The University of Chicago
Spatial Scaling of Species Composition: Body Masses of North American Land Mammals
Author(s): James H. Brown and Paul F. Nicoletto
Reviewed work(s):
Source: The American Naturalist, Vol. 138, No. 6 (Dec., 1991), pp. 1478-1512
Published by: The University of Chicago Press for The American Society of Naturalists
Stable URL: http://www.jstor.org/stable/2462557 .
Accessed: 27/09/2012 13:49
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
.
The University of Chicago Press, The American Society of Naturalists, The University of Chicago are
collaborating with JSTOR to digitize, preserve and extend access to The American Naturalist.
http://www.jstor.org
Vol. 138, No. 6
The AmericanNaturalist
December 1991
SPATIAL SCALING OF SPECIES COMPOSITION: BODY MASSES OF
NORTH AMERICAN LAND MAMMALS
JAMES H. BROWN AND PAUL F. NICOLETTO
Departmentof Biology, Universityof New Mexico, Albuquerque, New Mexico 87131
SubmittedJuily24, 1989; Revised October 15, 1990; Accepted November2, 1990
mammalian
Abstract.-We describe the nonrandomassemblyof the NorthAmericanterrestrial
of body masses among
fauna based on body size and spatial scale. The frequencydistribution
species forthe entirecontinentalfaunawas highlymodal and rightskewed,even on a logarithmic
scale; the median size of the 465 species was approximately45 g. In contrast,comparable
frequencydistributionsfor24 small patches of relativelyhomogeneoushabitatwere essentially
uniform,withapproximatelyequal numbersof species in each logarithmicsize class; the median
sizes of the 19-37 species ranged fromapproximately100 to 2,500 g. Frequency distributions
for 21 biomes (large regions of relativelysimilarvegetation)were intermediatebetween the
continentaland local assemblages. This patternof assemblyindicatesthatspecies of modal size
(20-250 g) tend not to coexist in local habitatpatches and theyreplace each other more frequentlyfromhabitat to habitatacross the landscape than species of relativelylarge or small
to produce
size. We hypothesizethat three mechanismsare necessary and possibly sufficient
this result: competitiveexclusion of species of similarsize withinlocal habitats,differential
extinctionof species of large size with small geographicranges, and greaterspecialization of
modal-sized species owing to energeticand dietaryconstraints.
The diversityand compositionof biotas vary with spatial scale. The increase
in species richnesswithsample area has been quantifiedby species-area relationships for nonisolated sites withincontinents,as well as for islands or insular
habitatpatches and continentsof varyingsizes (see, e.g., MacArthurand Wilson
1967; Flessa 1975; Schoener 1976; Connor and McCoy 1979; Brown and Gibson
1983; Brown 1986). Here, we address the question of whetherthere are other
changes in the compositionof the biota as species diversityincreases with increasing spatial scale.
Because different
scales (see,
processes affectbiotic compositionon different
e.g., Orians 1980; Ricklefs 1987), we would expect these processes to resultin
predictable changes in the attributesof species. On the largest scales, entire
continentsor large geographicregionswithincontinents,species-levelprocesses
such as colonization,extinction,and speciationaffectbiotic composition.On the
smallestscales, withinsmall patches of homogeneoushabitat,ecological interactions of species with each other and with the abiotic environmentdetermine
whichcombinationsof species coexist. It is also on these small scales thatmicroevolutionaryprocesses of naturalselectionand geneticdriftoperate withinpopulations. On intermediatescales, both macroscopic and microscopic processes
cause changes in species compositionacross the landscape. However, processes
Am. Nat. 1991. Vol. 138, pp. 1478-1512.
CO1991 by The Universityof Chicago. 0003-0147/91/3806-0009$02.00.
All rightsreserved.
BODY SIZE AND SPECIES COMPOSITION
1479
80
70
70
60-
50.
o
E
403
30
20
10
0
2
4
6
8
10
12
14
Lo92 BodyMass (g)
16
18
20
FIG. 1.-Frequencydistribution,
on a log)scale,ofthebodymassesofthe465speciesof
modal,right-skewed
shape.
terrestrial
NorthAmericanmammals.Note thecharacteristic
operatingon disparate scales are partiallycoupled. Macroscopic biogeographic
and evolutionaryprocesses affectmicroscopiccommunitystructurebecause local
ecological communitiesare assembled from continentaland regional species
pools. Conversely,microscopicecological and evolutionaryprocesses affectthe
compositionof continentaland regionalassemblages because large-scale biotas
reflectthe cumulative effectsof phenomena that occur in many local communities.
In this article we use the distributionof body sizes among species of North
Americanland mammalsto assess one aspect of variationin faunalcomposition
withspatial scale. Body size is an easily measuredvariable thatis closely correlated with many aspects of morphology,physiology,behavior, and ecology
throughallometricrelations (Peters 1983; Calder 1984; Schmidt-Nielsen1984;
Zeveloffand Boyce 1988). In 1959, Hutchinsonand MacArthurcalled attention
of body sizes among species of NorthAmericanmammals(fig.
to the distribution
1). On a logarithmicscale, this distributionis highlymodal and skewed toward
have since been foundfora
largerbody sizes. Qualitativelysimilardistributions
wide range of organisms,frombirds and insects (May 1978, 1988; Morse et al.
1988) to bacteria (Bonner 1988).
of body sizes among species
We characterizethe variationin the distributions
withrespect to spatial scale by comparingthe frequencydistributionsforthree
scales: the entireNorth Americancontinent,regionalbiomes, and local habitat
patches. We reject the null hypothesisthat the sizes of mammalsco-occurring
on successivelysmallerspatial scales are randomsubsamplesofthelargerspecies
pools. We propose mechanistichypothesesto account forthe observed pattern
of faunalassemblage.
METHODS
NorthAmericanmammalsfor
We compiled species lists (App. A) of terrestrial
threedifferent
spatial scales: (1) the entireNorthAmericancontinent,including
Mexico, (2) 21 biomes as definedand mapped by Dasman (1975), and (3) 24 small
patches of relativelyhomogeneoushabitat.The NorthAmericanspecies list was
1480
THE AMERICAN NATURALIST
taken fromHall (1981) and supplementedwithMexican species fromRamirezPulido et al. (1986). Species listsforthe biomes were compiledby usingtherange
maps in Hall (1981) to determinethe occurrenceof species in the biomes mapped
by Dasman (1975). Species lists for the local habitatswere obtained fromthe
literatureor fromcolleagues who contributedunpublisheddata fromintensive
fieldstudies. We triedto ensure thatthe local habitatsrepresenteda small area
(usually 10-1,000 ha) of uniformgeology and vegetationand that the faunal list
included all species thatutilizedthe habitatat the timeof colonizationby European humans. The analyses included all species of native mammalsexcept bats,
pinnipeds,cetaceans, and the sea otter.
A singlevalue of body mass was assignedto each species whereveritoccurred;
we ignored intraspecificgeographicvariation.The body masses were obtained
fromfieldguides (Burtand Grossenheider1976; Whitaker1980),fromG. Ceballos
(personal communication)formostMexican species, and, in thefew cases where
masses were unavailable, fromestimationbased on comparisons with closely
related species of similarhead and body length.
The analysis focused on the frequencydistributionsof species in logarithmic
(log2; see App. B forsize categories)body-sizecategories.We used log2intervals
data and
because most previous analyses have been based on log-transformed
because base 2 divides thefaunaintoa convenientnumberofcategories(see, e.g.,
Preston 1962). Using logarithmicsize categoriesmakes our analyses insensitiveto
small errors in assigning body masses to species (such as may be caused by
intraspecificgeographicvariation).
The distributionsof body masses among species forbiomes and local habitats
were compared with null models that assumed that species were assembled at
randomfromappropriatelarger-scalespecies pools. Each of the 21 biome distributionswas compared withthe North Americandistributionby drawingat random 500 timesfromthe continentalspecies pool the same numberof species as
occurred in that biome. The median was calculated for each of the 500 simulations; the numberof simulatedmedians out of 500 was compared with the observedvalue to evaluate the nullhypothesisthatthebiome was a randomsubsample of the continentalspecies pool. A similarprocedurewas used to compare the
distributionsfor24 local habitatsand 24 simulations,each consistingof 500 random draws fromthe biome distributionsin which each of those habitats was
located.
Inspection suggested that the North American distributionwas highlymodal
and that the distributionsfor biomes and local habitatpatches became progressivelymoreuniform.We quantifiedthischangein shape by comparingall distributions to a log-uniform
distribution,withthe same range as the North American
distribution,by using the Kolmogorov-SmirnovDn statisticand test.
If the distributionsof body sizes change with spatial scale, there must be a
differential
replacementof certain size classes between biomes and especially
size classes should show differbetween local habitats.In otherwords, different
ent degrees of beta diversity.There are two ways species can change statusfrom
absent to present (or vice versa) between habitats. A species may be replaced
because the border of its geographicrange has been crossed. Alternatively,a
BODY SIZE AND SPECIES COMPOSITION
1481
TABLE 1
SUMMARY STATISTICS FOR FREQUENCY DISTRIBUTIONS OF LOG2 OF BODY MASSES (IN GRAMS) FOR
MAMMALS OF NORTH AMERICA AND 2i NORTH AMERICAN BIOMES
Biome
Number
Region
All
NorthAmerica
1
Sitkan
2
Oregonian
3
Yukon taiga
4
Canadian taiga
S
Eastern forest
6
Austroriparian
7
Californian
8
Sonoran
Chihuahuan
9
10
Tamauilipan
11
Great Basin
12
Alaskan tundra
13
Canadian tundra
14
Grasslands
15
Rocky Mountains
16
Sierra-Cascade
17
Madrean-Cordilleran
18
Campechean
19
Guerreran
Sinaloan
20
Yucatecan
21
Biome
mean
N
464
46
77
46
72
74
59
88
102
113
66
95
37
38
115
110
108
182
95
119
84
51
82.2
Median Minimum Maximum
Interquartile Standard
Range
Skewness
6.4
10.2
8.1
10.1
8.0
8.4
8.4
7.0
7.6
7.3
10.0
8.1
10.8
10.6
8.5
7.6
7.1
7.1
8.2
7.6
7.6
11.6
1.6
2.3
2.8
1.6
1.6
1.6
1.6
2.6
2.6
2.3
2.3
1.6
2.3
2.3
2.3
1.6
2.3
2.3
2.3
2.3
2.6
2.8
18.9
18.9
18.9
18.8
18.9
18.9
18.9
18.9
18.9
18.9
18.7
18.9
18.8
18.8
18.9
18.9
18.9
18.9
18.2
18.2
17.0
18.2
4.1
7.8
6.5
8.9
8.4
7.2
7.6
6.3
7.1
6.4
7.7
6.4
8.5
8.6
7.1
6.6
6.4
6.2
6.4
5.9
7.0
5.4
9.2
.4
1.9
.3
1.4
1.3
1.2
2.8
2.5
2.9
.3
2.3
.0
.4
1.7
2.3
2.8
3.4
1.3
2.4
2.0
- .5
8.6
2.2
18.7
7.0
1.5
species withspecialized habitatrequirementsmaynotoccur in some ofthehabitat
typeswithinits geographicrange. For each of the 24 local habitats,we compiled
two frequencydistributionsof body sizes, one containingthose species in the
continentalspecies pool whose geographicrangesdid not includethe habitatand
the othercontainingthose species in thecontinentalpool whose geographicrange
did include the site but thatdid not occur in thathabitat.Then we summedeach
kindof listwithredundancy(countingeach species repeatedly,as manytimesas
forspecies exhibit occurred)to determinethe overall shapes of thedistributions
itingthe two kinds of replacementbetween habitats.For most habitatsthis was
because the authorsof the local species lists included informastraightforward,
tionon the occurrenceof species in the otherhabitattypeswithinthatlocal area.
In threecases, these local lists were not available and we used otherpublished
sources to determinewhich species had geographicranges thatincludedthe site
but were foundin adjacent habitats.
RESULTS
ofbody masses of NorthAmericanterrestrial
The frequencydistributions
mammals varied with the spatial scale of the sample: the medians and interquartile
ranges increased and the skewness decreased fromcontinentto biome to local
habitatpatch (tables 1 and 2). The distributionfor the entirecontinentalfauna
O2
-?
u
Z
~ ~ ~
O._
>~=
U)~
00
0
\\O
dNC\
v00,0
00ct
\ \ 0\
c~cc
C00
'D
U
'D
CLN
CL
P.
0
0
O
0
o0
0
0
00 0
0
00
0*
03
0
\C
0
-
000
ON
0
00
Oo0o 00
00
sON
0
r
_-_
00
0
r
0
0
0
0
0
0
0
0
C ll
0
0
r
-o
0
E
??
--
0
-
r
0
S
05
D0
r
00
S
)
0\
x(7
0
0
0
krZ
kr)rqU
CI;
~~~~~~~~~o ~ ~
-_
00or \
S
ON
700 000000
cr\
--
\p\
<0
-~~~~
-~~~~~~
~~~~~03
Z~~~~~~~~~~~~~~~~~~~~,=
3~
NS
O 00 N
ol-
-M
F0
Q
00
N
E
~ ~~ ~ ~=-k\~ --~ ~~~~~~\
to
0
c
0
00
O O
z
7~~~~
00
0?
0z
D00
0
00
C)
x00
-?
\
(\
C
V)
v~-)I
I-
C
\OO0
C
1"
0I
"I
"t
I
c~~~~~0
Li~~~~~~~~~~~~~~~~~~~C
R C'soooo
R
5t
LL
5
z'
00
0xS.-
z~~~
~~~~
EN
0 0 0 000
00
~~~~4 0~~~~~~~~'
~ ~ ~~~0
S0
mt
rKi~~~~
~ ~
~~0
o6c~~~~i
~
?
o6R
tVUc
--
c
00t
1-t~~~C'
~
~
4Q
0&66
~
4I
t
i
t 4
.
.
0000
O
00
-
O
00~~o
0 00
0
XoON
t?F
Xo
0
-~~~~~~~~~~.
t.
C1
i't~
0N0
0U
O3
00
0
CL
)\
0C7
)
-
00000
0
00s
z
0
O r-ON
N5ONO
c
t
~
rn"
-)\
4 -Y
-
0
.
--\
(1fA 0
0t
0t
0
c
00O
BODY SIZE AND SPECIES COMPOSITION
?
Ql)I
V)
1483
20
CANADIAN
TAIGA
CO., MN
WASHINGTON
20
EASTERNFOREST
GREENMTS.,Vr
SONORM
COCHISECO., AZ
0
20
4-
z
20
CHIHUAUAN
CO., NM
BERNAULLO
GRASSLANDS
KS
KONZAPRAIRIE,
10
_._
20
101
0
4
8
12
16
20 o
4
8
Lo92 Body Mass (g)
12
16
20
FIG. 2.-Frequency distributions,on a log2scale, of the body masses of terrestrialmammals inhabitingfivebiomes and fivelocal patches of uniformhabitatwithinthose biomes.
Note that the biome distributionsare all highlymodal, but the habitatdistributionsare all
nearlyuniformon the logarithmicscale.
was highlymodal and significantly
rightskewed; the mode was in size class 5,
forall 24 local habitats
approximately45 g (fig. 1). In contrast,the distributions
had an approximatelyequal numberof species in each logarithmicsize category
were all statisticallyindistinguishable
fromlog-uniform
(fig.2); these distributions
distributions(table 3). The distributionsfor the biomes were intermediatebetween those forthe continentand those forthe local habitats(fig.2); nine of the
fromlog-uniform
distributions(table 3).
21 biomes did not differsignificantly
Neitherthe biomes nor the local habitatswere randomsamples of the species
pools on the next largerscale, continentand biomes, respectively.All 21 biome
distributionshad medians significantly
(P < .05; <25 of 500 simulations)larger
than the North American distribution.Nine of the 24 local habitatpatches had
medians significantly
largerthan the simulatedmedians (P < .05), and an addi-
TABLE 3
COMPARISONSOF THE DISTRIBUTIONSOF BODY SIZES FOR NORTH AMERICA,2I BIOMES, AND 24 LOCAL
TEST
HABITAT PATCHES WITH A LOG-UNIFORMDISTRIBUTION,USINGTHE KOLMOGOROV-SMIRNOV
NorthAmerica
Biome:
Sitkan
Oregonian
Yukon taiga
Canadian taiga
Eastern forest
Austroriparian
Californian
Sonoran
Chihuahuan
Tamauilipan
Great Basin
Alaskan tundra
Canadian tundra
Grasslands
Rocky Mountains
Sierra-Cascade
Madrean-Cordilleran
Campechean
Guerreran
Sinaloan
Yucatecan
Locality:
White Sands, N.Mex.
BernalilloCo., N.Mex.
Deep Canyon, Calif.
Deep Canyon, Calif.
Deep Canyon, Calif.
Cochise Co., Ariz.
Barge Canal, Fla.
Barge Canal, Fla.
Barge Canal, Fla.
Chamela, Jalisco, Mexico
Chamela, Jalisco, Mexico
Green Mountains,Vt.
Ligonier Valley, Pa.
Cook Co., Minn.
Cook Co., Minn.
Animas Mountains,N.Mex.
Sagehen, Calif.
Cascade Mountains,Oreg.
Cascade Mountains,Oreg.
Cascade Mountains,Oreg.
Lac Qui Parle Co., Minn.
WashingtonCo., Minn.
Norman Co., Minn.
Konza Prairie,Kans.
Dn
P value
.32
.0000
.10
.16
.11
.16
.13
.14
.23
.21
.22
.20
.16
.08
.11
.15
.17
.20
.20
.21
.22
.18
.16
.9999
.0464
.9992
.0618
.1889
.2009
.0001
.0001
.0001
.4693
.0116
1.0000
.9999
.0128
.0034
.0003
.0001
.0001
.0001
.0079
.1626
Habitat
Dn
P value
Chihuahuandesert
Riparianforest
Pine forest
Pinion/Juniper
Agave/Ocotillo
Chihuahuandesert
Longleaf pine
Scrub oak
Pine flatwoods
Deciduous forest
Evergreenforest
Spruce/Fir
Deciduous forest
Aspen/Birch
Whitepine
Oak forest
Jeffrey
pine
Alpine tundra
Sage
Ponderosa pine
Upland prairie
Shrub swamp
Willow swamp
Tallgrass prairie
.18
.12
.13
.16
.17
.21
.19
.13
.13
.24
.19
.07
.14
.15
.18
.16
.12
.17
.13
.08
.09
.10
.11
.19
.2078
.9999
.9999
.9999
.4057
.1157
.5303
.9996
.9999
.0876
.2611
1.0000
.9991
.9998
.4121
.4658
.9999
.4053
.5224
1.0000
1.0000
.9999
.9999
.2009
were compared witha log-uniform
distribuNOTE.-The NorthAmericanand biome distributions
and the local habitatswere
tion withthe same range as the observed NorthAmericandistribution,
distributionwiththe same rangeas the biome in whichtheyoccurred.For
comparedto a log-uniform
each comparisonthe Kolmogorov-SmirnovDn statisticand the probabilityvalue are given.
BODY
SIZE
AND SPECIES
COMPOSITION
1485
TABLE 4
RESULTS OF SIMULATIONS TO EVALUATE THE NULL HYPOTHESIS THAT THE BODY-SIZE DISTRIBUTIONS
FOR EACH OF THE 24 LOCAL HABITATS ARE A RANDOM SUBSET OF THE BIOME POOL
IN WHICH THAT HABITAT Is LOCATED
MEDIAN
HABITAT
WhiteSands, N.Mex.
BernalilloCo., N.Mex.
Deep Canyon, Calif.
Deep Canyon, Calif.
Deep Canyon, Calif.
Cochise Co., Ariz.
Barge Canal, Fla.
Barge Canal, Fla.
Barge Canal, Fla.
Chamela, Jalisco, Mexico
Chamela, Jalisco, Mexico
Green Mountains,Vt.
LigonierValley, Pa.
Cook Co., Minn.
Cook Co., Minn.
Animas Mountains,N.Mex.
Sagehen, Calif.
Cascade Mountains,Oreg.
Cascade Mountains,Oreg.
Cascade Mountains,Oreg.
Lac Qui Parle Co., Minn.
WashingtonCo., Minn.
Norman Co., Minn.
Konza Prairie,Kans.
Observed
Simulated
Proportionof
SimulationsThat Are
Less than Observed
Value
8.4
12.0
9.7
9.8
8.6
7.6
7.6
10.7
10.8
12.0
11.8
10.1
8.8
8.9
7.6
12.0
9.4
8.4
8.1
10.4
11.0
10.8
11.2
8.5
7.29
7.54
7.38
7.46
7.29
7.09
7.12
8.61
8.59
7.78
7.77
7.72
8.80
7.23
6.69
8.38
7.16
6.99
6.99
7.35
8.42
7.36
8.53
7.70
.088
.001
.096
.048
.156
.386
.736
.060
.090
.001
.001
.068
.384
.382
.432
.004
.002
.168
.168
.001
.058
.004
.001
.238
NOTE.-The probabilityof failingto reject each null hypothesisis given as the proportionof the
500 simulationsthatare less than the observed value formedians.
tional six patches were marginallysignificantly
larger(P < .10) thanthe random
draws fromthe biome in which theywere located (table 4). The medians forthe
local habitat patches were all significantly
larger than the medians of random
draws fromthe NorthAmericandistribution.
The onlyway to account fortheprogressiveflattening
of thefrequencydistributions fromcontinentalto biome to local scales is by highspatial replacementof
species in the modal size classes. This is reflectedin the limitedhabitatdistributions and small geographicrangesof these species comparedto theirlargerrelatives. The vast majorityof species in the smallersize categories(less than class
12, approximately1 kg) occurred in fouror fewerbiomes, whereas the majority
of species in the largersize categoriesoccurredin eightor more biomes (fig.3).
An even clearerpatternis apparentforthe areas of geographicranges: a substantial proportionof the species in the smaller size categories (less than class 13)
whereas all species in the largersize classes
had rangesof less than 100,000kmi2,
had ranges thatexceeded this area (fig.4).
1486
THE AMERICAN NATURALIST
30
Body Mas
I g
a16
10
1024 - 8192 g
25 -68
2015
410
2.
500
Cfl 100
la8-128g
10
1192 -6553a g
4-
0
.
40-
4-
E
20-
2-
6b
0
ie
12 - 1024 g
)65536 g
50 -68
FIG.3.Feunydsrbtosofseisiaifrn
oysiecassacrigth
40304202-
100
0
2
4
6
10 12 1'4 16S 16 20
C -.
0
2
4
6
16 10 12 14 16a 163 20
Numberof Biomes
ofspeciesindifferent
FIG. 3.-Frequencydistributions
to the
body-sizeclassesaccording
numbersof biomeswheretheyoccur.Notethatspeciesof smallsize tendto occurin few
ofspeciesinthelargersize classesinhabit
biometypes,whereasthemajority
manybiomes.
The compositionof species may vary among habitatsacross the landscape for
two reasons: because these species have limitedgeographicranges or because
theydo not occur in all the habitatswithintheirgeographicranges.It is of interest
to assess the contributionsof these two phenomena to the replacement of
different-sized
species among local habitats. For each local habitat,we determinedthe identityand body mass of those species in the continentalspecies pool
thatwere absent because theyeitherhad a geographicrangethatdid not encompass the site or had a geographicrange thatincluded the site but did not occur
in thathabitat.These are illustratedin figure5, whichshows thefrequencydistributions of body masses of the species in these two categories summed with
in these distriburedundancyforthe 24 local habitats.Note the generalsimilarity
tions, althoughthere were proportionatelymore medium-sizedspecies (classes
BODY SIZE AND SPECIES COMPOSITION
30 3-30169
BodyM,,a: (
3
1487
9
1024-8192
25-
~~~~~~~~25-
20-
~~~~~~~~20-
15 -
~~~~~~~~~15-
-10
100,
~~~~~~~~~~5-
5-
70
5
60
10
10-8
- 182 g
oL.-I
LA
0)8
Q~50-
~~~~~~~~~~~~~~~~~6
40
0
9
65536
30 -4
20-
E
Z
2
10
o 050 -20128 -1024 g
)65536 g
40-
15-
3010205-
10.
0
1
100
10,000 1,000,000
0.
1
6
100
10,000
,0,000
Geographic Range Class (km2)
ofspeciesindifferent
FIG. 4.-Frequencydistributions
body-sizeclassesaccording
to the
frommapsinHall 1981).Notethat
areasoftheirgeographic
ranges(measured
byplanimetry
manyspeciesofsmallsize haverestricted
geographic
ranges,whereasmostspeciesoflarge
size are distributed
overlargeareas.
9-13) in category2. Species in the modal size classes replaced each othermore
rapidlyacross the landscape both because they had more restrictedgeographic
ranges and because theyoccurredin a smallerproportionof habitattypeswithin
theirgeographicranges.
DISCUSSION
Patterns
We have shown what appears to be a generalpatternof spatial scalingof North
Americanterrestrialmammalianassemblages withrespectto body size. The distributionof body masses forthe continentas a whole is highlymodal and skewed
THE AMERICAN NATURALIST
1488
1800
RANGESDO
GEOGRAPHIC
DNOT
INCLUDESITES
1600
700
14001200 1000
800-
U)
Q)600-
ci) 4000.200
C)
0
0
ci) 80-0
E
RANGES
~~~~~~~~GEOGRAPHIC
SITES
INCLUDE
~~~~~~~DO
70-
D60-
5Q403020-
100
0
2
4
6
8
1
12 14 1'6 1'8 20
Log2 Body Mass (g)
FIG. 5.-Frequency distributions
of body masses summedwithredundancy(see text) for
the two classes of species that did not occur in the 24 local habitatpatches: those whose
geographicranges did not include the local sample sites (top) and those whose geographic
ranges did encompass the sites even thoughthey did not occur in the habitats sampled
(bottom). Note that both distributionsare modal and rightskewed.
toward largersize classes (fig. 1). The frequencydistributionsfor small patches
of homogeneous habitathave ranges of body sizes similarto those of the North
American distribution,but they contain an approximatelyequal numberof species in each logarithmicsize class (fig. 2). The distributionsfor biomes, which
are intermediatein area and habitatdiversitybetweenthe continentand the local
habitats,are intermediatein shape between the NorthAmericanand the habitat
distributions.These patternscharacterizethe compositionof the mammalianfaunas on these scales before the impact of European humans. Hunting,trapping,
habitatchanges, and other human impacts withinthe last three centurieshave
caused the local extirpationof a few species; humanintroductions
of alien species
have resultedin thecolonizationof some additionalspecies. These recentchanges
BODY SIZE AND SPECIES COMPOSITION
1489
in species compositionwere not sufficient,
however,to change the shapes of the
above distributionsto any substantialextent.
The North American mammalianfauna experiencedadditionalchanges in the
period 500-20,000 yr ago when catastrophicextinctionsoccurred (Martin and
Klein 1984). Althoughthe causes of these losses are hotlydebated, humansmay
have played a substantialrole. It is clear thatlarge species sufferedmore extinctions (Martin 1984; Webb 1984). These extinctionswere not sufficient,
however,
to affectthe qualitative patternsreportedhere. Despite the numberof species
and genera that were lost, the frequencydistributionof body masses for North
American fauna prior to the Quaternaryextinctionsstill has a mode and right
skew similarto the distributionforthe contemporaryfauna (R. Rusler, personal
communication).Similarly,the additionof species known or presumedto have
become extinctwithinthe last 20,000 yrto thelocal habitatswould have extended
the rangeto largersize categoriesin some cases but would not have substantially
alteredthe shape of the frequencydistributions.
The shape of the continentaldistribution,
firstemphasized by Hutchinsonand
MacArthur(1959), appears to be typicalof the frequencydistributionsof body
sizes of species in diverse taxa fromlarge geographicregions(May 1978, 1988;
Bonner 1988). An explanationforthis patternin mammals,and perhaps in other
with
organisms,musttake intoaccount thechanges in thefrequencydistributions
spatial scale. The continentalmammalianfauna is not simplythe sum of species
assemblages on smallerspatial scales, and thefaunasof local habitatsand biomes
are notjust randomsubsamples of the species pools on large spatial scales. Species near the modal size of approximately45 g replace each other frequently
between habitats and biomes, whereas species of large size tend to have large
geographic ranges and to occur in a large proportionof the habitats (figs. 3
and 4).
Hypotheses
We offerthefollowinghypothesesto explainthispattern:(1) competitiveexclusion tends to preventlocal coexistence of similar-sizedspecies with similarreextinctionof species of large size withsmall
source requirements,(2) differential
limit
the
numberof large species in the continental
tends
to
geographicranges
constraints
cause modal-sized species to be
and
allometric
fauna,
(3)
energetic
more specialized in theiruse of resources than largerspecies. We develop each
of these hypothesesin more detail below.
Competitiveexclusion among similar-sizedspecies.-Since the faunas of local
habitatscontainfewermodal-sized species than randomsamples of assemblages
fromlargerspatial scales, itappears thatsome local process preventscoexistence
of species of similar size. Competitiveexclusion is the most likely process to
have thiseffect.If such competitionoccurs, however,its impactshouldbe limited
to those species thatshare requirementsforthe same limitedresources. Competitive exclusion should occur withinbut not betweenfeedingguilds. There is substantialevidence for interspecificcompetitionamong species of mammals and
other organisms with similar diets (see, e.g., Hutchinson 1959; McNab 1963;
Rosenzweig 1966; MacArthur1972; Brown 1975, 1987; Pacala and Roughgarden
1490
THE AMERICAN NATURALIST
1982, 1985; Brown and Munger 1985; Grant 1986). In at least some of these
cases, competitionappears to account forthetendencyoflocal communitiesto be
composed of species that are more differentin body size than expected from
randomassemblages drawn fromappropriatespecies pools (see, e.g., Schoener
and Boecklen 1981; Bowers and Brown 1982;
1970, 1984; Brown 1973; Simberloff
Brown and Bowers 1985; Hopf and Brown 1986). It is hypothesizedthat the
distributionson the scale of habitatsreflectthe factthatlocal assemlog-uniform
blages are composed of multipleguilds, each of which tends to have its own
log-uniformdistribution.For example, if mammals are divided into two very
general trophicgroups, herbivoresand carnivores, most habitats contain representativesfromeach group that vary in size fromless than 20 g to more than
100 kg.
Large species with small geographic ranges have high extinctionprobabilities.-The fact that large species exhibitlow beta diversity(low replacement
between habitatsand geographicregions)suggeststhatsome large-scaleprocess
preventsthe accumulationof highdiversityof large mammalsin the continental
fauna. We hypothesizethatthisprocess is the selectiveextinctionof species with
large body sizes and small geographicranges. Individualsof large size have large
resource requirements,and as a resulttheyrequirelarge home rangesand occur
at low populationdensities (McNab 1963; Schoener 1968; Parra 1978; Harestad
and Bunnell 1979; Damuth 1981; Peters 1983; Peters and Raelson 1984). As a
consequence, species of large size withsmall geographicrangeshave small total
susceptibleto extinction(MacArthur
populationsizes and should be differentially
and Wilson 1967; Brown 1981; Pimm et al. 1988; Schoener and Spiller 1988). If
extinctionis caused exclusively by demographicvariation (in the absence of
major environmentalchange), criticalpopulation sizes for persistenceare very
individuals
small,and even smallgeographicrangeswould oftencontainsufficient
to avoid extinction.If, on the otherhand, extinctionis caused by environmental
variation,even relativelylarge, dispersed populationsmay be at risk (Goodman
1987).
We hypothesizethatenvironmentalchanges have been a major cause of mamaffectedthe species
malian extinctionsand thatthese changes have differentially
in each size class withthe smallestgeographicranges. The observationthatthe
minimumsize of geographicranges of large species is largerthan thatof modalsized species (Brown 1981; Rapoport 1982; Brown and Maurer 1987, 1989; see
also fig.4) is consistentwiththishypothesis,as is the observationthatmammals
and otherorganismsof large size have been differentially
susceptibleto historical
perturbationsthatcaused mass extinctions(Martinand Klein 1984; Webb 1984).
Large mammals may also have lower speciationrates than theirsmallerrelatives. Large species tend to have greater vagilityand broader environmental
of
tolerances,and these should resultin less isolationand geneticdifferentiation
populations and in lower speciation rates than in small species. While lower
speciation rates mightcontributeto the low diversityof large mammals in the
continentalfauna, they are not sufficientto account for the failureto observe
large species withsmall geographicranges.
BODY SIZE AND SPECIES COMPOSITION
1491
Specialization of modal-sizedspecies.-A correlateof body size thatis related
to bothof the above hypothesesis the apparentlygreaterspecializationof modalsized species. Not only does the greatestnumberof mammalianspecies occur in
the size range of 20-250 g, but species withinthis size range exhibit greater
turnoveramong habitatsand on theaverage have smallergeographicranges(figs.
limitedby varia3, 4). This suggeststhatmodal-sizedspecies are morefrequently
tion in the physical environment,the presence of otherorganisms,or both processes than theirlargerrelatives. We hypothesizethatthis specialization is not
simplya consequence of geographicand habitatdistribution(in which case the
above argumentswould be circular) but is caused by allometricconstraintson
physiologyand energetics.
It is well-knownthatenergeticand basic nutrientrequirements(D) of individuals scale as a fractionalexponentof body mass (M) as M067to M075(Peters 1983;
Calder 1984). Althoughthe followingqualitativeargumentdoes not depend on
the precise value of the exponent,we will assume thatthe daily energydemand
of free-livingmammalsscales as
D
= c1M0.75
(1)
where cl is a constant(Nagy 1987). The rate of food intakein a varietyof mammals also scales as M075 (Calder 1984). However, larger animals have larger
and longeralimentarytracts;in both herbivorousand carnivorousmammalsgut
capacity (A) scales as approximately
A
= c2MI0,
(2)
where c2 is another constant (Calder 1984). As a consequence, largeranimals
retainfood in the gutfora longerperiod; turnovertime(T) forgutcontentsscales
as approximately
T = c3M0 25,
(3)
where C3 is anotherconstant(Calder 1984). This enables largeranimals to ingest
poorer-qualityfood and, by subjectingit to digestionfora longerperiod, stillto
extract sufficientenergy and nutrientsto meet their requirements.Thus, diet
qualityscales inverselyand digestiveefficiencyscales directlywithbody size.
The food quality(Q) necessaryto meet requirementscan be predictedfromthe
ratioof metabolicdemand to gut capacity; it should scale as
Q = DIA = c4M-025,
(4)
where C4 is another constant. The best data to evaluate this predictionare for
ruminantmammals(Hoppe 1977; Sibly 1981; du Toit and Owen-Smith1989). The
predictedvalue of -0.25 is very close to those (-0.20 to -0.27) estimatedby
McNaughtonand Georgiadis (1986). We also used data on 14 species of African
ruminantsfromHoppe (1977; cited in Sibly 1981) to obtain anotherestimateof
the scalingexponentforfood quality.We performeda regressionof the reciprocal
contentsdivided by metabolicdemand against body mass and
of reticulo-rumen
obtained an exponentof -0.351 ? 0.088.
1492
THE AMERICAN NATURALIST
We hypothesize that these physiological constraintson food quality force
foods and to restricttheirforaging
smalleranimals to specialize in higher-quality
to habitats where suitable foods are available in sufficientsupply. Even small
omnivorous species should be subject to these allometricrelationsand should
restricttheirdiets to foods of high energeticand nutritionalvalue. Other contraits(see,
straintsof body size, such as those on reproductiveand life-history
e.g., Eisenberg 1981; Peters 1983; Calder 1984), increase the nutritionaldemands
on smaller organismsand tend to reinforcethe selection to specialize on foods
and habitats.Anothermechanismmustbe hypothesizedto account forthe small
number of species (and the low population densities of these species) in the
classes (see Brown and Maurer 1987, 1989; Dial and Marzluff
smaller-than-modal
1988).
The specialization hypothesis developed here is not independent of the
competitive-exclusionand differential-extinction
hypotheses presented above.
Rather,the nutritionalconstraintsand the resultingspecializationprovideanother
level of explanationforthe pervasive effectsof body size on resourceuse, habitat
selection, population regulation,and distributionallimitsthat ultimatelydetermine small-scale species interactionsand large-scalespecies dynamics.
Allocation of Food and Space among Species
Regardless of whetherthe above hypothesesare necessary and sufficientto
explain the empiricalpatterns,the spatial scaling of the body-size distributions
has importantimplicationsfor the allocation of food and space among species
and hence forthe assembly of continentalbiotas. First,the natureof ecological
communitiesdepends on the spatial scale of study.The largerand moreheterogeneous the area sampled (and these will tend to be correlated),the more species
compositionwill reflectbeta diversity(replacementof species between habitats)
relativeto alpha diversity(numberof coexistingspecies withina habitat).
Furthermore,not only the numberof species but also the kindsof species that
actually and potentiallyinteractvary with spatial scale. The limitednumberof
species that coexist in small patches of relativelyhomogeneous habitattend to
be of differentsizes. Because of the energeticconstraintsoutlined above, this
food resources or using
will tend to resultin coexistingspecies' using different
the same resources in different
ways. Many of the coexistingspecies of similar
size are in different
trophicguilds. In terrestrialmammals,carnivoresand herbivores span virtuallythe entirerangeof body sizes. Althoughspecies in the same
guildsmay coexist locally and competeforfood (especially iftheyare of different
body sizes), oftenthe most severe competitionwill be among species (of similar
sizes) thatrarelyencountereach otherbecause theyoccur in adjacent but largely
nonoverlappinghabitatsand geographicregions.
The spatial scaling of body size emphasizes the importanceof beta diversity
(Cody 1975; Wilson and Shmida 1984). The highfrequencyof species in themodal
size classes on the biome or continentalscale reflectsthe frequentreplacement
of these species among local habitats.This turnovercan be of two types. On the
one hand, it may reflectcrossingthe bordersof geographicranges. On the other
BODY SIZE AND SPECIES COMPOSITION
1493
hand, it may reflectthe fact that species do not occur in all habitats in their
geographicranges. The distributionof body sizes of species in the formercategory(fig.5, top) is virtuallyidenticalin shape to theNorthAmericandistribution.
The species in the lattercategory (fig. 5, bottom)also exhibit a highlymodal
but are composed of relativelymorerepresentativesin size
body-sizedistribution
classes 9-13 (500 g-16 kg), manyof whichappear to be omnivoreswithrelatively
large geographicranges.
Generality
Since the presentanalyses are exclusivelyof data forNorthAmericanterrestrialmammals,it is reasonable to ask how general the resultsare. Of course it
would be desirable to obtain and analyze comparable data for other groups of
organisms in other geographic regions. Until this is done, we can make two
comments.
First, the right-skeweddistributionforthe North Americanmammalfauna is
typical of the shapes of frequencydistributionsof body-size measurementsfor
many other taxa fromlarge geographicareas. For example, May (1978, 1988),
Bonner (1988), Morse et al. (1988), and Brown and Maurer (1989) presentexamples of qualitativelysimilar size distributionsfor a wide varietyof organisms
frombacteria to insects to birds. In addition,Rusler (1987) examinedfrequency
distributionsof body masses among species and genera of mammalson different
continents.She found similarright-skeweddistributionsforall the largercontinents with diverse faunas. We cannot claim that these biotas exhibitthe same
patternof spatial scaling withincontinentswithoutanalyses of data for smaller
scales; but it seems unlikelythattherewould be so much similarityin the largescale patternsunless the underlyingmechanismsand small-scale patternswere
also similar.
Second, none of the mechanismsthatwe have proposed above to account for
thepatternsin NorthAmericanmammalsis inherently
specificeitherto mammals
or to the North American continent.Allometricscaling of morphologicaland
physiologicalvariables withbody size exhibitssimilarexponentsin a wide variety
of animals, includingvertebratesand invertebrates,endothermsand ectotherms
(Peters 1983; Calder 1984). The relationshipof body size to populationdensity,
area of geographicrange, dietaryspecialization,and otherattributesof species
thataffectthe assemblyof bothlocal communitiesand continentalbiotas appears
to be similarin the organismsthathave been studiedto date (see data on beetles,
birds,and mammalsin Brown and Maurer [1987, 1989] and Morse et al. [1988]).
The generalityof these relationshipsmakes us optimisticthatboththepatternsof
spatial scaling and the mechanisticprocesses thatproduce themare also general.
The constraintson the body sizes of species that occur togetheron different
spatial scales can be thoughtof as one importantcomponentof the rules for
assemblingcontinentalbiotas. Althoughcompleteassemblyruleswould be based
on othervariables in additionto body size, it is clear thatsize, and traitsthatare
correlatedwith size, profoundlyaffectsthe compositionof species assemblages
(see also Cody 1975; Diamond 1975; Brown 1981; Brown and Maurer 1987, 1989;
1494
THE AMERICAN NATURALIST
Dial and Marzluff1988). Because of its pervasiveinfluenceon physiology,behavior, and ecology, body size influencesthe coexistenceof species in local habitats,
the turnoverof species across the landscape, and the colonization, speciation,
and extinctionof species on continentalscales.
Additionalassembly rules will be requiredto account fordifferencesin biotas
among continentsand between continentsand islands. The importanceof body
size is also seen in the divergence of insular populationsfromtheir mainland
relatives(see, e.g., formammals,Foster 1964; Lomolino 1985)and in the systematic variationin the frequencydistributionof size among species or genera in
continentalfaunas as a functionof the area of the land mass (see, e.g., Brown
1986; Rusler 1987; see also Van Valen 1973).
Implications for Conservation
spatial scales also have important
The rules forassemblingspecies on different
implicationsfor the maintenanceof diversityand for the disassemblyof biotas
duringenvironmentalperturbations.Such disassemblyhas occurrednaturallyin
the past, most dramaticallyduringepisodic mass extinctions(Martinand Klein
1984; Jablonski1986; Raup 1986). At present,such disassemblyis occurringin
response to the multipleeffectsof the growinghumanpopulation(Wilson 1988).
Losses of species duringpast mass extinctionsand in response to the impactsof
modernhumansare nonrandomwithrespectto body size and otherattributesof
species (see, e.g., Diamond 1984; Martin1984; Webb 1984). For example, of the
species known to have disappeared duringthe last threecenturiesfromthe 24
local habitatsthatwe studied, 11 of 13 are largerthan2 kg and seven are larger
than 35 kg. From the patternsof faunal assembly and these kinds of data on
known extinctions,it should be possible to predictwhich species will be most
vulnerableto perturbations(Arita et al. 1990).
Most currentconservation strategiesfocus on effortsto preserve individual
endangeredspecies and threatenedhabitats.These procedures,by themselves,
will be inadequate to preventwholesale extinctionscaused by humanactivities.
inhabit
It is estimatedthatbetween5 and 30 millionspecies oforganismscurrently
the earth (Erwin 1983; May 1988; Wilson 1988). It will be impossibleto identify
which ones are endangered,much less to develop recoveryplans forthem,on a
to inventoryall kinds
species-by-speciesbasis. It will be almost equally difficult
of habitats.
Rules fortheassemblyand disassemblyof biotas suggestalternativesto present
approaches to conservationbiology.The statisticalpatternsof bioticcomposition
to extinction
on different
spatial scales providea basis forassessing vulnerability
on the basis of body size and othervariables, such as area of geographicrange
(e.g., see Brown and Maurer 1987, 1989; Pimmet al. 1988; Aritaet al. 1990). Not
only can these assessments identifyclasses of species that may be vulnerable,
theycan pointto the spatial scales and habitattypes thatwarrantattention.For
sizes, dispersions,and habitatswillbe differentially
example, reservesof different
effectivein enhancingthe survivalof different
species, dependingon theirbody
sizes and otherecologically relevantattributes.In addition,body size and other
BODY SIZE AND SPECIES COMPOSITION
1495
easily measured variables providea way of initiallyassessing the ecological roles
of species. The patternsof species replacementwithinand between large areas,
such as biomes and continents,suggestthatthereis considerablecomplementarityin these roles. As humanimpactsinevitablyincrease and nativespecies inevitablybecome extinct,the only way to restorelost diversityand ecological functions will be throughintroductionof alien species. Assembly rules can be used
to identifycandidate species forintroductionto fillmissingecological roles. For
example, Janzen (1982) has advocated the use of ferallivestockto replace seed
dispersal, browsing,and grazing roles of large native mammalsthat have been
extirpatedfromtropicalreserves.
CONCLUSIONS
We conclude thatprocesses operatingover a wide rangeof spatial scales-from
interactionsthataffectcoexistence withinlocal habitatsto colonizainterspecific
of species over
tion, speciation,and extinctionevents thataffectthe distribution
the continent-interact to determinethe compositionof the biota at all scales,
fromlocal to continental(Cody 1975; Orians 1980; Brown 1981; Rapoport 1982;
Ricklefs1987; Brown and Maurer 1987, 1989). All of these processes are reflected
in the body sizes of co-occurringspecies, because of the pervasive influenceof
size on manyaspects of physiology,behavior,and ecology. Both the microscopic
perspectivesof physiological,population,and communityecology and of microevolutionand the macroscopic perspectivesof biogeographyand macroevolution
are requiredifwe are to understandcompletelythe compositionof biotas on any
spatial scale.
ACKNOWLEDGMENTS
We are especially gratefulto the colleagues who gave us species lists of mammals forlocal habitatpatches fromtheirunpublishedfieldnotes or fromthe "gray
literature"of environmentalimpact studies: E. Birney,G. Ceballos, J. Cook, J.
Findley, E. Fink, S. Humphery,D. Kaufman, T. Lacher, J. Merritt,and R.
Timm. We thankM. Taper for statisticaladvice and numerousothercolleagues
fordiscussions that helped us to formulatethe ideas presentedhere. M. Boyce,
D. Glazier, L. Hawkins, A. Kodric-Brown,B. Maurer, K. Schoenly, M. Taper,
and an anonymous reviewermade valuable commentson an earlierdraftof the
manuscript.The researchwas supportedby National Science Foundationgrants
BSR-8718139 and BSR-8807792 to J.H.B.
<~~~~~~~~~~~~~~~~~~~~~~~~C
oc
o~~~~~~~~~C
l
:
cq r
C,
~~~~~N
:
F
m
ON
ON
cKIno
C4cO bX
CNCN
??-4
mm
tm
-4
X t b o o ?Obm^
~~~~~~00
_~~~~~~~~I
N
O~~~~~~~~~~~~~~~~C
00
0q
0
4
4U
3
obbN
r- r- r- cq
"t
?
<
'IC
.W
.t
z
2
N
NOOOOOOON
G
E
;
t~~~~t
149
z
CNOm
Ci
06
O
o
-
-N
-
ON
0
o
o
000~~~~~0
00
CN
06 CN
aN
e ti
0^
a',
06
r
t
CNN
?
-
f.~~~O
F N F F mPF F OFFFF
m~~~~~~~0
m t m 00 ? t) t bm
~~ ~
o
ON~~~~~~~~~~~~~~~_
0
Xt t
a
:t
~~
t0
SSst
o -
z
0So0
-
o
ZtLLtULLOtLLaRK
O
?
-
ei
^
ttE >- X
x~~~~~Z
Cq
0e
etw
o
-
1497
F
m>?
z oo
0-
S
C)
S
0aaRRaa;.
00~~~19
> 0t
Z0mt}
(N
(NN
CO
-~~~~Cl
CN~
(N
(N
(N
o~~~~~~c
0
~
(N
^^
c^
~ ~ ~~~~~~~~~~0
(N
~~~~~~~~~00
00
--4N
cqCN
N
N -
00~~~~~~~~~~
0
00
-00
N(
,,t .0,,X
,-"X
t
V
SN
06
U
C
~~~~~~~~~~1ON
v
E
< - t,
4>
X
r
~~~~~~0o~~~z
-0N
~~00
Con
2t
-
009
at
0t
ON~~~~~~~~~~~,
0
o
00
rq
06
00
016
06
if) if)
rq
\C
06 --l C4
06 CD
4
r-, cr
4
06
(4
cq
00
rn
e
cq
C-,j
C) oo C) C)
-zt \c -- -- oo
oo oo
-, C- C-
-.4
C'
C-
00
06
06
4 oo oo
c"Ic"Ic",
C',Ir,,
-'t
, C'l
,
(,j
C) C) C) C) W)
Go
("I
m m
-
-.4-
kr 06
rl
if) r- r4
r,IC,-l "t
(-,I oo
-.,
-.4
m
\,c CD CD 00
m m C) 00 C)
r-
00
rn
_ZZ.
rn
rn
:z
rn
rn
rn
rn
zt
rn
rn
Zt
rn
tz (::)
L. . -
to
CS,
(4i
-4
0-40-41-40-4N-41-4114
1499
rn rn
rn
:..
;Z
rz
tq
rz
(4i
rz -Z
4.,
rn Z
06~~
--
ON0
-
-
~
t~
--
r0
"0~~~~~r
-
00
e
-
ei
0r~c
--el-.
-4
00
Cl~~~~~~~t
ON
S
-~~~~8
-(:
64t
H
-
e4~~~~~~10
0
e-.,
--
0
r
-
0
ei
-'
c
r~--~
-
-
06
"t _I
"o
-
-
o~~
vo~-ov~
o o
o
NC-nt
Dx
~~~
C7
oo
--
-
N
oo
-
b
- -,
~~~~e
rieori
ON ?
O00
Ot?O^m
00
oo
v
O
^
0
m t
0O
6
ONONON0
-
C
o
0
N0
C
0i
_~~P,
rn
,l
E
?t????
ON
O000
O0
,
%%-000000O00000000000o0000
r0
-
NN
oo
oo vo^ oo oo?
ON
-
0
m
^
ON
O~
00 m^m
no
00;;
rn150
1501
(4j '..
~ ~
c)
~
~
~
~
~
c"I~
i(
rq
EO
_e
to
m
trrre
C
-, '
t oN oo rE oo
~to
-
~
-
D
o m o but- o
oo l-
t rE
-'
0r)C,0
'x
m
-o
ON l--O
~
1t=1e
-"
rc
ON
o
oo
c
-
-
--,'
N=
-
00 oo oN l0-
oo rE 1-
(
E
ooNool
t
_
1502~~~
1502
t
-(N
C
-
---
(-
C
-(
1f
0l
N
Cl
00
~
Cl~ Cl
-O
Cl
'in
00
00
-
o
eotr
N
-
N
-
-
000
ON
l
az00
olt
~ ~ ~~~0
Cl~
Cl
Oo
oN
-
~ ~ ~ ~ ~ ~ ~ ~ ~~~~~~~~~0
~~ ~~ ~ ~~~~~~~
~ ~ ~ ~ ~~~~~~O
0
"
r
CC
ON~~~~~~~~~~~~~
r--~0
C~~,
00 r-
oo
06
?
~~~~00
6
~ ~ ~ ~ ~~ ~ ~ ~ ~ ~ ~ ~
N
O~~~~~
C
00CC
Clr'
C
Cl C
ON r'
^
Cl
Cl r'
t
00
rv
Cr
N~t~ 1C
O~~~~~~-
r
NN
O
~O
00
l
-
l
t t'
- t
O~hNCCC~N0
NCC
i
N
-4
1-6
N
r'~
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~15
C4.
C'l~~~~~~~~~~~4
_Z
tJ tl
4
,0
~O
tc
C4
1503
C4
~
'
lC
~
o
N
oo
ON
-
ON
~~~~~~~~~~~
~
?
-^
00
~
c
-
~
NN
0
e-
CN
06
S
&
00
0
0
0
-
ON
-n -
\~ 000
S-
~
0
ON
00
000
V)
r
-IN 0
-X
O
-- 000
00
0
-
.
00
00
--
.
S c1504
0~~~~~t0t
to-z
E EEE
1504
E
z Q
00
-
00
00~~~~~~~~~~~~~~~~~~0
06~~~~~
&
00-
00
ON
-a
-4-
0000)C)00
4
W) W
~~~tot~rei
-
\t
00
ot
00
ONN-
00 00
00
ei
\ct N rf)-
\c
r-~00
-Jz
tc .
~~~~~~~~~~4
(::
:
.
o :
1505
if r-r
-
O
N
0000----
C'
~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~~~~~~~~~~~~0
0
~~~~~
00
-~~~~0
-
~~~r~~00
CCl
r-
H
~ ~ ~
-0
r
-r
ON
r-
r- (ON
)0
r-m 0
-
-
&
-00
)W
2t7,
,
-
-
O ON
ON
~~~~~~~~- Z~~~~~
~~~~~~~~~~z
0
-
-
Z
*
t
tE
t
r
m
,t-
-Z
~~~~~~-~
ot
\ON
~ ~ ~~ ~ ~ ~ ~ ~ ~~~~~~~~0
c~~~~~~~
~~~~~~O
U
oo
W
tk
00
-
C4~~~~~~~~~~~~~~~~~C
(O
0r"
Z&0
&o
O
,
OS
1
"
z
tkz .
-~~~~~~~~~~-5zz
0
1506
06
ei
o
~~~~~~~~~~~~~~
o~~
N
c kr
Cl
1
-
C-mP
-)
ei
-0
06
4C
Cl
00
CN
fc
C
)
~~~~~~~~~~~~~~~~~~~~~o
o
C0 o
t
00t
oo
\c
)m
10001
mPN?t
001~
) 00
0
b~~~~~
-t
~
0~
-
010
t- t- N'-a 00\
o
1000
o
to
aE
-aE t
0
151
07
0
0
t Cl 1t N000'00
0
O
00
\z
~ -~~~~
0
z.
,Z
cn
to
1507
0
cn
-n*
..
Cl
O
C
DO
Ct~~~~~
|
Cl
Cl~~~~C
- r
-0
-O
N0 oo
00
0
00
?
,C
o
~~ ~
~~O
06
a.,00
r-
m-4m
~~~~~c-tt-t
-
t
-
-
00
ClN
0)
cpl
-
-l -
- 00 .FFt00
cr-
c)
~~00
z~t
Ez U
A~~~~~~(Z
-<
:
9
9 9
,
9 9U)
1--)
Cl'-ClCl
Cl0000
coo
n
^to0
Cl
0C
mlSON
XXX$zt
U)U) >
>
RN1 N1 N N
BODY
SIZE
AND SPECIES
APPENDIX
COMPOSITION
1509
B
TABLE Bi
LOG,
BODY-SIZE
CLASSES
FOR NORTH
AMERICAN
MAMMALS,
MAXiMUM BODY MASS
IN EACH
AND
THE
MINIMUM,
MIDPOINT,
AND
CLASS
MASS
(g)
SIZE CLASS
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Minimum
Midpoint
Maximum
1
2
4
8
16
32
64
128
256
512
1,024
2,048
4,096
8,192
16,384
32,768
65,536
131,072
262,144
524,288
1.4
2.8
5.7
11.3
22.6
45.3
90.5
181
362
724
1,448
2,896
5,793
11,585
23,170
46,341
92,682
185,364
370,728
741,455
1.9
3.9
7.9
15.9
31.9
63.9
127.9
255.9
511.9
1,023.9
2,047.9
4,095.9
8,191.9
16,383.9
32,767.9
65,535.9
131,071.9
262,143.9
524,287.9
10,485,503
LITERATURE CITED
Anonymous. 1976. Cross Florida barge canal restudyreport:wildlifestudy.Florida Game and Fresh
Water Fish Commission4:D88-D95.
1987. Plants and animals of White Sands. WhiteSands National Monument,N.Mex.
Arita,H. T., J. G. Robinson, and K. H. Redford. 1990. Rarityin neotropicalforestmammalsand its
ecological correlates. ConservationBiology 4:181-192.
Bonner, J. T. 1988. The evolutionof complexityby means of naturalselection. PrincetonUniversity
Press, Princeton,N.J.
Bowers, M. A., and J. H. Brown. 1982. Body size and coexistence in desert rodents: chance or
communitystructure?Ecology 63:391-400.
Brown, E. R. 1985. Managementof wildlifeand fishhabitatsin forestsof westernOregon and Washington.U.S. Departmentof Agriculture,Forest Service, PacificNorthwestDivision.
Brown, J. H. 1973. Species diversityof seed-eatingdesert rodentsin sand dune habitats. Ecology
54:775-787.
1975. Geographical ecology of desert rodents. Pages 315-341 in M. L. Cody and J. M.
Diamond, eds. Ecology and evolution of communities.Harvard UniversityPress, Cambridge,Mass.
1981. Two decades of homage to Santa Rosalia: towarda generaltheoryof diversity.American Zoologist 21:877-888.
model and the complexities
1986. Two decades of interactionbetweenthe MacArthur-Wilson
of mammaliandistributions.Biological Journalof the Linnean Society 28:231-251.
1987. Variation in desert rodent guilds: patterns,processes, and scales. Pages 185-203 in
J. H. R. Gee and P. S. Giller,eds. Organizationof communities:past and present.Blackwell
Scientific,Oxford.
1510
THE AMERICAN NATURALIST
Brown, J. H., and M. A. Bowers. 1985. On the relationshipbetween morphologyand ecology:
Auk 102:251-269.
communityorganizationin hummingbirds.
Brown, J. H., and A. C. Gibson. 1983. Biogeography.Mosby. St. Louis.
Brown, J. H., and B. A. Mauier. 1987. Evolution of species assemblages: effectsof energeticconof the NorthAmericanavifauna. Ameristraintsand species dynamicson the diversification
can Naturalist130:1-17.
1989..Macroecology: the division of food and space among species on continents.Science
(Washington,D.C.) 243:1145-1150.
Brown, J. H., and J. C. Munger. 1985. Experimentalmanipulationof a desert rodentcommunity:
food additionand species removal. Ecology 66:1545-1563.
Boston.
Burt,W. H., and R. P. Grossenheider.1976.A fieldguide to the mammals.HoughtonMifflin,
Calder, W. A., 111.1984. Size, function,and lifehistory.Harvard UniversityPress, Cambridge,Mass.
Ceballos, G., and A. Miranda. 1986. Los mamifetosde Chamela, Jalisco. Universidad Nacional
Aut6noma de Mexico, Ciudad Universitaria,Mexico.
over MediterCody, M. L. 1975.Towards a theoryof continentalspecies diversities:birddistributions
ranean habitatgradients.Pages 214-257 in M. L. Cody and J. M. Diamond, eds. Ecology
and evolutionof communities.Harvard UniversityPress, Cambridge,Mass.
Cook, J. A. 1986. The mammalsof the Animas Mountainsand adjacent areas, Hidalgo County,New
Mexico. Occasional Papers of the Museum of SouthwesternBiology, Universityof New
Mexico 4:1-45.
Connor, E. F., and E. D. McCoy. 1979. The statisticsand biologyof the species-area relationship.
American Naturalist113:791-833.
Damuth, J. 1981. Populationdensityand body size in mammals.Nature (London) 290:699-700.
Dasmann, R. 1975. Biogeographicalprovinces. InternationalUnion forConservationof Nature and
Natural Resources, Occasional Papers 18:1-5.
Dial, K. P., and J. M. Marzluff. 1988. Are the smallest organisms the most diverse? Ecology
69:1620-1624.
Diamond, J. M. 1975. Assembly of species communities.Pages 342-444 in M. L. Cody and J. M.
Diamond, eds. Ecology and evolution of communities.Harvard UniversityPress, Cambridge,Mass.
1984. Historic extinction:a Rosetta Stone for understandingprehistoricextinctions.Pages
824-862 in P. S. Martinand R. G. Klein, eds. Quaternaryextinctions.Universityof Arizona
Press, Tucson.
du Toit, J. T., and N. Owen-Smith.1989. Body size, populationmetabolism,and habitatspecialization
among large Africanherbivores.AmericanNaturalist133:736-740.
Eisenberg, J. F. 1981. The mammalianradiations.Universityof Chicago Press, Chicago.
Erwin, T. L. 1983. Beetles and other insects of tropical rain forestcanopies at Manaus, Brazil,
sampled by insecticidalfogging.Pages 59-75 in S. L. Sutton, T. C. Whitmore,and A. C.
Chadwick, eds. Tropical rain forest:ecology and management.Blackwell, Edinburgh.
Finck, E. J., D. W. Kaufman,G. A. Kaufman,S. K. Gurtz,B. K. Clark, L. J. McLellan, and B. S.
Clark. 1986. Mammals of the Konza Prairieresearchnaturalarea, Kansas. PrairieNaturalist
18:153-166.
Flessa, K. W. 1975. Area, continentaldriftand mammaliandiversity.Paleobiology 1:189-194.
Foster, J. B. 1964. The evolutionof mammalson islands. Nature (London) 202:234-235.
Goodman, D. 1987. The demographyof chance extinctions.Pages 11-34 in M. E. Soule, ed. Viable
populations. CambridgeUniversityPress, Cambridge,Mass.
Grant,P. R. 1986. Ecology and evolutionof Darwin's finches.PrincetonUniversityPress, Princeton,
N.J.
Hall, E. R. 1981. The mammalsof NorthAmerica. Vols. 1 and 2. 2d ed. Wiley, New York.
Harestad, A. S., and F. L. Bunnell. 1979. Home range and body weight: a re-evaluation.Ecology
60:389-402.
Hink, V. C., and R. D. Ohmart. 1984. Middle Rio Grande biological survey.Final report,ArmyCorp
of Engineerscontractno. DACW47-81-C-0015.
Hopf, F. A., and J. H. Brown. 1986. The bullseye methodfor testingforrandomnessin ecological
communities.Ecology 67:1139-1155.
BODY SIZE AND SPECIES COMPOSITION
1511
and body weightin Africanruminants.Pages 141-150 in
Hoppe, P. P. 1977. Rumen fermentation
Proceedings of 12thCongress of Game Biologists,Atlanta,Ga.
Hutchinson,G. E. 1959. Homage to Santa Rosalia, or why are there so many kinds of animals?
AmericanNaturalist93:145-159.
Hutchinson,G. E., and R. H. MacArthur.1959. A theoreticalecological model of size distributions
among species of animals. AmericanNaturalist93:117-125.
Jablonski,D. 1986. Causes and consequences of mass extinctions:a comparativeapproach. Pages
183-229 in D. K. Elliot, ed. Dynamics of extinction.Wiley,New York.
seed passage ratesin cows and horses,surrogatePleistocenedispersal
Janzen,D. H. 1982. Differential
agents. Oikos 38:150-156.
Lomolino, M. V. 1985. Body sizes of mammalson islands: the island rule reexamined. American
Naturalist125:310-316.
MacArthur,R. H. 1972. Geographicalecology. Harper & Row, New York.
MacArthur,R. H., and E. 0. Wilson. 1967. Theory of island biogeography.PrincetonUniversity
Press, Princeton,N.J.
Martin,P. S. 1984. Prehistoricoverkill:the global model. Pages 354-403 in P. S. Martinand R. G.
Klein, eds. Quaternaryextinctions.Universityof Arizona Press, Tucson.
Martin,P. S., and R. G. Klein. 1984. Quaternaryextinctions.Universityof Arizona Press, Tucson.
May, R. M. 1978. The dynamicsand diversityof insect faunas. Pages 188-204 in L. A. Mound and
N. Waloff,eds. Diversityof insectfaunas. Blackwell, Oxford.
1988. How many species are thereon earth? Science (Washington,D.C.) 241:1441-1449.
McNab, B. K. 1963. Bioenergeticsand the determinationof home ecological significanceof Bergmann's rule. Ecology 52:845-854.
McNaughton,S. J., and N. J. Georgiadis. 1986. Ecology of Africangrazingand browsingmammals.
Annual Review of Ecology and Systematics17:39-65.
Morrison,M. L., M. P. Yoder-Williams,D. C. Erman, R. H. Barrett,M. White, A. S. Leopold,
and D. A. Airola. Natural historyof vertebratesof Sagehen Creek Basin, Nevada County,
California.AgriculturalExperimentalStation,Universityof California,Division of Agricultureand Natural Resources.
Morse, D. R., N. E. Stork, and J. H. Lawton. 1988. Species number,species abundance and body
lengthrelationshipsof arboreal beetles in Bornean lowland rain foresttrees. Ecological
Entomology13:25-37.
Nagy, K. A. 1987. Field metabolicrateand foodrequirementscalingin mammalsand birds.Ecological
Monographs57:111-128.
Orians, G. H. 1980. Micro and macro in ecological theory.BioScience 30:79.
and interspecificcompetitionin twoPacala, S., and J. Roughgarden. 1982. Resource partitioning
species insularAnolis lizard communities.Science (Washington,D.C.) 217:444-446.
1985. PopulationexperimentswiththeAnolis lizards of St. Maartenand St. Eustatius (Neth.
Antilles). Ecology 66:129-141.
in herbivores.Pages 205-229 in G.
Parra, P. 1978. Comparison of foregutand hindgutfermentation
Montgomery,ed. The ecology of arborealfolivores.SmithsonianInstitutionPress, Washington, D.C.
Peters, R. H. 1983. The ecological implicationsof body size. Cambridge UniversityPress, Cambridge.
Peters,R. H., and J. V. Raelson. 1984. Relationsbetween individualsize and mammalianpopulation
density.AmericanNaturalist124:498-517.
Pimm, S. L., H. L. Jones, and J. Diamond. 1988. On the risk of extinction.American Naturalist
132:757-785.
Preston, F. W. 1962. The canonical distributionof commonness and rarity.Ecology 43:185-215,
410-432.
Ramirez-Pulido,J., M. C. Britton,A. Perdomo, and A. Castro. 1983. Gufa de los mamiferosde
Mexico. UniversidadAut6noma Metropolitana,Mexico City, Mexico.
Rapoport, E. H. 1982. Areography:geographicalstrategiesof species. Pergamon,Oxford.
Raup, D. M. 1986. Biological extinctionin earth history.Science (Washington,D.C.) 231:15281533.
1512
THE AMERICAN NATURALIST
Ricklefs,R. E. 1987. Communitydiversity:relativeroles of local and regional processes. Science
(Washington,D.C.) 235:167-171.
Rosenzweig, M. L. 1966. Communitystructurein sympatricCarnivora. Journalof Mammalogy
47:602-612.
of mammalianbody size analyzed by continent.M.S. thesis.
Rusler,R. 1987. Frequency distributions
Universityof Arizona, Tucson.
Ryan, R. M. 1968. Mammals of Deep Canyon Colorado desert, California.Desert Museum, Palm
Springs,Calif.
Schmidt-Nielsen,K. 1984. Scaling: why is animal size so important?CambridgeUniversityPress,
Cambridge.
Schoener, T. W. 1968. Sizes of feedingterritoriesamong birds. Ecology 49:123-131.
1970. Size patternsin West Indian Anolis lizards. II. Correlationswiththe sizes of sympatric
species-displacement and convergence.AmericanNaturalist104:155-174.
1976. The species-area relationwithinarchipelagos:models and evidence fromisland birds.
Pages 629-642 in Proceedingsof the 16thInternationalOrnithologicalCongress, Canberra.
1984. Size differencesamong sympatric,bird-eatinghawks: a worldwide survey. Pages
254-281 in D. R. Strong,D. Simberloff,L. G. Abele, and A. B. Thistle, eds. Ecological
communities:conceptual issues and the evidence. PrincetonUniversityPress, Princeton,
N.J.
Schoener,T. W., and D. A. Spiller. 1988. High populationpersistencein a systemwithhighturnover.
Nature (London) 330:474-477.
Sibly, R. M. 1981. Strategiesof digestionand defecation.Pages 109-139 in C. R. Townsend and P.
Calow, eds. Physiologicalecology. Sinauer, Sunderland,Mass.
Simberloff,D., and W. Boecklen. 1981. Santa Rosalia reconsidered: size ratios and competition.
Evolution 35:1206-1228.
Timm,R. M. 1975. Distribution,naturalhistory,and parasitesof mammalsof Cook County,Minnesota. Bell Museum of Natural HistoryOccasional Papers 14:1-56.
Van Valen, L. 1973. Patternand the balance of nature.EvolutionaryTheory 1:31-40.
Webb, S. D. 1984. Ten millionyears of mammalianextinctionsin NorthAmerica. Pages 189-210 in
P. S. Martin and R. G. Klein, eds. Quaternaryextinctions.Universityof Arizona Press,
Tucson.
Whitaker,J. O., Jr.1980. The Audubon Society fieldguide to NorthAmericanmammals.Chanticleer,
New York.
Wilson, E. 0. 1988. Biodiversity.National Academy, Washington,D.C.
Wilson, M. V., and A. Shmida. 1984. Measuringbeta diversitywithpresence-absencedata. Journal
of Ecology 72: 1055-1064.
Zeveloff,S. I., and M. S. Boyce. 1988. Body size patternsin NorthAmericanmammalfaunas. Pages
123-146 in M. S. Boyce, ed. Evolution of lifehistoriesof mammals.Yale UniversityPress,
New Haven, Conn.