Oecologia (Berlin) (1983) 60:89-96

Oecologia1983 9 Springer-Verlag

The effect of body size on animal abundance

Robert Henry Peters and Karen Wassenberg Department of Biology, McGill University, 1205 Ave. Dr. Penfield, Montreal, Qtt~bec, H3A 1Bt, Canada

Smnmary. Although it is a commonplace that sma11 animals are more abundant than large ones, few attempts have been made to quantify this and none for non-mammalian species. This study uses estimates of animal density and body mass culled from 12 journals published between 1961 and 1978 to test and extend Damuth's relationship between population density and body size of herbivorous mammals. In general, his analysis is supported, for density usually declines roughly as W -~ and poikilotherms maintain higher densities than homeotherms. However the residual variation is higher than Damuth's regressions might suggest and significant differences exist among animal groups. In particular, birds maintain much lower, and aquatic invertebrates much higher abundances than a general curve for all species would suggest. Carnivores are significantly rarer than herbivores. These relationships may be used to compare the average relative contributions of species of different size to community structure and function. Such relations also provide a necessary basis both for more complete empirical analyses of the determinants of animal abundance and for the construction of more realistic conceptual models in theoretical ecology.

A primary goal for ecology is prediction of the numerical density, biomass and biological activity of animal populations. Damuth (1979) took an important step in this direction when he showed the dramatic effect of body size on the average density of 307 species of herbivorous mammals. He found that body mass (W) alone explained 74% of the variation in population density and that density declined as W o.vs. Since biomass can be calculated as the product of individual size and population density, this implies that population biomass rises with body mass (i.e. biomass area- 1 =individual mass x density ~ W -~ x W= W~ Since individual biological rates rise approximately as W~ (Giinther 1975; Peters in press), this also implies that population activity per unit area, calculated as the product of individual rate and animal density, is unaffected by body size (i.e., rate area- 1 = rate individual- 1 x individuals area 1 ~ WO.Ts x W o.75 = WO). For Damuth (1979), the similarity in absolute value of the exponents of size in allometric equations for density and biological rate suggested a metabolic constraint on mammalian abundance. He further speculated that other
Offprint requests to: R.H. Peters

animals wouid be similarly constrained; this would imply that, in other animal groups, density would also decline approximately as W -~ and that poikilothermic species would maintain higher densities because of their lower rates of metabolism (Hemmingsen 1960; Robinson et al. 1983) and ingestion (Farlow 1976). Other investigations of the effect of body size on mammalian density have given more ambiguous support to the premise that density declines as W o.Ts. Mohr (1940) proposed that average density declines as W 1 for both carnivorous and herbivorous mammals. Mohr also reported differences of two orders of magnitude in the abundance of mammalian herbivores and carnivores of similar size. Since there is no such strong difference in individual metabolism associated with diet (McNab 1980), this difference in density would imply similarly great differences in both population metabolism and biomass among mammals as a function of trophic level. Peters and Raelson (in prep) also analyzed the effect of body size on the density of carnivorous and herbivorous mammals. They found, like Mohr (1940), that densities of predatory mammals declined approximately as W-1. This exponent also applied to a global relation for herbivores comparable to those discussed by Mohr (1940) and Damuth (1979). When herbivore populations were analyzed by geographic region, the exponent of mass was significantly less steep (i.e. closer to zero) than - I in all cases and, in all cases but one (temperate herbivores), it was significantly less steep than -0.75. Like Mohr (1940), Peters and Raelson also found that herbivorous mammals maintained higher densities than carnivorous mammals of similar size. There are, therefore, good reasons to suspect that Damuth's scheme may not be generally applicable. On the other hand, animal populations are subject to so many factors that biases within the data sets, not necessarily associated with size, may well modify the general trend. This is especially likely if the data sets are relatively small or cover only a limited range of body size, because even a small amount of bias in such data can markedly change the statistics generated by least squares regression. Given these sources of possible error and considering the potential importance of generalizations about animal abundance, Damuth's proposals should be examined as thoroughly as possible. The purpose of this paper is to test Damuth's argument by examining data relating both homeotherm and poikilotherm densities to body mass. Under Damuth's hypothe-

90 sis, all populations would have a slope of approximately - 0 . 7 5 and poikilotherms would be markedly more abundant than homeotherms. Under Mohr's suggestion, slopes would approximate - 1 . Even if no single pattern can be identified, such relationships may serve to indicate consistencies within different animal groups and to indicate the relative importance of other factors as residual variation. To this end, data describing density for birds, mammals, vertebrate poikilotherms and invertebrates were collected from the literature and regressed against body size.
Methods

These data were collected as part of a larger literature survey directed towards the construction of allometric relations. In all, 12 journals pulished between 1961 and 1978, inclusive, were surveyed. For present purposes, relevant journals are Acta Theriologica, Copeia, Ecological Monographs, Ecology, Journal of Animal Ecology, Journal of Mammalogy, Journal of Wildlife Management, Oecologia, Oikos, and Physiological Zoology. Comparative Physiology and Biochemistry and the Journal of Protozoology were also surveyed but yielded no data to address the present problem. This study was restricted to temperate populations, which effectively restricted it to North America and Europe. This restriction was applied because Peters and Raelson (in preparation) found substantial biogeographic differences in the densities of holarctic, paleotropical and neotropical mammals of similar size. Because tropical sites were only sparsely represented in the present collection, their inclusion might only increase variation in the data set. Tropical data were therefore excluded from the analysis. With this exception, no attempt was made to select the data both because judgements as to the adequacy of so diverse and extensive a set of observations are beyond our capacities and because we wished to represent fully the amount of variation in contemporary estimates. This approach therefore shows the full extent of our uncertainty in estimating animal abundance from body size because methodological shortcomings, specific characteristics, and environmental effects appear as residual variation. Although we attempted to include all relevant data from the surveyed journals, some data of interest were undoubtedly overlooked. Mean densities were calculated for each species in each paper. Reported means were preferred but, if these were not available, means of tabulated data were calculated. When only maximum and minimum densities were given, the midpoint of the range was taken as mean density. The extremes were also recorded to give a measure of the amount of variation which could be expected within a single study of one species. Average fresh body masses were taken from the same source, if available, or from other sources included in the survey if not. Any error introduced into the estimation of the independent variable by this approximation is small relative to variations in animal density and is unlikely to effect the analysis (Ricker 1973). Each datum was classed as invertebrate, poikilothermic vertebrate, mammal or bird. Mammals and birds were further subdivided into herbivores or carnivores (including omnivores). Many of the invertebrates could not be readily classed according to diet, but invertebrates were classified according to habitat as terrestrial or aquatic. All vertebrate

poikilotherms in the survey were carnivorous and, because there were so few vertebrate poikilotherms in the survey, no attempt was made to subdivide that class any further. In order to determine if each group required separate regressions or if some classes could be lumped together, analysis began by dividing the data set into two parts, poikilotherms and homeotherms coded as a " d u m m y " variable (C) with a value of unity or zero. Next, stepwise multiple regression of the decimal logarithm of animal density (D, in number km -z) against the dummy variable and the logarithm of body mass (W, in kg fresh mass) was performed to determine if the dummy variable explained a significant ( P < 0.05) proportion of the variation in D. The model employed was therefore: logao D loglo a+b loglo W + c C

(1)

where a, b and c are fitted constants. When the dummy variable proved significant, each subset was in turn regressed to determine if vertebrate and invertebrate poikilotherms required separate analyses and if birds and mammals should also be analyzed separately. Each homeotherm group was then analyzed as carnivore or herbivore and invertebrates as terrestrial or aquatic. This analysis, which is equivalent to an analysis of variance for difference in elevation, showed that all classes required different regressions. Consequently, predictive relations for each group were generated by regressing loglo D against loglo W as: loglo D = l o g l o a+b loglo W. (2)

This transforms to the familiar power equation widely used in allometry:

D=a W b.

(3)

For comparative purposes, regressions were also performed for all animals, for all birds, all mammals and all invertebrates. All analyses used a commercial statistical package which employs the principle of least squares (Hellwig and Council 1979).
Results and discussion

The data are too numerous to list here but are shown in Fig. 1. The results of regression analysis are shown in Table 1 which lists the information necessary to construct confidence limits around each regression.

The exponents of body mass

With the notable exception of birds, which are discussed below, animal density declines with increasing animal size. In other words, the exponents of body mass or slopes [b in Eq. (2) and (3)] are always significantly less than zero. One would expect this both from general knowledge and from previous work (Damuth 1979; Ghilarov 1967; Mohr 1940; Peters and Raelson in preparation). The slopes in Table 1 are typically somewhat more shallow (i.e. closer to zero) than - 1 , indicating that density falls somewhat more slowly than body mass rises. For vertebrate poikilotherms, mammals, and terrestrial invertebrates the 95% confidence limits include the --0.75 value favoured by Damuth (I979) and the upper limit for slope of aquatic invertebrates ( - 0 . 7 2 ) is not far off. The slope for all animals was significantly steeper than -0.75. Consequently

91

Aquatic invertebrates 1012

E 6
r

".,..:.:::. :.
~z~ z~ z~

9
9 o

Terrestrials Mammals Birds Vertebrate

invertebrates

10 8
9

>,.
IZ ILl C~
10 4

poikilotherms

9 9

Lx

zx zx
~ 9

Lx

Ao 9

zx=

z~

<
Z

..I
1.0 -0.98 ..
o

D = 32W r2= lo-4 o.81

I ! i

1 0- 9

10-6

1 0- 3

1.0

10 3

BODY MASS

(kg)

Fig. 1. The relationship between animal body size and species abundance. Data sources used in this figure are: for mammals; Barkalow et al. 1970; Bobek 1969, 1973; Brand 1976; Bujalska and Rydtkowski 1966; Chitty and Phipps 1966; Dunford 1977; Gorecki 1977; Goszczynski 1977; Grodzinski et al. 1966; Holmes and Sanderson 1965; Jezierski 1967; King 1975; Marburger and Thomas 1965; Mazurkiewicz 1972; McCaffery 1976; Meslow and Kieth 1971; Montgomery et al. 1975; Myrcha 1968; Nellis and Keith 1976; Nixon and McClain 1969; Nixon et al. 1974, 1975; Nosek et al. 1972; Petrusewicz et al. 1972; Puglisi et al. 1974; Rich9 1965; Robinette etal. 1973; Rusch etal. 1972; Rysekowski etal. 1973; Sanderson 1961; Smith etal. 1971; Smyth 1968; Storm 1972; Urban 1970; Wagner and Stoddart 1972; Wilkinson et al. 1976; for birds, Anderson and Stirling 1974; Anderson 1974; Boeker and Bolen 1972; Dunn 1977; Ell 1974; Emlen 1972; Hanson and Miller 1961; Miller and Hatfield 1974; Nellis et al. 1972; Rusch and Keith 1971; Rusch et al. 1972; Yeaton 1974; for vertebrate poikilotherms, Barbault 1976; Burton and Likens 1975; Griffiths 1973; Heckrotte 1967; Kennedy 1977; Paragamian and CoNe 1975; for invertebrates: Abrahamson and Goldman 1970; Benke 1976; Block 1966; Brian et al. 1965, 1967; Browne 1978; Burnett 1970; Dixon 1966; Griffiths 1973; Hinton 1971; Jacobs 1977; Jennings and Barkhan 1975; Kowal and Crossley 1971; Lebrun 1969; Mason 1970; Mispagel 1978; Negus 1966; Nielson 1961; Nosek et al. 1972; Paine 1976; Phillipson et al. 1977, 1978; Sunderland et al. 1976; Sutton 1968; Ward 1975; Wiegert 1965; Williamson et al. 1977

Table 1. Statistics describing regression of loglo density (D, animals km -2) on loglo body size (IV, kg fresh mass). Listed values are the intercept (log a) and its standard error, the slope (b) and its standard error, the coefficient of determinism (r2), the geometric means of the independent (log ~ and dependent (log /5) variables, the standard deviation of log D (SD.w) , the sum of squares of deviations in log W (Ztog w2) the number of observations (n) and the range of the independent variable, log W Group Animals Mammals Herbivorous mammals Carnivorous mammals Birds Herbivorous birds Carnivorous birds Vertebrate poikilotherms Invertebrates Aquatic invertebrates Terrestrial invertebrates log a_+SE 1.51 +0.14 1.99_+0.13 2.33-+0.08
1.11 -+ 0 . 3 2

b,+SE -0.98+0.029 -0.77+0.09 -0.61 -+0.07 -0.94-+0.26 -0.19-+0.14 0.21 -+0.20 -0.52,+0.18 -0.77-+0.30 -0.54-+0.038 --0.58-+0.07 --0.69-+0.04

r2 0.81 0.57 0.62 0.56 0.03 0.05 0.18 0.42 0.55 0.54 0.72

log W -4.1 -0.44 -0.48 0.23 -1.3 -1.0 -1.4 -2.8 -6.4 -5.1 -7.2

iog D 5.5 2.3 2.6 0.89 1.0 1.2 0.96 4.0 18.4 8.3 18.4

SD. w

Zlog w2 2330 100 83.6 17.9 25.7 9.9 15.1 32.1 838.0 294.0 484.0

n 91 57 68 12 60 22 38 10 62 56 06

Range - 1 1 to 2.3 - 2 to 2.3 -2.1 to 2.3 - 2 to 1.0 - 2 to 0.6 - 2 to 0.6 - 2 to 0.6 - 4 to 0 - 1 1 to - 1 - 7 to - 1 - 1 1 to - 2

1.4 0.90 0.64

1.i

0.79_+0.20 1.40-+0.24 0.25_+0.28 1.81 - + 0 . 9 8 4.93-+0.26 5.37_+0.39 3.47_+0.30

0.71 0.63 0.70 1.7 1.1 1.2 0.88

92

1012 2=r

6
10 8

~
10 4

Aquatic

invertebrates

>1O9 Z LU
J

~
Carnivorous

ertebrate

polk

<

z <

m
1.0
i i

~~'~ marnmals~
i i i .3arni .....

Heri~; ~i
....

~ I
1.0

S I
10 3

I
1(3 3

Fig. 2. Confidence limits around the mean trends describing the effect of animal size on species abundance for significantly different groupings in Table 1. The solid line indicates the mean trend for all animals. Two panels have been used to separate the various vertebrate curves

169

16 6

ld 3 BODY

1.0

1 03

MASS

(kg)

the uncertainty associated with any single estimate of slope is usually large enough that Damuth's hypothesis cannot be rejected although some departures from this value should be expected. The slopes for carnivorous mammals and birds are steeper than those for corresponding herbivores. This can be explained by reference to the rates of production of the potential prey populations relative to the demands of their predators. Larger predators eat larger prey (Hespenheide 1973; Rosenzweig 1966; Schoener 1968). If the curve for all animals is taken to reflect potential prey density, then the biomass of each prey would be essentially independent of prey size. However, since production per unit mass declines with increasing size (Banse and Mosher 1980; Blueweiss et al. 1978; Farlow 1976), the rate of production of each prey population must decline too. Thus, at steady state, each population of larger prey supplies less resource to their larger predators per unit time. This shortfall is compounded by the rapid decline in number of species of larger size (May 1978). The net effect is to provide a decreasing resource base to larger carnivores. Table I gives a different picture for birds. Both the regression for all birds and that for herbivorous birds do not explain a significant amount of variation in avian density and the slopes in these regressions are not significantly different from zero. These groups can be as effectively described by the mean observed densities (11 km -2 and 15 k m - 2 respectively) as by the regressions. Even the weak relationship for carnivorous birds exists only because of the low density of a few large raptors. Newton (1979) has convincingly shown that raptor density declines with greater size. The lack of response for birds must be interpreted cautiously. Almost all avian data for this analysis were taken from a single study of overwintering birds in Texas (Emlen 1972) and it is certainly possible that those sites or species were not representative. In addition, the limited size range of birds in this survey makes identification of any trend less probable. However, the size range of carnivorous mammals is only slightly wider yet a highly significant relationship was identified in that case. The avian relationships would bear further analyses with more extensive data.

Elevations of the regressions

Discussion of the relative contributions of the different groupings in Table 1 must also consider variations in the elevations of the curves. The simplest statistic for such comparisons is the intercept but these can only be compared when the intercept lies close to the mean of the data collected. This is the case for mammals. Global and North American estimates of the biomass of herbivorous mammals range from 95 individuals km -2 (Damuth 1979) to 500 (Mohr 1940), so the value reported here, 214, is in rough agreement. Mammalian carnivore densities range from 1 (Mohr 1940) to 36 (Peters and Raelson, in preparation) again bracketing the value reported here. Further discussion of the relative contributions of the different groupings in Table 1 is based on the elevation of each curve at the midpoint of each regression. This must be compared with some standard which is here provided by the general regression for all data solved at the average body mass of each group. In order to present a symmetrical value, the logarithm of the ratio of the mean group density to the density predicted by the general regression was calculated. Thus, when the density of the subgroup lies an order of magnitude below the general line the ratio's logarithm is - i. When the subgroup density is ten times greater than the general regression, the logarithm is 1. When the logarithms of the ratios are compared in this way (Table 2), it is apparent that the mean of most groups lies within an order of magnitude of the general curve. Carnivorous mammals, and vertebrate poikilotherms lie somewhat below the general curve (Fig. 2); terrestrial invertebrates and herbivorous mammals lie somewhat above it. In contrast, all avian densities lie between one and two orders of magnitude below the line and aquatic invertebrates almost two orders of magnitude above it. Given the scatter of individual points (Fig. 1) and the substantial standard deviations associated with the mean values (Table 1), one is tempted to use the general curve for groups other than aquatic invertebrates or birds, since only an approximation can be provided at best. This similarity in the density of most groups is also

93 Table 2. Comparisons of animal densities at the mean body size of each animal group (P/, kg) as predicted by the specific relations for density (/), number km -2) and as predicted by the general relation for all animals (/), number kin-x). Population biomass (B, kg km -2) is calculated as /)-W, and population respiration rate (R_, in Watts km -2) is calculated as B R where R is the specific standard metabolic rate (Watts kg-1). For homeotherms, R=0.131 e~ -~ and for poikilotherms R=0.071 e~176 0.2~ where T is body temperature in degrees celsius (Robinson et al. 1983) Group Animals Mammals Herbivorous mammals Carnivorous mammals Birds Herbivorous birds Carnivorous birds Vertebrate poikilotherms Invertebrates Aquatic invertebrates Terrestrial vertebrates L5 3.2 x 105 210 490 7.9 11 15 9.2 9600 2.5 x 108 2.2 x l0 s 2.9 x 10s /) 3.2 x 105 85 120 19 550 330 700 19000 6.4 x 107 3.2 x 106 3.8 x l0 s LS//) 1 2.5 4.0 0.42 0.020 0.046 0.013 0.51 3.9 67 0.75 log(/)//)) 0 0.39 0.61 -0.38 -1.7 - 1.3 - 1.9 -0.29 0.59 1.8 -0.12 l~ 8.4 x 10-5 0.37 0.26 1.7 0.055 0.092 0.043 0.0015 3.7 x 10- 7 7.9 x 10 .6 6.0 x 10 s B 26 78 130 46 1.5 1.4 0.40 14 93 1700 17 R 1.9 4.4 4.7 3.2 8.5 7.6 9.0 0.94 6.9 3.3 Rp 49 340 600 150 5.1 11 3.6 13 640 5700 190

11

a p p a r e n t when the biomass o f the average p o p u l a t i o n is calculated (Table 2). Excepting birds and aquatic invertebrates, average biomass only ranges from 14 kg km - 2 for vertebrate poikilotherms to 126 kg k m -2 for herbivorous mammals. The biomass o f the average bird population, however, is less than one tenth that of vertebrate poikilotherms, and that o f the average aquatic invertebrate is almost ten times higher than herbivorous mammals. Thus in terrestrial systems, n o n - a v i a n p o p u l a t i o n s contribute approximately equally to c o m m u n i t y biomass. This will be modified a m o n g animals which are larger or smaller than the mean for each group but, as Fig. 2 shows, these differences are likely to be substantial only for larger herbivorous m a m m a l s and terrestrial invertebrates. Iverson (1982) lists a series o f m a x i m u m biomass estimates for single species which m a y be qualitatively comp a r e d with the trends in Table 2. His d a t a show that the m a x i m u m biomass for vertebrate poikilotherms are usually less than those for homeotherms and that birds maintain much lower biomasses than mammals. These trends would be expected from Table 1. His values for average turtle biomasses, however, are considerably higher than would be suggested by the regression for poikilotherm vertebrates (Table 1). W i t h o u t m o r e data, it is difficult to ascribe this difference to biases in the small set o f d a t a from poikilothermic vertebrates used here, to real differences between Iverson's largely aquatic animals and these, or to some unique characteristics o f turtles. Certainly the b r o a d limits for vertebrate poikilotherms (Fig. 2) leave considerable r o o m for improvement.

Variation in elevation
Finally, some c o m m e n t should be made regarding sources o f variation in these data. The s t a n d a r d deviation in the

d a t a at the mean density and body mass of the collection (SDw in Table 1) is an a p p r o p r i a t e measure o f scatter in the data. F r o m this statistic, one can calculate confidence limits for individual estimates as + t e S~y. In other words, a p p r o x i m a t e l y 95% o f all d a t a points lie within the limits defined as the mean + 2 SDW. Since these analyses refer to d a t a transformed to decimal logarithms, 4 SDw is an estimate o f the number o f orders o f magnitude between the upper and lower confidence limits. This ranges from 2.5 (330 times) for herbivorous birds to 5.8 (575,000 times) in the case o f all animals. It should be stressed that these wide estimates apply to predictions for individual points, confidence limits for the mean value in equally large d a t a sets are much smaller. In the case for all animals, the confidence limits a r o u n d the predicted mean differ by only (4 SDw ~ i / n = ) 0.38 orders of magitude or 2.2 times. The average value of SDW in Table 1 is 0.93 which corresponds to a 3.7 order of magnitude difference (5000 fold) between the upper and lower confidence limits a r o u n d individual points. This variation can be attributed to any or all of a long list o f factors addressed in p o p u l a t i o n biology: weather, food, other organisms, habitat ( A n d r e w a r t h a 1961), biotic potential, age structure, death rate, equilibrium level ( K o r m o n d y 1969), density dependent a n d independent factors (Lack 1954) and so forth. This study does n o t address these variables, but it can give some measure o f the a m o u n t o f variation to be expected in a single study and that which m a y be attributed to variation a m o n g species, sites, investigators, and methods. To this end, the average value o f the range from loglo (maximum D) to loglo (minimum D) was calculated for invertebrates (0.979, n = 56), birds (0.842, n = 85) and m a m m a l s (0.982, n = 37). The average over all species was 0.935 ( S D = 0 . 6 9 ) orders o f magnitude (10 fold). Thus, only one quarter o f the average variation o f a b o u t 3.7 orders of magnitude can be attributed

94 to variations within single populations. This must be addressed by studying local and temporal factors within populations. The remainder will require examination of differences among sites, species, and methods. The logarithmic transformation used in these regressions tends to stabilize the variance around the regressions (Fig. 1) and implies that variation is a roughly constant proportion of the mean. If the data are " b a c k transformed" to their antilogarithms, variation would rise with mean density. Comparison of the mean and variance of the untransformed data would therefore suggest that populations of smaller organisms, which have higher means and even higher variances, are more aggregated than those of larger species. As a result, smaller species may appear more patchily distributed than large. Platt and Denman (1980) suggest that a similar size dependent trend will exist in temporal patches, at least in pelagic ecosystems. populations of mammals and terrestrial invertebrates contribute similarly to community processes. Vertebrate poikilotherms have relatively low individual rates despite their sizable densities because the specific metabolism of larger poikilotherms is quite small. Birds too are minor contributors to community process, despite their high specific rates, because they maintain such low densities. Aquatic invertebrates maintain large biomass and high specific metabolic rates because of their small size. Together, these produce very large population metabolic rates.

The role of allometric relations

Allometric relations serve as standards of comparison for measured values. For example, a field biologist who has estimated the density of a given population may wonder if the study species is particularly common or rare. This involves comparison with some 'usual value'. Such a value is given by allometric relationships. Alternatively one may need to approximate community biomass from some minimal piece of information. This can be calculated from the frequency of the species sizes in the community and regressions of density on species size. Given the paucity of methods to obtain such values, allometric relations provide the best alternative. It is also important to see these regressions as starting points for more comprehensive empirical relationships involving other independent variables. One might expect, for example, that site productivity, species diversity, and climatic stability are also determinants of population density. However, because body size explains so much of the variance in abundance (Table 1), deeper analysis must also treat the effects of size or risk confounding the influence of other, more ecological variables. Fortunately, computer packages have rendered such multifactorial analysis relatively easy. Finally, these regressions contribute to a growing body of empirical relationships which describe the gross patterns of nature. Further conceptual argument and speculation can then be restricted within these empirical bounds; this would seem an effective vehicle for the combination of contemporary 'theoretical ecology' with the results of observation and experiment.

Population contributions to community biomass

The relative constancy of the slopes in Table 1 may indicate a basic similarity among animal communities. The average contribution of individual populations to the biomass (the structure) of the community can be calculated as the product of body mass and the density-body size relation. As the slopes in these relations usually lie between - 0 . 5 and - 1 , biomass should vary as W~ to W~ Excluding birds, the mean of all slopes in Table 1 is only -0.74. This suggests that biomass should rise as W~ However, the wide scatter in Fig. 1 and the variation among slopes in Table 1 shows that this average contribution is subject to a great deal of uncertainty. The significantly steeper slope for all organisms shows that these considerations apply, at best, only to groups of similar organisms and the low slopes for birds suggests that even similar organisms need not follow this average pattern.

Contributions to community processes and functions

As indicated in the introduction, the contribution of an average population to the rate of a community process may be calculated as the product of density and individual rate. In a survey of 297 allometric relations describing various biological rates, Peters (in press) found that, on average, individual rate rises as W~ Thus, average population rates should be independent of size, as suggested by Damuth (1979). Again this lack of trend would be a statistical average and much variation can be expected around it. The contribution of the average populations to community process can also be represented as the product of animal biomass and specific rate of that process. Since most biological rates are proportional to metabolic rate (Hemmingsen 1960; Humphreys 1979; Kajak 1980; Lavigne 1982; Peters in press), the contribution of an average population to community process is proportional to the population's metabolic rate. Consequently, the specific standard metabolic rate for an animal of mean body mass for each grouping was calculated using regressions relating metabolic rate to body mass and temperature (Robinson et al. 1983) assuming average body temperatures of 20 ~ C for poikilotherms, 38 ~ C for mammals, and 41 ~ C for birds. Population metabolism was then calculated as the product of biomass and specific metabolism. These calculations (Table 2) suggest that, on average,

Acknowledgements. This study was financed by funds from the National Research Council of Canada and the McGill Computing Centre.
References

Abrahamson SAA, Goldman CR (1970) Distribution, density and production of the crayfish Pacifastacus leniusculus Dana in Lake Tahoe, California - Nevada. Oikos 21 : 83-91 Anderson DR, Sterling RT (1974) Population dynamics of molting pintail drakes banded in south central Saskatchewan. J Wildl Manag 38:266-274 Anderson M (1976) Populatin ecology of the long-tailed skua (Stereorarius longicaudus Vieill). J Anita Ecol 45:537-559 Andrewartha HG (1961) An introduction to the study of animal populations. University of Chicago Press, Chicago Banse K, Mosher S (1980) Adult body mass and annual production/biomass relationships of field populations. Ecol Monogr 50: 355-379 Barbault R (1976) Population dynamics and reproductive patterns of three African skinks. Copeia 76:483-490 Barkalow FS, Hamilton RB, Soots RF (1970) The vital statistics of an unexploited gray squirrel population. J Wildl Manag 34: 489-500

95 Benke AC (1976) Dragonfly production and prey turnover. Ecology 57:915-927 Block W (1966) Seasonal fluctuations and distribution of mite populations in moorland soils, with a note on biomass. J Anita Ecol 35 : 487-503 Blueweiss L, Fox H, Kudzma V, Nakashima D, Peters R, Sams S (1978) Relationships between body size and some life history parameters. Oecologia (Berlin) 37:257-272 Bobek B (1973) Net production of small rodents in a deciduous forest. Acta Theriol 18 : 403-434 Bobek B (1969) Survival, turnover and production of small rodents in a beech forest. Acta TherioI 14:191-2t0 Boeker EL, Bolen EB (1972) Winter golden eagle populations in the south west. J Wildl Manag 36:477-484 Brand CJ, Keith LB, Fisher CA (1976) Lynx responses to changing snowshoe hare densities in central Alberta. J Wildl Manag 40 :416-428 Brian MV, Elnes G, Kelley AF (1967) Populations of the ant Tetramorium caespitum Latreille. J Anim Ecol 36:337-342 Brian MV, Hibble J, Stradling DJ (1965) Ant pattern and density in a southern English heath. J Anim Ecol 34:545-555 Browne RA (1978) Growth, mortality, fecundity, biomass and productivity of four lake populations of the prosobranch snail, Viviparus georgianus. Ecology 59 : 742-750 Bujalska G, Rydtkowski L (1966) Estimation of the reproduction of the bank vole under field conditions. Acta Theriol 11 : 351-361 Burnett T (1970) Effect of simulated natural temperatures on an acarine predator-prey population. Physiol Zool 43:155-165 Burton TM, Likens GE (1975) Salamander populations and biomass in the Hubbard Brook experimental forest, New Hampshire. Copeia 75 : 541-546 Chitty D, Phipps E (1966) Seasonal changes in survival in mixed populations of two species of vole. J Anita Ecol 35:313331 Damuth J (1981) Population density and body size in mammals. Nature 290 : 699-700 Dixon AFG (1966) The effect of population density and nutritive status of the host on the summer reproductive activity of the sycamore aphid, Drepanosiphum platanoides (Schr.). J Anim Ecol 35 : 105-112 Dunford C (1977) Behavioral limitation of round-tailed ground squirrel density. Ecology 58:1254-1268 Dunn E (1977) Predation by weasels (Mustela nivalis) on breeding tits (Parus spp.) in relation to the density of tits and rodents. J Anim Ecol 46:633-652 Ellison LN (1974) Population characteristics of Alaskan spruce grouse. J Wildl Manag 38:383 395 Emlen JT (1972) Size and structure of a wintering avian community in southern Texas. Ecology 53:317 329 Farlow JO (1976) A consideration of the trophic dynamics of a late Cretaceous large-dinosaur community (Oldman Formation). Ecology 57:841-857 Ghilarov MS (1967) Abundance, biomass and vertical distribution of soil animals in different zones p. 611-630. In: Petrusewicz K (ed). Secondary productivity of terrestrial ecosystems. Institute of Ecology. Polish Academy of Sciences Warsaw Gorecki A (1977) Energy flow through the common hamster population. Acta Theriol 22: 25-66 Goszczynski J (1977) Connections between predatory birds and mammals and their prey. Acta Theriol 22:399-430 Griffiths D (1973) The structure of an acid moorland pond community. J Anita Ecology 42:263-283 Grodzinski W, Gorecki A, Janas J, Migula P (1966) Effects of rodents on the primary productivity of alpine meadows in Bieszczady Mountains. Acta Theriol 11:419-431 Gfinther B (1975) Dimensional analysis and theory of biological similarity. Physiol Rev 55:659-699 Hanson WR, Miller RJ (1961) Edge types and abundance of bob~ whites in southern Illinois. J Wildl Manag 25:71-76 Heckrotte C (1967) Relations of body temperature, size, and crawling speed of the common garter snake, Thamnophis s. sirtalis. Copeia 67 : 759-763 Hellwig JT, Council KA (1979) (eds) SAS user's guide, 1979 edition. SAS Institute Inc. Cary. N.C. Hemmingsen AM (1960) Energy metabolism as related to body size and respiratory surfaces, and its evolution. Rep. Steno Hosp. 9:1-110 Hespenheide HA (1973) Ecological inferences from morphological data. Ann Rev Ecol Syst 4:213-219 Hinton JM (1971) Energy flow in a natural population of Neophilaenus lineatus (Homoptera). Oikos 22:155-17t Holmes ACV, Sanderson GC (1965) Populations and movements of opossums in east-central Illinois. J Wildl Manag 29 : 287-295 Humpbreys WF (1979) Production and respiration in animal populations. J Anita Ecol 48:427-453 Iverson JB (1982) Biomass in turtle populations; a neglected subject. Oecologia (Berlin) 55 : 69-76 Jacobs J (1977) Coexistence of similar zooplankton species by differential adaptation to reproduction and escape in an environment with fluctuating food and energy densities. II. Field data analysis of Daphnia. Oecologia (Berlin) 30:313-329 Jennings TJ, Barkham JP (1975) Slug populations in mixed deciduous woodland. Oecologia (Berlin) 20:279-286 Jezierski W (1967) Four cases of "Home Instinct" in the European hare. Acta Theriol 12:173-175 Kajak A (1980) Invertebrate predator subsystems, p. 539-589 In: Breymeyer AI, Van Dyne GM (eds). Grasslands systems analysis and man. Cambridge University Press. London Kennedy SE (1977) Life history of the Leon Springs pupfish, Cyprinodon bovinus. Copeia 77:93-103 King CM (1975) The home range of the weasel (Mustela nivalis) in an English woodland. J Anita Ecol 44:639-668 Kormondy EJ (1969) Concepts of ecology. Prentice-Hall Inc. Englewood Cliffs, New Jersey. 209p Kowal NE, Crossley DA (1971) The ingestion rates of microarthropods in pine mor, estimated with radioactive calcium. Ecology 52: 444-452 Lack D (1954) The natural regulation of animal numbers. Oxford University Press, London Lavigne DM (1982) Similarity in energy budgets of animal populations. J Anim Ecol 51 : 195-206 Lebrun PH (1969) Ecologic et Biologie de Nothrus palustris (C.L. Koch) (Acari Oribatei) - Densit~ et structure de la poulation. Oikos 20: 34-40 Margurger RG, Thomas JW (1965) A die-off in white-tailed deer of the central mineral region of Texax. J Wildl Manag 29 : 706-716 Mason CF (1970) Snail populations, beech litter production and the role of snails in litter decomposition. Oecologia (Berlin) 5:215-239 May RM (1978) The evolution of ecological systems. Sci Am 239 (3):16~175 Mazurkiewicz M (1972) Density and weight structure of populations of the bank vole in open and enclosed areas. Acta Theriol 17:455-465 McCaffery KR (1976) Deer trail counts as an index to populations and habitat use. J Wildl Manag 40 : 308-316 McNab BK (1980) Food habits, energetics and the population biology of mammals. Am Nat 116 : 106-124 Meslow EC, Keith LB (1971) A correlation analysis of weather versus snowshoe hare population parameters. J Wildl Manag 35:1-15 Miller RS, Hatfield JP (1976) Age ratios of sandhill cranes. J Wildl Manage 38:234-242 Mispagel ME (1978) The ecology and bioenergetics of the acridid grasshopper, Bootettix punctatus on creosotebush, Larrea tridentata, in the northern Mojave desert. Ecology 59 : 779-788 Mohr CO (1940) Comparative populations of game, fur and other mammals. Am Midl Nat 24:581-584 Montgomery SD, Whelan JB, Mosby HS (1975) Bioenergetics of a woodlot gray squirrel population. J Wildl Manag 39:709 717

96 Myrcha A (1968) Caloric value and chemical composition of the body of the European hare. Acta Theriol 13:65-71 Negus CL (1966) A quantitative study of growth and production of unionid mussels in the River Thames at Reading. J Anim Ecol 35 : 513-532 Nellis CH, Keith LB (1976) Population dynamics of coyotes in central Alberta 1964-68. J Wildl Manag 40:389-399 Nellis CH, Wetmore SP, Keith LB (1972) Lynx-prey interactions in central Alberta. J Wildl Manag 36 : 320-329 Nielson CO (1961) Respiratory metabolism of some populations of enchytraeid worms and free-living nematodes. Oikos 12:17-35 Nixon CM, McClain MW (1969) Squirrel population decline following a late spring frost. J Wildl Manag 33 : 353-357 Nixon CM, McClain MW, Donohoe RW (1975) Effects of hunting and mast crops on a squirrel population. J Wildl Manag 39:1-25 Nixon CM, Donohoe RW, Nash T (1974) Overharvest of fox squirrels from two woodlots in Western Ohio. J Wildl Manag 38: 67-80 Nosek J, Kozuch O, Chmela J (1972) Contribution to the knowledge of home range in common shrew, Sorex araneus. Oecologia (Berlin) 9 : 59-63 Newton I (1979) Population ecology of raptors. Buteo books Vermillion, S.D. Paine RT (1976) Size-limited predation: an observational and experimental approach with the Mytilus-Pisaster interaction. Ecology 57:858-873 Paragamian VL, Coble DW (1975) Vital statistics of smallmouth bass in two Wisconsin rivers, and other waters. J Wildl Manag 39:201-210 Peters RH (in press) The ecological implications of body size. Cambridge University Press, New York Petrusewicz K, Markov G, Gliwicz J, Christov L (1972) A population of Clethrionomys glareolus pirinus on the Vitosha Mountain, Bulgaria. IV Production. Acta Theriol 17:347-353 Phillipson J, Abel R, Steel J, Woodell SRJ (1977) Nematode numbers, biomass and respiratory metabolism in a beech woodland - Wytham Woods, Oxford. Oecologia (Berlin) 27 : 141-155 Phillipson J, Abel R, Steel J, Woodell SRJ (1978) Earthworm numbers, biomass and respiratory metabolism in a beech woodl a n d - Wytham Woods, Oxford. Oecologia (Berlin) 33:291-309 Puglisi MJ, Lindzey JS, Bellis ED (1974) Factors associated with highway mortality of white-tailed deer. J Wildl Manag 38 : 799-807 Richens VB (1965) An evaluation of control on the Wasatch pocket gopher. J Wildl Manag 29:413-425 Ricker WE (1973) Linear regressions in fishery research. J Fish Res Board Can 30:409-434 Robinette WL, Baer CH, Pillmore RE, Knittle CE (1973) Effects of nutritional change on captive mule deer. J Wildl Manag 37:312-326 Robinson WR, Peters RH, Zimmerman J (1983) Body size and temperature effects on metabolic rates of organisms. Can J Zool 61 : (in press) Rosenzweig ML (1966) Community structure in sympatric Carnivora. J Mammal 47 : 602-612 Rusch DH, Keith LB (1971) Seasonal and annual trends in numbers of Alberta ruffed grouse. J Wildl Manag 35:803-822 Rusch DH, Meslow EC, Doerr PD, Keith LB (1972) Response of great horned owl populations to changing prey densities. J Wildl Manag 36:282-296 Rysekowski L, Goszozynski J, Truszkowski J (1973) Trophic relationships of the common vole in cultivated fields. Acta Theriol 18:125-165 Sanderson GC (1961) Estimating opossum populations by marking young. J Wildl Manag 25 :20-27 Schoener TW (1968) Sizes of feeding territories among birds. Ecology 49:123-141 Smith MH, Blessing R, Chelton JG, Gentry JB, Golley FB, McGinnis JT (1971) Determining density for small mammal populations using a grid and assessment lines. Acta Theriol 16:105-125 Smyth M (1968) The effects of the removal of individuals from a population of bank voles. Clethrionomys glareolus. J Anim Ecol 37:167-183 Storm GL (1972) Daytime retreats and movements of skunks on farmlands in Illinois. J Wildl Manag 36:3145 Sunderland KD, Hassall M, Sutton SL (1976) The population dynamics of Philoseia museorum (Crustacea, Oniscoidea) in a dune grassland ecosystem. J Anim Ecol 45:487-506 Sutton SL (1968) The population dynamics of Trichoniscus pusillus and Philoscia muscorum (Crustacea, Oniscoidea) in limestone grassland. J Anim Ecol 37:425-444 Urban D (1970) Racoon populations, movement patterns, and predation on a managed waterfowl marsh. J Wildl Manag 34: 372-382 Wagner FH, Stoddart LC (1972) Influence of coyote predation on black-tailed jackrabbit populations in Utah. J Wildl Manag 36:329-342 Ward JV (1975) Bottom fauna substrate relationships in a northern Colorado trout stream: 1945 and 1974. Ecology 56:1429-1434 Wiegert RG (1965) Energy dynamics of the grasshopper populations in old field and alfalfa field ecosystems. Oikos 16:161-176 Wilkinson PF, Shank CC, Penner DF (1976) Muskox-Caribou summer range relations on Banks Island, N.W.T. J Wildl Manag 40:151-162 Williamson P, Cameron RAD, Carter MA (1977) Population dynamics of the land snail Cepaea nemoralis L. : A six year study. J Anita Ecol 46:181-194 Yeates GW (1973) Nematoda of a Danish beech forest. II production estimates. Oikos 24:179-185 Yeaton RI (1974) An ecological analysis of chaparral and pine forest bird Communities on Santa Cruz Island and mainland California. Ecology 55: 959-973

Received April 28, 1983