The animal species­body size distribution of
Marion Island
Kevin J. Gaston*
, Steven L. Chown
, and Richard D. Mercer§
*Biodiversity and Macroecology Group, Department of Animal and Plant Sciences, University of Sheffield, Sheffield S10 2TN, United Kingdom; Department
of Zoology, University of Stellenbosch, Pivate Bag X1, Matieland 7602, South Africa; and §Department of Zoology and Entomology, University of Pretoria,
Pretoria 0002, South Africa
Edited by Robert May, University of Oxford, Oxford, United Kingdom, and approved October 11, 2001 (received for review June 29, 2001)
Body size is one of the most significant features of animals. Not
only is it correlated with many life history and ecological traits, but
it also may influence the abundance of species within, and their
membership of, assemblages. Understanding of the latter processes
is frequently based on a comparison of model outcomes
with the frequency of species of different body mass within natural
assemblages. Consequently, the form of these frequency distributions
has been much debated. Empirical data usually concern
taxonomically delineated groups, such as classes or orders,
whereas the processes ultimately apply to whole assemblages.
Here, we report the most complete animal species­body size
distribution to date for those free-living species breeding on
sub-Antarctic Marion Island and using the terrestrial environment.
Extending over 15 orders of magnitude of variation in body mass,
this distribution is bimodal, with separate peaks for invertebrates
and vertebrates. Under logarithmic transformation, the distribution
for vertebrates is not significantly skewed, whereas that for
invertebrates is right-skewed. Contrary to expectation based on a
fractal or pseudofractal environmental structure, the decline in the
richness of species at the smallest body sizes is a real effect and not
a consequence of unrecorded species or of species introductions to
the island. The scarcity of small species might well be a consequence
of their large geographic ranges.
Systematic variation in the numbers of species of different
body sizes is one of the most readily apparent features of the
structure of animal assemblages. Indeed, species­body size
distributions for such assemblages have been documented over
a period of nearly 100 years, with more than 300 now in the
literature. The rate at which they are produced continues to grow
(1, 2). Nevertheless, understanding of the form taken by these
distributions remains wanting. Perhaps foremost, this lack is
because almost all documented examples of species­body size
distributions are for selected higher taxa typically at the level of
classes or orders (e.g., carabid beetles, birds, mammals), and not
for all, or even the bulk, of the species that compose entire
animal assemblages. The precise form of the distributions for
such higher taxa varies, particularly with spatial scale (3­12).
However, in terrestrial systems they are typically strongly rightskewed,
with many more small-bodied than large-bodied species.
Moreover, the smallest body-size class is seldom the most
speciose, despite the argument that the fractal or pseudofractal
nature of the environment provides progressively more opportunities
for smaller organisms (12­16). The paucity of species in
the smallest size class has been attributed to a variety of causes,
including resource distributions, size-biases in speciation and
extinction, and optimization of energy allocation (8, 12, 17­26).
It has also been attributed to the influence of species in other
taxonomic groups, where such body sizes are closer to being
modal and which may thus be better exploiters of, or competitors
for, the available resources (5, 27, 28).
The principal, and influential, exception to the preoccupation
in the ecological literature with species­body size distributions
for particular higher taxa has been May's (15, 29) estimate of the
form of the species­body size distribution of all known terrestrial
animal species [see also ref. 30; Fenchel (31) conducted a similar
exercise for aquatic species]. Although explicitly somewhat
speculative, in that it involved significant approximation and
extrapolation, this distribution was also strongly right-skewed
and again revealed a paucity of species in the smallest size class.
However, in common with many other species­body size distributions,
a major concern in interpreting this one is the effect of
the absence of a considerable proportion of extant species (29).
Newly described species in higher taxa tend to be smaller-bodied,
on average, than those already known (32­37), and some of the
hyperdiverse and more poorly known taxa are small-bodied (e.g.,
nematodes, mites). Debate has thus persisted as to the overall
richness of small-bodied species relative to those of larger body
size (15, 29, 31, 38­40).
Given that the paucity of small-bodied species for individual
higher taxa need not necessarily translate into such a paucity of
species in the smallest body-size classes for entire terrestrial
animal assemblages, it remains an open question what form
species­body size distributions for the latter might actually take.
In this paper, we report such a distribution for the free-living
Animalia assemblage of Marion Island.
Data
Marion Island is the larger of the two sub-Antarctic islands that
make up the Prince Edward archipelago (46o
53 S, 37o
57 E),
which lies 2300 km south-east of Cape Town. It has a surface
area of 290 km (2), a maximum elevation of 1230 m, a cool
hyperoceanic climate (mean annual air temperature 5° C), and
a tundra-type biome (for details see refs. 41 and 42). The flora
and fauna of the island have been subject to prolonged and
intensive study (41­47). This study has culminated in a 4-year
(1996­1999) survey of the invertebrate assemblages associated
with each of the broad land types, with replicated standardized
field sampling conducted continuously for periods of a year or
more in each type (48­50). This study has resulted in a remarkably
full understanding of the composition of the fauna.
Although clearly very isolated, the animal assemblage of
Marion Island is far from being a closed one (42, 51, 52). Here,
we limit attention to those free-living (i.e., nonparasitic) species
that breed on the island and use the terrestrial environment, thus
excluding those that solely exploit freshwater and littoral habitats.
This assemblage constitutes a well-defined one using a
common spatial resource, and thus is potentially subject to any
constraints on the frequency of species of different body sizes
that might be posed by the fractal or pseudofractal nature of the
terrestrial environment (12­16). The assemblage includes species
of birds and mammals that breed on the island but may
forage largely or exclusively at sea; these species have particularly
important terrestrial ecosystem functions on the island (e.g., in
This paper was submitted directly (Track II) to the PNAS office.
To whom reprint requests should be addressed. E-mail: k.j.gaston@sheffield.ac.uk.
The publication costs of this article were defrayed in part by page charge payment. This
article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C.
§1734 solely to indicate this fact.
www.pnas.org cgi doi 10.1073 pnas.251332098 PNAS December 4, 2001 vol. 98 no. 25 14493­14496
ECOLOGY
nutrient cycling, ecosystem engineering; refs. 51, 53, and 54).
Significantly, the only major taxonomic component of the assemblage
for which the number of species on the island has not
been well characterized is the nematodes (see below). The
precise taxonomic identities of some of the species in some other
invertebrate groups (especially the mites) are yet to be resolved
(some may be new to science; refs. 55 and 56), but the specieslevel
composition is well understood, and it is very unlikely that
significant numbers of species remain to be discovered.
Defined as above, the entire animal assemblage comprises 153
recorded species, including representatives of many groups that
are commonly ignored in studies of community ecology and
biodiversity (e.g., Collembola, Acari). The body mass of adults
of each species that has been distinguished was determined in
one of the following ways: by direct measurement, where possible
averaging over multiple individuals [Collembola, Psocoptera,
Hemiptera (in part), Thysanoptera, Coleoptera, Diptera,
Lepidoptera, Hymenoptera, Mesostigmata (in part), Prostigmata
(in part), Cryptostigmata (in part), Astigmata (in part),
Araneae (in part), Mollusca], by using appropriate regressions to
estimate body mass from body-length measurements [Mesostigmata
(in part), Prostigmata (in part), Cryptostigmata (in part),
Araneae (in part)], from the literature [Ixodoidea (in part), Aves
(in part)], from the unpublished records of appropriate experts
[Aves (in part), Mammalia], and by extrapolation from the body
masses of related species of similar dimensions [Hemiptera (in
part), Siphonaptera, Cryptostigmata (in part), Astigmata (in
part), Ixodoidea (in part), Tardigrada, Annelida].
Results and Discussion
The animal species recorded from Marion Island vary in body
mass by over 15 orders of magnitude, from 0.00000045 to
1,350,000 g. The species­body size distribution based on logarithmically
transformed body masses is approximately bimodal,
with the second mode comprising vertebrates and the first mode
comprising invertebrates (Fig. 1). We are not aware of such a
marked size discontinuity having previously been documented
for animal assemblages, although it may be influenced by the
limited number of species in this particular case.
The logarithmically transformed body-size distribution for
vertebrates is not significantly skewed [g1 0.792, n 33; using
the test given by Sokal & Rohlf (57) t 1.936, not significant
(NS)]. The logarithmically transformed body-size distribution
for invertebrates is significantly right-skewed (g1 0.558, n
120; t 2.525, P 0.05), and such a pattern can be interpreted
in two different ways (12). It could be taken to mean that there
are relatively few very large invertebrate species in this assemblage,
in comparison to a log-normal distribution covering the
same range of body-mass values. Alternatively, the right skew
could be interpreted as a consequence of more very large species
in comparison to a log-normal distribution with the same mode
and maximum.
If, for the Marion Island species­body size distribution (as
depicted in Fig. 1), the two size classes to the left of the modal
class were to contain the same numbers of species as that class
(i.e., there were to be no paucity of species in the smallest size
classes), then at least 53 terrestrial animal species would have to
be missing from the faunal inventory for this assemblage. If these
classes were to contain more species than the modal class (in line
with the decrease in richness to the right of the mode), then 100
species would have to be missing. The only plausible contributor
to these numbers of species would be the nematodes. As
previously mentioned, there are no data for nematodes for
Marion Island. However, on the climatically similar Macquarie
Island, 26 nematode species have been recorded (58), whereas
on the more extreme, maritime Antarctic Signy Island, 27 species
have been found (59). Boag & Yeates (60) also have recently
compiled available information on the species richness of soil
nematodes in terrestrial ecosystems. They estimate conservative
average values for local or -diversity of 49 species in grassland
vegetation, 53 in cultivated soil, and 23 in polar vegetation. The
numbers of species to be found in samples from Marion Island
seem likely to be in the region of 20 to 30 species. Collections that
may be regarded as reflecting regional diversity give figures of 80
species in the entire sub-Antarctic Antarctic region (59, 61), 36
species from fen vegetation (Wicken Fen, England), 41 from
maritime turf (Orkney Island), 70 from tussock grasslands (New
Zealand), 75 from alpine vegetation (Austria), 75 from Ellesmere
Island (Canada), and 89 species from sites in Spitzbergen
(60). Given the generally depauperate state of the biota of
Southern Ocean islands, a figure of substantially less than 100
species thus seems probable for the whole of Marion Island.
The extent to which nematodes will offset the otherwise low
numbers of very small species in the Marion Island assemblage
obviously depends on their body masses. Information here is
limited. However, such data are available for the 27 species
found on Signy Island (59), spanning the size range 0.00000004
to 0.00002164 g. Of these species, 3 would fall within the
invertebrate modal size class in Fig. 1, 9 within the size class to
its left, a further 10 species within the size class to the left of the
previous one, and 5 would occupy the vacant size class to the
extreme left. Similarly, Yeates (62) gives data on the live body
masses of eight species of soil nematodes in New Zealand. These
body masses range from 0.0000000711 to 0.000009112 g and span
the three extreme left-hand size classes in Fig. 1, including the
one that is otherwise unoccupied. Yeates (63) gives data on the
body masses (averaged over all stages) of the 29 most common
species of litter and soil nematodes in a beech forest. These body
masses span 0.000000025 to 0.000005441 g (median 0.000000132
g), and nearly half (13 28) would fall in a more extreme size class
than the most left-hand occupied one in Fig. 1. Thus, it seems
highly unlikely that the nematodes of Marion Island will remove
the pattern of decline in species richness of the smallest bodied
species observed in the overall animal assemblage.
Given their high global species richness, and the relatively
poor state of their taxonomy in most regions, the possible role
of mites in determining the shape of animal assemblage speciesbody
size distributions has been highlighted (29, 40); of an
estimated global total of at least some hundreds of thousands,
less than 50,000 species of mites have been described (40, 64),
and samples from some habitats support the notion that mites
are a hyperdiverse group (40, 65, 66; but see also ref. 67). The
detailed knowledge of the mite fauna of Marion Island provides
an opportunity to examine their influence on animal speciesFig.
1. Frequency distribution of logarithmically transformed body masses of
free-living animal species using Marion Island to breed; the distribution
excludes those that solely exploit freshwater and littoral habitats.
14494 www.pnas.org cgi doi 10.1073 pnas.251332098 Gaston et al.
body size distributions. As defined here, mites comprise 50 of the
species (33%; 42% of the invertebrates) in the Marion Island
assemblage. These are significantly smaller-bodied than the
other invertebrates; the average body size of the mite species was
smaller than the average in each of 100,000 random draws of 50
from the 120 invertebrate species. Although additional cryptic
species of mites yet may be discovered on Marion Island, these
are likely to be very few and would most probably be found in
the genus Algophagus (D. J. Marshall, unpublished work). The
known species in this genus fall within the 4 to 5, or modal,
size class (log-transformed mass) for the Marion Island assemblage,
and so any such additions will not significantly alter these
results. Thus, mites make a substantial contribution to the overall
shape of the invertebrate component of the species­body size
distribution for the Marion Island assemblage.
In common with many faunas (particularly on islands), a
significant proportion of those on Marion Island have been
introduced by human activities (68). For 21 (14%) of the species
presently on the island, the probability of introduction is sufficient
to regard them as alien (some past introductions have since
been removed or eradicated; ref. 68 and references therein).
Although some of these species are having a marked, and
arguably escalating, impact on components of the native biota,
there is no evidence that, as yet, they have caused extinctions on
the island (69­73).
Only one of the introductions (the mouse Mus musculus) is
a vertebrate, and it is the smallest-bodied of those vertebrates
in the Marion Island assemblage. Ignoring this species, the
smallest bodied vertebrate is the indigenous gray-backed
stormpetrel Garrodia nereis (38 g), which is close to the lower
observed size limit for pelagic foraging seabirds (although the
Least Petrel Halocyptena microsoma is somewhat smaller at
20 g; ref. 74) and perhaps close to the functional limit that is
possible (75, 76).
Lawton and Brown (77) reported a positive relationship
across a wide variety of organisms between the probability of
establishment of an invader accidentally or intentionally introduced
to the British Isles and its body size, and the opposite
trend for data for various insect orders. The 20 species of
invertebrates that are alien to Marion Island are on average
larger-bodied than the other invertebrate species; the average
body size of the alien species was larger than the average in
97,276 of 100,000 random draws of 20 from the 120 invertebrate
species (one-tailed test, P 0.05). However, this difference
was not significant with a two-tailed test, suggesting
either that there is no significant body-size difference in the
introduced and indigenous species, or that there are differences
that are the consequence of specific species or groups of
species. Indeed, the large-bodied slug and calliphorid fly (see
ref. 46) are likely to have contributed substantially to the
significance of the one-tailed test. Nevertheless, following
Lawton & Brown's (77) example of separating the species into
higher taxa provides an indication that the second explanation
is also appropriate. In this case, the 10 alien insect species
were, on average, smaller than the other insect species (smaller
on 95,258 of 100,000 random draws; one-tailed test, P 0.05).
On the contrary, the single alien mite species has a larger body
mass than the average mass of all mite species on the island,
but not significantly so (larger on 35,973 of 100,000 random
draws; NS), and the six alien collembolan species are likewise
on average larger-bodied than the average of all Collembola,
but again not significantly so (larger on 71,900 of 100,000
random draws, NS). This result suggests that for alien invertebrate
species, there may be a threshold body size on either
side of which the relationship between body size and successful
establishment differs. One explanation for this possibility may
be that for very small invertebrates, such as mites and springtails,
physiological characteristics that allow individuals to
endure an oceanic crossing by ship in a dry hold (e.g., lipid
content and desiccation resistance), and which are often
positively related to body size (78), are likely to be most
significant. For the larger insects, on the other hand, body size
may already be large enough to ensure survival. In this case,
the higher abundances (12) and the decline in probability of
detection and subsequent eradication by humans associated
with small body size are likely to be most significant.
Irrespective of whether this hypothesis is correct, the removal
of these alien invertebrate species makes no marked difference
to the estimated numbers of species that would have to be
missing from the faunal inventory of the Marion Island assemblage
if the two size classes to the left of the modal size class were
to contain the same number of species as that modal class (47
rather than 53). However, if these classes were to contain more
species than the modal class in line with the decrease in richness
to the right of the mode, then, in the absence of the introduced
species, rather more species would have to be missing.
In short, it seems that there is a genuine paucity of species in
the smallest body-size classes in the animal assemblage that
exploits the land of Marion Island, supporting the notion that
such a pattern is not solely a feature of the species­body size
distributions of individual higher taxonomic groups. The shape
of the distribution thus cannot result from any fractal or pseudofractal
structure of the terrestrial environment. Why should it
take the form that it does? Most discussion of the determinants
of the shapes of species­body size distributions has centered on
individual classes or orders, and increasingly on the tradeoff
between, and size dependence of, reproductive rate (production)
and mortality (21, 23, 25). It is, however, questionable how
readily such mechanisms will extrapolate to whole assemblage
body-size distributions, comprising groups with fundamentally
different patterns of physiology and life history.
One possibility that has been suggested is that species in taxa
that are particularly small-bodied tend to be widely distributed,
abundant, and good at dispersal, and that these characteristics
effectively depresses their species richness (by limiting the
opportunities for speciation). Such an argument has been used
to explain why the global species richness of protists is not as
great as their small body sizes might have suggested (31, 39, 79,
80; see also discussion of ref. 11 with regard to mites). Some
support for this notion derives from the observation that of the
indigenous collembolan species from Marion Island, 3 are
endemic and 7 are widely distributed across the sub-Antarctic
(the alien species are typically quite cosmopolitan; ref. 81), and
that of 49 indigenous mite species, only 14 are endemic to the
island, although this figure may be inflated because of poor
sampling elsewhere (55, 56, 82). Likewise, of the 40 nematode
species recorded in the maritime Antarctic, only 5 are endemic
to single islands (61).
This argument rests on several assumptions, most notably
that widely distributed species have the lowest probability of
speciation, and that there are straightforward relationships
between body size, range size, abundance, and dispersal ability.
In the former case, the evidence is inconclusive (83), whereas
in the latter, the relationships are complex and scaledependent
(12, 84). Thus, in this case, and like most other
explanations for the species­body size distribution, there is
some way to go before all of the links in the causal chain of
reasoning are empirically verified. In so doing, recognition
should be given to the fact that different processes may be
important at different scales, and that no single explanation is
likely to account for patterns at intraspecific, interspecific, and
whole-assemblage scales.
We thank the many people who have contributed to understanding of the
composition of the fauna of Marion Island, particularly D. J. Marshall,
P. Greenslade, C. Ha¨nel, J. Barendse, and A. G. A. Gabriel. M. N. Bester
Gaston et al. PNAS December 4, 2001 vol. 98 no. 25 14495
ECOLOGY
and D. Nel kindly provided information on body sizes. G. W. Yeates
generously helped our understanding of the masses of nematodes. R. M.
May and two anonymous referees provided valuable comments on the
manuscript. The Department of Environmental Affairs and Tourism
(South Africa) provided financial and logistic support. This paper is a
contribution within the Scientific Committee on Antarctic Research
(SCAR) Regional Sensitivity to Climate Change (RiSCC) program.
K.J.G. is a Royal Society University Research Fellow.
1. Loder, N. (1997) Ph.D. thesis (University of Sheffield, U.K.).
2. Gaston, K. J. & Blackburn, T. M. (1999) Oikos 84, 353­368.
3. Yeates, G. W. (1982) Zool. Anz. Jena 208, 92­96.
4. Eadie, J. M., Broekhoven, L. & Colgan, P. (1987) Am. Nat. 129, 1­17.
5. Morse, D. R., Stork, N. E. & Lawton, J. H. (1988) Ecol. Entomol. 13, 25­37.
6. Basset, Y. & Kitching, R. L. (1991) Ecol. Entomol. 16, 391­402.
7. Brown, J. H. & Nicoletto, P. F. (1991) Am. Nat. 138, 1478­1512.
8. Blackburn, T. M. & Gaston, K. J. (1994) Trends Ecol. Evol. 9, 471­474.
9. Brown, J. H. (1995) Macroecology (Univ. Chicago Press, Chicago).
10. Novotny´, V. & Kindlmann, P. (1996) Oikos 75, 75­82.
11. Osler, G. H. R. & Beattie, A. J. (1999) Pedobiologia 43, 401­412.
12. Gaston, K. J. & Blackburn, T. M. (2000) Pattern and Process in Macroecology
(Blackwell Scientific, Oxford).
13. Morse, D. R., Lawton, J. H., Dodson, M. M. & Williamson, M. H. (1985) Nature
(London) 314, 731­733.
14. Lawton, J. H. (1986) in Insects and Plant Surfaces, eds. Juniper, B. E. &
Southwood, T. R. E. (Edward Arnold, London), pp. 317­331.
15. May, R. M. (1986) Ecology 67, 1115­1126.
16. Williamson, M. H. & Lawton, J. H. (1991) in Habitat Structure: The Physical
Arrangement of Objects in Space, eds. Bell, S. S., McCoy, E. D. & Mushinsky,
H. R. (Chapman & Hall, London), pp. 69­81.
17. Stanley, S. M. (1973) Evolution (Lawrence, Kans.) 27, 1­26.
18. Stanley, S. M. (1979) Macroevolution Pattern and Process (Johns Hopkins Univ.
Press, Baltimore).
19. Warren, P. H. & Lawton, J. H. (1987) Oecologia 74, 231­235.
20. Maurer, B. A., Brown, J. H. & Rusler, R. D. (1992) Evolution (Lawrence, Kans.)
46, 939­953.
21. Brown, J. H., Marquet, P. A. & Taper, M. L. (1993) Am. Nat. 142, 573­584.
22. Dixon, A. F. G., Kindlmann, P. & Jarosik, V. (1995) Ecol. Entomol. 20, 111­117.
23. Chown, S. L. & Gaston, K. J. (1997) Funct. Ecol. 11, 365­375.
24. Jablonski, D. (1997) Nature (London) 385, 250­252.
25. Koslowski, J. & Weiner, J. (1997) Am. Nat. 149, 352­380.
26. Tokeshi, M. (1999) Species Coexistence: Ecological and Evolutionary Perspectives
(Blackwell Scientific, Oxford).
27. Darlington, P. J., Jr. (1971) Bull. Mus. Comp. Zool. 142, 129­337.
28. Erwin, T. L. (1982) Bull. Biol. Soc. Wash. 5, 189­224.
29. May, R. M. (1978) in Diversity of Insect Faunas, eds. Mound, L. A. & Waloff,
N. (Blackwell Scientific, Oxford), pp. 188­204.
30. Tilman, D. & Pacala, S. (1993) in Species Diversity in Ecological Communities:
Historical and Geographical Perspectives, eds. Ricklefs, R. E. & Schluter, D.
(Univ. Chicago Press, Chicago), pp. 13­25.
31. Fenchel, T. (1993) Oikos 68, 375­378.
32. Gaston, K. J. (1991) Ecol. Entomol. 16, 505­508.
33. Gaston, K. J. & Blackburn, T. M. (1994) Biodiv. Lett. 2, 16­20.
34. Patterson, B. D. (1994) Biodiv. Lett. 2, 79­86.
35. Patterson, B. D. (2000) Div. Distrib. 6, 145­151.
36. Blackburn, T. M. & Gaston, K. J. (1995) J. Biogeog. 22, 7­14.
37. Allsopp, P. G. (1997) J. Biogeog. 24, 717­724.
38. May, R. M. (1999) Philos. Trans. R. Soc. London B 354, 1951­1959.
39. Finlay, B. J., Esteban, G. F. & Fenchel, T. (1996) Nature (London) 383,
132­133.
40. Walter, D. E. & Behan-Pelletier, V. (1999) Annu. Rev. Entomol. 44, 1­19.
41. Gremmen, N. J. M. (1981) The Vegetation of the Subantarctic Islands Marion and
Prince Edward (Dr. W. Junk Publishers, The Hague, The Netherlands).
42. Smith, V. R. (1987) South Afr. J. Sci. 83, 211­220.
43. Crafford, J. E., Scholtz, C. H. & Chown, S. L. (1986) South Afr. J. Antarct. Res.
16, 42­79.
44. Chown, S. L. & Language, K. (1994) Afr. Entomol. 2, 57­60.
45. Ha¨nel, C., Chown, S. L. & Davies, L. (1998) Afr. Entomol. 6, 366­369.
46. Ha¨nel, C. & Chown, S. L. (1999a) South Afr. J. Sci. 95, 87­112.
47. Ha¨nel, C. & Chown, S. (1999b) An Introductory Guide to the Marion and Prince
Edward Island Special Nature Reserves 50 years After Annexation (Department
of Environmental Affairs and Tourism, Pretoria).
48. Mercer, R. D., Chown, S. L. & Marshall, D. J. (2000) Polar Biol. 23, 775­784.
49. Barendse, J. & Chown, S. L. (2001) Polar Biol. 23, 346­351.
50. Gabriel, A. G. A., Chown, S. L., Barendse, J., Marshall, D. J., Mercer, R. D.,
Pugh, P. J. A. & Smith, V. R. (2001) Ecography 24, 421­430.
51. Smith, V. R. (1978) Oecologia 32, 239­253.
52. Burger, A. E. (1985) in Antarctic Nutrient Cycles and Food Webs, eds. Siegfried,
W. R. Condy, P. R. & Laws, R. M. (Springer, Berlin), pp. 582­591.
53. Burger, A. E., Lindeboom, H. J. & Willams, A. J. (1978) South Afr. J. Antarc.
Res. 8, 59­70.
54. Panagis, K. (1984) South Afr. J. Sci. 80, 30.
55. Marshall, D. J., Gremmen, N. J. M., Coetzee, L., O'Connor, B. M., Pugh,
P. J. A., Theron, P. D., Uekermann, E. A. (1999) Polar Biol. 21, 84­89.
56. Marshall, D. J., O'Connor, B. M. & Pugh, P. J. A. (2001) Polar Biol. 24, 101­104.
57. Sokal, R. R. & Rohlf, F. J. (1981) Biometry: The Principles and Practice of
Statistics in Biological Research (Freeman, New York).
58. Bunt, J. S. (1954) Aust. J. Zool. 2, 264­274.
59. Maslen, N. R. (1981) Brit. Antarc. Surv. Bull. 53, 57­75.
60. Boag, B. & Yeates, G. W. (1998) Biodiv. Conserv. 7, 617­630.
61. Maslen, N. R. (1979) Brit. Antarc. Surv. Bull. 49, 207­229.
62. Yeates, G. W. (1973a) New Zealand J. Sci. 16, 711­725.
63. Yeates, G. W. (1973b) Oikos 24, 179­185.
64. Hammond, P. M. (1992) in Global Biodiversity: Status of the Earth's Living
Resources, ed. Groombridge, B. (Chapman & Hall, London), pp. 17­39.
65. Walter, D. E., Seeman, O., Rodgers, D. & Kitching, R. L. (1998) Aust. J. Ecol.
23, 501­508.
66. Walter, D. E. & Proctor, H. C. (1998) Biotropica 30, 72­81.
67. Osler, G. H. R. & Beattie, A. J. (2001) Aust. Ecol. 26, 70­79.
68. Chown, S. L., Gremmen, N. J. M. & Gaston, K. J. (1998) Am. Nat. 152, 562­575.
69. Cooper, J. & Fourie, A. (1991) Bird Conserv. Int. 1, 171­175.
70. Chown, S. L. & Smith, V. R. (1993) Oecologia 96, 508­516.
71. Cooper, J., Marais, A. V. N., Bloomer, J. P. & Bester, M. N. (1995) Marine
Ornithol. 23, 33­37.
72. Ha¨nel, C. & Chown, S. L. (1998) Polar Biol. 20, 99­106.
73. Huyser, O., Ryan, P. G. & Cooper, J. (2000) Biol. Conserv. 92, 299­310.
74. Warham, J. (1990) The Petrels: Their Ecology and Breeding Systems (Academic,
London).
75. Pennycuick, C. J. (1987) in Comparative Physiology: Life in Water and on Land,
eds. Dejours, P., Bolis, L., Taylor, C. R. & Weibel, E. R. (Springer, Berlin), pp.
371­386.
76. Warham, J. (1996) The Behaviour, Population Biology and Physiology of the
Petrels (Academic, London).
77. Lawton, J. H. & Brown, K. C. (1986) Philos. Trans. R. Soc. London B 314,
607­617.
78. Hadley, N. F. (1994) Water Relations of Terrestrial Arthropods (Academic, San
Diego).
79. Finlay, B. J., Esteban, G. F., Olmo, J. L. & Tyler, P. A. (1999) Ecography 22,
138­144.
80. Finlay, B. J. & Clarke, K. J. (1999) Nature (London) 400, 828.
81. Gabriel, A. G. A. (1999) M.Sc. thesis (Univ. of Durban-Westville, Durban).
82. Pugh, P. J. A. (1993) J. Nat. Hist. 27, 323­421.
83. Chown, S. L. & Gaston, K. J. (2000) Trends Ecol. Evol. 15, 311­315.
84. Chown, S. L. (1997) in The Biology of Rarity, eds. Kunin, W. E. & Gaston, K. J.
(Chapman & Hall, London), pp. 91­109.
14496 www.pnas.org cgi doi 10.1073 pnas.251332098 Gaston et al.