Content uploaded by Masahiko Kitahara
Author content
All content in this area was uploaded by Masahiko Kitahara on Feb 08, 2019
Content may be subject to copyright.
Ecological Research (2003) 18, 503 –522
Blackwell Science, LtdOxford, UKEREEcological Research0912-38142003 Ecological Society of JapanSeptember 2003185503522Original Article
Diversity and rarity hotspots of butterfliesM. Kitahara and M. Watanabe
*Author to whom correspondence should be
addressed. Email: mkita@yies.pref.yamanashi.jp
Received 30 January 2002.
Accepted 6 January 2003.
Diversity and rarity hotspots and conservation of
butterfly communities in and around the Aokigahara
woodland of Mount Fuji, central Japan
Masahiko Kitahara* and Maki Watanabe
Department of Animal Ecology, Yamanashi Institute of Environmental Sciences, Kenmarubi,
Fujiyoshida, Yamanashi 403-0005, Japan
In central Japan, Aokigahara woodland is considered to be one of the most natural areas around
Mount Fuji and a core area in the conservation of the biodiversity of Mount Fuji. We chose butterflies
as an indicator species of biodiversity and examined six communities in and around the woodland in
2000 using transect counts to examine and search for diversity and rarity hotspots and their asso-
ciated landscapes. The results showed that butterfly species richness and species diversities H¢ 1/l
were significantly higher in forest-edge sites than in forest-interior and/or open-land sites, and vari-
ation in the total number of species among these three landscape types was well accounted for by eco-
logically specialist species, such as landscape specifics, oligovoltines, narrow diet feeders and low-
density species. Thus, the species regarded as vulnerable to extinction, including Red List species,
were observed more often in forest-edge sites than in forest-interior and/or open-land sites. As a
result, in the study area, diversity and rarity hotspots were found in forest-edge landscapes. The rea-
sons why butterfly diversity and rarity hotspots were established in forest-edge landscapes were ana-
lyzed and interpreted from several points of view, including disturbance level, landscape elements
and plant species richness. From these results, and the fact that some species were confined to forest-
interior sites, we conclude that it is very important to conserve and manage forest-edge habitats (con-
sidered to be semi-natural) as well as forest-interior habitats (considered to be the most natural) to
maintain the diversity of butterfly communities and preserve the various types of threatened species
in and around the Aokigahara woodland.
Key words: butterfly community; conservation; diversity hotspots; forest edge; rarity hotspots.
INTRODUCTION
What we discover, identify and define as a biolog-
ical hotspot, that is, a geographic area with high
concentrations of species, endemic species, rare or
threatened species and/or high levels of threat to
species survival (Myers 1988; Mittermeier et al.
1998; Reid 1998), at a geographic location of con-
cern is one of the most important steps in conserv-
ing and maintaining local biodiversity, in
prioritizing areas for potential protection and in
the establishment of nature reserves in local eco-
systems (Primack 1993, 1995; Hunter 1996; Pres-
sey 1996; Mittermeier & Forsyth 1997; Myers
1997). In the analysis of hotspots at a local scale/
level, it is very important to specify the type of
natural environment or landscape that is associated
with these biological hotspots. In general, this
attempt is very useful in assigning and setting
conservation priorities to local hotspots, which
produce a profound effect on the conservation of
local biodiversity (Washitani 1999).
Until recently it has been stated that insects
respond more rapidly to disturbance than verte-
brates and, therefore, have potential as early indi-
cators of environmental change (Kremen 1992;
Kremen et al. 1993; Hamer et al. 1997). Among
insects, butterflies are believed to be the most
504 M. Kitahara and M. Watanabe
suitable for indirect measures of environmental
variation because of their high sensitivity to local
weather, climate, light levels and other parameters
that are affected by habitat change (Ehrlich et al.
1972; Weiss et al. 1987; Hill et al. 1995; Blair &
Launer 1997; Wood & Gillman 1998). They are
also very suitable for studies examining the struc-
ture and dynamics of populations and communi-
ties (Ehrlich 1992). In addition, the taxonomy and
life histories of most Japanese butterfly species are
already well known, and the adults of many species
can be reliably identified in the field. Thus, diurnal
Lepidoptera make ideal study subjects for diversity
hotspot analyses of local biological communities.
In the present study, we chose the Aokigahara
woodland on the northwestern foot of Mount Fuji,
central Japan, which is considered to be one of the
most natural areas around Mount Fuji and a core
area in the conservation of biodiversity at Mount
Fuji, as our study site. This study attempted to
discover and identify butterfly hotspots (diversity
and/or rarity (endangered species) hotspots) from
among the representative landscapes present in
and around the Aokigahara woodland. Many stud-
ies of butterfly communities have suggested that
butterfly diversity and species richness is high at
forest edges or transitional and ecotonal areas com-
pared to open-land and woodland areas (e.g.
Erhardt 1985; Leps & Spitzer 1990; Ishii et al.
1993, 1995; Spitzer et al. 1993; Kitahara & Fujii
1994; Ishii 1996a; Yata 1996; Tashita & Ichimura
1997; Natuhara et al. 1999; Balmer & Erhardt
2000; Schneider & Fry 2001). Thus, we aimed to
confirm and clarify changing patterns in the struc-
ture of butterfly communities among selected rep-
resentative landscapes, and to examine whether the
above-mentioned butterfly diversity pattern is
observed in the present study area, with a view to
formulating conservation strategies for commu-
nity diversity.
METHODS
Study sites
This study was carried out at an altitude of approx-
imately 1000 m in and around the Aokigahara
woodland on the northwestern foot of Mount Fuji
(altitude 3775.6 m) in central Japan. We selected
three representative types of landscapes in and
around the woodland: (i) forest interior (FI); (ii)
forest edge (FE) of the woodland; and (iii) open
land (OL) outside the woodland. We established
two census sites (1 and 2) in each landscape type.
Thus, all six census sites (named FI-1, FI-2, FE-1,
FE-2, OL-1 and OL-2) were within the study area.
This design made it possible to, at least in part,
differentiate the effects of landscape or environ-
mental differences on butterfly communities from
the effects of other physical factors. That is, all six
study sites were similar in terms of altitude and
topography (almost flat or gently sloping land),
and were located inside an area measuring 2.63 km
east to west and 1.38 km north to south in or
around the eastern part of the Aokigahara
woodland.
The characteristics of the six study sites, mainly
their vegetation, are outlined in Table 1. Sites FI-
1 and FI-2 were in the forest interior of the eastern
part of the Aokigahara woodland. At site FI-1 we
established a fixed census route (300 m in length)
along a path crossing the site’s interior to examine
butterflies present mainly in the forest understory.
At site FI-2 we used an artificial tower that stood
in the forest. A fixed census point was established
on the table of the tower, which was above the can-
opy of the forest (approximately 18 m above the
ground), to examine butterflies present mainly
above or near the forest canopy. These sites were
both situated inside the Aokigahara woodland,
which consisted of a continuous and extensive nat-
ural and primary forest (3000 ha in area) that grew
on the Aokigahara lava flow from the crater of
Mount Nagao, which is located halfway up Mount
Fuji and formed in the AD 864 eruption (Tsuya
1971). It is believed that this forest has never been
subjected to large-scale human disturbance. The
average tree age is approximately 150 years and
the highest age ever recorded of the representative
dominant tree species is 356 years for Tsuga siebol-
dii and 240 years for Chamaecyparis obtusa (Seido
1991). Almost no human land use or disturbance
were found in these two woodland interior sites.
Sites FE-1 and FE-2 were at the eastern forest
edge of Aokigahara woodland. In each site we
established a fixed census route (300 m in length)
along the forest boundary, bordering other land-
scapes (treeless areas). A small part of the route at
site FE-1 ran through a grassland near the forest
Diversity and rarity hotspots of butterflies 505
Table 1 Characteristics and vegetation of the six study sites
Study
site
Altitude
(m)
Landscape and
landscape element
(open land)
Main plant (Phanerogamae) species
Type
No. plant
(Phanerogamae)
species TotalTrees Small trees and shrubs Herbs
FI-1 1030 Woodland
(forest understory)
Quercus mongolica
var. crispula
Clethra barbinervis
Acanthopanax
sciadophylloides
Acer sieboldianum
Pinus densiflora
Chamaecyparis obtusa
Ilex pedunculosa
Acer micranthum
Sorbus americana
ssp. japonica
Rhus trichocarpa
Enkianthus campanulatus
Rhododendron dilatatum
Skimmja japonica f. repens
Polygonum cuspidatum
Maianthemum dilatatum
Oplismenus undulatifolius
Corydalis incisa
Artemisia princeps
Trees
Shrubs
Perennials
Annuals
Others
21
15
9
1
3
49
FI-2 1070 Woodland
(forest canopy)
Tsuga sieboldii
Chamaecyparis obtusa
Clethra barbinervis
Acer distylum
Quercus mongolica
var. crispula
Betula grossa
Pinus parviflora
Ilex pedunculosa
Euonymns macropterus
Acer tschonoskii
Pieris japonica
Acer micranthum
Prunus jamasakura
Ilex macropoda
Not known Not known
FE-1 1025 Woodland Quercus mongolica var.
crispula
Quercus serrata
Castanea crenata
Pinus densiflora
Larix kaempferi
Alnus hirsuta
Magnolia obovata
Abies firma
Prunus incisa
Lonicera japonica
Malus toringo
Rosa multiflora
Deutzia crenata
Hydrangea paniculata
Miscanthus sinensis
Boehmeria tricuspis ssp.
paraspicata
Cirsium nipponicum var.
incomptum
Lysimachia clethroides
Agrimonia pilosa
Sanguisorba officinalis
Vicia cracca
Vicia unijuga
Picris hieracioides ssp. japonica
Erigeron annuus
Trees
Shrubs
Perennials
Annuals
Others
16
15
48
18
9
106
Open land
secondary
grassland
conifer
plantations
vegetable plots
abandoned
arable land
sparse forest
506 M. Kitahara and M. Watanabe
FE-2 1010 Woodland Quercus serrata
Pinus densiflora
Larix kaempferi
Alnus hirsuta
Acer capillipes
Zelkova serrata
Prunus maximowiczii
Clethra barbinervis
Acer crataegifolium
Lindera obtusiloba
Enkianthus campanulatus
Rhododendron dilatatum
Euonymns macropterus
Spiraea japonica
Ligustrum obtusifolium
Miscanthus sinensis
Oplismenus undulatifolius
Campanula punctata
Cirsium nipponicum var.
incomptum
Kalieris pinnatifida
Lysimachia clethroides
Polygonum cuspidatum
Trifolium repens
Astilbe microphylla
Picris hieracioides ssp. japonica
Trees
Shrubs
Perennials
Annuals
Others
31
20
55
20
10
136
Open land conifer
plantations
secondary
grassland
sparse forest
bare site
OL-1 990 Open land
athletic
fields and open
areas with
grassland
Oxalis corniculata
Taraxacum officinale
Geranium nepalense
Vicia cracca
Trifolium repens
Cerastium fontanum ssp.
japonica
Poa annua
Ambrosia artemisiifolia var.
elatior
Kummerovia striata
Trees
Shrubs
Perennials
Annuals
Others
0
1
23
27
1
52
OL-2 1025 Open land
farmland
consisting of
cabbage,
potato, and
strawberry
plots
Morus australis
Pinus densiflora
Cornus controversa
Salix bakko
Pieris japonica
Celastrus orbiculatus
Rosa multiflora
Rorippa indica
Miscanthus sinensis
Plantago asiatica
Oxalis corniculata
Taraxacum officinale
Rumex crispus ssp. japonicus
Agrimonia pilosa
Calystegia japonica
Trifolium pratense
Vicia unijuga
Hemerocallis fulva var.
Kwanso
Trees
Shrubs
Perennials
Annuals
Others
6
9
24
18
3
60
Study
site
Altitude
(m)
Landscape and
landscape element
(open land)
Main plant (Phanerogamae) species
Type
No. plant
(Phanerogamae)
species TotalTrees Small trees and shrubs Herbs
Table 1 Continued.
Diversity and rarity hotspots of butterflies 507
boundary. In general, both sites comprised natural
forest (usually along one side of the census route),
secondary (seminatural) grassland, conifer planta-
tions, vegetable plots, abandoned farmland and
scattered forest (usually on the other side of the
census route). As these sites were in the same
woodland, the forest structure and composition of
the sites were basically similar to the two forest
interior sites; however, they were further charac-
terized by a greater presence of broad-leaved decid-
uous trees that resulted from good sunlight along
the forest edge. There were no major differences in
the landscape and environmental structure
between FE-1 and FE-2, except for some differ-
ences in the component vegetation species. At
these two sites, various types of human land use
(e.g. plantations and vegetable plots) and distur-
bance (e.g. mowing and cultivation) were found in
most places except the forested parts. Thus,
approximately half the area at each site (the
wooded parts) was relatively undisturbed, whereas
the other half (the treeless parts) had moderate
human disturbance. In general, the human
disturbance level at both sites can be considered
to be intermediate.
Open land was located outside the eastern edge
of the Aokigahara woodland. Most areas of this
open land, except near the woodland edge, were
reclaimed as farmland, recreational and resort areas
with athletic fields, and a summer cottage zone.
Sites OL-1 and OL-2 were located in the open land
and we established a fixed census route (300 m in
length) along a path running through each site.
Site OL-1 comprised mainly athletic fields and
open areas with tracts of turf for track and field
sports and events. There were almost no large
trees, small trees or shrubs at this site. Site OL-2
comprised farmland consisting of cabbage, potato,
strawberry and Sanguisorba officinalis (for use in
flower arrangements) fields. OL-2 contained some
species of trees, shrubs and herbs in the grasses
around the agricultural plots and along the farm
roads. High levels of human land use (e.g. sports
and recreational areas, farmland) and disturbance
(e.g. trampling, mowing, pesticide spraying, fer-
tilizer application, cultivation) were observed in
most areas of these two reclaimed open-land sites.
The human disturbance level at these two sites
was considered to be the highest of all three
landscape types.
Census method
Intensive regular censuses were carried out twice
per month during the adult flight season (from
April to November 2000), from 10:00 to 15:00 h
local time under fair weather conditions.
Data were collected at all sites except FI-2 using
the line transect method (Pollard 1977, 1984;
Thomas 1983; Pollard & Yates 1993), which is
extensively used to survey and monitor butterfly
populations and communities (e.g. Shreeve &
Mason 1980; Erhardt 1985; Warren et al. 1986;
Yamamoto 1988; Ishii 1993; Pollard & Yates
1993). It is a method of considerable value when
investigating differences in species abundance
among sites (Gall 1985; New 1991). Walking at a
steady pace along the transect line, we recorded the
number of adult individuals of each butterfly spe-
cies sighted within a belt approximately 10 m
wide and 300 m long (~3000 m2 in area). Indi-
viduals that could not be identified immediately
on sight were often netted and released after
identification.
Data were collected from site FI-2 using the
visual observation method in a fixed area (South-
wood 1978). We covered approximately the same
survey area (~3000 m2) as at the other sites and
spent a standard time (~15 min) estimated from
the surveys at the other study sites so that data
obtained using this method were directly compa-
rable with data from the other sites. We recorded
all adult butterflies within a circular area, with a
radius of approximately 30 m, near and above the
forest canopy around the fixed census point on the
table of the tower. We often used binoculars to
help identify distant individuals. We recorded the
weather conditions and human disturbance-related
events, such as mowing and cultivation, at each
site during each census.
We conducted a vegetation survey along each
transect route using the belt-transect method
(Lincoln et al. 1998) at all sites surveyed using the
transect method. Vegetation surveys were con-
ducted twice (12 June and 27 August 1999) for
each transect route, and we recorded all species of
plants (belonging to Phanerogamae) sighted
within a belt approximately 10 m wide along
each route. The vegetation at site FI-2 was
inferred from Ohtsuka (T. Ohtsuka, unpubl. data,
2001).
508 M. Kitahara and M. Watanabe
Data analysis
We analyzed the butterfly community structure at
each site using the following ecological parame-
ters: population density, average population den-
sity, total population density, species richness,
species diversity and species evenness. We used the
population density of each butterfly species at each
study site, which we obtained as follows, to ana-
lyze butterfly communities. First, we determined
the monthly abundance of each species at each site
during the study period (i.e. the mean number of
adult individuals obtained from the two transect
surveys in each month). However, in September,
October and November, transect surveys were only
conducted once per month. Thus, in these months
monthly abundance was the number of adults from
the one survey. To obtain a monthly mean abun-
dance, we averaged the monthly abundance in
months when a butterfly species was recorded.
Excluding months when no butterflies were
observed minimized the effect of different volt-
inism among butterfly species on the yearly abun-
dance estimate (i.e. monthly mean abundance).
Finally, we obtained the population density (no.
Adults month-1 ha-1) by dividing the monthly
mean abundance by 0.3 ha (the area of each census
site). Average population density was obtained by
averaging the population densities of sites only
where a specific species was found. Exclusion of
study sites where no individuals were observed
minimized the effect of different distribution pat-
terns (widespread or restricted) of butterfly species
on the average population density. As the applica-
tion of different monitoring methods may restrict
the comparison of results among the study sites,
despite the fact that the areas of census was almost
the same (0.3 ha) at all sites, we note that there
may be a possibility that the population values
recorded at FI-2 are more or less underestimated
when compared to the other sites because of sam-
pling errors related to the specific monitoring
method (the visual point observation method in a
fixed area) used at FI-2.
The list of butterfly species observed in the
present study and their population density values
at each site and average population density are
shown in Appendix I.
Total population density was calculated as the
population density of all component species at
each site. Species richness was expressed as the
number of species found at each study site during
the observation period. Species diversity at each
site was expressed by both the Shannon–Wiener
function, , where
S is the number of
species at each site and pi is the proportion of the
population density of the ith species at each site, and
the Simpson’s index of diversity (Simpson 1949),
1/
l
, , where ni is the
population density of the ith species at each site,
and N is the total population density of all com-
ponent species at each site. Species evenness was
expressed by the Shannon equitability index,
J
¢
= H
¢
ln-1 S, where H
¢
is the Shannon–Wiener
function and S is the number of species at each site.
The ecological characteristics of the component
species used in the analyses were as follows: average
population density, local distribution pattern, vol-
tinism, potential larval diet breadth, geographic
range size in Japan, Red List category in Japan and
the type of larval host plant. The local distribution
pattern of each species was expressed by the num-
ber of census sites at which the species was recorded
(i.e. 1–6). Voltinism was defined as the number of
generations per year based mainly on Unno and
Aoyama (1981), and complemented the actual data
of seasonal fluctuations in the observed number of
individuals of each species in the present study.
Potential larval diet breadth was expressed as the
number of all larval host-plant species ever
recorded in Japan according to Endo and Nihira
(1990). The geographic range size of each species in
Japan was represented by the number of
10 km ¥ 10 km squares in which the species was
recorded in a report on the distribution patterns of
Japanese animals and plants (Butterfly volume,
Nature Conservation Bureau 1993). This report
includes data on butterflies that were collected and
recorded by 421 volunteer specialists, and plotted
on a grid of 10 km ¥ 10 km cells that cover almost
the entire country. Most of the distribution records
(89.33%) of butterflies used in the analyses were
collected from 1980 onwards, with the rest
(10.67%) completed before 1979. Component spe-
cies observed in this study were compared to those
on the Red Data list of Japan (Ministry of the Envi-
ronment of Japan (hereafter M.E.J.) 2000) and spe-
¢-Â
Hpp
i
i
S
i
=
=1
ln
l=-
()
-
()
=
Ânn NN
ii
i
S
11
1
/
Diversity and rarity hotspots of butterflies 509
cies that corresponded to any of the Red List
categories of Japan were determined. The type of
larval host plant was classified as either grassy
(herbaceous) or woody based on Unno and Aoyama
(1981). The ecological characteristics of each com-
ponent species are shown in Appendix I.
We chose species vulnerable to extinction from
the component species to examine which sites or
landscapes were associated with vulnerable species.
Referring to Primack (1993), we treated Red List
species of Japan, univoltine species, diet-specialist
species, site-specific species, narrow-geographic-
range species and low-population-density species
as species vulnerable to extinction. Among these
species, diet-specialist species were defined as those
whose larvae had been reported to feed on five or
less plant species based on Endo and Nihira (1990),
site-specific species were those recorded at only
one census site in the present study, narrow-
geographic-range species were those recorded in
£500 10 km ¥ 10 km blocks in Japan based on a
report on the distribution patterns of Japanese but-
terflies (Nature Conservation Bureau 1993), and
low-population-density species were those with an
average population density less than 2.00 (no. Indi-
viduals month-1 ha-1) in the present study.
To examine differences in the community indi-
ces and the observed numbers of species vulnerable
to extinction among the landscape types, we used
ANOVAS and Bonferroni tests on the annual values
of the respective indices at each site. However, in
cases where the mean value in a particular land-
scape type was zero, we used a t-test. To examine
the impact on changes in species composition
among landscape types, we calculated the coeffi-
cient of determination (r2) for the number of all
species and for species belonging to each category
associated with several ecological characteristics in
each landscape. To clarify the relationships
between community attributes and the threat-
ened/vulnerable species, we conducted correlation
analyses between species richness or diversity and
the number of several threatened species among
the communities at all six census sites.
To investigate similarities and differences in spe-
cies composition among census sites, we analyzed
the community data using principal components
analysis (PCA) based on the variance–covariance
matrix, using the program EXCEL-multivariate
analysis ver. 3.0 (Esumi 1998). A 57 (species) by six
(census sites) matrix based on the population den-
sity of each species at each site was subjected to this
analysis. To clarify the characteristics of site-specific
species, we conducted a Spearman rank correlation
analysis between the number of sites at which a spe-
cies was observed and the value of several ecological
characteristics for all component species.
RESULTS
Community indices and species vulnerable
to extinction at each site
Table 2 shows the values of various community
indices and the numbers of several types of species
vulnerable to extinction at each census site. The
mean values of species richness, species diversity
H
¢
and 1/l were significantly different among the
three landscape types. That is, species richness,
species diversity H
¢
and 1/l were higher in forest-
edge sites than in forest-interior and/or open-land
sites. Whereas, the mean values of total population
density and evenness were not significantly differ-
ent among the landscape types.
Red List species authorized by M.E.J. (2000)
were only recorded in two forest-edge sites and at
OL-2. From the other species vulnerable to extinc-
tion, univoltine, larval-diet specialist and low-
population-density species were significantly more
abundant in forest-edge sites than in forest-
interior and open-land sites. Whereas, site-
specific (endemic) species were only present in two
forest-edge sites and at FI-1.
Differences in species composition among
the landscape types
Table 3 shows changes in species composition
among the three landscape types for several eco-
logical factors. The values of the coefficient of
determination (r2) and the maximum difference in
the number of species among the landscapes indi-
cate that variations in the total number of species
among the three landscape types were accounted
for more by the numbers of species observed in
only one landscape type, oligovoltines (uni- or
bivoltines), narrow diet breadth (corresponding
to 1–9 or 10–29 species of larval potential
host plants) and low population densities
510 M. Kitahara and M. Watanabe
Table 2 The values of various community indices and the numbers of several types of species vulnerable to extinction at each census site in the three landscape
types in and around the Aokigahara woodland
Landscape type
Census site
Open land Forest edge Forest interior ANOVA or
t-test
Bonferroni
test
OL-1 OL-2 Mean FE-1 FE-2 Mean FI-1 FI-2 Mean
(a) Community index
Total population density 44.04 137.81 90.93 215.39 180.98 198.19 22.59 39.44 31.02 F = 8.389, P = 0.0591
Species richness (S) 13 24 18.5 44 42 43.0 10 11 10.5 F = 27.310, P = 0.0119 FE, FI* FE, OL*
Species diversity (H¢) 2.440 2.825 2.633 3.357 3.370 3.363 2.149 2.149 2.149 F = 30.232, P = 0.0103 FE, FI*
Species diversity (1/l) 12.460 18.227 15.343 22.926 23.973 23.450 10.409 8.137 9.273 F = 15.360, P = 0.0265 FE, FI*
Evenness ( J¢) 0.951 0.889 0.920 0.887 0.902 0.894 0.933 0.896 0.915 F = 0.405, P = 0.6985
(b) Species vulnerable to extinction†
Red List spp‡031.5 4 4 4.0 0 0 0 t = - 1.667, P = 0.2375 –
Univoltine spp.‡375.0 20 18 19.0 6 5 5.5 F = 36.048, P = 0.0080 FE, FI* FE, OL*
Larval-diet specialist spp.‡354.0 9 10 9.5 2 2 2.0 F = 36.200, P = 0.0079 FE, FI* FE, OL*
Site-specific (endemic) spp.‡000.0 6 6 6.0 4 0 2.0 t = 2.000, P = 0.1835
Narrow-geographic-range spp.‡063.0 13 14 13.5 5 2 3.5 F = 9.152, P = 0.0528 –
Low-population-density spp.‡132.0 16 13 14.5 5 3 4.0 F = 31.824, P = 0.0096 FE, FI* FE, OL*
†Reference to Primack (1993).
‡See text for exact criteria.
*P < 0.0167.
Diversity and rarity hotspots of butterflies 511
(corresponding to <2 or from 2 to less than 5) than
by the number of species observed in two land-
scape types, multivoltines (more than bivoltines),
broad diet breadth (corresponding to ≥30 species
of host plants) and high population densities (cor-
responding to ≥5), respectively. Thus, the high
species richness in forest-edge sites was closely
related to the high numbers of species observed in
only one landscape type, and the number of oligo-
voltine species, narrow-diet-breadth species and
low-population-density species.
Relationship between butterfly species
richness and voltinism, larval host-plant
type and plant species richness
As expected from the level of human disturbance
at each landscape type, open-land (more disturbed)
and forest-interior (less disturbed) communities
had higher proportions of multivoltine (more than
bivoltine) and univoltine species, respectively
(Fig. 1a). In forest-edge communities on both less-
disturbed (woodland parts) and more-disturbed
(treeless parts) areas, the highest numbers of all
categories of species (univoltines, bivoltines and
multivoltines) were detected. In particular,
marked increases in the number of univoltine or
bivoltine species contributed to the high species
richness in forest-edge communities.
Similarly, as expected from landscape types,
open-land (more grassy) and forest-interior (wood-
land) communities had higher proportions of spe-
cies of larval grass feeders and larval tree feeders,
respectively (Fig. 1b). In forest-edge communities
in both woodland and treeless grassland areas, the
highest number of species of both tree and grass
feeders were detected. In particular, marked
increases in the number of species of grass feeders
contributed to the high species richness in forest-
edge communities.
Butterfly species richness was significant and
positively correlated with plant (Phanerogamae)
species richness at each census site (Fig. 1c).
Relationship between species richness
and diversity and species vulnerable
to extinction
All correlations of species richness and diversity
(H
¢
) with numbers on the Red List, univoltine,
larval-diet specialist, narrow-geographic-range
and low-population-density species except
Table 3 Number and frequency (%) of butterfly species belonging to each of several ecological characteristics in the
three landscape types at the census sites
Open-land
sites (26 spp.)
Forest-edge
sites (53 spp.)
Forest-interior
sites (16 spp.) r2
(a) No. landscape types observed
38 (30.8%) 8 (15.1%) 8 (50.0%) –
218 (69.2%) 21 (39.6%) 4 (25.0%) 0.666
10 (0.0%) 24 (45.3%) 4 (25.0%) 0.834
(b) Voltinism (no. generations per year)
≥314 (53.8%) 16 (30.8%) 5 (33.3%) 0.671
24 (15.4%) 13 (25.0%) 2 (13.3%) 0.992
18 (30.8%) 23 (44.2%) 8 (53.3%) 0.931
(c) No. potential host-plant species
≥30 5 (19.2%) 6 (11.5%) 2 (12.5%) 0.736
10–30 8 (30.8%) 22 (42.3%) 6 (37.5%) 0.978
1–10 13 (50.0%) 24 (46.2%) 8 (50.0%) 0.998
(d) Average population density (no. month-1 ha-1)
≥512 (46.2%) 14 (26.4%) 5 (31.3%) 0.709
2–5 10 (38.5%) 19 (35.8%) 5 (31.3%) 0.990
0–2 4 (15.4%) 20 (37.7%) 6 (37.5%) 0.863
512 M. Kitahara and M. Watanabe
Fig. 1. Relationship between species richness (total
number of species) and (a) the number of species with
different voltinism (univoltines: r = 0.961, P = 0.0007;
bivoltines: r = 0.988, P = 0.0001; multivoltines (more
than bivoltines): r = 0.844, P = 0.032), (b) with differ-
ent host-plant types (tree feeders: r = 0.813,
P = 0.0493; grass feeders: r = 0.969, P = 0.0003) and
(c) with plants (belonging to Phanerogamae) (r = 0.924,
P = 0.0225) in each butterfly community at each census
site studied. The FI-2 site is excluded from the analysis
of (c) because of the lack of the data on the number of
plant species.
Table 4 Correlation coefficient of species richness (total no. species) or species diversity (H¢) of each butterfly community with the number of the various species
vulnerable to extinction at each of the six census sites
Red List
spp.
Univoltine
spp.
Larval-diet-
specialist spp.
Site-specific
(endemic) spp.
Narrow-geographic-
range spp.
Low-population-
density spp.
Species richness
(total no. species)
0.957*** 0.961*** 0.990*** 0.734 0.936** 0.908**
Species diversity (H¢)0.965*** 0.903** 0.986*** 0.654 0.888*0.831*
*P < 0.05, **P < 0.01, ***P < 0.001.
Diversity and rarity hotspots of butterflies 513
site-specific species were positive (Table 4). This
indicates that butterfly communities with high
species richness and diversity had many more
species vulnerable to extinction than communities
with low species richness and diversity.
Ordination of butterfly communities
The distribution of the butterfly communities at
the six sites on major- and minor-axes planes using
PCA based on the variance–covariance matrix is
shown in Fig. 2. There were two distinct groups:
(i) group A distributed on the left side of the first
axis; and (ii) group B distributed sparsely on the
right side of the first axis. This suggests that the
butterfly community in and around the Aokiga-
hara woodland was represented by forest-edge and
forest-interior communities, according to species
composition and occurrence patterns.
The cumulative contribution by the first and
second principal components was 83.6%. In the
first axis, the values of eigenvectors almost had a
bias toward positive, and the sites FE-1, FE-2 and
OL-2, which showed higher total population den-
sity, had higher eigenvector values (>0.5), whereas
sites FI-1 and FI-2, which showed lower total pop-
ulation density, had lower eigenvector values (<0)
(Fig. 2). These results suggest that the first axis
reflects total population density of the butterfly
communities. In the second axis, the eigenvector
values of sites OL-1 and OL-2, which were highly
disturbed grassland areas, were positive and those
of sites FI-1 and FI-2, which were woodland areas,
were almost zero. In contrast, the value for site FE-
1, which included semi-natural grassland areas,
was negative (Fig. 2). Thus, we suggest that the
second axis reflects a gradient of human distur-
bance from highly disturbed grassland via wood-
land (no grasses) to semi-natural (secondary)
grassland. This suggestion is highly supported by
the species composition of the butterfly communi-
ties at each census site (Table 5). That is, the two
OL sites, which were distributed on the positive
side of the second axis, had higher proportions of
grass-feeding species, most of which were multi-
voltine grass feeders associated with more dis-
turbed grassland habitats, whereas the two FE
sites, which were distributed on the more negative
Fig. 2. Scattergram of the six study sites along the
first and second principal component axes of the prin-
cipal components analysis (PCA) ordination.
Table 5 Number and its proportion of species of larval grass or tree feeders and those of univoltine or multivoltine
grass feeders in each census site. The larval hostplant type (grass or tree) of each component species was examined
based on the literature (Unno and Aoyama 1981) (see Appendix I)
Site
order†Site
Number of species (proportion)
Larval grass
feeders (%)
Univoltine grass
feeders (%)
Multivoltine grass
feeders (%)
Unknown
(%)
Larval tree
feeders (%) Total
1OL-2 20 (83.3) 6 (30.0) 14 (70.0) 4 (16.7) 24
2OL-1 8 (61.5) 1 (12.5) 7 (87.5) 5 (38.5) 13
3FI-2 3 (27.3) 1 (33.3) 1 (33.3) 1 (33.3) 8 (72.7) 11
4FI-1 2 (20.0) 0 (0.0) 2 (100) 8 (80.0) 10
5FE-2 27 (64.3) 11 (40.7) 15 (55.6) 1 (3.7) 15 (35.7) 42
6FE-1 28 (63.6) 13 (46.4) 15 (53.6) 16 (36.4) 44
†Site order from positive to negative on the second principal component of the principal components analysis (cf. Fig. 2).
514 M. Kitahara and M. Watanabe
side of the second axis, also had higher proportions
of grass-feeding species, but they included more
univoltine grass feeders associated with less dis-
turbed (i.e. secondary) grassland habitats. In com-
parison with these sites, the two FI sites were
distributed between the above two site groups on
the second axis, and had much lower proportions
of grass-feeding species. Most of the component
species at the FI sites were tree feeders associated
with woodland habitats.
Characteristics of site-specific species
Figure 3 shows the relationship between the num-
ber of census sites at which a butterfly species was
observed and several ecological traits. Significant
positive correlations were detected between the
number of sites and average population density,
number of generations per year and number of
10-km-blocks in Japan in which the species was
recorded. The correlation between the number of
sites and the number of potential host-plant spe-
cies was not significant. These results suggest that
site-specific and site-restricted species were closely
associated with relatively low population densities,
a few generations per year and small geographic
range sizes in Japan.
DISCUSSION
Diversity, structure and hotspots of
butterfly communities
This study showed that butterfly species richness
and species diversities, H
¢
and 1/l, were signifi-
cantly higher in forest-edge sites than in forest-
interior and/or open-land sites (Table 2), indicat-
ing that in and around Aokigahara woodland
there were butterfly diversity hotspots at the for-
est edges. Relatively high species diversity and
richness at forest edges or transitional and
ecotonal areas from open land to woodland have
been reported in many butterfly studies (e.g.
Erhardt 1985; Leps & Spitzer 1990; Ishii et al.
1993, 1995; Spitzer et al. 1993; Kitahara & Fujii
1994; Ishii 1996a; Yata 1996; Tashita & Ichimura
1997; Natuhara et al. 1999; Balmer & Erhardt
2000; Schneider & Fry 2001). Thus, this trend
appears to be a general and consistent one in but-
terfly diversity patterns in these landscapes. In
general, this trend can also be understood as an
edge effect (i.e. high species diversities and densi-
ties are recognized in boundaries between differ-
ent habitat or landscape types) (Odum 1971;
Fagan et al. 1999). In addition, this study also
showed that variation in the total number of spe-
cies among the three landscape types was well
accounted for by species that were landscape spe-
cific, oligovoltine, and with a narrow diet breadth
and low population density (Table 3). In other
words, these types of species contributed to the
differences in the number and composition of spe-
cies among the landscape types, whereas other
species types did not. When we examine the char-
acteristics of the former species based on the r/K
concept (MacArthur & Wilson 1967; Pianka
1970, 1988; Gadgil & Solbrig 1972; MacArthur
1972; Southwood et al. 1974; Southwood 1977,
1981, 1988) or the generalist/specialist concept
(sensu Odum 1989; Novotny 1991; Kitahara &
Fujii 1994, 1997; Leps et al. 1998), these species
correspond to K-type (with lower r) or ecological
specialist species. The tendency for K-type or eco-
logically specialist species to contribute to differ-
ences in the number and composition of species
among communities has been shown in several
butterfly community studies (Spitzer et al. 1993;
Kitahara & Fujii 1994; Ishii 1996a; Kitahara
et al. 2000; Kitahara & Sei 2001). Thus, this ten-
dency also appears to be a general and consistent
trend in the spatial changing patterns of butterfly
communities.
Why is butterfly species richness high in forest-
edge landscapes? We answered this question from
the following three points of view (Fig. 1). First,
forest-edge areas included more-disturbed (mainly
treeless open land) and less-disturbed (mainly
woodland) parts. Thus, it is predicted that forest-
edge communities show high species richness
because of the inclusion of both r- and K-type spe-
cies, compared with communities in only open-
land or woodland areas. Indeed, our study con-
firmed the high species richness of both multivol-
tines (r-type species) and oligovoltines (K-type
species) in forest-edge sites and, therefore, verified
the above prediction. Similarly, we confirmed that
species richness in forest-edge sites, resulting from
the inclusion of both grassland and woodland spe-
cies, was high compared to the richness observed
Diversity and rarity hotspots of butterflies 515
in open-land or woodland areas only. Second, but-
terflies are herbivores and their diet resources are
almost entirely dependent on specific plants for
both larval and adult stages. Thus, it is predicted
that the number of species within a butterfly com-
munity in a certain area is strongly affected by the
number of plant species. As this prediction sug-
gests, our study revealed a strong positive correla-
tion between butterfly and plant species richness at
each census site and, thus, the high butterfly spe-
cies richness at forest-edge sites was closely linked
to a high plant species richness. Finally, the high
butterfly species richness at forest-edge sites prob-
ably led to the high butterfly species diversity.
In the present study, the various types of species
vulnerable to extinction, including Red List spe-
cies, were more often observed at forest-edge sites
than in forest-interior or open-land sites (Table 2).
Thus, similar to the diversity hotspots, there were
butterfly rarity hotspots in areas at the forest edges
in and around the Aokigahara woodland and,
therefore, both hotspots occurred in the same land-
scape area. Moreover, we suggest that because
strong positive correlations were detected between
the numbers of almost all types of species vulner-
able to extinction and butterfly species richness
and diversity (H
¢
) (Table 4), the high richness of
species vulnerable to extinction in forest-edge sites
significantly contributed to the high total species
richness and diversity.
Why were there butterfly rarity hotspots in for-
est-edge areas? In Japanese and British butterfly
communities (Asher et al. 2001), species that are
vulnerable to extinction are characteristic of semi-
natural grasslands or coppice woodlands (Ishii
1996b). In particular, in and around Mount Fuji,
Fig. 3. Relationship between the number of sites in which component species were observed and (a) average pop-
ulation density (rs = 0.530, P < 0.0001), (b) number of generations per year (rs = 0.333, P < 0.05), (c) number of
potential host-plant species (rs = 0.072, P > 0.05) and (d) number of 10-km-blocks where the species was recorded in
Japan (rs = 0.472, P < 0.001). Each dot and bar represents mean ± SD, respectively.
516 M. Kitahara and M. Watanabe
most of the recorded Red List species authorized
by M.E.J. (2000) are semi-natural grassland or
open-forest species (Sei 1988; Kitahara 1999). In
addition, various types of species treated as vulner-
able to extinction in the present study are charac-
teristic of semi-natural grassland or woodland
habitats according to Fukuda et al. (1982, 1983,
1984a,b). Thus, we suggest that in the area of the
present study, butterfly rarity hotspots formed at
forest-edge sites, which included areas of semi-nat-
ural grassland with a low frequency of human
impact and woodland areas with almost no human
impact.
Conservation implications for butterfly
diversity and rarity
The present study showed that both butterfly
diversity and rarity hotspots were detected in for-
est-edge landscapes (Table 2). In addition, site-
specific and landscape-specific species, which are
considered to have a high conservation value
because of their endemism to specific habitats,
were observed only in the areas of forest edges and
interiors (Tables 2,3). From these results we con-
clude that, at least in this area, the conservation of
forest-edge habitats in addition to woodland hab-
itats is very important for the maintenance of but-
terfly species diversity and richness and for the
preservation of rare or threatened species. More-
over, the results of the PCA suggest that all six
butterfly communities were divided into two com-
munity groups represented by forest-edge and for-
est-interior communities (Fig. 2). It is likely that
the communities at OL-1 and OL-2 belonged to a
different group because the differences in their
total population densities and species composition
resulted from the different types of human distur-
bance at these sites. Thus, the need to conserve
both forest-edge and woodland habitats can be
supported by the species grouping and composi-
tion of the butterfly communities observed in the
present study. Accordingly, in and around the
Aokigahara woodland, forest-edge and woodland
areas have great potential value and, therefore,
should be considered to be priority areas for but-
terfly conservation.
In the present study, species observed at fewer
sites (i.e. restricted species) were associated with
traits that reflect a vulnerability to extinction,
such as lower population densities, fewer genera-
tions per year and smaller geographic range sizes
(Fig. 3). Thus, these species appear to be represen-
tative of species prone to extinction and, therefore,
can be thought of as species with high values of
conservation or priority species of concern for con-
servation. The positive relationship between dis-
tribution and abundance has been well shown for
many organisms, and is considered to be a general
community pattern (Hanski 1982; Brown 1984;
Bock 1987; Gaston & Lawton 1988, 1990; Han-
ski et al. 1993; Lawton 1993; Gaston 1994,
1996). In general, it is suggested that species
prone to extinction have similar characteristics.
This should be emphasized and is very important
in considering and planning butterfly conserva-
tion. We need to monitor and manage extinction-
prone species carefully to maintain local butterfly
diversity.
ACKNOWLEDGEMENTS
We are most grateful to Professor K. Fujii of the
University of Tsukuba for his continuous advice
and encouragement throughout the study, and
also to Professor Emeritus Y. Ito of Nagoya Uni-
versity for helpful comments on the use of the
diversity index. Special thanks are also due to Dr
M. Iriki (Director) and other members of the
Yamanashi Institute of Environmental Sciences,
in particular Drs H. Imaki, Z. Jiang, and H.
Ueda of the Laboratory of Animal Ecology for
their discussion, help and understanding, and Drs
T. Ohtsuka and T. Nakano from the Laboratory of
Plant Ecology for providing us with their unpub-
lished data on the vegetation of the Aokigahara
woodland. Finally, we wish to thank Dr H.
Yasuda and two referees who gave invaluable and
constructive comments on the manuscript, which
contributed greatly to the improvement of the
contents.
REFERENCES
ASHER J., WARREN M., FOX R., HARDING P.,
JEFFCOATE G. & JEFFCOATE S. (2001) The Mil-
lennium Atlas of Butterflies in Britain and Ireland.
Oxford University Press, Oxford.
Diversity and rarity hotspots of butterflies 517
BALMER O. & ERHARDT A. (2000) Consequences
of succession on extensively grazed grasslands for
central European butterfly communities: rethink-
ing conservation practices. Conservation Biology 14:
746–757.
BLAIR R. B. & LAUNER A. E. (1997) Butterfly
diversity and human land use: species assemblages
along an urban gradient. Biological Conservation 80:
113–125.
BOCK C. E. (1987) Distribution–abundance rela-
tionships of some Arizona landbirds: a matter of
scale? Ecology 68: 124–129.
BROWN J. H. (1984) On the relationship between
abundance and distribution of species. American
Naturalist 124: 255–279.
EHRLICH P. R. (1992) Population biology of check-
erspot butterflies and the preservation of global
biodiversity. Oikos 63: 6–12.
EHRLICH P. R., BREEDLOVE D. E., BRUSSARD
P. F. & SHARP M. (1972) Weather and the regulation
of subalpine populations. Ecology 53: 243–247.
ENDO S. & NIHIRA I. (1990) Larval Food of Japanese
Butterflies. Group ‘Tamamushi’, Tokyo (in Japanese).
ERHARDT A. (1985) Diurnal Lepidoptera: sensitive
indicators of cultivated and abandoned grassland.
Journal of Applied Ecology 22: 849–861.
ESUMI (1998) EXCEL-Multivariate Analysis, Version
3.0. Esumi Inc., Tokyo (translated into Japanese).
FAGAN W. F., CANTRELL R. S. & COSNER C.
(1999) How habitat edges change species interac-
tions. American Naturalist 153: 165–182.
FUKUDA H., HAMA E., KUZUYA T., TAKAHASHI
A., TAKAHASHI M., TANAKA B., TANAKA H.,
WAKABAYASHI M. & WATANABE Y. (1982) The
Life Histories of Butterflies in Japan, Vol. 1. Hoikusha,
Osaka (in Japanese with English summary).
FUKUDA H., HAMA E., KUZUYA T., TAKAHASHI
A., TAKAHASHI M., TANAKA B., TANAKA H.,
WAKABAYASHI M. & WATANABE Y. (1983)
The Life Histories of Butterflies in Japan, Vol. 2.
Hoikusha, Osaka (in Japanese with English
summary).
FUKUDA H., HAMA E., KUZUYA T., TAKAHASHI
A., TAKAHASHI M., TANAKA B., TANAKA H.,
WAKABAYASHI M. & WATANABE Y. (1984a)
The Life Histories of Butterflies in Japan, Vol. 3.
Hoikusha, Osaka (in Japanese with English
summary).
FUKUDA H., HAMA E., KUZUYA T., TAKAHASHI
A., TAKAHASHI M., TANAKA B., TANAKA H.,
WAKABAYASHI M. & WATANABE Y. (1984b)
The Life Histories of Butterflies in Japan, Vol. 4.
Hoikusha, Osaka (in Japanese with English
summary).
GADGIL M. & SOLBRIG O. T. (1972) The concept
of r- and K-selection: evidence from wild flowers
and some theoretical considerations. American
Naturalist 106: 14–31.
GALL L. F. (1985) Measuring the size of Lepi-
dopteran populations. Journal of Research on the Lep-
idoptera 24: 97–116.
GASTON K. J. (1994) Rarity. Chapman & Hall, Lon-
don.
GASTON K. J. (1996) The multiple forms of the
interspecific abundance–distribution relationship.
Oikos 76: 211–220.
GASTON K. J. & LAWTON J. H. (1988) Patterns in
the distribution and abundance of insect popula-
tions. Nature 331: 709–712.
GASTON K. J. & LAWTON J. H. (1990) Effects of
scale and habitat on the relationship between
regional distribution and local abundance. Oikos
58: 329–335.
HAMER K. C., HILL J. K., LACE L. A. & LANGAN
A. M. (1997) Ecological and biogeographical
effects of forest disturbance on tropical butterflies
of Sumba, Indonesia. Journal of Biogeography 24:
67–75.
HANSKI I. (1982) Dynamics of regional distribu-
tion: the core and satellite species hypothesis.
Oikos 38: 210–221.
HANSKI I., KOUKI J. & HALKKA A. (1993) Three
explanations of the positive relationship between
distribution and abundance of species. In: Species
Diversity in Ecological Communities: Historical and
Geographical Perspectives (eds R. E. Ricklefs & D.
Schluter), pp. 108–116. University of Chicago
Press, Chicago.
HILL J. K., HAMER K. C., LACE L. A. & BANHAM
W. M. T. (1995) Effects of selective logging on
tropical forest butterflies on Buru, Indonesia. Jour-
nal of Applied Ecology 32: 754–760.
HUNTER M. L. Jr (1996) Fundamentals of Conserva-
tion Biology. Blackwell Science, Cambridge.
ISHII M. (1993) Transect counts of butterflies. In:
Decline and Conservation of Butterflies in Japan II (eds
O. Yata & K. Ueda), pp. 91–101. The Lepidopter-
ological Society of Japan & the Nature Conserva-
tion Society of Japan, Osaka (in Japanese with
English summary).
ISHII M. (1996a) Species diversity of butterfly com-
munities in different environment of forests in
southern Osaka. In: Decline and Conservation of But-
terflies in Japan IV (eds B. Tanaka & Y. Arita), pp.
63–75. The Lepidopterological Society of Japan,
Osaka (in Japanese with English summary).
ISHII M. (1996b) Decline and conservation of but-
terflies in Japan. In: Decline and Conservation of
518 M. Kitahara and M. Watanabe
Butterflies in Japan III (eds S. Ae, T. Hirowatari, M.
Ishii & L. P. Brower), pp. 157–167. The Lepi-
dopterological Society of Japan, Osaka.
ISHII M., HIROWATARI T. & FUJIWARA S. (1995)
Species diversity of butterfly communities in ‘Mt
Mikusa coppice for Zephyrus’. Japanese Journal of
Environmental Entomology and Zoology 7: 134–146
(in Japanese with English abstract).
ISHII M., UEDA K. & SHIGEMATSU T. (1993)
[Conservation of the Nature of ‘Satoyama’.] Tsukiji
Shokan Publishers, Tokyo (in Japanese).
KITAHARA M. (1999) Species diversity of butterfly
communities in a variety of woodlands at the
northern foot of Mt Fuji, central Japan. Japanese
Journal of Environmental Entomology and Zoology 10:
11–29 (in Japanese with English abstract).
KITAHARA M. & FUJII K. (1994) Biodiversity and
community structure of temperate butterfly spe-
cies within a gradient of human disturbance: an
analysis based on the concept of generalist vs. Spe-
cialist strategies. Researches on Population Ecology
36: 187–199.
KITAHARA M. & FUJII K. (1997) An island bio-
geographical approach to the analysis of butterfly
community patterns in newly designed parks.
Researches on Population Ecology 39: 23–35.
KITAHARA M. & SEI K. (2001) A comparison of the
diversity and structure of butterfly communities
in semi-natural and human-modified grassland
habitats at the foot of Mt Fuji, central Japan.
Biodiversity and Conservation 10: 331–351.
KITAHARA M., SEI K. & FUJII K. (2000) Patterns
in the structure of grassland butterfly communi-
ties along a gradient of human disturbance: fur-
ther analysis based on the generalist/specialist
concept. Population Ecology 42: 135–144.
KREMEN C. (1992) Assessing the indicator proper-
ties of species assemblages for natural areas mon-
itoring. Ecological Applications 2: 203–217.
KREMEN C., COLWELL R. K., ERWIN T. L., MUR-
PHY D. D., NOSS R. F. & SANJAYAN M. A. (1993)
Terrestrial arthropod assemblages-their use in con-
servation planning. Conservation Biology 7: 796–808.
LAWTON J. H. (1993) Range, population abun-
dance and conservation. Trends in Ecology and Evo-
lution 8: 409–413.
LEPS J. & SPITZER K. (1990) Ecological determi-
nants of butterfly communities (Lepidoptera,
Papilionoidea) in the Tam Dao Mountains, Viet-
nam. Acta Entomologica Bohemoslovaca 87: 182–194.
LEPS J., SPITZER K. & JAROS J. (1998) Food
plants, species composition and variability of the
moth community in undisturbed forest. Oikos 81:
538–548.
LINCOLN R., BOXSHALL G. & CLARK P. (1998)
A Dictionary of Ecology, Evolution and Systematics,
2nd edn. Cambridge University Press,
Cambridge.
MACARTHUR R. H. (1972) Geographical Ecology:
Patterns in the Distribution of Species. Harper & Low,
New York.
MACARTHUR R. H. & WILSON E. O. (1967) The
Theory of Island Biogeography. Princeton University
Press, Princeton.
MINISTRY OF THE ENVIRONMENT OF JAPAN (2000)
[Red Data List of Insects in Japan.] http://
www.biodic.go.jp/ Biodiversity Center of Japan,
Japan.
MITTERMEIER R. A. & FORSYTH A. B. (1997) Set-
ting priorities for biodiversity conservation. In:
Principles of Conservation Biology, 2nd edn (eds G.
K. Meffe, C. R. Carroll & Contributers), pp. 141–
142. Sinauer Associates Inc., Sunderland.
MITTERMEIER R. A., MYERS J. B., THOMSEN
G. A. & OLIVIERI S. (1998) Biodiversity hotspots
and major tropical wilderness areas: approaches to
setting conservation priorities. Conservation Biology
12: 516–520.
MYERS N. (1988) Threatened biota: ‘hotspots’ in
tropical forests. Environmentalist 8: 1–10.
MYERS N. (1997) Global biodiversity II: losses and
threats. In: Principles of Conservation Biology, 2nd
edn (eds G. K. Meffe, C. R. Carroll & Contribut-
ers), pp. 123–158. Sinauer Associates Inc.,
Sunderland.
NATUHARA Y., IMAI C. & TAKAHASHI M. (1999)
Pattern of land mosaics affecting butterfly assem-
blage at Mt Ikoma, Osaka, Japan. Ecological
Research 14: 105–118.
NATURE CONSERVATION BUREAU (1993) [Report on
the 4th Basic Research in Natural Environmental Con-
servation, with Special Reference to the Distribution
Patterns of Japanese Animals and Plants (Volume But-
terflies)]. Environmental Agency Government of
Japan, Tokyo (in Japanese).
NEW T. R. (1991) Butterfly Conservation. Oxford
University Press, Melbourne.
NOVOTNY V. (1991) Effect of habitat persistence
on the relationship between geographic distribu-
tion and local abundance. Oikos 61: 431–433.
ODUM E. P. (1971) Fundamentals of Ecology, 3rd edn.
W. B. Sanders Co., Philadelphia.
ODUM E. P. (1989) Ecology and Our Endangered Life-
Support Systems. Sinauer Associates Inc., Sunderland.
PIANKA E. R. (1970) On r- and K-selection. Amer-
ican Naturalist 102: 592–597.
PIANKA E. R. (1988) Evolutionary Ecology, 4th edn.
Harper & Row, New York.
Diversity and rarity hotspots of butterflies 519
POLLARD E. (1977) A method for assessing changes
in the abundance of butterflies. Biological Conser-
vation 12: 116–134.
POLLARD E. (1984) Synoptic studies on butterfly
abundance. In: The Biology of Butterflies (eds R. I.
Vane-Wright & P. R. Ackery), pp. 59–61. Aca-
demic Press, London.
POLLARD E. & YATES T. J. (1993) Monitoring But-
terflies for Ecology and Conservation. Chapman &
Hall, London.
PRESSEY R. L. (1996) Protected areas: where should
they be and why should they be there. In: Conser-
vation Biology (ed. I. F. Spellerberg), pp. 171–185.
Longman, Essex.
PRIMACK R. B. (1993) Essentials of Conservation Biol-
ogy. Sinauer Associates Inc., Sunderland.
PRIMACK R. B. (1995) A Primer of Conservation Biol-
ogy. Sinauer Associates Inc., Sunderland.
REID W. V. (1998) Biodiversity hotspots. Trends in
Ecology and Evolution 13: 275–280.
SCHNEIDER C. & FRY G. L. A. (2001) The influ-
ence of landscape grain size on butterfly diversity
in grasslands. Journal of Insect Conservation 5: 163–
171.
SEI K. (1988) [The butterflies unable to have inhabited in
the Mt. Fuji area.] Tsukiji Shokan, Tokyo (in Japanese).
SEIDO K. (1991) Age of the hinoki (Chamaecyparis
obtusa) natural stands and their distribution on the
Aokigahara Woodland, Mt. Fuji. Transactions of the
Meeting in Kanto Branch of the Japanese Forestry Soci-
ety 42: 33–36 (in Japanese).
SHREEVE T. G. & MASON C. F. (1980) The number
of butterfly species in woodlands. Oecologia 45:
414–418.
SIMPSON E. H. (1949) Measurement of diversity.
Nature 163: 688.
SOUTHWOOD T. R. E. (1977) Habitat, the templet
for ecological strategies? Journal of Animal Ecology
46: 337–365.
SOUTHWOOD T. R. E. (1978) Ecological Methods,
with Particular Reference to the Study of Insect Popu-
lations. Chapman & Hall, London.
SOUTHWOOD T. R. E. (1981) Bionomic strategies
and population parameters. In: Theoretical Ecology:
Principles and Applications, 2nd edn (ed. R. M.
May), pp. 30–52. Blackwell Scientific Publica-
tions, Oxford.
SOUTHWOOD T. R. E. (1988) Tactics, strategies,
and templets. Oikos 52: 3–18.
SOUTHWOOD T. R. E., MAY R. M., HASSELL
M. P. & CONWAY G. R. (1974) Ecological strat-
egies and population parameters. American Natu-
ralist 108: 791–804.
SPITZER K., NOVOTNY V., TONNER M. & LEPS
J. (1993) Habitat preferences, distribution and
seasonality of the butterflies (Lepidoptera, Papil-
ionoidea) in a montane tropical rain forest, Viet-
nam. Journal of Biogeography 20: 109–121.
TASHITA M. & ICHIMURA T. (1997) Environmen-
tal evaluation by butterfly communities according
to altitude in Nagano Prefecture. Japanese Journal
of Environmental Entomology and Zoology 8: 73–88
(in Japanese with English abstract).
THOMAS J. A. (1983) A quick method for estimat-
ing butterfly numbers during surveys. Biological
Conservation 27: 195–211.
TSUYA H. (1971) Topography and geology of Vol-
cano Mt. Fuji. In: Report on the synthetic academic
surveys of Mt. Fuji, pp. 1–149. Fujikyuko Co.,
Fujiyoshida-city (in Japanese with English
summary).
UNNO K. & AOYAMA J. (1981) [The Butterflies of
Japan.] Shougakukan, Tokyo (in Japanese).
WARREN M. S., POLLARD E. & BIBBY T. J. (1986)
Annual and long-term changes in a population of
the wood white butterfly Leptidea sinapis. Journal of
Animal Ecology 55: 707–719.
WASHITANI I. (1999) [Introduction to New Ecology:
Ecology for Biological Conservation.] Kyouritsu
Shuppan Co., Tokyo (in Japanese).
WEISS S. B., WHITE R. R., MURPHY D. D. &
EHRLICH P. R. (1987) Growth and dispersal of
larvae of the checkerspot butterfly Euphydryas
editha. Oikos 50: 161–166.
WOOD B. & GILLMAN M. P. (1998) The effects of
disturbance on forest butterflies using two meth-
ods of sampling in Trinidad. Biodiversity and Con-
servation 7: 597–616.
YAMAMOTO M. (1988) On the research method for
butterfly assemblages. Special Bulletin of the Lepi-
dopterological Society of Japan 6: 191–210 (in Japa-
nese with English abstract).
YATA O. (1996) Butterfly community in laurel-
leaved forest of Yamada park, Kitakyushu. In:
Decline and Conservation of Butterflies in Japan IV
(eds B. Tanaka & Y. Arita), pp. 49–56. The Lepi-
dopterological Society of Japan, Osaka (in Japa-
nese with English summary).
520 M. Kitahara and M. Watanabe
Appendix I List of butterfly species observed in the present study, their population densities, and their characteristics
Population density
Potential
larval
diet
breadth†
Larval
hostplant
type‡
Local
distribution
pattern§
Geographic
range
size¶
Status in
Red Data
list of
Japan††
Species OL-1 OL-2 FE-1 FE-2 FI-1 FI-2 Average Voltinism
Hesperiidae
Erynnis montanus 3.33 5.83 4.17 1.67 1.67 3.33 1 6 W 5 636
Daimio tethys 3.33 3.33 2 6 H 1 781
Leptalina unicolor 1.11 11.11 3.33 5.19 1 7 H 3 256 NT
Thymelicus sylvaticus 1.67 1.67 1.67 1 6 H 2 177
Ochlodes venatus 1.67 1.67 1 14 H 1 352
Ochlodes ochraceus 7.41 9.72 8.56 1 12 H 2 500
Potanthus flavus 1.67 1.67 1 21 H 1 494
Pelopidas jansonis 1.67 1.94 1.81 2 7 H 2 285
Parnara guttata 2.50 15.00 8.89 15.83 10.56 3£28 H 4 898
Papilionidae
Parnassius glacialis 4.72 20.83 18.89 14.81 1 5 H 3 599
Graphium sarpedon 1.67 1.67 2 14 W 1 738
Papilio machaon 2.50 2.50 3£42 H 1 1098
Papilio macilentus 2.22 2.22 2 5 W 1 639
Papilio protenor 1.11 1.11 2 15 W 1 871
Papilio bianor 1.67 1.39 2.50 1.85 2 10 W 3 1152
Pieridae
Leptidea amurensis 2.59 4.17 3.38 2 1 H 2 144 VU
Eurema hecabe 3.33 5.00 7.04 9.03 3.33 5.55 3£30 W 5 301
Gonepteryx rhamni 3.33 3.33 1 1 W 1 47 NT
Gonepteryx aspasia 1.67 1.67 1 4 W 1 377
Colias erate 9.37 11.76 4.44 4.63 7.55 3£21 H 4 1179
Anthocharis scolymus 1.67 1.67 1 18 H 1 704
Diversity and rarity hotspots of butterflies 521
Population density
Potential
larval
diet
breadth†
Larval
hostplant
type‡
Local
distribution
pattern§
Geographic
range
size¶
Status in
Red Data
list of
Japan††
Species OL-1 OL-2 FE-1 FE-2 FI-1 FI-2 Average Voltinism
Pieris rapae 3.15 13.49 2.92 2.41 5.49 3£25 H 4 1209
Pieris melete or napi 3.33 15.78 15.19 13.67 2.22 10.04 3£16.5 H 5 902
Lycaenidae
Narathura japonica 2.22 1.67 1.94 1 13 W 2 556
Antigius attilia 1.67 1.67 1 11 W 1 561
Chrysozephyrus smaragdinus 5.00 5.00 1 14 W 1 345
Favonius jezoensis 0.83 1.39 1.11 1 8 W 2 260
Rapala arata 1.39 1.48 1.44 2 25 W 2 695
Callophrys ferrea 1.67 1.67 1.67 1 18 W 2 597
Lycaena phlaeas 1.94 6.56 5.37 2.01 3.97 3£5H 4 1179
Lampides boeticus 6.67 6.67 6.67 3£41 H 2 461
Pseudozizeeria maha 6.67 3.33 1.67 3.89 3£2H 3 905
Celastrina argiolus 1.67 3.06 1.32 1.67 2.22 1.99 3 32 H 5 1276
Celastrina sugitanii 1.67 1.67 1 3 W 1 325
Everes argiades 2.22 1.39 2.50 2.92 2.26 3 31 H 4 946
Plebejus argus 4.44 15.83 1.53 7.27 1 38 H 3 274 NT
Lycaeides subsolanus 1.39 11.11 6.25 1 9 H 6 86 VU
Curetis acuta 1.39 10.00 5.69 2 10 W 2 672
Libytheidae
Libythea celtis 4.17 1.39 1.48 3.06 4.81 3.06 2.99 1 4 W 6 820
Danaidae
Parantica sita 1.11 8.33 4.72 ? 12 H 2 660
Nymphalidae
Argyronome ruslana 1.67 2.50 2.08 1 1 H 2 408
Appendix I List of butterfly species observed in the present study, their population densities, and their characteristics
Appendix I Continued.
522 M. Kitahara and M. Watanabe
Population density
Potential
larval
diet
breadth†
Larval
hostplant
type‡
Local
distribution
pattern§
Geographic
range
size¶
Status in
Red Data
list of
Japan††
Species OL-1 OL-2 FE-1 FE-2 FI-1 FI-2 Average Voltinism
Argynnis paphia 3.33 3.33 5.83 4.44 3.33 4.06 1 6 H 5 791
Nephargynnis anadyomene 2.78 1.11 1.94 1 2 H 2 410
Fabriciana adippe 2.78 2.04 2.41 1 – H 2 589
Limenitis camilla 1.57 3.15 2.36 2 8 W 2 879
Limenitis glorifica 1.67 1.67 2 4 W 1 361
Neptis sappho 1.48 8.19 6.39 1.11 1.67 3.77 3 22 W 5 1139
Neptis philyra 1.11 2.22 1.11 1.11 1.39 1 10 W 4 392
Neptis rivularis 1.67 1.67 1 7 W 1 108
Polygonia c-aureum 1.67 11.98 7.66 9.44 3.33 6.82 3 2 W 5 784
Polygonia c-alubum 1.67 1.67 2 9 W 1 414
Kaniska canace 1.39 3.33 1.94 2.22 2 13 H 3 1048
Inachis io 1.67 2.50 1.25 1.81 2 7 H 3 426
Cynthia cardui 4.03 1.48 0.56 2.02 3_ 9 H 3 635
Vanessa indica 0.83 0.83 3 13 H 1 933
Satyridae
Ypthima argus 3.61 3.75 2.78 3.38 3 12 H 3 1084
Minois dryas 14.17 22.50 13.33 16.67 1 6 H 3 544
† Number of larval potential host-plant species cited from the literature (see text).
‡ W (wood), H (herbaceous).
§ Number of study sites observed.
¶ Number of 10–10 km2 in which the species was recorded (see text).
†† VU: vulnerable species, NT: semivulnerable species.
Appendix I List of butterfly species observed in the present study, their population densities, and their characteristics
Appendix I Continued.
