reduction in diversity following
habitat disturbance or fragmentation (Bowman et al. 1990;
Daily & Ehrlich 1995; Hill et al. 1995), whereas others
have concluded that disturbance results in an increase in
diversity (Kremen 1992; Spitzer et al. 1993, 1997; Brown
& Hutchings 1997; Hamer et al. 1997).
There are many possible explanations for these con-
trasting results. For instance, if diversity is highest at in-
termediate levels of disturbance (Horn 1975; Connell
1978), then any change in diversity is likely to reflect
not only the severity of disturbance but also the initial
condition of the habitat before the disturbance event.
The effects of disturbance may also vary geographically,
for instance in relation to initial species diversity (May
1972) or tree species composition (e.g., proportion of
fruiting species). A further possibility that has not been
considered properly (but see May 1994) is that recorded
effects are influenced by the spatial scale at which spe-
cies assemblages are sampled.
The effects of forest disturbance on Lepidoptera have
been studied at a wide range of spatial scales, deter-
mined by sampling technique (light traps, fruit-baited
traps, hand netting, or observations along transects) and
by differences in protocol for a given technique (traps
set different distances apart, transects of varying lengths,
or habitat patches of different size). Generally, individ-
ual studies have been restricted to a single spatial scale,
and this restriction has been assumed (usually implicitly)
to be sufficient to control for scale dependence. This is
an invalid assumption, however, if disturbance alters the
pattern and scale of environmental heterogeneity within
forests over the spatial scales used to sample species.
Any such alteration will be likely to affect the rate at
which species composition changes with increasing area
sampled within each habitat. Under these circumstances,
comparisons of species diversity in disturbed and undis-
turbed forest could depend heavily on the spatial scale at
which communities are sampled. This may be a particular
problem in tropical forests, which under natural condi-
tions have high heterogeneity in terms of vegetation struc-
ture and plant and animal species composition (Erwin
1982; Huston 1994; He et al. 1996), much of which may
be lost following disturbance (Malcolm 1994; Van Valken-
burg & Ketner 1994; Hill et al. 1995).
We reviewed the published literature to examine how
the probability of finding decreased or increased species di-
versity in response to disturbance has varied with sampling
scale and method, degree of disturbance, elevation of study
site, and location of study. In addition, we examined the ef-
fects of spatial scale on species richness and evenness of
butterflies in logged and unlogged lowland monsoon forest
on the island of Buru, Maluku Province, Indonesia.

Methods

Previous Studies

To examine how reported effects of forest disturbance
on species diversity in previous studies might be related
to spatial scale and sampling technique, we used the
BIDS ISI database (Institute for Scientific Information,
United Kingdom) to search about 2500 international
peer-reviewed journals for each year from 1981 through
1998. We then used reference lists in studies located by
this method to search for further studies published prior
to 1981. We searched for comparisons of butterfly or
moth diversity in tropical forests differing in degree of
disturbance. We restricted our analysis to those studies
or parts of studies that were within forest, and we ex-
cluded data that compared forest with nonforest habi-
tats such as farm fallow and grassland. Multiple studies
from a single location and sampling technique (e.g.,
Spitzer et al. 1993, 1997) were included only once. Wood
and Gillman (1998) recorded separate results from two
different sampling techniques, and we included these
data separately. In total, we extracted the following
information from 15 independent data sets: impact of
disturbance (diversity increased vs. diversity decreased
or unaffected), main sampling technique (attractant trap

la perturbación fueron fuertemente influenciados por la escala espacial a la cual los ensamblajes de especies
fueron muestreados. Los estudios futuros necesitan tomar en consideración esto examinando explícitamente
los efectos de la perturbación a diferentes escalas espaciales. Un problema emergente se debe a que la rel-
ación entre la escala y la diversidad es probable que difiera entre taxones en relación a su movilidad. Esto
puede, hasta cierto punto, explicar porqué los efectos medidos de la perturbación difieren entre taxones rela-

tivamente móviles e inmóviles.