Download Species number, species abundance and body

Ecological Entomology (1988) 13, 25-37
Species number, species abundance and body length
relationships of arboreal beetles in Bornean
lowland rain forest trees
D. R. MORSE, N. E. STORK* and J. H. LAWTON? The Computing Laboratory,
The University, Canterbury, *Department of Entomology, British Museum (Natural
History), and ?Department of Biology, University of York
ABSTRACT. 1. The relationships between number of species, abundance
per species, and body length are examined for 859 species of beetles in
samples of arthropods collected from ten Bornean lowland forest trees by
insecticide fogging. Similar relationships are examined for different feeding guilds of these beetles, and for those beetles from different species of
trees.
2. The data are used to construct four interrelated graphs, namely
species:abundance, species :body length, population abundance:body length
and total number of individuals:body length distributions.
3. In contrast to a number of previous studies, no consistent linear
relationship between population density and body length was found for the
Bornean beetles and it is suggested that, as in birds, the added dispersal
ability of flight reduces critical population densities necessary for persistence in small species. Previous relationships between body weight and
population abundance may also be artefacts of the way in which data were
gathered.
4. Despite large samples, we failed to locate the mode in plots of the
number of species in each abundance category (species: abundance
distribution).
5. Species:body length and total number of individuals:body length
plots were similar to those found in previous studies, although using data
for Coleoptera alone may have produced a steeper decline in the total
number of individuals as body size increases than is apparent in samples of
all arthropods.
6. We present the first three-dimensional graph relating numbers of
species, body lengths and population abundances. The surface of this
three-dimensional relationship is relatively simple.
Key words. Species abundance, body length, beetles, Borneo, tree
canopy, rain forest, insecticide fogging.
Correspondence: Mr D. R. Morse, The Computing Laboratory, The University, Canterbury, Kent CT2 7NF.
25
26
D.R. Morse, N . E. Stork
utidJ. H . Lawtoti
Introduction
The structure of animal communities can be
described in a number of ways; the simplest is
the number of coexisting species. More subtle
descriptions of samples of organisms from a
community include: the number of species in
each abundance category (the so-called
species: abundance distribution. or species :
frequency distribution. e.g. Preston. 1938;
Williams. 1953; May, 1Y75: Southwood. 1978:
Sugihara. 1980); the number of species of
different body sizes (species: body-size distribution) (e.g. Hemmingsen. 103.1; Hutchinson &
MacArthur. 1959: May. 1978. 1986; Griffiths.
1986); and the abundance of each species versus
body size (population abundance; body size relationship) (e.g. Damuth. 1981; Peters. 1983:
Peters & Raelson, 1984: Peters & Wassenberg.
'T\
1983; Brown & Maurer, 1986). Summing
individuals for all species of one body size yields
a fourth pattern. the total number of individuals
of a particular size, irrespective of species (total
number of individuals: body size, or if biomass
rather than number is used. total biomass:body
size relationship, e.g. Janzen & Schoener, 1968;
Janzen, 1973; Morse eral., 1985; Griffiths, 1986;
Rodriguez & Mullin. 1986; Strayer. 1986).
Each of these relationships has its own literature, and theoretical explanations for particular
patterns (loc. cit.). But, as Fig. 1 makes plain,
they are interrelated. For example. fixing any
two of the basic relationships in Fig. 1 (i)
(species:abundance, species:body size. or
population abundance :body size) automatically
defines the limits of the third (Harvey &
Lawton, 1Y86). A growing number of papers
consider relationships between two of the dis-
NUMBER OF SPECIES
LOG ABUNDANCE
PER SPECIES
TOTAL INDIVIDUALS
Y
SPECIES
FIG. 1 . Generalized relationship between community variables. The upper (3-D) figure (i) illustrates
relationships between three distributions. In the Y X , plane is the species:abundance distribution, here
illustrated for convenience as a log-normal distribution (Preston. 1967). The YZ plane defines the
species: body size distribution. and the ZX, plane the population abundance:body size relationship.
Summing individuals for all species of one body size gives the projected ZX? plane in (ii). Focusing on (i),
the question is. what does the full. three-dimensional surface (iii) look like in real communities'?
Species nbundnncelsize in canopy beetles
tributions (e.g. Janzen, 1973; Griffiths, 1986;
Strayer, 1986; Harvey & Godfrey, 1987). but
because data are almost always presented in two
dimensions (for example species: abundance distributions are presented by summing across
individuals of all body sizes), development of a
unifying theory is hampered by ignorance about
the shape of the full three-dimensional surface
depicted in Fig. 1. May (1978) is one of the few to
have realized that all these community patterns
are interrelated. The surface defined by Fig. 1
will not necessarily be simple, and may differ
from community to community. For example,
attempts to predict observed numbers of insect
species (S) in different body size classes, knowing only the total number of individuals of each
body size (N), using the empirical relationship
S a p 2 5 , derived from Preston’s log-normal distribution of species abundances (May, 1978), fail
(Lawton, 1986). One possible explanation is that
different species abundance distributions hold
for species of different body size classes.
The present work is the first to establish
empirical relationships between the variables
depicted in Fig. 1, for an assemblage of species
collected at one place and time, namely adult,
arboreal beetles sampled from the canopy of
rain forest trees using knockdown insecticides
(Stork, 1987a, b; and unpublished). Our purpose is to establish what the patterns look like,
not to develop theories to explain the patterns;
the end result is the three-dimensional surface
depicted in Fig. 7. As will become apparent, the
data are not ideal, but they are the best currently
available, based on samples of 23,000
individuals and approximately 3000 species of
arthropods. Although full information on body
lengths, numbers of species and numbers of
individuals are only available for the Coleoptera
in these samples, the data nevertheless consist of
859 species and nearly 4000 individuals. We are,
however, aware that the Coleoptera may represent a biased sample of the relationships displayed by the entire arthropod assemblage of the
canopy. We have attempted to discover whether
this bias might be serious by reanalysing
Janzen’s (1973) sweep net samples from the
understorey of tropical forest, for total
arthropods and the Coleoptera alone. We return
to the problem of taxonomic and other sampling
biases in the Discussion. Finally, as a first step in
understanding the biological basis underpinning
the patterns revealed by the data, we have
27
analysed not only the total collection of Coleoptera, but also different feeding guilds, and the
species associated with different species of trees.
Methods
Samples of arthropods were collected by fogging
the canopy of trees with a synthetic pyrethroid
insecticide in an area of lowland rain forest near
Bukit Sulang, Brunei (Borneo) in September
1982. Full details of the techniques used will be
given elsewhere (Stork, unpublished). Ten trees
were chosen for sampling; trees 1-4, Shorea
johorensis Foxwood; trees 5-6, S.macrophylla
(De Vriese) [Dipterocarpaceae]; trees 7-8, Pentaspadon motleyi Hook [Anacardiaceae]; tree 9,
Number of individuals
1000.
100
10
1
1000
100
10
1
4
8
12
16
20 24
Body length
28
3 2 36
4 0 44
mm
FIG. 2. Histograms of the number of individuals of all
arthropods in different 1 mm size classes (i.e. graphs
in the Z X , plane of Fig. l(ii) for sweep samples of
understorey vegetation from (a) Taboga primary
riparian vegetation and (b) Osa secondary vegetation,
Costa Rica (from Janzen, 1973). The shaded bars
represent the number of beetles of different body
lengths.
28
D . R . Morse, N . E . Stork and J . H . Lawton
Casranopsis sp. [Fagaceae]; and tree 10. unidentified species and family.
The insect samples were sorted to orders and
for most groups, to morpho-species (for
simplicity, referred to as species from here on).
The 859 species (3919 individuals) of adult Coleoptera were assigned to four guilds, herbivores
(384 species), predators (200). scavengers (120)
and fungivores (141) (guild assignments as in
Stork (1987a), but with the fungivores separated
from the scavengers); a further fourteen species
could not be confidently assigned to guilds.
We used body length as a measure of the size
of each species since length is considerably
easier to measure than weight, and for
arthropods the two measures are highly correlated (Rogers el a l . , 1976,1977; Schoener, 1980).
Measurements of the body lengths of species
were taken from the hind-most tip of the abdomen or the elytra to the most forward part of the
head (excluding the antennae) using a dissecting
microscope with an eyepiece graticule. For most
species of beetle in the samples. body length
varies little. the smallest individual usually not
being more than 10% shorter than the longest.
Mean body lengths. therefore, were estimated
Number of species
from a maximum sample of ten specimens per
species (or all available specimens of species
represented by less than ten individuals) and
grouped into a series of length classes arranged
on a logarithmic scale.
Results
Analysis of'Janzen's Costa Rican datu
In Figs. 2(a-b) total numbers of individuals
are plotted against log body length classes, for
the arthropods in sweep samples from two tropical lowland sites in Costa Rica. In both samples
the Coleoptera comprise a major part of the
smaller arthropods but are almost totally lacking
in the larger size classes; hence the rate of
decline from the mode in number of individuals
is much steeper for the Coleoptera than for all
arthropods combined.
Species :abundance distribution
The species: abundance distribution for all
individuals combined is shown in Fig. 3(a). The
most striking feature of the graph is the number
t
--
i.
=I3
21
lo,
8-0
0&3;4
0
62
0
133
496
1
I
I
0
(a)
Number of Indlviduals per species
(b)
100
200
300
400
500
600
700
800
Species ranked in order 01 abundance
FIG. 3. (a) Species:abundance distributions for all the beetles in ten Bornean tree samples (1.e. the YX,
plane of Fig. 1 ) . (b) The same data as (a), plotted as abundance (on a log scale) versus species rank.
Numbers are given where data points in each block are too numerous to illustrate.
Species abundancelsize in canopy beetles
29
TABLE 1. Equations of regression lines fitted through the species rankabundance plots of Fig. 3(b) for the beetles collected from three different
species of tree. The regression lines were fitted by ordinary least squares
techniques through the data after it had been logarithmically transformed.
All regression lines are significant at the 0.1% level. Also shown are the
mean numbers of species and individual beetles found on each tree.
Tree species
Slope
SE
Intercept
SE
Shorea johorensis
Shorea macrophylla
Pentaspadon motleyi
-0.2810
-0.2747
-0.1780
0.0031
0.0058
0.0069
2.1546
1.8313
1.0176
0.0215
0.0352
0.0362
Tree species
Mean no. per tree
Shorea johorensis
Shorea macrophylla
Pentaspadon motleyi
Species
Individuals
175.5
143.0
72.0
466.0
513.0
112.0
of rare species, 58% of the species in the samples
being represented by single individuals. We
have not attempted to fit a log-normal distribution to the data, because it is clear that we have
not yet discovered the mode of the distribution
(parameters estimated for log-normal distributions without a mode are probably meaningless
(Hughes, 1986)). Instead, in order to compare
samples from the different tree species, we
replotted the data (Fig. 3b) as a rank-abundance
graph (e.g. Williamson, 1973). The equation of
the line fitted to the data in Fig. 3(b) is:
abundance=470 * rank-0.96(P<O.OOl)
Similar plots for samples taken from the
different species of tree show a linear relationship between species-rank and species-abundance on a double logarithmic plot, but their
slopes are different (Table 1) (the data from
trees 9 and 10 were not sufficient to justify separate analyses). The slopes of the lines through the
rank abundance data for beetles from Shorea
johorensis and S. macrophylla are very similar.
However, that for Pentaspadon motleyi is
approximately half the value of the slopes for
either Shorea species, and the intercept for P.
motleyi is lower.
Species:body length distributions
In Fig. 4(a) the number of species is plotted
against body length class. The maximum number of species occurs in size class 3 (a body length
of about 2 mm). The upper tail of the distribution from size class 5 (a body length of about
3 mm) shows a near linear decline, indicating a
power-law relationship between number of
species and body length. May (1978) provides a
theoretical argument for the expected shape of
the upper tail of this distribution (quantitative
theoretical predictions about the shape of the
full distribution have not been made). Following
May (1978) we have therefore fitted a regression
line to the upper (right-hand) part of the distribution (taking the mid-point of each size class
on the abscissa); the fitted line has a slope of
-2.64 (k0.38). Equivalent data for the different
guilds of beetles are in Figs. 4(b-e). It is clear
that each guild contributes differentially to the
overall picture in Fig. 4(a). The most speciesrich size classes are in different positions for each
guild and the shapes of the distributions appear
to be quite different.
Population abundance:body length distributions
In Fig. S(a) the number of individuals per
species is plotted against the body length of each
species on double logarithmic axes. The resulting scatter-plot has two peaks the lower of which
(corresponding to body lengths of 0.7-1.4 mm
and population abundances of up to fifty
individuals) is comprised mainly of fungivores
(Fig. 5d) and the upper (corresponding to body
lengths of 3-5 mm and population abundances
of up to 200 individuals) mainly of herbivores
(Fig. 5b). There are no consistent relationships
between size and population abundances in Figs.
5(a-e) of the form proposed by Peters (1983)
and Peters & Wassenberg (1983). Significant but
30
D.R . Morse. N . E . Stork and J . H. Lawton
a
100
I
-
m
.-0u
a
.
a
c
15
a
10-
5
2
:
.
n
-
1,
0
-2
-2
i!
0
loo3
b
0
1 1 1 1
I
1
1
4
~
I
1
8
2
8
10
12
14
16
Body length c l a s s
16
14
I
I
1
1
I
1
1
12
1
16
1
I
1
1
1
20
I
24
1
1
1
28
B o d y length mm
C
k
Herbivores
-
Fungivores
-
Scavengers
-1
r-
-
4
Body length classes
FIG. 4. Histograms of number of species in different length classes (i.e. the Y Z plane of Fig. 1) for (a) all
beetles in the Bornean samples. and for the following guilds of beetles: (b) herbivores, (c) predators, (d)
fungivors. (e) scavengers. ( N . B . In order to display the dataeconomically, the Y axisof the Y Z plane of Fig.
1 is here logarithmically transformed. The scale converting length classes to actual length is displayed below
Fig. 4 (a). For cxample. beetles in length class 14 are hetween 16.4 and 20 mm long; classes are in groups of
5*10& length.)
Species abundancelsize in canopy beetles
a
0.
100
..
... ...
.
... ...
.. ... .......
.
..
.
.
.....
....
.....
......
.
........
........
.
*
I
-
- *
10
*
.
0
.
.:..A.
w..
. w .
.w No-.....
..m..--
.
n
o
.
.
.
7
.
.
7
"
1
10
a
..
.....
... . .
....
. ............
0
100
. ...-... -.'.. .
. ... . . . .
.... ..-. ..
.
d
10
' e
..
. .
...
......
......
.- .. .
........- ...
.........--.-. . ".
10.
.. ..... .
..........
...-... . . . .
.....
..-. ...
7-
31
32
D . R . Morse, N . E . Stork and J . H . Lawton
Total number 01 individuals 01 a11 species
Total number of individuals: body length
distributions
1
2
0
2
8
10
12
14
16
Body length class
FIG.6. Histogram or the total number of individuals in
different length classes (i.e. the Z X z plane of Fig. 1
(ii)). The scale converting length classes to actual
length is displayed below Fig. 4(a).
very weak relationships were found in the predators (which have a positive. rather than negative
slope, as expected; log length=0.100 log abundance+0.325, P<O.OOl). and the fungivores
(log length= -0.129 log abundance+0.229,
P<O.Ol) but were not found in the other guilds,
nor for all of the species when pooled. In both
the predator and fungivore guilds the relationships are weak; only 4% and 0.7% of the
variance is explained, respectively.
The total number of individuals in the
different body length classes are shown in Fig. 6.
The upper tail of the distribution from modal
length class 6 (a body size of c . 3 . 5 4 mm) shows
an almost linear decline in the number of
individuals with increasing body length. Similar,
but less clear-cut, patterns are evident in the
individuals: body lengths graphs for each guild
and in the samples taken from the different
species of tree (not illustrated). The distributions for individual guilds tend to be skewed
relative to the total distribution, with modes in
different body length classes (Table 2). Least
squares regressions fitted through the upper tails
of the total individuals: length distributions
(when the data have been logarithmically transformed) are in Table 2. After excluding the
scavengers from the analysis, because the
residual variance after regression was not homogeneous with that of other guilds (Bartlett’s test
for homogeneity of variances, ,yz (with 3
df)=20.54, P<O.OOl), the slopes for the other
guilds were found to differ significantly from one
P<O.OOl). The
another (ANOVA, F’2,18=21.85,
slopes for the individual species of tree also
differed significantly from one another
(ANOVA, F2,y=30.91, P t O . O O 1 ) . The slope for
Penfaspadon motleyi is lower by a factor of more
than 2.5 than the slopes of the lines through the
data for either of the two Shorea species (again,
data for trees 9 and 10 were too sparse to justify
separate analyses).
TABLE 2. Equations of regression lines fitted through the upper parts of the
total individua1s:body length graphs for different subsets of the data (i.e. slopes
to the right of the mode in Fig. 6). The regressions were fitted by ordinary leastsquares techniques after the data had been logarithmically transformed. All data
from and to the right of the mode of the distribution were included. The
geometric mean of the limits on each size class was used for the abscissa. All
regressions are significant at the 0.5% level.
Data set
SE
Slope
~
Intercept’
SE
All species
Herbivores
Predators
Fungivores
Scavengers
Shorea johorensis
S . macrophylla
Pentaspadon motleyi
-4.14
-4.63
-3.70
-2.27
-2.54
-4.32
-4.55
-1.56
Modal
size-class
__-_
~~
0.369
0.213
0.238
0.248
0.376
0.379
0.455
0.208
5.20
5.28
4.28
2.13
3.01
4.70
4.92
2.18
0.352
0.190
0.199
0.132
0.552
0.324
0.371
0.162
6
6
6
0
4
5
6
3
* log,, number of individuals at a body-size of 1 mm (based on extrapolation).
Species ahundancelsize in canopy beetles
33
e
FIG. 7. (a) A three-dimensional graph plotting number of species against each log body length class and
against total log abundance class for all beetles, and for each guild: (b) herbivores, (c) predators, (d)
fungivores, and (e) scavengers. The height of each intersection is proportional to the number of species
having a particular combination of body length in each length class (see Fig. 4(a) for the conversion scale)
and abundance in each octave on the conventional logz scale of species abundances (Preston, 1962).
Combining the variables
The
three-dimensiuna1 plots for
the
beetles, and for individual guilds, are in Figs.
7(a-e). Note that, unlike Fig. 4, the y-axis in Fig.
7 is not logarithmic.
The full surface is reasonably smooth,
although careful inspection shows three ‘ridges’
running roughly parallel to the X,-axis in the
YX, plane (Fig. 7a). These ‘ridges’are due to the
main and two subsidiary modes in the number of
species in different body length classes, with
different guilds contributing differentially to
(Figs. 7b-e).
each
Discussion
Confining the scope of this study to the Coleoptera has both advantages and disadvantages.
34
D.R . Morse, N . E. Stork and J . H.Larvton
The chief disadvantage is lack of generality. It is
difficult to extrapolate from beetles to complete
arthropod communities. Examination of total
number of individuals: body length distributions
for arthropods in sweep net samples from two
sites in Costa Rica (Fig. 2) reveals that the Coleoptera comprise the major part of the lower
body length classes although sweep netting captures a biased sample of arthropods (e.g.
Hespenheide, 1979). The beetles make up only
18.0% of the species and 16.5% of the
individuals in the complete arthropod samples
from the Bornean trees (Stork, unpublished).
However, they represent major proportions of
four of the guilds of arthropods most closely
associated with trees. For instance. the beetles
account for a mean of 79.2% of the species and
66.8% of the individual chewing herbivores. For
predators the equivalent figures are 50.4% and
45.6%. and for scavengers and fungivores combined, 40.7% and 37.9% respectively (Stork.
1987a). The Coleoptera clearly represent an
important cross-section, in terms of feeding
habits, of the arthropod fauna in the Brunei rain
forest canopy.
We are also aware that insecticide fogging as a
method of collecting is not without problems but
these are considerably less than for other
methods of sampling from the canopy. and
beetles appear to be particularly well sampled
(Stork, 1987a, and unpublished). Confining
attention to adult Coleoptera also has other
advantages. One of the problems of examining
body length relationships is that species in some
groups have individuals with a wide range of
body lengths. This is particularly true of Hemiptera and the orthopteroid orders where the
adults and nymphs often occur together, are
similar in appearance, and have similar feeding
habits. For instance, 95% of the Blattodea in the
samples were nymphs of a range of instars and
hence sizes. Even the adults of some species can
vary considerably in size. These problems are of
only minor relevance to this study for several
reasons. First, most of the adult beetles varied
little in size within species. Second. beetle larvae
are usually very different in appearance from the
adults and, in some groups, are found in
different habitats. Third, beetle larvae represented less than 5% of all Coleoptera in the
Bornean samples (and were excluded from our
analyses).
Taxonomic restrictions aside, there is one
other important way in which Fig. 7, and its
component parts (Figs. 3, 4 and 5) may give a
distorted or indeed totally false impression of
the relationships that exist between these variables in more natural and taxonomically diverse
faunal assemblages. Our analyses are based on
samples of selected trees (see Methods), and are
not stratified according to tree abundance, and
as we have indicated, different species of tree
have rather different patterns (e.g. Tables 1 and
2). Only further studies, designed to overcome
taxonomic and sampling biases, can reveal
whether the patterns in our total data are serious
distortions of patterns in real communities. At
the present time we can do no more than present
the data to illustrate the nature of the problem,
as a spur to others to gather more and better
data. The interpretations that follow must be
viewed in the light of these major caveats.
It is unclear which of the many available
models (see Southwood. 1978. for summary)
would be the most appropriate to fit the
species: abundance distribution (Fig. 3). Taylor
(1978) and Hughes (1986) point out that in small
samples both the logarithmic series and the
upper section of the log-normal distribution fit
equally well, which is also true of the data in Fig.
3(a). Since there is no mode or suggestion of a
mode. the true number of beetle species in the
Bornean lowland trees must be much greater
than the 859 species found, if they have a lognormal distribution. Because there is no mode in
Fig. 3(a), it isdifficult (e.g. Hughes, 1986) to test
the notion (see Introduction; Lawton, 1986) that
species of different sizes may have different
species:abundance distributions with modes in
different abundance classes. Fig. 7(a) simply
shows monotonic declines in the numbers of
species in each abundance category for beetles
of all sizes (i.e. ‘slices’ in the Y X , plane look
qualitatively similar at all points along the Z
axis).
The species: body length distributions for
beetles in Bornean rain forest (Figs. 4a-e) are
similar to those found by other authors (e.g.
Hemmingsen, 1934; Schoener & Janzen, 1968)
in that they approximately conform to a lognormal distribution. The slope of -2.64 (k0.38)
17132.6x46.9for the upper tail of the distribution
for the combined data when plotted on doublelogarithmic axes shows a reasonable fit to May’s
(1978) predicted value of approximately -2.
More recent calculations (May, 1986) based on
Species abundancelsize in canopy beetles
the fractal nature of plant surfaces (Morse et al.,
1985) imply slopes lying between -1.5 and
-3.0. Data from independent studies of tropical
canopy beetles (Erwin & Scott, 1980; Erwin,
1983) and other insect communities (Terakawa
& Ohsawa, 1981) also conform reasonably
closely to May’s (1978) prediction that the number of species scales as body length-’, above
some critical minimum body size (see Lawton,
1986). It is unclear whether the total number of
arthropod species declines below the modal size
class 3 in Fig. 4(a), or whether very small species
in groups other than Coleoptera would
markedly increase the total number of
arthropod species, below a length of about 2 mm
(see also May, 1978, 1986).
Several studies (Damuth, 1981; Peters, 1983;
Peters & Raelson, 1984; Peters & Wassenberg,
1983; Brown & Maurer. 1986) have found a
simple linear relationship between the population densities of individual species and body
weight. The graphs presented by these authors
are noticeably lacking in data for small, rare
species equivalent to a complete lack of points
close to the Y-axis on the Z X , plane of Fig. 1 (i.e.
there is a ‘hole’ in the data on the ‘floor’ of Fig.
l(i), furthest from the reader). Then, by definition, a Fig. 1style three-dimensional plot of their
data would result in an incomplete surface, with
two implications for the other faces of Fig. 1.
First, rare species in a species:frequency distribution (the Y X 1plane) would tend mainly to
be large individuals; secondly, only more abundant species would contribute to the shape of the
Y Z plane for small values of Z (small body size).
In marked contrast to these authors, we found a
considerable number of small rare species in the
Z X , plane (Fig. 5 ) and the three-dimensional
surface was complete (Fig. 7). There was no
consistent relationship between the average
abundance and body size of species (Fig. 5).
There are three possible reasons why the data
in Fig. S differ from that of earlier studies, two
biological, the other an artefact of the way in
which earlier data have been assembled. Both
biological reasons centre on species mobilities.
For example, it is interesting to observe that
both Peters & Wassenberg (1983) and Juanes
(1986) found only weak and inconsistent relationships between population density and body
size for birds. Perhaps the feature that sets both
birds and beetles apart from the other groups
analysed is that most of their species are winged,
35
highly mobile, and therefore able to encounter
one another and to mate at very low population
densities (see also Juanes (1986) for alternative
biological arguments). Second, high mobility
will tend to yield many ‘tourists’ (sensu Moran &
Southwood, 1982) in the samples, possibly
boosting the number of very rare species.
It is also possible that the clear correlations
that exist in earlier studies between population
density and body size are sampling artefacts;
many of these data were gathered from
published studies on the autecology of individual
species. Such species may tend to be both commoner and larger than average (few biologists
choose to work with small, rare species). Hence
data for smaller, rarer species may be under
represented in many earlier studies (but see
Brown & Maurer, 1986).
Summing individual population abundances
for species of a particular size in Fig. 5(a). yields
the data in Fig. 6 (see also Fig. 1, showing how
the Z X , plane gives rise to the Z X 2graph). Fig. 6
is again roughly log-normally distributed, with a
mode at size class 6. Although there is considerable variation in the rate of decline in the total
number of individuals with body length to the
right of this mode for various beetle guilds
(Table 2), the majority of those plots have slopes
much greater, sometimes by an order of magnitude, than has been found in other studies of
insect communities (Hijii, 1984; Kikuzawa &
Shidei, 1967; Terakawa & Ohsawa, 1981;
Lawton, 1986; Morse et al., 198s). On average,
the Bornean beetle data show that for an order
of magnitude decrease in body length, the total
number of individual beetles, summed over all
species, increases by a factor of approximately
10,000. However, analysis of Janzen’s (1973)
data (Fig. 2) indicates that restricting our study
to the Coleoptera may have had the effect of
greatly increasing the slope above the mode,
compared with similar slopes for entire
arthropod assemblages. Data for rain forest
beetles may not therefore be at variance with the
model proposed by Morse ef al. (1985) and
Lawton (1986), which predicts a slope of -3.25
to the right of the mode for data of the type
displayed in Fig. 6 and Table 2.
The end point of our analyses is the threedimensional surface displayed in Fig. 7(a). Small
ridges run across it, roughly parallel to the YX,
plane, for species in the smallest body length
classes; however, the bulk of the surface is
36
D.R . Morse, N . E . Stork arid J . H . Lawton
relatively smooth. Each guild (Fig. 7b-e) contributes to the shape of the overall surface in a
different way but we have no theoretical basis
for interpreting these patterns. It is possible that
they are artefacts produced by taking subsamples. Further progress is impossible without
even larger samples. which might also be
required to carry out separate analyses for each
species of tree.
It is unclear whether widening the taxonomic
base of our samples, or collecting at different
seasons, will alter the shape of the surface.
Future empirical and theoretical studies will
have to address both points. Because there are
no similar data in the literature, we cannot say
whether they are typical or unusual. There are
hints in the literature (e.g. Griffiths, 1086:
Strayer, 1986). that surfaces for other communities may be both more complex and more
irregular.
Acknowledgments
We are particularly grateful to colleagues at the
BM(NH), who will be fully acknowledged in a
later publication, for assisting Nigel Stork in
sortingthe Coleoptera to species. We thank Paul
Harvey, Bob May, Stuart Pimm and Mark Wetton for valuable discussion. Nigel Stork
acknowledges the permission of the Sultan of
Brunei to study in Brunei, assistance from Jaya
bin Sahat of the Brunei Museum, and fieldwork
support from members of the Leeds University
Expedition to Brunei. David Morse was supported by an NERC studentship. The insecticide
was provided by Wellcome Research Laboratories through Peter Chadwick. Peter Hammond.
Cliff Moran, Joe Perry and an anonymous
referee made helpful comments o n the
manuscript.
References
Brown. J.H. & Maurer. B.A. (1986) Body size.
ecological dominance and Cope’s rule. Nature.
324, 248-250.
Damuth, J . (1981) Population density and body size in
mammals. Narure, 230, 699-700.
Erwin. T.L. (1983) Beetles and other insects of tropical forest canopies at Manaus. Brazil, sampled by
insecticidal fogging. Tropical Rain Foresit Ecology
and management (ed. by S . L. Sutton. T. C. Whitmore and A. C. Chadwick), pp. 59-75. Special
publication no. 2 of the British Ecological Society.
Blackwell Scientific Publications. Oxford.
Erwin. T.L. & Scott, J.C. (1980) Seasonal and size
patterns, trophic structure and richness of Coleoptera in the tropical arboreal ecosystem: the fauna
of the tree Luehea seemanii Triana and Planch in
the Canal Zone of Panama. Coleopterists Bulletin,
34, 30-5322.
Griffiths, D. (1986) Size-abundance relations in communities. American Naiuralist. 127, 14C166.
Harvey. P.H. & Godfray. H.C.J. (1987) How species
divide resources. American Naturali.rt. 129, 31%
320.
Harvey, P.H. & Lawton. J.H. (1986) Patterns in three
dimensions. Naiure. 324, 212.
Hemmingsen, A.M. (1934) A statistical analysis of the
differences in body size of related species.
Videnskabelige Meddelelser Dansk Naricrhistorisk
Forening Kohenhavn. 98. 125-160.
Hespenheide. H.A. (1979) Are there fewer parasitoids in the tropics? Americun Nafurahf. 113,
766-769.
Hijii. N. (1983) Arboreal arthropod fauna in a forest.
11. Presumed community structures based on
biomass and number of arthropods in a
Chumaecyparis obrclsa Plantation. Japanese Journal of Ecology, 34, 187-193.
Hughes, R.G. (1986) Theories and models of species
abundance. American Naturalist, 128, 879-899.
Hutchinson, G.E. & MacArthur, R.H. (1959) A
theoretical ecological model of size distributions
among species of animals. American Naturalist. 93,
117-125.
Janzen, D.H. (1973) Sweep samples of tropical foliage
insects: description of study sites, with data on
species abundanccs and size distributions. Ecology. 54,659-686.
Janzen. D.H. & Schoener. T.W. (1968) Differences in
insect abundance and diversity between wetter and
drier sites during a tropical dry season. Ecology.
49,96110.
Juanes. F. (1986) Population density and body size in
birds. Americun Naturalisr. 128, 921-929,
Kikuzawa. K. & Shidei. T. (1967) On the biomass of
arthropods of the Japanese red pine forest in the
vicinity of Kyoto. Bulletin of Kyoto University
Forests. 39, 1-8.
Lawton. J.H. (1986) Surface availability and insect
community structure: the effects of architecture
and fractal dimension of plants. Insects and die
Planf Surface (ed. by B. E. Juniper and T. R. E .
Southwood). pp. 317-331. Edward Arnold,
London.
May. R.M. (1975) Patterns of species abundance arld
diversity. Ecology and Evolution of Communities
(ed. by M. L. Cody and J. M. Diamond). pp.
81-120. Harvard University Press. Cambridge.
Mass.
May. R.M. (1978) The dynamics and diversity of
insect faunas. Diversity o f h e c t Fauriu~(ed. by L.
A. Mound and N. Waloff). pp. 188-204. Symposia
of the Royal Entomological Society of London.
No. 9. Blackwell Scientific Publications, Oxford.
May, R.M. (1986) The search for patterns in the
balance of nature: advance and re!reats. Ecology.
67, 111-55-1126.
Moran. V.C. & Southwood, T.R.E. (1982) The guild
Species abundancelsize in canopy beetles
composition of arthropod communities in trees.
Journal of Animal Ecology, 51,289-306.
Morse, D.R.. Lawton, J.H., Dodson, M.M. &
Williamson, M.H. (1985) Fractal dimension of
vegetation and the distribution of arthropod body
lengths. Nature, 314,731-733.
Peters, R.H. (1983) The Ecological Implications of
Body Size. Cambridge University Press.
Peters, R.H. & Wassenberg, K. (1983) The effect of
body size on animal abundance. Oecologia, 60,
89-96.
Peters. R.H. & Raelson, J.V. (1984) Relations
between individual size and mammalian population density. American Naturalist, 124, 49&517.
Preston, F.W. (1948) The commonness and rarity of
species. Ecology, 29, 254-283.
Preston. F.W. (1962) The canonical distribution of
commonness and rarity. Ecology, 43,185-215 and
410-432.
Rodriguez, J. & Mullin, M.M. (1986) Relation
between biomass and body weight of plankton in a
steady state oceanic ecosystem. Lirnnology and
Oceanography, 31, 361-70.
Rogers, L.E.. Buschbom, R.L. & Watson, C.R.
(1977) Length-weight relationships of shrubsteppe invertebrates. Annals of the Entomological
Society of America, 10, 51-53.
Rogers. L.E., Hinds, W.T. & Buschbom, R.L. (1976)
A general weight vs. length relationship for insects.
Annals of the Entomological Society of America,
69, 387-389.
Schoener, T.W. (1980) Length-weight regressions in
tropical and temperate forest-understory insects.
Annals of the Entomological Society of America,
73, 106109.
Schoener, T.W. & Janzen, D.H. (1968) Notes on
environmental determinants of tropical versus
31
temperate insect size pattern. American Naturalist,
102, 207-224.
Southwood, T.R.E. (1978) Ecological Methods with
Particular Reference to the Study of Insect Populations. Chapman & Hall, London.
Strayer, D. (1986) The size structure of a lacustrine
zoobenthic community. Oecologia, 69,513-6.
Stork, N.E. (1987a) Guild structure of arthropods
from Bornean rain forest trees. Ecological
Entomology. 12, 69-80. (Erratum, 12,480.)
Stork. N.E. (1987b) Arthropod faunal similarity of
Bornean rain forest trees.-Ecological Entomolhgy,
12, 219-226.
Sugihara, G. (1980) Minimal community structure: an
explanation of species abundance patterns.
American Naturalist, 116, 770-87.
Taylor, L.R. (1978) Bates, Williams, Hutchinson-a
variety of diversities. Diversity of Insect Faunas
(ed. by L. A . Mound and N . Waloff), pp. 1-18.
Symposium of the Royal Entomological Society of
London, No. 9. Blackwell Scientific Publications,
Oxford.
Terakawa, N.I. & Ohsawa, N. (1981) On the arboreal
arthropods in the Inabu experimental forest of
Nagoya University. 11. The relationship between
body length and numbers of individuals. Japanese
Forestry (Society)Discussions, 91, 349-350.
Williams, C.B. (1953) l h e relative abundance of
different species in a wild population. Journal of
Animal Ecology, 22, 1431.
Williamson, M. (1973) Species diversity in ecological
communities. The Mathematical Theory of the
Dynamics of Biological Populations (ed. by M. S.
Bartlett and R. W. Hiorns). pp. 325-335.
Academic Press. London.
Accepted 6 June 1987