Ecology, 80(6), 1999, pp.

1806–1816
q 1999 by the Ecological Society of America

SPECIES IN THE TAIL OF RANK–ABUNDANCE CURVES

BRAD R. MURRAY,1,7 BARBARA L. RICE,1 DAVID A. KEITH,2 PETER J. MYERSCOUGH,3 JOCELYN HOWELL,4
ALEX G. FLOYD,5 KEVIN MILLS,6 AND MARK WESTOBY1
1School of Biological Sciences, Macquarie University, New South Wales 2109, Australia
2 National Parks and Wildlife Service, P.O. Box 1967, Hurstville, New South Wales 2220, Australia
3Institute of Wildlife Research, A12, University of Sydney, New South Wales 2006, Australia
4National Herbarium of New South Wales, Royal Botanic Gardens, Sydney,

New South Wales 2000, Australia

54/29 Curacao Street, Coffs Harbour, New South Wales 2450, Australia
6Ecological and Environmental Consultants, 222 North Curramore Road, Jamberoo,

New South Wales 2533, Australia

Abstract. At focal sites within dry sclerophyll woodland and temperate rain forest,
species were identified that were of low local abundance and hence in the tail of the rank–
abundance curve. We then asked the question: What proportion of tail species within a
given community are constitutive members of the tail everywhere throughout their geo-
graphical range, versus what proportion are found as substantially more abundant some-
where within their range? Out of 55 tail species identified from dry sclerophyll woodland
and 116 tail species identified from temperate rain forest, 91% and 95%, respectively, were
significantly more abundant at other locations (‘‘somewhere-abundant’’ species), versus 9%
and 5% ‘‘everywhere-sparse’’ species. Among eight attributes in dry sclerophyll woodland
and nine attributes in temperate rain forest compared between somewhere-abundant and
everywhere-sparse species, none discriminated consistently between the two groups of
species. The size and dispersal morphology of seeds, flowering and fruiting durations and
seasons, regeneration strategy after fire, size of geographical ranges, maximum plant height,
and size class revealed no consistent distinctions. For the small minority of species that
are everywhere-sparse, some general explanation may exist as to why they are in the tail
of rank–abundance curves, though none was located among the attributes compared in this
paper. For the majority of tail species that are somewhere-abundant, any explanation as to
why they are in the tail will need to account for different outcomes in different places.
Key words: abundance; geographical range; macroecology; plants; rain forest; rank-abundance
curve; sclerophyll woodland.

INTRODUCTION and on within-community explanations for their shapes

A fundamental question of ecology concerns the
abundance and distribution of species. In particular,
there has been considerable interest in the distribution
of abundance among species within local communities.
A common observation is that at a given location some
species will be very abundant, while many species will
have small- to moderate-sized populations (Darwin
1859, Preston 1948, MacArthur 1960, Whittaker 1965,
May 1975). Abundance distributions within commu-
nities have been represented usually as rank–abundance
or dominance–diversity curves, with one axis of the
curve representing species rank in a community and
the other representing species abundance.
Both theoretical and empirical studies have inves-
tigated the structure of rank–abundance curves for
many different communities. Theory has concentrated
primarily on the structure of rank–abundance curves
Manuscript received 12 January 1998; revised 25 June
1998; accepted 3 July 1998.
7 Present address: Centre for Plant Biodiversity Research,
CSIRO—Plant Industry, G.P.O. Box 1600, Canberra, ACT
2601, Australia. E-mail: B.Murray@pi.csiro.au
1806
(e.g., Preston 1962, May 1975, Pielou 1975, De Vita
1979, Stenseth 1979, Sugihara 1980, Hughes 1986,
Wilson 1991, Gregory 1994, Wilson et al. 1996). Em-
pirical work on local abundance has focused on fluc-
tuations in the numbers of individuals of a species
through time at one location, or among a small number
of communities, and has also tried to support or reject
the various theories regarding curve structure (e.g.,
Preston 1962, Sugihara 1980, Diamond and Case 1986,
Grubb 1986, Mitchley and Grubb 1986, Chesson and
Huntley 1989, Tokeshi 1990, Wilson 1991, Wilson et
al. 1996). This work has clarified a number of important
issues; however, broad generalizable patterns between
particular curve structures and particular underlying
mechanisms have yet to be found. The present paper
investigates rank–abundance curves from a different
perspective, employing a macroecological approach to
explore how the local abundance of species is related
to abundance throughout entire geographical ranges.
Some empirical studies have investigated species’
abundances across entire or nearly entire geographical
ranges (Hanski 1982, Hengeveld and Haeck 1982, Ra-
September 1999 TAIL SPECIES OF RANK–ABUNDANCE CURVES
FIG. 1. Two alternatives for patterns of
abundance throughout geographical ranges. In
both figures, six hypothetical species are illus-
trated. Solid lines represent species that have
sites of high abundance, while dotted lines rep-
resent species that are everywhere of low abun-
dance. (A) Most species (e.g., five out of six)
have sites of high abundance within their geo-
graphical ranges. (B) Most species are of low
abundance everywhere throughout their geo-
graphical range. In both figures, species 3 be-
haves differently than the other five species.
poport 1982, Brown 1984, Rabinowitz et al. 1986,
Schoener 1987, Maurer 1994, Brown et al. 1995). Gen-
erally, it has been observed that species are found in
low abundance at most sites, throughout their respec-
tive geographical ranges while achieving high abun-
dances only at a small number of sites (e.g., Brown et
al. 1995). The recurring rank-abundance pattern within
local communities of a small number of abundant spe-
cies together with a long tail of lower-abundance spe-
cies might be associated with a geographical-scale pat-
tern whereby most species are of low abundance in
most parts of their range but of higher abundance at
some locations (see Fig. 1A). Alternatively, the rank–
abundance pattern might be associated with most spe-
cies being of low abundance throughout their geo-
graphical ranges and thus in the tail of rank–abundance
curves wherever they occur (Fig. 1B).
We test which alternative best describes patterns of
abundance throughout geographical ranges for flowering
plant species of dry sclerophyll woodland and temperate
rain forest. We ask the simple but important question
whether most species in the tail of rank–abundance
curves can be found in high abundance at some other
location within their respective geographical ranges. In
addition we compare a total of 10 attributes between the
1807
two groups of species to assess any traits that consis-
tently distinguish somewhere-abundant from every-
where-sparse species through the ability to achieve high
abundances given the opportunity.
METHODS
Design
The basic approach was to locate a set of species
that contributed low cover, i.e., were in the tail of the
rank–abundance curve, within focal communities, and
to determine how many and which of these were sub-
stantially more abundant at other locations or in some
other environment throughout their geographical range,
versus how many and which occupied low cover wher-
ever they occurred. Only species whose entire geo-
graphical ranges fell within Australia were considered.
The methods necessarily made use of two types of ev-
idence. The study began at the level of the local com-
munity, first procuring abundance information for all
flowering plant species within focal sites by standard
quadrat methods of 1 ha in dry sclerophyll woodland.
Having identified certain species as being of low abun-
dance at focal sites, the second phase was to visit other
communities within the same general vegetation type
1808 BRAD R. MURRAY ET AL.
where the species were known or thought to occur. At
these secondary sites, similar sampling of abundance
occurred, which gave an initial listing of species that
were low in abundance at the focal sites and distinctly
greater in abundance at the secondary sites. For species
that still had not been found to be abundant anywhere
at this stage, no sampling procedure was capable of
assessing abundance throughout their geographical
ranges, which for many species extended far beyond
the particular vegetation type in which the focal site
was situated. Accordingly, the third phase of investi-
gation drew on a different type of evidence, the first-
hand knowledge of field-survey ecologists and herbar-
ium botanists having many years of experience in the
relevant regions, including authors D. A. Keith, P. J.
Myerscough, J. Howell, A. G. Floyd, and K. Mills.
These contributors and others (see Acknowledgments)
were surveyed to find out whether they had seen any
of the low-abundance species from within the focal
hectares in high abundance, and if so, where this was
observed. This evidence, drawn from personal knowl-
edge, field notes, and unpublished surveys, represented
cumulatively .200 person-yr of field botanical expe-
rience in the region. We consider it the most reliable
available with regard to abundance throughout wide
areas and whole geographical ranges. To assess the
generality of patterns found in dry sclerophyll wood-
land, the relationship between local abundance and
abundance throughout geographical ranges was also
investigated for flowering plant species of temperate
rain forest, using data on species abundance within
communities of 15 ha of undisturbed vegetation ob-
tained from Turner and Vernon (1994).
Dry sclerophyll woodland
Four focal sites, each a replicate of dry sclerophyll
woodland, were situated within the Sydney region,
New South Wales, Australia. Focal sites A, B, and D
were in Ku-ring-gai Chase National Park, while focal
site C was in Garigal National Park. All sites were on
infertile Hawkesbury sandstone. A focal site had to be
1 ha of vegetation that had not been burned or cleared
in any manner for a minimum of 10 yr. The site had
to appear homogeneous in the sense that there was no
particular tendency for one side to be different from
the other side. Focal sites were square (100 3 100 m).
Within them were located nine 2 3 5 m quadrats in a
3 3 3 grid. Each quadrat was ;30–40 m from the next.
The percent canopy cover of all flowering plant species
was estimated in each of the nine quadrats within the
hectare. Mean cover was calculated for each species
across the nine quadrats and used as an estimate of
canopy cover over the hectare. The rest of the hectare
was searched for species not found in the nine quadrats,
and canopy cover over the whole hectare was estimated
for these species.
Because maximum potential canopy area covered by
an individual varies between plant species, species
Ecology, Vol. 80, No. 6
were assigned to three size classes. Species were placed
in the large size class (size class 1) if their maximum
potential cover exceeded 2 3 2 m (mostly trees such
as Eucalyptus spp.), in size class 2 if maximum po-
tential canopy was between 0.2 3 0.2 m and 2 3 2 m
(mostly shrubs such as Grevillea spp.), and the small
size class (size class 3) if maximum potential canopy
never exceeded 0.2 3 0.2 m (mostly herbs, e.g., Lom-
andra spp.; and small woody dicotyledons, e.g., mem-
bers of the Euphorbiaceae). Given that the present
study compared abundances of species at a particular
site, between sites for each species, and between spe-
cies for geographical ranges, the quantification by size
class emphasizes the relative nature of comparisons.
For example, some species have very large individuals,
and thus the smallest percentage canopy cover an adult
can have is quite large.
Rank–abundance curves were created for each focal
site (Figs. 2A–D). A species was considered to be of
low abundance (i.e., in the tail) if it fell below a thresh-
old value of percent cover. Cutoff points were placed
subjectively where the tail section of curves became
perceptible. The threshold value was different for each
size class (size class 1 5 1% canopy cover, size class
2 5 0.4% canopy cover, and size class 3 5 0.01%
canopy cover). The average number of all species per
focal site was 66 (A, 76; B, 51; C, 69; D, 69), with a
total of 43, 20, 26, and 19 tail species at each focal
site, respectively. In all, 78 species were identified as
being of low abundance across the four focal sites, with
some species within the tail of the rank–abundance
curve at more than one focal site. Of the 78 tail species,
six were identifiable to genus only; these species were
discarded, as species-level identification was required.
Two of the 78 species were exotics (Chrysanthemoides
monilifera ssp. rotundata and Lantana camara), and
were given no further consideration, as abundances
within the native geographical ranges of species were
the topic. Two further species, Cassytha pubescens and
Themeda australis, were discarded because their geo-
graphical ranges extended outside Australia, to New
Zealand and New Guinea, respectively. This left 68
native species of low abundance to be studied.
Habitat descriptions and extent of geographical ranges
were obtained for all tail species. Sources included Aus-
tralian Government Publishing Service (1981), Harden
(1990–1993), Hnatiuk (1990), Robinson (1991), Benson
and McDougall (1993, 1994, 1995), Carolin and Tindale
(1994), Fairley and Moore (1995), and species lists ob-
tained from the Royal Botanic Gardens, Sydney (cour-
tesy of R. Coveny). From these sources it was possible
to construct a list of eight prospective locations within
the Sydney region where each of the focal species might
occur in high abundance. Abundances of focal species
at the eight prospective locations were determined by
estimating percent canopy cover. All cover values were
comparable directly to focal site abundances. A species
was considered to be significantly more abundant at a
September 1999 TAIL SPECIES OF RANK–ABUNDANCE CURVES 1809

FIG. 2. Rank–abundance curves for focal sites A, B, C, and D in dry scherophyll woodland. Abundance as percentage
canopy cover is on the y-axis, while species are ranked consecutively on the x-axis. Size classes (see Methods): 1 5 solid
diamonds; 2 5 open squares; 3 5 open circles. The cutoff points for tail species are marked for each size class on each
curve.

location compared to its abundance at a focal site if it
met both of two criteria. First, its abundance had to fall
above the threshold value established at the focal site
for the relevant plant size class. Second, abundance had
to be at least 10 times greater than that recorded at the
focal site. Tail species that were found to be more abun-
dant somewhere else within their geographical range
were termed somewhere-abundant species. If a species
was found to be much more abundant before all eight
locations were visited, it was not necessary to visit re-
maining locations, since this indicated that the species
could be classified as being more abundant elsewhere
(i.e., somewhere-abundant). Species that were abundant
during early post-fire years (various authorities; B. R.
1810 BRAD R. MURRAY ET AL.
Murray, personal observation) but not in established
vegetation 10 years or more after fire were categorized
as sometimes-abundant.
When the eight locations had been visited for all tail
species where necessary, and any species remained that
were not found to be more abundant, ecologists and
botanists experienced in the Sydney flora (including D.
A. Keith, P. J. Myerscough, and J. Howell) were con-
sulted further about remaining species. In particular,
people consulted were asked whether they had ever
found the species to be abundant anywhere within its
geographical range according to the criteria described
previously. Given the many combined years of field
experience of those consulted, this information was
more reliable than what could have been obtained by
any fresh sampling. Species that were not found to be
abundant in established vegetation (somewhere-abun-
dant species), nor identified as being abundant only
during early post-fire years (sometimes-abundant spe-
cies), were considered to be of low abundance through-
out their geographical range, and were termed every-
where-sparse.
Temperate rain forest vegetation
To provide a further test of the predictions, a second
vegetation type was examined. Turner and Vernon
(1994) provided information on abundances for a large
area of floristically and structurally well developed,
undisturbed temperate rain forest vegetation in New
South Wales. All flowering plant species at 46 sites
(totaling ;850 ha, median area 5 15 ha) were de-
scribed, and their abundances rated as one of four cat-
egories (very common, common, occasional, and rare)
at each site. These 46 sites were considered equivalent
to the four focal sites of dry sclerophyll woodland, in
that they represented sites at which local abundances
were estimated. We considered a classification of either
occasional or rare at any of the 46 sites as denoting
low local abundance (i.e., tail species, in our termi-
nology) at those particular sites. From the 46 rain forest
sites, 116 native species were identified as being tail
species at one site at least. As with the tail species of
dry sclerophyll woodland, the next step was to ask
whether the tail species of local sites in temperate rain
forest could be found in much greater abundance in
similar or other vegetation types within their respective
geographical ranges. This was assessed from infor-
mation also provided in Turner and Vernon (1994) (e.g.,
a species of low abundance at one of the 46 sites may
have been abundant at one or more of the remaining
sites listed) and from referencing other sources of abun-
dance information of temperate rain forest species (e.g.,
Mills and Jakeman 1995). Again, the third phase in
distinguishing somewhere-abundant from everywhere-
sparse species was consultation with botanists and ecol-
ogists (including A. G. Floyd and K. Mills) having
extensive field experience in this type of vegetation.
Ecology, Vol. 80, No. 6
Attribute comparisons
Attributes were compared between somewhere-
abundant and everywhere-sparse species. Data were
not available for all the attributes that might have been
desired. Those compared were dispersal morphology
(dry sclerophyll woodland [W], temperate rain forest
[R]), flowering length ( W, R), flowering season (W, R),
fruiting period (R), fruiting season (R), geographical
range (W, R), maximum potential height (W, R), plant
size class (W, R), regeneration after fire (W), and seed
size (W, R). Attribute information for comparisons for
one or both vegetation types was obtained from various
authorities and a plant attribute database established
and maintained in the School of Biological Sciences at
Macquarie University. Seed masses were expressed as
dry mass in milligrams for dry sclerophyll woodland,
and included embryo, endosperm, and seed coat struc-
tures, while any dispersal structures (e.g., elaiosomes)
were removed before weighing. For plants of temperate
rain forest, seed sizes were presented as largest linear
dimension (millimeters), rather than as mass, since
more data were available in this form from literature
sources. Dispersal morphologies were assessed using
structures present on the seeds indicative of a particular
mode of dispersal (e.g., Westoby et al. 1990). Flow-
ering information was a broad measure of the months
of the year spent flowering. These months were then
translated into seasons of the year spent flowering (e.g.,
a species that flowers from August to October flowers
in austral winter–spring). Species methods of regen-
eration after fire were recorded as fire-killed and relying
on a stored seed bank (obligate-seed-regenerator) or
capable of vegetative regeneration (resprouter). For
geographical range, the Census of Australian Plants
(Hnatiuk 1990) divides the Australian continent into
97 regions, based on biogeographical zones recognized
by the relevant herbaria in each state. Presence or ab-
sence of all species in each region is listed, so that the
number of regions occupied provides a broad, conti-
nental-scale measure of total geographical range.
Data analyses: cross-species and phylogenetic
approaches
Data for seed mass (but not seed length), geograph-
ical range, and plant height were all log-transformed
for statistical analyses (Sokal and Rohlf 1981). To iden-
tify which attributes discriminated most strongly be-
tween somewhere-abundant and everywhere-sparse
species, a stepwise discriminant function analysis
(DFA) was employed (Dixon 1992). While DFA was
used to look at cross-species comparisons and consider
questions about present-day ecological relationships
among species, phylogenetically independent contrasts
(PICS, Burt 1989) were also used to compare some-
where-abundant and everywhere-sparse species. Phy-
logenetic analyses test whether differences in the at-
tributes compared arise consistently across replicated
September 1999 TAIL SPECIES OF RANK–ABUNDANCE CURVES
phylogenetic divergences between somewhere-abun-
dant and everywhere-sparse species or clades. Pairs of
species for PICs were selected to be as closely related
as possible. In dry sclerophyll woodland, a total of four
contrasts was possible, two at species-within-genus
level, one at genera-within-family, and one at subclass-
es-within-class. For temperate rain forest, six contrasts
were established, two at species-within-genus level,
one at genera-within-family, two at families-within-or-
der, and one at orders-within-subclass.
RESULTS
Dry sclerophyll woodland
Of the 68 tail species studied, 13 species were found
to be significantly more abundant only at sites ,10 yr
since fire (e.g., Acacia suaveolens, Cassinia denticulata,
Gompholobium glabratum). These sometimes-abundant
species are probably abundant within focal site com-
munities 10 yr after fire, but in the form of a seed bank
rather than as above-ground cover. Of the remaining 55
species, 50 (91%) were found to be considerably more
abundant somewhere within their geographical range
(Table 1). Of these 50 somewhere-abundant species, the
higher-abundance sites for most (45) were in the same
vegetation type (dry sclerophyll woodland), while the
other five species had their higher-abundance sites in
habitat that was different from the focal sites. Because
the original search for these species as abundant at some
other location was not aimed at investigating all dry
sclerophyll woodland sites first, it is possible that further
searches may find these five species in high abundance
at some woodland sites.
Only five (9%) of the 55 tail species were found to
be abundant nowhere (Bossiaea ensata, Bossiaea scol-
opendria, Dampiera purpurea, Hovea linearis, Logan-
ia albiflora). All of these everywhere-sparse species
fell into size class 2, the intermediate-sized species.
Focal sites A, B, C, and D had three, zero, two, and
three everywhere-sparse species, respectively, within
the hectare, indicating that 96%, 100%, 97%, and 96%
of species within the tail of rank–abundance curves for
these four local communities (A, B, C, and D, respec-
tively) could be abundant elsewhere within their range.
Temperate rain forest vegetation
The proportion of tail species in temperate rain forest
that were found to be in high abundance at other lo-
cations was similar to the equivalent proportion in dry
sclerophyll woodland. In total, 110 temperate rain for-
est species (95%) were found to be abundant at one or
more locations, and were considered to be somewhere-
abundant (e.g., Acmena smithii, Geijera parviflora, Pit-
tosporum revolutum, and Syzygium australe). There
was no clear indication that any of the species found
to be in high abundance somewhere were actually
sometimes-abundant species, i.e., abundant only after
fire or other disturbances. Only six tail species (5%)
1811
were found nowhere in high abundance and were con-
sidered everywhere-sparse (Cayratia eurynema, Dy-
soxylum rufum, Emmenosperma alphitonioides, Pas-
siflora herbertiana ssp. herbertiana, Sambucus austra-
lasica, and Symplocos thwaitesii). Of the 110 some-
where-abundant species, 105 could be found in greater
abundance in similar vegetation to the original loca-
tions where they were identified as tail species (i.e.,
temperate rain forest). The other five species could only
be found in greater abundance in a different habitat.
Attribute comparisons
Flowering season duration was the best single dis-
criminator for dry sclerophyll woodland (DFA: F1,51 5
5.44, P , 0.05). Somewhere-abundant species on av-
erage flowered across a longer period of the year than
everywhere-sparse species. After this was entered into
the discriminant function, no further attributes were
found to be significant discriminators between the two
groups. The majority of both somewhere-abundant and
everywhere-sparse species in dry sclerophyll woodland
were dispersed by ants, were capable of vegetative re-
generation, and flowered in winter–spring. For rain for-
est, geographical range was found to be a significant
discriminator between the two groups (F1, 109 5 5.01, P
, 0.05), with somewhere-abundant species having more
widespread distributions. After this was entered into the
function, no further attributes were found to be signif-
icant discriminators between the two groups. Both
groups of temperate rain forest species were dispersed
primarily by vertebrates, flowered in spring and summer,
and had their fruit ripe mainly during summer–autumn.
Phylogenetic contrasts were generally consistent
with the cross-species patterns described by DFA. For
dry sclerophyll woodland, there was a not-quite-sig-
nificant trend for somewhere-abundant species to spend
a longer period of the year flowering compared with
everywhere-sparse species (t 5 2.18, P 5 0.058, df 5
3). This concurred with cross-species DFA analyses.
For temperate rain forest species, the only significant
result was for geographical range to be smaller in eve-
rywhere-sparse species (t 5 2.92, P 5 0.05, df 5 5),
again matching cross-species (DFA) comparisons.
For both dry sclerophyll woodland and temperate rain
forest, tables listing species used for phylogentically in-
dependent contrasts and their attributes are available in
Ecological Archives, along with a table listing all 116
somewhere-abundant and everywhere-sparse species of
temperate rain forest (see Appendix).
DISCUSSION
Patterns of abundance throughout
geographical ranges
How abundance is distributed among species within
local communities has been a much-investigated ques-
tion (e.g., Preston 1948, MacArthur 1957, Cohen 1968,
May 1975, De Vita 1979, Hughes 1986, Wilson et al.
1812 BRAD R. MURRAY ET AL. Ecology, Vol. 80, No. 6

TABLE 1. Somewhere-abundant species of dry sclerophyll woodland listed by size class. Abundances (percent canopy cover)
recorded at focal sites and abundance at the site where each species was found to be most abundant are shown.

Highest abundance
Abundance (% canopy (% canopy cover) at Vegetation type
Species cover) at focal sites different site at different site
Size class 1
Allocasuarina distyla 0.22 7.44 Woodland
Banksia ericifolia 0.001, 0.29 9.78 Woodland
Banksia integrifolia spp. integrifolia 0.11 9.12 Woodland
Ceratopetalum gummiferum 0.36 7.44 Woodland
Elaeocarpus reticulatus 0.18 k1.0 Woodland
Leptospermum laevigatum
Leptospermum polygalifolium spp. polygalifol- 0.25 k1.0 Heath
ium 0.32 5.0 Woodland
Persoonia pinifolia 0.03, 0.25 4.0 Woodland
Size class 2
Acacia myrtifolia 0.33, 0.06 6.67 Woodland
Baeckea diosmifolia 0.001 0.5 Woodland
Boronia ledifolia 0.01 2.5 Woodland
Bossiaea heterophylla 0.001, 0.17 2.0 Woodland
Callistemon linearis 0.39 k0.4 Heath
Caustis flexuosa 0.001, 0.33 k0.4 Woodland
Caustis pentandra 0.09 2.11 Woodland
Conospermum longifolium spp. longifolium 0.002 1.61 Woodland
Correa reflexa var. reflexa 0.08 k0.4 Woodland
Crowea saligna 0.13, 0.11 4.57 Woodland
Dianella caerulea var. caerulea 0.001 0.04 Woodland
Dianella caerulea var. producta 0.001 2.0 Woodland
Dianella revoluta var. revoluta 0.002 0.33 Woodland
Dillwynia retorta 0.17 7.27 Woodland
Epacris pulchella 0.25, 0.2 2.78 Woodland
Gompholobium grandiflorum 0.31 4.0 Woodland
Gonocarpus teucriodes 0.03, 0.01 0.89 Woodland
Grevillea buxifolia ssp. buxifolia 0.28, 0.11 3.78 Woodland
Grevillea sericea 0.26, 0.02 4.11 Woodland
Hakea dactyloides 0.33 9.78 Woodland
Hakea sericea 0.04 7.24 Woodland
Hibbertia aspera 0.15 k0.4 Woodland
Hibbertia bracteata 0.002 1.14 Woodland
Isopogon anemonifolius 0.17 5.67 Woodland
Lasiopetalum ferrugineum var. ferrugineum 0.34, 0.11 k0.4 Woodland
Leucopogon microphyllus var. microphyllus 0.002 0.44 Woodland
Lomandra longifolia 0.01 10.33 Woodland
Lomatia silaifolia 0.3, 0.01, 0.11 k0.4 Woodland
Micrantheum ericoides 0.01 1.94 Woodland
Monotoca scoparia 0.23, 0.37 k0.4 Woodland
Petrophile pulchella 0.001, 0.06, 0.02 4.5 Heath
Phebalium squamulosum ssp. squamulosum 0.2 k0.4 Woodland
Platysace linearifolia 0.07 2.94 Woodland
Pultenaea daphnoides 0.33 k0.4 Closed-forest
Pultenaea elliptica 0.001 1.11 Woodland
Tetratheca ericifolia 0.01 4.0 Woodland
Size class 3
Cassytha glabella forma glabella 0.001 0.11 Woodland
Hypolaena fastigiata 0.001 k0.01 Heath
Lepyrodia scariosa 0.001 2.94 Woodland
Patersonia glabrata 0.003 0.67 Woodland
Patersonia sericea 0.001 0.93 Woodland
Stipa pubescens 0.001 k0.01 Woodland

1996). Nevertheless, the relationship between abun-
dance within a local community and abundance else-
where throughout the complete geographical range of
species has been investigated rarely (see Brown et al.
1995). Most models that predict species abundance dis-
tributions at local scales invoke within-community
mechanisms alone, foregoing processes operating at
wider scales. Yet since around 96% to 100% of species
within the tail of a given rank–abundance curve can
be abundant at other locations or times (as shown for
dry sclerophyll woodland in the present study), expla-
nations as to why species are in the tail that are based
on within-community interactions without considera-
tion of wider-scale patterns will be incomplete for most
September 1999 TAIL SPECIES OF RANK–ABUNDANCE CURVES
species, because any explanation ought to account for
different outcomes in different places.
Of the 55 tail species identified in dry sclerophyll
woodland (and not classified as sometimes-abundant),
and the 116 tail species found in rain forest, 91% and
95%, respectively, were found to be substantially more
abundant at other sites. For the purpose of this study,
substantially more abundant meant that they were at
least 10-fold more abundant than at the focal site, and
also that their abundance exceeded the threshold de-
fined for their size-class. Under some circumstances,
1% cover could constitute ‘‘substantially more abun-
dant.’’ Hence, the results should not be interpreted as
meaning that all these species attained cover of 30–
50% somewhere.
Mainly, somewhere-abundant species were more
abundant at other locations within the same general
vegetation type. Of the two alternatives outlined in the
introduction, these results support the first (Fig. 1A) as
the best description of patterns of abundance through-
out geographical ranges. If this finding is combined
with the results of Brown et al. (1995), then the most
commonly observed distribution of species abundance
at a local scale (i.e., few abundant species and many
species of low abundance) arises as follows. First, it
is clear that the majority of species in the tail of rank–
abundance curves have locations where they can be
found in much greater abundance. Second, most if not
all species are of low abundance throughout much of
their geographical range (Brown et al. 1995). Thus, the
distinctive distribution of species abundance within lo-
cal communities simply reflects that sampling at any
given point within the landscape will capture most spe-
cies at low abundance and a few species at high abun-
dance. The question why certain species are found in
the tail of rank–abundance curves, so often asked from
a within-community perspective, can now be ap-
proached from a different perspective. Rather than fo-
cus on factors constraining abundance of particular spe-
cies within a site, it is more pertinent to ask how species
are able to switch positions in the rank–abundance
curve across wider scales. That is, what allows some-
where-abundant species to become abundant at some
sites within their geographical range? Any explanation
should explain also why the few everywhere-sparse
species are not able to become abundant anywhere
within their respective geographical ranges.
One can distinguish two ways a somewhere-abun-
dant species might achieve high abundance at some
sites but low at many sites: First, particular sites may
favor high abundance for that species on a continuing
basis (e.g., remnant populations, Eriksson 1996). Sec-
ond, somewhere-abundant species may achieve high
abundance opportunistically and temporarily at any of
a range of sites. Under the second scenario, somewhere-
abundant species would be characterized by attributes
that permitted them to become abundant given a suit-
able opportunity, and these attributes would be absent
1813
in everywhere-sparse species. These two scenarios are
of course extremes of a spectrum, rather than strict
alternatives. The real world might be somewhere be-
tween these two extremes, with some sites where a
species is nearly always abundant, others where it may
become abundant opportunistically quite often, and
others where the opportunity for high abundance occurs
relatively infrequently. In this scheme, sometimes-
abundant species represent a sampling or life-history
complication: Species conspicuous during early years
after fire in dry sclerophyll woodland are actually abun-
dant in later years, but as a seed bank rather than as
canopy cover.
There is some evidence that particular sites within
the geographical ranges of somewhere-abundant spe-
cies favor high abundance on a continuing basis. Using
long-term data (;30 yr) for North American birds,
Brown et al. (1995) showed that a few sites supported
consistently higher abundances of particular species
than much of the rest of their geographical ranges.
Many other studies, varying from few to many sites
and short to long time periods, have similarly indicated
conservation of high abundance by particular species
at certain sites (e.g., MacArthur 1972, Moyle and Von-
dracek 1985, Mitchley and Grubb 1986, Lawton and
Gaston 1989, Bengtsson et al. 1997). On the other hand,
a number of studies have found evidence that high
abundances are not maintained by species at particular
sites, or have found that differences in site environ-
mental conditions, latitudinal position, and climatic
conditions determine whether species’ abundances re-
main constant (Coull and Fleeger 1977, Järvinen 1979,
Taylor and Taylor 1979, Grubb et al. 1982, Grubb 1986,
Bethke 1993, Böhning-Gaese et al. 1994, Wilson et al.
1996). Evidence for high abundance being possible at
a number of locations throughout geographical ranges
can thus be found. In these cases, it would appear that
there was a shifting cloud of abundance (sensu Hubbell
and Foster 1986), with opportunities for high abun-
dance being taken as they arose across the landscape.
The concept of a shifting cloud of abundance relies
heavily on stochastic, or lottery models (e.g., Sale
1977, Chesson and Warner 1981, Fagerström 1988, Pa-
cala 1996). Rather than particular sites favoring high
abundance by certain species (e.g., Brown et al. 1995),
‘‘the relative abundance of . . . species . . . at any site
is largely a result of the chance recruitment of young
to that site and will change from time to time’’ (Sale
1977:354).
If continuing high abundance sites do exist for some-
where-abundant species, existing evidence suggests
they will most likely be found toward the centers of
species’ geographic ranges (Grinnell 1922, Bock et al.
1977, Hengeveld and Haeck 1981, 1982, Brown 1984,
Bart and Klosiewski 1989, Hengeveld 1994, Carey et
al. 1995). Therefore, testing the hypothesis that sites
of long-term high abundance exist for these vegetation
types could be carried out by first establishing long-
1814 BRAD R. MURRAY ET AL.
term monitoring sites towards the centers of geographic
ranges. It is also worth considering that high abun-
dances may be expected in conditions that are more
similar to those in evolutionary centers.
The proportion of everywhere-sparse species was
small but not dismissable. Five species of dry sclero-
phyll woodland (9%) and six species of temperate rain
forest (5%) were categorized as everywhere-sparse. If
these percentages can be extrapolated roughly across
all species in each different vegetation type, then the
absolute number of everywhere-sparse species may be
quite large. For example, ;2000 species in total are
found in dry sclerophyll woodland on Sydney sand-
stone (Carolin and Tindale 1994), so it might be esti-
mated that around 180 of these (9%) are everywhere-
sparse. This is a substantial number of species that are
apparently unable to make use of opportunities for high
abundance. Note that everywhere-sparse species are
not expected to be distinguished by attributes that en-
able them to persist at low abundances. Somewhere-
abundant species are similarly at low abundances
throughout most of their geographical ranges. There-
fore, the distinction in this study is not centered on the
persistence of everywhere-sparse species at low abun-
dances or on possible adaptations for existing at low
abundances, but rather on the potential for becoming
abundant in somewhere-abundant species versus the
absence of that potential in everywhere-sparse species.
None of the everywhere-sparse species identified in the
present study are considered rare or threatened (Briggs
and Leigh 1996). We have avoided the term rare here
for this reason, and also because ‘‘rare’’ has been used
in many different senses (see Rabinowitz et al. 1986,
Gaston 1994). It is possible that with even more com-
plete data about abundance throughout the geograph-
ical range, some everywhere-sparse species might be
reclassified to somewhere-abundant. The effect of this
would be to increase further our estimate of the pro-
portion of tail species that are somewhere-abundant.
Attribute comparisons
Of the attributes available to be compared between
somewhere-abundant and everywhere-sparse species,
only one was significantly different in each vegetation
type as cross-species tendencies and as replicated phy-
logenetic divergences. In dry sclerophyll woodland,
somewhere-abundant species flowered on average
across a longer time during the year than everywhere-
sparse species. In temperate rain forest, somewhere-
abundant species occurred throughout wider geograph-
ical ranges than everywhere-sparse species. Since nei-
ther pattern was repeated in the other vegetation type,
there was no consistent support for any differences be-
tween somewhere-abundant and everywhere-sparse
species in any of the attributes compared here. In sep-
arate field-studies on two phylogenetically-indepen-
dent comparisons (Murray and Westoby, unpublished
data), provisional evidence has been found for a lower
Ecology, Vol. 80, No. 6
potential for increase in everywhere-sparse species
compared to somewhere-abundant species.
Conclusion
The principal conclusion reported here is that most
(.90%) species in the tails of local rank–abundance
curves are substantially more abundant somewhere
else. If this pattern extends to other vegetation types
and to other taxa besides plants, it indicates that models
of population interactions within communities will not
by themselves provide a complete account of rank–
abundance patterns. The second principal conclusion
reported here is that there exists a small but still in-
teresting group of species that are everywhere sparse.
If this conclusion extends to other vegetation types and
taxa, it must be asked what attributes might account
for some species’ reaching high abundance at some
times or places, and for others’ being confined to low
abundance everywhere.
ACKNOWLEDGMENTS
We are very grateful to other botanists who shared their
wealth of field information regarding the abundances of nu-
merous plant species: D. Benson, F. Burrows, R. Coveny, A.
Downing, M. Dunlop, L. Rodgerson, L. McDougall, M. Reed,
J. Turner, S. Vernon, and G. Williams. Thanks also to Lesley
Hughes, Carlos Fonseca, Will Edwards, Bill Lee, Jeremy
Smith, and Rob Whelan for critiquing earlier drafts of this
work and to the Ecology Discussion Group for helpful com-
ments. Charles Canham and two anonymous referees provid-
ed encouraging and helpful advice, for which we thank them
wholeheartedly. This work was carried out while B. R. Mur-
ray was in receipt of an Australian Postgraduate Award, and
he would like to thank personally Darryl Nelson, Carlos Fon-
seca, Jake Overton, Chris Chambers, and Susan Murray for
their support and friendship. This is contribution 265 from
the Centre for Biodiversity and Bioresources, Macquarie Uni-
versity.
LITERATURE CITED
Australian Government Publishing Service. 1981. Flora of
Australia. Bureau of Flora and Fauna. Griffin Press Ltd.,
Canberra, Australia.
Bart, J., and S. P. Klosiewski. 1989. Use of presence–absence
to measure changes in avian density. Journal of Wildlife
Management 53:847–852.
Bengtsson, J., S. R. Baillie, and J. Lawton. 1997. Community
variability increases with time. Oikos 78:249–256.
Benson, D., and L. McDougall. 1993. Ecology of Sydney
plant species Part 1: ferns, fern-allies, cycads, conifers di-
cotyledon families Acanthaceae to Asclepiadaceae. Cun-
ninghamia 3:257–422.
Benson, D., and L. McDougall. 1994. Ecology of Sydney
plant species Part 2: dicotyledon families Asteraceae to
Buddlejaceae. Cunninghamia 3:789–1004.
Benson, D., and L. McDougall. 1995. Ecology of Sydney
plant species Part 3: dicotyledon families Cabombaceae to
Eupomatiaceae. Cunninghamia 4:217–432.
Bethke, R. W. 1993. Geographical patterns of persistence in
duck guilds. Oecologia 93:102–108.
Bock, C. E., J. H. Bock, and L. W. Lepthien. 1977. Abun-
dance patterns of some bird species wintering on the Great
Plains of the USA. Journal of Biogeography 4:101–110.
Böhning-Gaese, K., M. L. Taper, and J. H. Brown. 1994.
Avian community dynamics are discordant in space and
time. Oikos 70:121–126.
September 1999 TAIL SPECIES OF RANK–ABUNDANCE CURVES
Briggs, J. D., and J. H. Leigh. 1996. Rare or threatened
Australian plants. Revised edition 1995. CSIRO Publish-
ing, Victoria, Australia.
Brown, J. H. 1984. On the relationship between abundance
and distribution of species. American Naturalist 124:255–
279.
Brown, J. H., D. W. Mehlman, and G. C. Stevens. 1995.
Spatial variation in abundance. Ecology 76:2028–2043.
Burt, A. 1989. Comparative methods using phylogenetically
independent contrasts. Oxford Surveys in Evolutionary Bi-
ology 6:33–53.
Carey, P. D., A. R. Watkinson, and F. F. O. Gerard. 1995.
The determinants of the distribution and abundance of the
winter annual grass Vulpia ciliata ssp. ambigua. Journal of
Ecology 83:177–187.
Carolin, R. C., and M. D. Tindale. 1994. Flora of the Sydney
region. 4th edition. Reed, Chatswood, New South Wales,
Australia.
Chesson, P. L., and N. Huntley. 1989. Short-term instabilities
and long-term community dynamics. Trends in Ecology and
Evolution 4:293–298.
Chesson, P. L., and R. R. Warner. 1981. Environmental vari-
ability promotes coexistence in lottery competitive sys-
tems. American Naturalist 117:923–943.
Cohen, J. E. 1968. Alternative derivation of a species-abun-
dance relation. American Naturalist 102:165–172.
Coull, B. C., and J. W. Fleeger. 1977. Long-term temporal
variation and community dynamics of meiobenthic cope-
pods. Ecology 58:1136–1143.
Darwin, C. 1859. The origin of species by means of natural
selection. John Murray, London, UK.
De Vita, J. 1979. Niche separation and the broken-stick mod-
el. American Naturalist 114:171–178.
Diamond, J., and T. J. Case. 1986. Community ecology.
Harper and Row, New York, New York, USA.
Dixon, W. J. 1992. BMDP statistical software manual,
BMDP Release 7. University of California Press, Berkeley,
California, USA.
Eriksson, O. 1996. Regional dynamics of plants: a review
of evidence for remnant, source-sink and metapopulations.
Oikos 77:248–258.
Fagerström, T. 1988. Lotteries in communities of sessile or-
ganisms. Trends in Ecology and Evolution 3:303–306.
Fairley, A., and P. Moore. 1995. Native plants of the sydney
district, an identification guide. Kangaroo Press Ltd., Kent-
hurst, New South Wales, Australia.
Gaston, K. J. 1994. Rarity. Chapman and Hall, London, UK.
Gregory, R. D. 1994. Species abundance patterns of British
birds. Proceedings of the Royal Society of London 257:
299–301.
Grinnell, J. 1922. The role of the ‘‘accidental.’’ The Auk
39:373–380.
Grubb, P. J. 1986. Problems posed by sparse and patchily
distributed species in species-rich plant communities.
Pages 207–225 in J. M. Diamond and T. J. Case, editors.
Community ecology. Harper and Row, New York, New
York, USA.
Grubb, P. J., D. Kelly, and J. Mitchley. 1982. The control
of relative abundance in communities of herbaceous plants.
Pages 79–97 in E. I. Newman, editor. The plant community
as a working mechanism. Special Publications Series of the
British Ecological Society, 1. Blackwell Scientific Publi-
cations, Oxford, UK.
Hanski, I. 1982. Dynamics of regional distribution: the core
and satellite species hypothesis. Oikos 38:210–221.
Harden, G. J. 1990–1993. Flora of New South Wales. Royal
Botanic Gardens, Sydney. New South Wales University
Press, New South Wales, Australia.
Hengeveld, R. 1994. Biogeographical ecology. Journal of
Biogeography 21:341–351.
1815
Hengeveld, R., and J. Haeck. 1981. The distribution of abun-
dance. II. Models and implications. Proceedings of the
Koninklijke Nederlandse Akademie van Wetenschappen
C84:257–284.
Hengeveld, R., and J. Haeck. 1982. The distribution of abun-
dance. I. Measurements. Journal of Biogeography 9:303–
316.
Hnatiuk, R. J. 1990. Census of Australian vascular plants.
Australian Flora and Fauna Series No. 2. Australian Gov-
ernment Publishing Service, Canberra, Australia.
Hubbell, S. P., and R. B. Foster. 1986. Biology, chance, and
history and the structure of tropical rain forest tree com-
munities. Pages 314–329 in M. L. Cody and J. M. Diamond,
editors. Ecology and evolution of communities. Harvard
University Press, Cambridge, Massachusetts, USA.
Hughes, R. G. 1986. Theories and models of species abun-
dance. American Naturalist 128:879–899.
Järvinen, O. 1979. Geographical gradients of stability in
European land bird communities. Oecologia 38:51–69.
Lawton, J. H., and K. J. Gaston. 1989. Temporal patterns in
the herbivorous insects of bracken: a test of community
predictability. Journal of Animal Ecology 58:1021–1034.
MacArthur, R. H. 1957. On the relative abundance of bird
species. Proceedings of the National Academy of Sciences,
USA 43:293–295.
MacArthur, R. H. 1960. On the relative abundance of spe-
cies. American Naturalist 94:25–36.
MacArthur, R. H. 1972. Geographical ecology: patterns in
the distribution of species. Harper and Row, New York,
New York, USA.
Maurer, B. A. 1994. Geographical population analysis: tools
for the analysis of biodiversity. Blackwell Scientific Pub-
lications, Oxford, UK.
May, R. M. 1975. Patterns of species abundance and diver-
sity. Pages 81–119 in M. L. Cody and J. M. Diamond,
editors. Ecology and evolution of communities. Harvard
University Press, Cambridge, Massachusetts, USA.
Mills, K., and J. Jakeman. 1995. Rain forests of the Illawarra
District. Coachwood Publishing, Sydney, New South
Wales, Australia.
Mitchley, J., and P. J. Grubb. 1986. Control of relative abun-
dance of perennials in chalk grassland in southern England.
I. Constancy of rank order and results of pot- and field-
experiments on the role of interference. Journal of Ecology
74:1139–1166.
Moyle, P. B., and B. Vondracek. 1985. Persistence and struc-
ture of the fish assemblage in a small California stream.
Ecology 66:1–13.
Pacala, S. W. 1996. Models of plant coexistence. Pages 532–
555 in M. C. Crawley, editor. Plant ecology. 2nd edition.
Blackwell Scientific Publications, Oxford, UK.
Pielou, E. C. 1975. Ecological Diversity. John Wiley and
Sons, New York, New York, USA.
Preston, F. W. 1948. The commonness and rarity of species.
Ecology 29:254–283.
. 1962. The canonical distribution of commonness and
rarity: part II. Ecology 43:410–432.
Rabinowitz, D., S. Cairns, and T. Dillon. 1986. Seven forms
of rarity and their frequency in the flora of the British Isles.
Pages 182–204 in M. E. Soulé, editor. Conservation biol-
ogy, the science of scarcity and siversity. Sinauer Asso-
ciates, Inc., Sunderland, Massachusetts, USA.
Rapoport, E. H. 1982. Areography: geographical strategies
of species. Pergamon Press, Oxford, UK.
Robinson, L. 1991. Field guide to the native plants of Syd-
ney. Kangaroo Press, Kenthurst, New South Wales, Aus-
tralia.
Sale, P. F. 1977. Maintenance of high diversity in coral reef
fish communities. American Naturalist 111:337–359.
1816 BRAD R. MURRAY ET AL. Ecology, Vol. 80, No. 6

Schoener, T. W. 1987. The geographical distribution of rarity.
Oecologia 74:161–173.
Sokal, R. R., and F. J. Rohlf. 1981. Biometry: the principles
and practice of statistics in biological research. Freeman,
San Francisco, California, USA.
Stenseth, N. C. 1979. Where have all the species gone? On
the nature of extinction and the Red Queen hypothesis.
Oikos 33:196–227.
Sugihara, G. 1980. Minimal community structure: an expla-
nation of species-abundance patterns. American Naturalist
116:770–787.
Taylor, R. A. J., and L. R. Taylor. 1979. A behavioural model
for the evolution of spatial dynamics. Pages 1–27 in R. M.
Anderson, B. D. Turner, and L. R. Taylor, editors. Popu-
lation dynamics. Blackwell Scientific Publications, Oxford,
UK.
Tokeshi, M. 1990. Niche apportionment or random assort-
ment: species abundance patterns revisited. Journal of An-
imal Ecology 59:1129–1146.
Turner, J. C., and S. L. Vernon. 1994. Rain forest stands
between Barrington Tops and the Hunter River, New South
Wales. Cunninghamia 3:465–519.
Westoby, M., B. Rice, and J. Howell. 1990. Seed size and
plant growth form as factors in dispersal spectra. Ecology
71:1307–1315.
Whittaker, R. H. 1965. Dominance and diversity in land plant
communities. Science 147:250–260.
Wilson, J. B. 1991. Methods for fitting dominance/diversity
curves. Journal of Vegetation Science 2:35–46.
Wilson, J. B., T. C. E. Wells, I. C. Trueman, G. Jones, M. D.
Atkinson, M. J. Crawley, M. E. Dodds, and J. Silvertown.
1996. Are there assembly rules for plant species abun-
dance? An investigation in relation to soil resources and
successional trends. Journal of Ecology 84:527–538.

APPENDIX
Tables presenting the species used for phylogenetically independent contrasts and their attributes are available in ESA’s
electronic data archive. Separate tables are posted for dry scherophyll woodland (Ecological Archives E080-011-A1) and for
temperate rain forest (Ecological Archives E080-011-A2). A third table, listing all 116 somewhere-abundant and everywhere-
sparse species of temperate rain forest examined is also provided (Ecological Archives E080-011-A3), along with a bibli-
ography (Ecological Archives E080-011-A4).