Climate history, human impacts and global body size of Carnivora (Mammalia: Eutheria) at multiple evolutionary scales

Abstract

Aim One of the longest recognized patterns in macroecology, Bergmann’s rule, describes the tendency for homeothermic animals to have larger body sizes in cooler climates than their phylogenetic relatives in warmer climates. Here we provide an integrative process-based explanation for Bergmann’s rule at the global scale for the mammal order Carnivora.Location Global.Methods Our database comprises the body sizes of 209 species of extant terrestrial Carnivora, which were analysed using phylogenetic autocorrelation and phylogenetic eigenvector regression. The interspecific variation in body size was partitioned into phylogenetic (P) and specific (S) components, and mean P- and S-components across species were correlated with environmental variables and human occupation both globally and for regions glaciated or not during the last Ice Age.Results Three-quarters of the variation in body size can be explained by phylogenetic relationships among species, and the geographical pattern of mean values of the P-component is the opposite of the pattern predicted by Bergmann’s rule. Partial regression revealed that at least 43% of global variation in the mean phylogenetic component is explained by current environmental factors. In contrast, the mean S-component of body size shows large positive deviations from ancestors across the Holarctic, and negative deviations in southern South America, the Sahara Desert, and tropical Asia. There is a moderately strong relationship between the human footprint and body size in glaciated regions, explaining 19% of the variance of the mean P-component. The relationship with the human footprint and the P-component is much weaker in the rest of the world, and there is no relationship between human footprint and S-component in any region.Main conclusions Bergmannian clines are stronger at higher latitudes in the Northern Hemisphere because of the continuous alternation of glacial–interglacial cycles throughout the late Pliocene and Pleistocene, which generated increased species turnover, differential colonization and more intense adaptive processes soon after glaciated areas became exposed. Our analyses provide a unified explanation for an adaptive Bergmann’s rule within species and for an interspecific trend towards larger body sizes in assemblages resulting from historical changes in climate and contemporary human impacts.

Full text

Page 1

ORIGINAL
ARTICLE
Climate history, human impacts and
global body size of Carnivora (Mammalia:
Eutheria) at multiple evolutionary scales
Jose ´ Alexandre Felizola Diniz-Filho1*, Miguel A´ngel Rodrı ´guez2,
Luis Mauricio Bini1, Miguel A´ngel Olalla-Tarraga3, Marcel Cardillo3,4,
Joa ˜o Carlos Nabout5, Joaquı ´n Hortal3and Bradford A. Hawkins6
1Departamento de Ecologia, Instituto de
Cie ˆncias Biolo ´gicas, Universidade Federal de
Goia ´s, Goiania, GO, Brazil,2Departamento de
Ecologı ´a, Universidad de Alcala ´, Alcala ´ de
Henares, Madrid, Spain,3NERC Centre for
Population Biology/Division of Biology,
Imperial College at Silwood Park, Ascot, UK,
4Centre for Macroevolution and Macroecology,
School of Biology, Australian National
University, Canberra, ACT, Australia,
5Programa de Po ´s-Graduac ¸a ˜o em Cie ˆncias
Ambientais, CIAMB, UFG, Goia ˆnia, GO,
Brazil and6Department of Ecology and
Evolution, University of California, Irvine,
CA, USA
*Correspondence: Jose ´ Alexandre Felizola Diniz-
Filho, Departamento de Ecologia, Instituto de
Cie ˆncias Biolo ´gicas, Universidade Federal de
Goia ´s, Cx.P. 131, 74001-970 Goiania,
GO, Brazil.
E-mail: diniz@icb.ufg.br
ABSTRACT
Aim One of the longest recognized patterns in macroecology, Bergmann’s rule,
describes the tendency for homeothermic animals to have larger body sizes in
cooler climates than their phylogenetic relatives in warmer climates. Here we
provide an integrative process-based explanation for Bergmann’s rule at the
global scale for the mammal order Carnivora.
Location Global.
Methods Our database comprises the body sizes of 209 species of extant
terrestrial Carnivora, which were analysed using phylogenetic autocorrelation and
phylogenetic eigenvector regression. The interspecific variation in body size was
partitioned into phylogenetic (P) and specific (S) components, and mean P- and
S-components across species were correlated with environmental variables and
human occupation both globally and for regions glaciated or not during the last
Ice Age.
Results Three-quarters of the variation in body size can be explained by
phylogenetic relationships among species, and the geographical pattern of mean
values of the P-component is the opposite of the pattern predicted by Bergmann’s
rule. Partial regression revealed that at least 43% of global variation in the mean
phylogenetic component is explained by current environmental factors. In
contrast, the mean S-component of body size shows large positive deviations
from ancestors across the Holarctic, and negative deviations in southern South
America, the Sahara Desert, and tropical Asia. There is a moderately strong
relationship between the human footprint and body size in glaciated regions,
explaining 19% of the variance of the mean P-component. The relationship with
the human footprint and the P-component is much weaker in the rest of the
world, and there is no relationship between human footprint and S-component
in any region.
Main conclusions Bergmannian clines are stronger at higher latitudes in
the Northern Hemisphere because of the continuous alternation of glacial–
interglacial cycles throughout the late Pliocene and Pleistocene, which generated
increased species turnover, differential colonization and more intense adaptive
processes soon after glaciated areas became exposed. Our analyses provide a
unified explanation for an adaptive Bergmann’s rule within species and for an
interspecific trend towards larger body sizes in assemblages resulting from
historical changes in climate and contemporary human impacts.
Keywords
Anthropogenic effects, Bergmann’s rule, body size, Carnivora, climate, Cope’s
rule, human footprint, phylogenetic effects, phylogenetic eigenvector regression.
Journal of Biogeography (J. Biogeogr.) (2009) 36, 2222–2236
2222
www.blackwellpublishing.com/jbi
doi:10.1111/j.1365-2699.2009.02163.x
ª 2009 Blackwell Publishing Ltd

Page 2

INTRODUCTION
The integration of ecology and evolutionary biology within the
macroecology research programme (Brown, 1995; Gaston &
Blackburn, 2000; Blackburn & Gaston, 2003) is shedding new
light on patterns known since the 19th century, especially those
related to ‘latitudinal’ diversity gradients and ecogeographical
and evolutionary rules such as Bergmann’s rule (Gaston et al.,
2008). Bergmann’s rule states that, within groups of phylo-
genetically related homeothermic animals, organisms living in
colder climates are generally larger than those living in warmer
climates. Many recent papers have discussed the ecological and
evolutionary mechanisms that may explain this pattern, both
at intraspecific (i.e. across populations) and at interspecific
(i.e. across multispecies assemblages) levels (Partridge &
Coyne, 1997; Blackburn et al., 1999; Ashton et al., 2000;
Freckleton et al., 2003; Meiri & Dayan, 2003; Meiri et al., 2004,
2007; Rodrı ´guez et al., 2006, 2008; Olalla-Ta ´rraga & Rodrı ´-
guez, 2007; Pincheira-Donoso et al., 2008; Olson et al., 2009).
A common view is that Bergmannian patterns can be
generated by adaptive processes operating in a strict Darwinian
sense, so that, within populations of species that inhabit cooler
environments, large-bodied individuals are favoured owing to
their lower surface area-to-volume ratios and, hence, their
higher body-heat retention (heat conservation hypothesis),
and/or because they metabolize fat stores at lower weight-
specific rates and thus may cope better with resource shortages
(resource availability hypothesis; e.g. Rodrı ´guez et al., 2006;
but see Guillaumet et al., 2008). Thus, assuming that these
selective processes drive body size variation within species,
interspecific Bergmannian clines would arise because assem-
blages in cold areas would tend to be composed of large-
bodied individuals of each species, and/or of larger species
generated by adaptive processes creating new large-bodied taxa
in the cooler parts of ancestral species ranges (see also Davies
et al., 2007, for potential mechanisms, including biotic inter-
actions). In the latter case, Bergmannian patterns would be
associated with positive deviations (increases) from ancestral
body sizes expressed independently in different species inhab-
iting a region.
On the other hand, geographical patterns of body size
variation at the interspecific or assembly level (sensu Gaston
et al., 2008; see also Blackburn & Hawkins, 2004) could also be
generated by other processes directly shaping the body size
frequency distributions (BSFDs) of assemblages (Blackburn &
Hawkins, 2004; Diniz-Filho, 2004; Smith et al., 2004; Olalla-
Ta ´rraga et al., 2006; Diniz-Filho et al., 2007; Olalla-Ta ´rraga &
Rodrı ´guez, 2007; Ramirez et al., 2008; Rodrı ´guez et al., 2008),
in addition to the classical adaptive processes at the population
level described above. Under this view, assemblage-level body
size patterns could arise from ‘species sorting’ mechanisms,
including selective extinctions and/or contractions or expan-
sions of species geographical ranges.
Species sorting processes may also explain Cope’s rule, that
is, the trend towards increasing body size within a lineage
through evolutionary time (see Alroy, 1998; Demetrius, 2000).
For example, in the cool and less stable climates of the
Northern Hemisphere, the adaptive advantages of large body
size have potentially triggered the evolution of hypercarnivory
(i.e. dietary specialization in which diet is composed of at least
70% flesh; see Van Valkenburgh, 1999, 2007). However, this
increase in body size may also lead to high extinction rates and
species turnover, which could explain the right-skewed BSFD
observed in carnivoran faunas in North America (e.g. Smith
et al., 2004; Van Valkenburgh et al., 2004). Thus, this process
may shift the BSFD independently of intraspecific adaptive
responses to climate, and would be better viewed as cladoge-
netic mechanisms creating assemblage patterns that result
from a life-history threshold related to resource use. Moreover,
higher extinction rates biased towards large-bodied species
would generate lower mean body sizes for assemblages in the
Northern Hemisphere and hence give rise to inverse Berg-
mannian patterns (see Hunt & Roy, 2006). That is, these
mechanisms predict an inverse outcome to that generated by
the selective pressures that past climates may have exerted on
the body sizes of individual species.
Finally, recent human impacts may also have disrupted the
BSFD of faunas by shifting geographical ranges and eliminat-
ing some species, especially large-bodied ones, from local
assemblages. Therefore, human impact should be considered as
an additional explanation for broad-scale patterns in body size.
Indeed, recent analyses suggest that extinction risks are
predicted by an interaction between intrinsic biological traits
and exposure to external effects (such as human occupation;
Cardillo et al., 2004, 2005). Large-bodied mammal species (i.e.
those with body masses larger than 3 kg) tend to have
relatively higher extinction risks owing to their life-history
traits, which means that human impacts can alter the BSFD at
an assemblage level in a similar way to changes driven by
macroevolutionary processes (i.e. they may increase the
extinction rates of large-bodied species and generate a right-
skewed BSFD). These assemblage-level patterns would be a
consequence of the reduction and fragmentation of species
geographical ranges that have increased extinction rates at local
or regional spatial scales. We might expect to find a disturbed
BSFD in highly impacted regions as a result of the selective loss
of large-bodied species, especially on the east coast of North
America and in Europe (Sanderson et al., 2002; Haberl et al.,
2004).
Here we investigate the extent to which these multiple
processes can be integrated to explain global geographical
patterns in body size of the eutherian mammalian order
Carnivora. Specifically, we use phylogenetic comparative
methods to decouple geographical variation in species body
sizes into within-species and macroevolutionary (interspecific)
components that can be explained by recent and past
adaptations to environmental conditions, macroevolutionary
trends, and patterns generated by modern human impacts.
Carnivora are an ideal group for macroecological and macro-
evolutionary analyses, as their geographical distributions and
phylogenetic relationships are well known (Bininda-Emonds
et al., 1999; Johnson et al., 2006), and they show large
Global patterns in Carnivora body size
Journal of Biogeography 36, 2222–2236
ª 2009 Blackwell Publishing Ltd
2223

Page 3

variation in body sizes and life-history traits (Gittleman, 1985,
1986; Webster et al., 2004; Friscia et al., 2007). Furthermore,
they have been extensively studied for intraspecific Bergman-
nian patterns (Meiri et al., 2004, 2007, 2009) and for
macroevolutionary trends in body size (Kelt & Brown, 1998;
Van Valkenburgh et al., 2004). Finally, because of their
relatively large body size, life-history characteristics and
ecological specialization, Carnivora usually have high extinc-
tion rates (Werdelin & Lewis, 2005; Leonard, 2007) and may
be particularly sensitive to recent human impacts (Cardillo
et al., 2004), being thus an important group for biodiversity
conservation.
MATERIALS AND METHODS
Body size data
Our database comprises the 209 species of extant terrestrial
Carnivora that are native to either continental regions
(Australia was excluded) or the islands of Baffin, Tierra del
Fuego, Great Britain and Sumatra, which we consider large
enough and close enough to mainlands that macroecological
and macroevolutionary patterns are not affected by island
effects (see Meiri et al., 2005) (for a list of all study species see
Appendix S1 in Supporting Information). The body size of
each species was measured as its log10-transformed mean body
mass (in grams) as provided in Meiri et al. (2005) and Smith
et al. (2003). Because we are primarily interested in broad-
scale comparative patterns, we did not differentiate between
males and females, although some species are sexually dimor-
phic. Species geographical ranges (extents of occurrence) were
obtained from a global database of mammal distributions
(Sechrest, 2003; Grenyer et al., 2006), and the presence/
absence of each species was mapped onto an equal-area grid
(Behrmann global projection) comprising 12,580 cells of
96.5 · 96.5 km each (i.e. of about one-degree resolution),
which constituted the grain size of our study.
Following the standard approach for analyses of body size at
the assemblage level (see Blackburn & Hawkins, 2004; Olalla-
Ta ´rraga et al., 2006; Diniz-Filho et al., 2007; Olalla-Ta ´rraga &
Rodrı ´guez, 2007; Ramirez et al., 2008; Rodrı ´guez et al., 2008),
we generated the variable ‘mean body size’ by averaging the
(log10-transformed) body sizes of the species present in each
grid cell. This variable, partitioned into its phylogenetic and
specific components (see below), constituted the basis of our
interspecific (assemblage) analysis of the spatial variation of
body size at the global extent (see Fig. 1). Results from analyses
based on medians and means of untransformed body masses
are similar because body size is log-normally distributed.
Environmental predictors
Initially, five environmental variables were generated for each
cell in the grid: (1) mean annual temperature (TEMP); (2)
annual precipitation (PREC); (3) annual actual evapotranspi-
ration (AET, following Ahn & Tateishi’s, 1994, formulation);
(4) the global vegetation index (GVI – an indicator of standing
plant biomass obtained from radiometer data from the NOAA
polar-orbiting environmental satellites and related to the
density and greenness of the plant canopy, total standing
biomass, green leaf-area index and percentage vegetation
cover); and (5) range in elevation (RELEV – the difference
between maximum and minimum elevations within each grid
cell, reflecting mesoscale climatic gradients). Details and data
sources for these environmental variables are given by Rodrı ´-
guez et al. (2005, 2006, 2008) and Olalla-Ta ´rraga & Rodrı ´guez
(2007). These variables can be explicitly linked to hypotheses
(see Rodrı ´guez et al., 2006, 2008) previously developed to
explain associations between climate and body size, including
heat conservation (TEMP), heat dissipation (PREC and AET),
resource availability (GVI), and habitat availability (RELEV).
However, a principal components analysis on these variables
revealed that they could be reduced to two main dimensions,
based on the broken-stick distribution of eigenvalues (Jackson,
1993). Because most ecological interpretations for Bergmann’s
rule are related to temperature (Rodrı ´guez et al., 2006, 2008),
we used TEMP and RELEV to express the main directions of
variation in the data (see Table S1 in Appendix S2).
To examine anthropogenic-driven disruptions of geograph-
ical ranges that may also have affected body size frequency
distributions at the assemblage scale (see Cardillo et al., 2004,
2005), we included human footprint (HUMANS) as an
explanatory variable. This variable consisted of cell averages
of the biome-normalized footprint values generated by
Sanderson et al. (2002) at 1-km resolution by combining
global records of population density, land use, transport access
(roads, rivers, etc.) and electrical power infrastructure
(data available at http://www.ciesin.columbia.edu/wild_areas/;
accessed October 2008).
We hypothesized that mean body size gradients would be
stronger in the Northern Hemisphere because of greater
climatic instability throughout the late Pliocene and Pleisto-
cene (e.g. Arau ´jo et al., 2008). Thus, we partitioned the global
dataset according to the stability of temperatures since the Last
Glacial Maximum (LGM), which we calculated as the differ-
ence between current mean annual temperatures and those of
21,000 years ago as estimated by the ECHAM3 palaeoclimatic
model (see Arau ´jo et al., 2008; Braconnot et al., 2007; and
http://pmip.lsce.ipsl.fr). Before temperature stability was cal-
culated, original palaeo-temperature data were downscaled to
the 96.5-km resolution used herein using a mean-mobile
technique to generate a continuous downscaled temperature
surface from the centroids of the original surface, ensuring that
the main geographical trends in temperature were retained in
the downscaled data. Temperature stability was subjected to a
K-means non-hierarchical clustering (Legendre & Legendre,
1998) that generated a two-cluster solution encapsulating both
the dramatic environmental changes experienced in the north
since the LGM (n = 2249 cells) and the relatively greater
climatic stability of the rest of the world (n = 10,324 cells). For
simplicity, we refer to these groups of cells found by K-means
clustering as GLACIATED and NON-GLACIATED regions.
J. A. F. Diniz-Filho et al.
2224
Journal of Biogeography 36, 2222–2236
ª 2009 Blackwell Publishing Ltd

Page 4

Finally, we divided the 209 species into two body size classes,
distinguishing species weighing less or more than 3 kg, a size-
threshold above which extinction risk has been predicted to
increase sharply owing to intrinsic and extrinsic factors
(Cardillo et al., 2005). We calculated for each cell in our
global grid the species richness of large-bodied (> 3 kg) and
small-bodied (< 3 kg) species and correlated the richness
patterns with the human footprint.
Phylogenetic comparative analyses
Phylogenetic patterns in species body sizes were initially
evaluated using Moran’s I coefficients (Gittleman & Kot, 1990;
Gittleman et al., 1996; Diniz-Filho, 2001; Pavoine et al., 2008),
given by
? ?P
where n is the number of species, yiand yjare body size values
for species i and j, ? y is the average body size across all species
I ¼
n
S
i
P
jyi? ? y
P
ðÞ yj? ? y
Þ2
??wij
iyi? ? y
ð
"#
;
and wij is an element of the matrix W. In matrix W, wij
elements are equal to 1 for all i, j species pairs within a given
phylogenetic distance interval, and to 0 otherwise. S indicates
the number of pairs of species connected in the W matrix. The
value expected under the null hypothesis of the absence of a
phylogenetic autocorrelation is )1/(n ) 1), and the statistical
significance can be established under a normal approximation
(Legendre & Legendre, 1998) or using randomization (Manly,
1998). High positive Moran’s I coefficients indicate that species
separated by a given distance in the phylogeny are similar for
the trait under study, whereas high negative values indicate
that species pairs are dissimilar.
calculated for eight distance classes, connecting in the matrix
W pairs of species situated at increasing intervals of c. 5 Myr.
Thus, a series of Moran’s I coefficients is obtained, which when
plotted against phylogenetic distances generates a correlogram.
The W matrices were obtained by deconstructing a pairwise
patristic distance matrix derived from the Carnivora supertree
(Bininda-Emonds et al., 1999). Although other phylogenies for
some particular groups of Carnivora are available (e.g. Johnson
Moran’s I coefficients were
Figure 1 Representation of the approach used to map body size components of 209 species of Carnivora. The first part of the figure
shows the procedures involved in the phylogenetic eigenvector regression (PVR): (1) back-transform the phylogeny into a phylogenetic
distance matrix; and (2) double-centre this matrix and compute a principal coordinates analysis (PCoA). The transition between the first
and second parts illustrates how the eigenvectors were selected. The second part shows how phylogenetic (P) and specific (S) components
were estimated: (1) run a multiple linear regression between body size (matrix Y) and the selected eigenvectors (X); and (2) save the
estimated values and residuals, as they represent the phylogenetic (P) and specific (S) components, respectively. These values (in addition to
original body size) are used to estimate means per cell. Then, as indicated in the last part, the vectors containing different components of
body size are regressed against environmental factors (E) in an explicit spatial context.
Global patterns in Carnivora body size
Journal of Biogeography 36, 2222–2236
ª 2009 Blackwell Publishing Ltd
2225

Page 5

et al., 2006), we use the Bininda-Emonds et al. (1999)
supertree because it provides a complete phylogeny of species.
Moreover, incorporating these new phylogenies would require
creating a new supertree, which is beyond the scope of this
paper. We believe that the effects of any inaccuracies in this
supertree on our results are minimal. Indeed, recent phyloge-
netic autocorrelation analyses of body size variation based on
the new molecular phylogeny of felids (Johnson et al., 2006)
and the supertree (Bininda-Emonds et al., 1999) gave almost
identical results (Diniz-Filho & Nabout, 2009).
Once phylogenetic patterns in body size are described using
phylogenetic autocorrelation (Moran’s I), it is possible to use
linear models to partition the total variation (T) of a trait (i.e.
body size in this case) into (1) a phylogenetic component (P),
which expresses the expected species trait values if current
values are entirely determined by the value in all species
weighted by their phylogenetic relationships, and (2) a unique,
or specific, component (S), which expresses deviations from the
phylogenetic expectation arising from measurement error and
the independent evolution of each species after speciation
events (see Cheverud et al., 1985; Gittleman & Kot, 1990;
Gittleman & Luh, 1992; Fig. 1). Although the interpretation of
evolutionary mechanisms underlying these two components
remains controversial (see Westoby et al., 1995; Desdevises
et al., 2003), they can be useful for disentangling patterns
shared by all species in a clade (the P-component) and
independent patterns of each species (the S-component).
Moreover, any correlation between the S-components of
different traits is a good estimate of the ‘input’ correlation
(sensu Martins & Garland, 1991), which is the correlation
between simultaneous changes of two traits at each time-step of
a phylogeny. This is indeed the correlation estimated by all
phylogenetic comparative methods, including Felsenstein’s
(1985) independent contrasts method (see Martins et al.,
2002, and Diniz-Filho & To ˆrres, 2002, for a comparative
evaluation of the statistical performance of different methods
in estimating this parameter under alternative evolutionary
models).
The partitioning of T into P- and S-components can be
done using a range of statistical techniques, including auto-
regressive models (Cheverud et al., 1985; Gittleman & Kot,
1990) or mixed models (Lynch, 1991). Here we use phyloge-
netic eigenvector regression (PVR) (Diniz-Filho et al., 1998),
which uses a principal coordinates analysis (PCoA) (see
Legendre & Legendre, 1998) to describe the phylogenetic
structure among taxa by a set of eigenvectors (X) extracted
from a matrix of pairwise phylogenetic distances among
species (Fig. 1). Formally, the response variable Y (body size)
is regressed against the phylogenetic structure following the
model
Y = Xb + e,
where X is the matrix with eigenvectors and b are the
regression coefficients of each eigenvector on Y, so that Xb
corresponds to the P-component (the Y-values estimated by
the PVR model) and the model residuals e correspond to the
S-component. The coefficient of determination (R2) of this
linear model measures the amount of variation in body size
explained by the phylogeny (phylogenetic inertia, or signal;
Diniz-Filho et al., 1998, 2007; see also Freckleton et al., 2002).
Finally, a phylogenetic correlogram was used to test the
assumption that phylogenetic effects were not present in the
model residuals, as suggested by Gittleman & Kot (1990).
PVR is part of a family of eigenvector-based techniques
whose main purpose is to describe spatial (Borcard &
Legendre, 2002; Griffith, 2003; Borcard et al., 2004; Diniz-
Filho & Bini, 2005; Griffith & Peres-Neto, 2006) or phyloge-
netic (Diniz-Filho et al., 1998; Desdevises et al., 2003) pat-
terns. In terms of phylogenetic distances, each eigenvector
numerically expresses groups of species that are similar at
distinct phylogenetic levels, and the first eigenvectors tend to
describe variation among deeper nodes of the phylogeny.
Successive increases in the number of eigenvectors tend to
better approximate the relationship among species, and this
can be evaluated by a matrix correlation between the original
phylogenetic distances and pairwise distances in the reduced
eigenvector space. The advantage of this approach, however, is
that the phylogenetic relationships are now expressed by a set
of eigenvectors (and not as pairwise distances) that can be used
directly in any form of linear or non-linear model (e.g. see
Legendre et al., 2005, for a similar reasoning in community
ecology).
We performed the PCoA by extracting eigenvalues and
eigenvectors from the double-centred (see Legendre & Legen-
dre, 1998; Desdevises et al., 2003) phylogenetic distance matrix
derived from the Carnivora supertree (Bininda-Emonds et al.,
1999). We used for further modelling those eigenvectors with
significant correlations (P < 0.01) with species body sizes
(Griffith, 2003; see Table S2 in Appendix S2). The values
estimated by a multiple regression analysis of body size against
these eigenvectors represent the vector with the phylogenetic
components (P), whereas the residuals of this regression model
give the vector with the specific components (S) (see Fig. 1).
The interpretation of the S-component as reflecting unique
components of species variation independently of trait values
in other species is analogous to evaluating the departure of
trait values in each species from its ancestral states. This was
explicitly verified by correlating the S-component with the
size change index (SCI) of Webster et al. (2004) for 117
species common to both studies. This index was derived by
modelling ancestral body sizes across the Carnivora phylogeny
and by directly comparing evolution from the most recent
common ancestor (MRCA) and extant species. Moreover,
published measures of the intraspecific form of Bergmann’s
rule for 42 species (i.e. the correlations between body size and
latitude provided in Meiri et al., 2004, 2007) were correlated
with the S-component values of PVR that we obtained for
these species. [Note that Meiri et al. (2004, 2007) used
latitude to approximate Bergmann’s rule, and this may be
problematic when dealing with regions in which latitude is
not a good surrogate for temperature. Even so, this may be a
reasonable general proxy for intraspecific Bergmann gradients
in the group.]
J. A. F. Diniz-Filho et al.
2226
Journal of Biogeography 36, 2222–2236
ª 2009 Blackwell Publishing Ltd

Page 6

Based on the phylogeny, we also computed the age of the
root from each species to its MRCA with other living species,
and calculated a mean age for each cell. We randomized
species MRCA values to generate a null distribution of these
values within each cell, and then analysed the deviation
between observed and null MRCA values, to establish if, on
average, species in a given cell are younger than expected by a
random association of species (see Greve et al., 2008).
Geographical analyses
As for the case of mean body size, we also calculated the mean
values of P and S in each cell (Diniz-Filho et al., 2007; see also
Fig. 1), and regressed each of these new variables against the
set of environmental predictors (TEMP, RELEV and human
footprint). Because relationships of both mean P- and mean
S-values with TEMP were curvilinear at the global scale, we
introduced a squared term of TEMP in the global models.
We also partitioned the effects of these environmental
predictors and human footprint on body size components (i.e.
mean P and mean S) using a series of partial regressions based
on the adjusted R2(Legendre & Legendre, 1998). The relative
strengths of the predictor variables were assessed according
to their mean standardized regression coefficient in the full
models. These coefficients were generated as weighted averages
(we used the Akaike weight index as the weighting variable; see
Burnham & Anderson, 2002) of the coefficients of all possible
models that can be obtained with the analysed predictors. This
procedure tends to avoid problems in finding minimum
adequate models because of high levels of uncertainty in
predictor choice (Diniz-Filho et al., 2008). Because of the large
number of cells, even within regions, it is difficult to deal with
statistical inaccuracies caused by spatial autocorrelation in the
analysisofenvironmentaldriversofbodysizepatterns.Although
problemsduetoautocorrelationinestablishingTypeIerrorsare
well known (Legendre & Legendre, 1998), unbiased estimates of
regression coefficients are obtained at very large sample sizes,
and thus the relative importance of predictors can be safely
established by ordinary least squares (OLS), especially
when uncertainty is reduced by using model averaging (see
Diniz-Filho et al., 2008; see also Hawkins et al., 2007).
All geographical analyses were performed for the global data
and for data divided into GLACIATED and NON-GLACI-
ATED regions. All analyses were performed in sam 3.0 (Rangel
et al., 2006), freely available at http://www.ecoevol.ufg.br/sam.
RESULTS
Phylogenetic patterns in body size
The Moran’s I correlogram for species body sizes shows that
closely related species are very similar (I = 0.988), and this
similarity tends to decrease with time, stabilizing after c.
20 Myr (Fig. 2). Thus, there is a strong phylogenetic signal in
the data, and a large proportion of the variation in Carnivora
body size (R2= 78.2%) can be explained by the PVR model.
The phylogenetic structure was expressed by 21 eigenvectors
extracted from the phylogenetic distances that are significantly
correlated with body size (see Table S2 in Appendix S2). The
pairwise similarity between species in this reduced dimensional
space formed by the 21 eigenvectors was strongly correlated
(r = 0.992) with the original phylogenetic distances, so there
was almost no loss of phylogenetic information during the
eigenanalysis.
The residual variation of the PVR constitutes the specific (S)
component and expresses the amount of body size evolution
that cannot be predicted by phylogenetic relationships, plus
measurement error of species traits. More importantly,
Moran’s I coefficients for the S-component were not statisti-
cally significant and, therefore, this component can be
interpreted as phylogenetically independent body size variation
among species (Fig. 2).
The interpretation of the S-component as indicating unique
species variation independently of other species is supported
by the high correlation between S-component values and the
size change index generated by Webster et al. (2004) for 117
species (r = 0.76; P << 0.001; Fig. 3). Confirming the findings
of Freckleton et al. (2003), there was no significant relation-
ship between the published intraspecific Bergmann’s rule
strengths for 42 species (Meiri et al., 2004, 2007) and species
body size values (T variation; r = 0.23; P = 0.135; Fig. 4a).
However, there was a low but statistically significant correla-
tion between S-component values and the intraspecific Berg-
mann’s rule (r = 0.30; P = 0.047; Fig. 4b). This indicates that
intraspecific geographical variation in body size consistent with
Bergmann’s rule tends to emerge when a species deviates
significantly from its ancestral body size.
Geographical patterns in mean body size
Maps for mean body size (Fig. 5a) and its mean P-component
(Fig. 5b) are very similar across the world, and the correlation
Figure 2 Phylogenetic correlograms built with Moran’s I coeffi-
cients for the body size of 209 Carnivora species (solid circles)
based on Bininda-Emonds et al.’s (1999) supertree and for the
residuals of the phylogenetic eigenvector regression (PVR) model
(crosses).
Global patterns in Carnivora body size
Journal of Biogeography 36, 2222–2236
ª 2009 Blackwell Publishing Ltd
2227

Page 7

between the two variables is very high (r = 0.902), which was
expected given the strong phylogenetic signal detected by PVR
and Moran’s I coefficients. On average, large-bodied species
tend to occur in the tropics, and smaller species in temperate
regions (see Fig. 5a). Partial regressions revealed that c. 43%
of the global variation in the mean phylogenetic component
is explained by the combination of current environmental
factors, with high positive standardized regression coefficients
for both temperature (including a squared term) and range in
elevation, a surrogate for mesoscale climatic gradients
(Table 1).
In contrast, the mean S-component of body size (Fig. 5c)
shows large positive deviations from ancestors across the
Holarctic and in northern sub-Saharan Africa, and negative
deviations in southern South America, the Sahara Desert, and
tropical Asia. In addition, although the explanatory power of
the environmental model for this component is relatively low
at the global scale (c. 13%), the coefficient for temperature
is negative (Table 1), as expected under climatically driven
adaptive processes generating Bergmannian gradients within
lineages (Diniz-Filho et al., 2007).
The patterns differed when glaciated and non-glaciated
regions were analysed separately. The proportion of the
variance in the mean S-component explained by our environ-
mental predictors was 39% in the glaciated regions but only
5% in non-glaciated regions, with the highest regression
coefficients corresponding to temperature and having a
negative sign in both cases (Table 1). The comparison of
mean MRCA values between glaciated and non-glaciated areas
(Fig. 6) shows that, in glaciated regions, species tend to be
younger than expected by chance, as would be expected if there
had been more faunal turnover.
The influence of climate independent of human effects in
the phylogenetic (mean P) component is much lower in
glaciated (6%) than in non-glaciated (31%) areas, and this
component is positively related to temperature in both regions
(see Table 1). This can be interpreted as indicating stronger
environmentally driven selection of lineages in areas unaffected
by glaciation. However, it should also be noted that the
regression coefficients of the model for the phylogenetic
component reveal a stronger and negative impact of human
footprint in glaciated regions (see Table 1), which may have
influenced our ability to identify the signal left by environ-
mental selection on the lineages. Indeed, partial regressions
suggest that this may well be the case, as they showed that
shared effects between human footprint and climate described
nearly half (20%) of the total variance explained by the model
(46%). This is consistent with the strong correlation between
temperature and the human footprint (r = 0.787 in glaciated
regions, as compared with a much lower correlation in non-
glaciated areas: r = 0.158), and it may explain why tempera-
ture, being negatively correlated with the mean phylogenetic
component in glaciated areas (r = )0.457), had a positive
(a)
(b)
Figure 4 Relationships between within-species Bergmann’s rule
(r from Meiri et al., 2004) and (a) total variation (T) and (b)
specific (S) components from phylogenetic eigenvector regression
(PVR), for 42 species of Carnivora common to this study and
Meiri et al.’s (2004). The correlation with the T-component is not
significant (r = 0.23; P = 0.135), but that with the S-component is
(r = 0.30; P = 0.047).
Figure 3 Correlation between S (from phylogenetic eigenvector
regression) and Webster et al.’s (2004) size change index (SCI), for
117 species of Carnivora common to this study and Webster
et al.’s (2004) (r = 0.76; P < 0.001).
J. A. F. Diniz-Filho et al.
2228
Journal of Biogeography 36, 2222–2236
ª 2009 Blackwell Publishing Ltd

Page 8

coefficient in the multiple regression. Moreover, these results
are consistent with temperature-driven selective processes of
the lineages giving rise to Bergmannian patterns in the
glaciated areas.
Although most of the variation of the mean P-component in
glaciated areas is explained by independent and overlapping
effects of the human footprint and climate, the effects of the
human footprint on this component are much weaker in the
rest of the world, as is the relationship of the human footprint
and the S-component, regardless of the area (Table 1). A
deconstructive approach of species richness provides a better
visualization of the relationship between body size patterns
and human occupation. Species richness for Carnivora
weighing less or more than 3 kg is positively correlated with
the human footprint both globally (r = 0.276 and 0.191,
respectively) and non-glaciated areas (r = 0.225 and 0.198).
However, for glaciated regions, although the human footprint
is more strongly positively correlated with the richness of small
species (r = 0.552), the correlation with the richness of larger
species is negative (r = )0.223). This suggests that dense
human occupation in previously glaciated areas promotes the
presence of small-bodied species while it limits the presence of
larger species (Fig. 7).
DISCUSSION
Global geographical patterns in assemblages
As found in many studies, across-lineage body size variation in
terrestrial Carnivora is strongly phylogenetically conserved
(Gittleman et al., 1996; Diniz-Filho & To ˆrres, 2002; Freckleton
& Jetz, 2009). Consequently, maps of mean body size
and mean P-component are quite similar, showing that,
on average, large-bodied assemblages occur in the tropics
and small-bodied assemblages occupy temperate regions (see
Fig. 5). Although a recent study by Olson et al. (2009)
(a)
(b)
(c)
(d)
Figure 5 Geographical patterns of (a) mean
body size and its (b) phylogenetic and (c)
specific (S) components for the 209 extant
species of terrestrial Carnivora native to
continental regions or the large islands of
Baffin, Tierra del Fuego, Great Britain and
Sumatra. The insert (d) shows the glaciated
and non-glaciated regions considered in the
analyses.
Global patterns in Carnivora body size
Journal of Biogeography 36, 2222–2236
ª 2009 Blackwell Publishing Ltd
2229

Page 9

supported Bergmann’s rule for birds globally, our findings on
Carnivora body size variation are counter to Bergmann’s rule.
This inverted pattern was also observed for non-volant New
World mammals by Rodrı ´guez et al. (2008). This is certainly
the main pattern emerging from our analyses. We also found a
strong phylogenetic basis for BSFD changes at the faunal level,
in agreement with the proposition that temperature-driven
Bergmannian trends may occur within cold regions but not
across the entire globe (Rodrı ´guez et al., 2006; see also
Blackburn & Hawkins, 2004). As described below, under-
standing these global patterns thus requires integrating argu-
ments involving adaptation to past and recent climates,
selective species extinctions, and variable species turnover
rates in different parts of the world.
A potential explanation for this inverse pattern uses the logic
of the macroevolutionary models developed to explain the
evolution of body size for North American species, which have
been based on a life-history threshold that results in high
extinction rates in large-bodied species (generating a higher
relative diversification rate in small-bodied species; Van
Valkenburgh et al., 2004; see also Carbone et al., 1999;
Carbone & Gittleman, 2002). We can hypothesize that trends
towards larger body sizes related to Bergmannian adaptation to
cooler environments (expressed in the S-component) and the
evolution of hypercarnivory have generated a higher turnover
of species and higher taxa through time in these climatically
unstable northern areas (Alroy, 1998; Kelt & Brown,1998; see
also Fig. 5). Because both a higher turnover of large-bodied
species and greater diversification rates of small-bodied species
would tend to generate lower mean body sizes across
assemblages, they can explain the smaller average body size
of the species pools in colder regions, and, hence, converse
Bergmann trends at the global scale. This interpretation is
supported by the comparison of mean MRCA values of
glaciated and non-glaciated areas (see Fig. 6), which shows
that species tend to be younger in glaciated regions than
expected by chance: this is consistent with there having been
more faunal turnover.
Within colder regions, long-term adaptation to changing
(generally cooling) environments and niche conservatism may
have initially triggered the evolution of past larger body sizes in
ancestral species, with other taxa subsequently diversifying
from them. Under a niche conservatism model (see Wiens &
Donoghue, 2004; Diniz-Filho et al., 2007; Losos, 2008), these
new taxa would have continued being large-bodied. That is,
contemporary mean body size patterns can be interpreted both
as a consequence of adaptive changes tracking climatic events
in the past and, to a large extent, as a consequence of
phylogenetic inertia in body size and niche conservatism after
past adaptation (Diniz-Filho & Bini, 2008). If this occurred
during the progressive cooling of the late Cenozoic and the
Pleistocene in the northern parts of the world, it would have
generated a gradient of selective advantage for large body sizes,
thus establishing a potential link between Cope’s and Berg-
mann’s rules (Hunt & Roy, 2006).
Despite the fact that no relationship was found between the
P-component and the human footprint (see below), it is
important to note that the footprint reflects current patterns of
interference on faunal patterns and not historical ones. Past
Table 1 Adjusted coefficients of determination of climatic variables and human footprint (HUMANS) predicting values of phylogenetic
(P) and specific (S) components from phylogenetic eigenvector regression (PVR) analysis for 209 Carnivora species, and partition of their
independent and overlapped effects by partial regression, for global data and for glaciated and unglaciated regions. Climatic variables used
were mean annual temperature (TEMP) and range in elevation (RELEV), given as the difference between maximum and minimum
elevations within each grid cell and reflecting mesoscale climatic gradients.
DatasetComponentR2
Partial R2
Standardized regression coefficients
Climate
Shared
(Climate + HUMANS) HUMANSTEMPTEMP2
RELEV HUMANS
GlobalP
S
P
S
P
S
0.435
0.127
0.455
0.386
0.308
0.046
0.434
0.116
0.059
0.137
0.307
0.045
)0.019
0.004
0.203
0.249
)0.025
)0.005
0.020
0.007
0.193
0.000
0.026
0.006
0.191
)0.646
0.172
)0.615
0.599
)0.203
0.499
0.414
–
–
–
–
0.37
)0.14
0.247
)0.011
0.341
)0.188
)0.162
0.100
)0.712
)0.025
)0.167
0.076
Glaciated
Unglaciated
Figure 6 Box-plot of most recent common ancestor (MRCA)
values derived from the Carnivora supertree, expressed as Z-
deviations from a null distribution within cells of glaciated
(n = 2256 cells) and non-glaciated (n = 10,324 cells) regions of
the world (see Fig. 5).
J. A. F. Diniz-Filho et al.
2230
Journal of Biogeography 36, 2222–2236
ª 2009 Blackwell Publishing Ltd

Page 10

megafaunal extinctions may have been more intense in
northern parts of the world, in Europe and North America
(as well as in Australia), but there is evidence for anthropo-
genically driven extinction events in the Pleistocene mammal
megafauna in different regions of the world (Martin, 1984;
Alroy, 2001; Lyons et al., 2004; but see de Vivo & Carmignotto,
2004). Furthermore, the persistence of large mammals in
Africa and southern Asia suggests a weaker effect of human-
caused Pleistocene extinctions in these regions, usually attrib-
uted to a longer co-existence of humans and these faunas
(Nieto et al., 2005; Leonard, 2007). The survival of these large
mammals has often been explained either as a result of a long
co-evolutionary history with humans and the consequent
development of anti-predatory behaviour (Martin, 1984), or
by modification in vegetation cover (increases of open
savanna) in response to climate change during the Holocene
(Cristoffer & Peres, 2003; de Vivo & Carmignotto, 2004).
Whether or not these explanations hold remains to be seen,
but, if large-bodied animals have experienced lower levels of
Pleistocene extinction in the tropics, this purely historical
event could explain in part why both mean body size and its
P-component increase towards warmer areas, although this
would not necessarily explain the large differences in the
observed MRCA values (which are better explained by the
previously described faunal turnover and macroevolutionary
patterns).
On the other hand, the map for the mean S-component of
body size differs substantially from the patterns found globally
for both mean body size and the mean P-component. This
component expresses the average amount of anagenetic
evolution of body size independently of ancestral values (see
Fig. 3). The explanatory power of the environmental model for
this component is relatively low at the global scale, reinforcing
the overall conclusions of Meiri et al. (2004) that Carnivora do
not provide strong evidence for Bergmann’s rule. Even so, it is
important to note that, globally, the regression coefficient for
(a)
(b)
(c)
(d)
Figure 7 Spatial patterns of species richness
of (a) large-bodied carnivore species (> 3 kg;
n = 108 species), (b) small-bodied carnivore
species (< 3 kg; n = 101 species), and (c) the
human footprint. The insert (d) shows the
glaciated and non-glaciated regions consid-
ered in the analyses.
Global patterns in Carnivora body size
Journal of Biogeography 36, 2222–2236
ª 2009 Blackwell Publishing Ltd
2231

Page 11

temperature is negative and relatively high (Table 1), indicat-
ing that in colder regions species tend to be larger than
expected by phylogenetic expectations (i.e. they have positive
S-components, indicating an increase with respect to ancestral
species). This can be expected under adaptive processes
generating Bergmannian gradients within lineages. However,
the pattern is not globally consistent because in cooler regions
of southern South America species are smaller than expected
based on the body size of their ancestors.
Contrasting glaciated and non-glaciated regions
The comparison between glaciated and non-glaciated regions
revealed interesting patterns that could not be perceived in the
global analysis. For the P-component, most of the variation in
glaciated regions is explained by the human footprint and its
overlap with climate, so that cooler regions with less dense
human occupation still harbour Carnivora assemblages
with larger mean body sizes (see below). In contrast, the
P-component for the non-glaciated regions, which cover most
of the world, follows the inverse Bergmannian pattern observed
at the global scale, which can be interpreted as reflecting less
faunal turnover in these areas, as previously discussed.
Although global patterns in body size do not follow
Bergmann’s rule, and adaptation detected at this scale for the
S-component is relatively weak, if we focus on the cooler
regions of the world (see Fig. 5c) and, particularly, on the
glaciated regions, the rule does apply in an adaptive sense, and
the S-component increases northwards and is negatively
correlated with temperature. Indeed, 39% of the variation in
the mean S-component is explained by the environmental
variables in glaciated regions, against < 5% in the non-
glaciated ones. In both areas, but especially in the glaciated
north, the data indicate that, within each species, average
deviations from ancestral body sizes are negatively associated
with temperature, and thus can be interpreted as an adaptive
process generating intraspecific Bergmannian gradients. Cou-
pled with the correlation between the S-component and the
independentlymeasured intraspecific
strengths (see Fig. 4), this supports an adaptive interpretation
for these gradients, although the overall amount of explanation
with respect to total body size variation is relatively small.
Bergmann’s rule
Recent human impacts
Multiple mechanisms can affect the BSFD across geographical
and taxonomic scales. However, an important aspect of our
analysis for the glaciated regions is that it suggests a strong
effect of the human footprint on body size, explaining
independently almost 20% of the variance of the mean P-
component (Table 1). The effect of the human footprint on
the P-component is much less important in the rest of the
world, as is the association between the human footprint and
the S-component, regardless of the area. This would be
expected if human effects were causing a phylogenetically
autocorrelated pattern of extinctions, probably as a result of
phylogenetic patterns in ecological and life-history traits that
are related to extinction risk (see Purvis et al., 2000).
Human impacts affect assemblage patterns by disrupting the
BSFD but not the unique (S) component of species body size,
thus creating more asymmetry in the statistical distribution of
body sizes. Biased extinctions of large-bodied species caused by
humans happened recently in both Europe and North America
(Leonard, 2007), and have begun to impoverish the fauna in
many parts of the world. Carnivores may be particularly
sensitive to human impacts (Cardillo et al., 2004) because of
their life-history traits, and this sensitivity will be greater in
large-bodied species owing to allometric scaling of these traits
(Cardillo et al., 2006). More specifically, it has been found that
intrinsic factors predict greater extinction risk in species
weighing more than 3 kg, and, above this size, susceptibility to
both intrinsic and external threats (such as those caused by
human occupation) increases sharply. The non-additive effects
of intrinsic and extrinsic factors are also supported by our
analyses for previously glaciated areas, as we found that,
although intense human occupation is positively associated
with the richness of Carnivora species weighing < 3 kg, it is
negatively associated with that of larger species (see Fig. 7).
CONCLUSIONS
Our analyses indicate that global patterns in body size must be
understood as resulting from a geographically structured
combination of evolutionary processes operating across multi-
ple spatio-temporal scales. The relationship between environ-
mental predictors and the phylogenetic expectation of body
size suggests a connection between Bergmann’s and Cope’s
rules, driving body size evolution. Recent human impacts were
detected at the assemblage level mainly for temperate North
America and Europe, and thus our analyses indicate that
humans can modify the BSFD of entire faunas rapidly within
ecologicaltime-scalesthrough
events. How the extirpation of large-bodied species in faunas
with little or no ecological redundancy (owing to the low
number of species) will affect food-webs and ecosystem
functioning remains a speculative issue, but it is undoubtedly
a theme that deserves further attention. On the other hand,
adaptive processes within species, related to climatic variation,
could also be detected for the northern part of the world. Thus,
the multiple explanations for global body size gradients in
Carnivora, based on human impacts and climate changes,
highlight the need for a broad understanding of evolutionary
mechanisms acting at different scales and of their association
with biogeographical dynamics and contingencies in the
history of the Earth.
anthropogenic extinction
ACKNOWLEDGEMENTS
We thank Kate Jones, Shai Meiri, Gavin Thomas, Jordi
Bascompte, Walter Jetz, Robert Freckleton, Thomas Lewin-
sohn, Rob Whittaker and two anonymous reviewers for
discussions and suggestions that improved previous versions
J. A. F. Diniz-Filho et al.
2232
Journal of Biogeography 36, 2222–2236
ª 2009 Blackwell Publishing Ltd

Page 12

of the manuscript. We also thank the Paleoclimatic Modelling
Intercomparison Project II (PMIIP) for allowing access to the
ECHAM3 palaeoclimatic reconstructions. Work by J.A.F.D.-F.
and L.M.B. has been continuously supported by many CNPq
fellowships and grants, whereas work by M.A.R. and M.A.O.-
T. was supported by the Spanish Ministry of Science and
Innovation (grant CGL2006-03000/BOS and FPU fellowship
AP2005-0636, respectively). J.H. was supported by the UK
Natural Environment Research Council. J.C.N. received a
doctoral scholarship from CAPES.
REFERENCES
Ahn, C.-H. & Tateishi, R. (1994) Development of a global
30-minute grid potential evapotranspiration data set. Jour-
nal of the Japan Society of Photogrammetry and Remote
Sensing, 33, 12–21.
Alroy, J. (1998) Cope’s rule and the dynamics of body mass
evolution in North American mammals. Science, 280, 731–
735.
Alroy, A. (2001) A multispecies overkill simulation of the end-
Pleistocene megafaunal mass extinction. Science, 292, 1893–
1896.
Arau ´jo, M.B., Nogue ´s-Bravo, D., Diniz-Filho, J.A.F., Haywood,
A.M., Valdes, P.J. & Rahbek, C. (2008) Quaternary climate
changes explain diversity among reptiles and amphibians.
Ecography, 31, 8–15.
Ashton, K.G., Tracy, M.C. & de Queiroz, A. (2000) Berg-
mann’s rule valid for mammals? The American Naturalist,
156, 390–415.
Bininda-Emonds, O.R.P., Gittleman, J.L. & Purvis, A. (1999)
Building large trees by combining phylogenetic information:
a complete phylogeny of the extant Carnivora (Mammalia).
Biological reviews of the Cambridge Philosophical Society, 74,
143–175.
Blackburn, T.M. & Gaston, K.J. (eds) (2003) Macroecology:
concepts and consequences. Cambridge University Press,
Cambridge.
Blackburn, T.M. & Hawkins, B.A. (2004) Bergmann’s rule and
the mammal fauna of northern North America. Ecography,
27, 715–724.
Blackburn, T.M., Gaston, K.J. & Loder, N. (1999) Geographic
gradients in body size: a clarification of Bergmann’s rule.
Diversity and Distributions, 5, 165–174.
Borcard, D. & Legendre, P. (2002) All-scale spatial analysis of
ecological data by means of principal coordinates of
neighbour matrices. Ecological Modelling, 153, 51–68.
Borcard, D., Legendre, P., Avois-Jacquet, C. & Tuomisto, H.
(2004) Dissecting the spatial structure of ecological data at
multiple scales. Ecology, 85, 1826–1832.
Braconnot, P., Otto-Bliesner, B., Harrison, S., Joussaume, S.,
Peterchmitt, J.-Y., Abe-Ouchi, A., Crucifix, M., Driesschaert,
E., Fichefet, Th., Hewitt, C.D., Kageyama, M., Kitoh, A.,
Laı ˆne ´, A., Loutre, M.-F., Marti, O., Merkel, U., Ramstein, G.,
Valdes, P., Weber, S.L., Yu, Y. & Zhao, Y. (2007) Results of
PMIP2 coupled simulations of the Mid-Holocene and Last
Glacial Maximum – Part 1: experiments and large-scale
features. Climate of the Past, 3, 261–277.
Brown, J.H. (1995) Macroecology. University Chicago Press,
Chicago.
Burnham, K.P. & Anderson, D.R. (2002) Model selection and
multimodel inference:apractical
approach, 2nd edn. Springer-Verlag, New York.
Carbone, C. & Gittleman, J.L. (2002) A common rule for the
scaling of carnivore density. Science, 295, 2273–2276.
Carbone, C., Mace, G.M., Roberts, S.C. & Macdonald, D.W.
(1999) Energetic constraints on the diet of terrestrial car-
nivores. Nature, 402, 286–288.
Cardillo, M., Purvis, A., Sechrest, W., Gittleman, J.L., Bielby,
J. & Mace, G.M. (2004) Human population density and
extinction risk in the world’s carnivores. PLoS Biology, 2,
909–914.
Cardillo, M., Mace, G.M., Jones, K.E., Bielby, J., Bininda-
Emonds, O.R.P., Sechrest, W., Orme, C.D.L. & Purvis, A.
(2005) Multiple causes of high extinction risk in large
mammal species. Science, 309, 1239–1241.
Cardillo, M., Mace, G., Gittleman, J.L. & Purvis, A. (2006)
Latent extinction risk and the future battlegrounds of
mammal conservation. Proceedings of the National Academy
of Sciences USA, 103, 4157–4161.
Cheverud, J.M., Dow, M.M. & Leutenegger, W. (1985) The
quantitative assessment of phylogenetic constraints in
comparative analyses: sexual dimorphism in body weight
among primates. Evolution, 39, 1335–1351.
Cristoffer, C. & Peres, C.A. (2003) Elephants versus butterflies:
the ecological role of large herbivores in the evolutionary
history of two tropical words. Journal of Biogeography, 30,
1357–1380.
Davies, T.J., Meiri, S., Barraclough, T.G. & Gittleman, J.L.
(2007) Species co-existence and character divergence across
carnivores. Ecology letters, 10, 146–152.
Demetrius, L. (2000) Directionality theory and the evolution
of body size. Proceedings of the Royal Society B: Biological
Sciences, 267, 2385–2391.
Desdevises, Y., Legendre, P., Azouzi, L. & Morand, S. (2003)
Quantifyingphylogenetically
variation. Evolution, 57, 2647–2652.
Diniz-Filho, J.A.F. (2001) Phylogenetic autocorrelation under
distinct evolutionary processes. Evolution, 55, 1104–1109.
Diniz-Filho, J.A.F. (2004) Macroecology and the hierarchical
expansion of macroevolutionary theory. Global Ecology and
Biogeography, 13, 1–5.
Diniz-Filho, J.A.F. & Bini, L.M. (2005) Modelling geograph-
ical patterns in species richness using eigenvector-based
spatial filters. Global Ecology and Biogeography, 14, 177–
185.
Diniz-Filho, J.A.F. & Bini, L.M. (2008) Macroecology, global
change and the shadow of forgotten ancestors. Global
Ecology and Biogeography, 17, 11–17.
Diniz-Filho, J.A.F. & Nabout, J.C. (2009) Modeling body size
evolutionin Felidaeunder
hypotheses. Genetics and Molecular Biology, 32, 170–176.
information-theoretic
structuredenvironmental
alternative phylogenetic
Global patterns in Carnivora body size
Journal of Biogeography 36, 2222–2236
ª 2009 Blackwell Publishing Ltd
2233

Page 13

Diniz-Filho, J.A.F. & To ˆrres, N.M. (2002) Phylogenetic com-
parative methods and the geographic range size – body size
relationship in new world terrestrial carnivora. Evolutionary
Ecology, 16, 351–367.
Diniz-Filho, J.A.F., Sant’Ana, C.E.R. & Bini, L.M. (1998) An
eigenvector method for estimating phylogenetic inertia.
Evolution, 52, 1247–1262.
Diniz-Filho, J.A.F., Bini, L.M., Rodrı ´guez, M.A´., Rangel,
T.F.L.V.B. & Hawkins, B.A. (2007) Seeing the forest for the
trees: partitioning ecological and phylogenetic components
of Bergmann’s rule in European Carnivora. Ecography, 30,
598–608.
Diniz-Filho, J.A.F., Rangel, T.F.L.V.B. & Bini, L.M. (2008)
Model selection and information theory in geographical
ecology. Global Ecology and Biogeography, 17, 479–488.
Felsenstein, J. (1985) Phylogenies and the comparative meth-
od. The American Naturalist, 125, 1–15.
Freckleton, R.P. & Jetz, W. (2009) Space versus phylogeny:
disentangling phylogenetic and spatial signals in compara-
tive data. Proceedings of the Royal Society B: Biological Sci-
ences, 276, 269–278.
Freckleton, R.P., Harvey, P.H. & Pagel, M. (2002) Phylogenetic
analysis and comparative data: a test and review of evidence.
The American Naturalist, 160, 712–726.
Freckleton, R.P., Harvey, P.H. & Pagel, M. (2003) Bergmann’s
rule and body size in mammals. The American Naturalist,
161, 821–825.
Friscia, A.R., Van Valkenburgh, B. & Biknevicius, A.R. (2007)
An ecomorphological analysis of extant small carnivorans.
Journal of Zoology, 272, 82–100.
Gaston, K.J. & Blackburn, T.M. (2000) Pattern and process in
macroecology, 1st edn. Blackwell Scientific, Oxford.
Gaston, K.J., Chown, S.L. & Evans, K.L. (2008) Ecogeograph-
ical rules: elements of a synthesis. Journal of Biogeography,
35, 483–500.
Gittleman, J.L. (1985) Carnivore body size: ecological and
taxonomic correlates. Oecologia, 67, 540–554.
Gittleman, J.L. (1986) Carnivore life-history patterns: allo-
metric, phylogenetic, and ecological associations. The
American Naturalist, 127, 744–771.
Gittleman, J.L. & Kot, M. (1990) Adaptation: statistics and a
null model for estimating phylogenetic effects. Systematic
Zoology, 39, 227–241.
Gittleman, J.L. & Luh, H.-K. (1992) On comparing compar-
ative methods. Annual Review of Ecology and Systematics, 23,
383–404.
Gittleman, J.L., Anderson, C.G., Kot, M. & Luh, H.-K. (1996)
Phylogenetic lability and rates of evolution: a comparison
of behavioral, morphological
Phylogenies and the comparative method in animal behavior
(ed. by E.P. Martins), pp. 166–205. Oxford University
Press, Oxford.
Grenyer, R., Orme, C.D.L., Jackson, S.F., Thomas, G.H., Da-
vies, R.G., Davies, T.J., Jones, K.E., Olson, V.A., Ridgely,
R.S., Rasmussen, P.C., Ding, T.-S., Bennett, P.M., Black-
burn, T.M., Gaston, K.J., Gittleman, J.L. & Owens, I.P.F.
and life-history traits.
(2006) Global distribution and conservation of rare and
threatened vertebrates. Nature, 444, 93–96.
Greve, M., Gaston, K.J., van Rensburg, B.J. & Chown, S.L.
(2008) Environmental factors, body size distributions and
spatial variation in body size of local avian assemblages.
Global Ecology and Biogeography, 17, 514–523.
Griffith, D.A. (2003) Spatial autocorrelation and spatial filter-
ing: gaining understanding through theory and visualization.
Springer-Verlag, New York.
Griffith, D.A. & Peres-Neto, P. (2006) Spatial modeling in
ecology: the flexibility of eigenfunction spatial analyses.
Ecology, 87, 2603–2613.
Guillaumet, A., Ferdy, J.-B., Desmarais, E., Godelle, B. &
Crochet, P.A. (2008) Testing Bergmann’s rule in the pres-
ence of potentially confounding factors: a case study with
three species of Galerida larks in Morocco. Journal of Bio-
geography, 35, 579–591.
Haberl, H., Erb, K.H., Plutzar, C.H., Fischer-Kowalski, M. &
Krausmann, F. (2004) Human appropriation of net primary
production (HANPP) as indicator for pressures on bio-
diversity. Sustainable development: how to measure progress
through indicators (ed. by B. Moldan, T. Hak and P. Bour-
deau), pp. 271–288. Island Press, Washington DC.
Hawkins, B.A., Diniz-Filho, J.A.F., Bini, L.M., De Marco, P. &
Blackburn, T.M. (2007) Red herrings revisited: spatial
autocorrelation and parameter estimation in geographical
ecology. Ecography, 30, 375–384.
Hunt, G. & Roy, K. (2006) Climate change, body size evo-
lution, and Cope’s rule in deep-sea ostracodes. Proceedings
of the National Academy of Sciences USA, 103, 1347–
1352.
Jackson, D.A. (1993) Stopping rules in principal components
analysis: a comparison of heuristical and statistical ap-
proaches. Ecology, 74, 2204–2214.
Johnson, W.E., Eizirik, E., Pecon-Slattery, J., Murphy, W.J.,
Antunes, A., Teeling, E. & O’Brien, S.J. (2006) The late
Miocene radiation of modern Felidae: a genetic assessment.
Science, 311, 73–77.
Kelt, D.A. & Brown, J.H. (1998) Diversification of body sizes:
patterns and processes in the assembly of terrestrial mammal
faunas. Biodiversity dynamics: turnover of populations, taxa
and communities (ed. by M.L. McKinney and J.A. Drake),
pp. 109–131. Columbia University Press, New York.
Legendre, P. & Legendre, L. (1998) Numerical ecology, 2nd edn.
Elsevier Science, Amsterdam.
Legendre, P., Borcard, D. & Peres-Neto, P.R. (2005) Ana-
lyzing beta diversity: partitioning the spatial variation of
community composition data. Ecological Monographs, 75,
435–450.
Leonard, J.A. (2007) Megafaunal extinctions and the disap-
pearance of a specialized wolf ecomorph. Current Biology,
17, 1146–1150.
Losos, J.B. (2008) Phylogenetic niche conservatism, phylo-
genetic signal and the relationship between phylogenetic
relatedness and ecological similarity among species. Ecology
Letters, 11, 995–1003.
J. A. F. Diniz-Filho et al.
2234
Journal of Biogeography 36, 2222–2236
ª 2009 Blackwell Publishing Ltd

Page 14

Lynch, M. (1991) Methods for the analysis of compara-
tive data in evolutionary biology. Evolution, 45, 1065–
1080.
Lyons, S.K., Smith, F.A. & Brown, J.H. (2004) Of mice, mas-
todons and men: human-mediated extinctions on four
continents. Evolutionary Ecology Research, 6, 339–358.
Manly, B.F.J. (1998) Randomization, bootstrap and Monte Carlo
methods in biology. Chapman & Hall, London.
Martin, P.S. (1984) Prehistoric overkill: the global model.
Quaternary extinctions: a prehistoric revolution (ed. by P.S.
Martin and R.G. Klein), pp. 354–403. University of Arizona
Press, Tucson.
Martins, E.P. & Garland, T., Jr (1991) Phylogenetic analyses of
the correlated evolution of continuous characters: a simu-
lation study. Evolution, 45, 534–557.
Martins, E.P., Diniz-Filho, J.A.F. & Housworth, E.A. (2002)
Adaptive constraints and the phylogenetic comparative
method: a computer simulation test. Evolution, 56, 1–13.
Meiri, S. & Dayan, T. (2003) On the validity of Bergmann’s
rule. Journal of Biogeography, 30, 331–335.
Meiri, S., Dayan, T. & Simberloff, D. (2004) Carnivores, biases
and Bergmann’s rule. Biological Journal of the Linnean
Society, 81, 579–588.
Meiri, S., Simberloff, D. & Dayan, T. (2005) Insular carnivore
biogeography: island area and mammalian optimal body
size. The American Naturalist, 165, 505–514.
Meiri, S., Yom-Tov, Y. & Geffen, E. (2007) What determines
conformity to Bergmann’s rule? Global Ecology and Bio-
geography, 16, 788–794.
Meiri, S., Guy, D. & Simberloff, D. (2009) Global change and
carnivore body size: data are stasis. Global Ecology and
Biogeography, 18, 240–247.
Nieto, M., Hortal, J., Martı ´nez-Maza, C., Morales, J., Ortiz-
Jaureguizar, E., Pelaez-Campomanes, P., Pickford, M., Pra-
do, J.L., Rodrı ´guez, J., Senut, B., Soria, D. & Varela, S.
(2005) Historical determinants of mammal diversity in
Africa: evolution of mammalian body weight distribution in
Africa and South America during Neogene and Quaternary
times. African biodiversity: molecules, organisms, ecosystems
(ed. by B.A. Huber, B.J. Sinclair and K.-H. Lampe), pp. 287–
296. Springer, New York.
Olalla-Ta ´rraga, M.A´. & Rodrı ´guez, M.A´. (2007) Energy and
interspecific body size patterns of amphibian faunas in
Europe and North America: anurans follow Bergmann’s
rule, urodeles its converse. Global Ecology and Biogeography,
16, 606–617.
Olalla-Ta ´rraga, M.A´., Rodrı ´guez, M.A´. & Hawkins, B.A. (2006)
Broad scale patterns of body size in squamate reptiles of
Europe and North America. Journal of Biogeography, 33,
781–793.
Olson, V.A., Davies, R.G., Orme, C.D.L., Thomas, G.H., Meiri,
S., Blackburn, T., Gaston, K.J., Owens, I.P.F. & Bennett,
P.M. (2009) Global biogeography and ecology of body size
in birds. Ecology Letters, 12, 249–259.
Partridge, L. & Coyne, J.A. (1997) Bergmann’s rule in ecoto-
therms: is it adaptive? Evolution, 51, 632–635.
Pavoine, S., Ollier, S., Pontier, D. & Chessel, D. (2008) Testing
for phylogenetic signal in phenotypic traits: new matrices of
phylogenetic proximities. Theoretical Population Biology, 73,
79–91.
Pincheira-Donoso, D., Hodgson, D.J. & Tregenza, T. (2008)
The evolution of body size under environmental gradients in
ectotherms: why should Bergmann’s rule apply to lizards?
BMC Evolutionary Biology, 8, Article 68 (doi: 10.1186/1471-
2148-8-68).
Purvis, A., Agapow, P.-M., Gittleman, J.L. & Mace, G.M.
(2000) Nonrandom extinction and the loss of evolutionary
history. Science, 288, 328–330.
Ramirez, L., Diniz-Filho, J.A.F. & Hawkins, B.A. (2008) Par-
titioning phylogenetic and adaptive components of the
geographical body-size pattern of New World birds. Global
Ecology and Biogeography, 17, 100–110.
Rangel, T.F.L.V.B., Diniz-Filho, J.A.F. & Bini, L.M. (2006)
Towards an integrated computational tool for spatial anal-
ysis in macroecology and biogeography. Global Ecology and
Biogeography, 15, 321–327.
Rodrı ´guez, M.A., Belmontes, J.A. & Hawkins, B.A. (2005) En-
ergy, water and large-scale patterns of reptile and amphibian
species richness in Europe. Acta Oecologica, 28, 65–70.
Rodrı ´guez, M.A´., Lo ´pez-San ˜udo, I.L. & Hawkins, B.A. (2006)
The geographic distribution of mammal body size in
Europe. Global Ecology and Biogeography, 15, 173–181.
Rodrı ´guez, M.A´., Olalla-Ta ´rraga, M.A´. & Hawkins, B.A. (2008)
Bergmann’s rule and the geography of mammal body size in
the Western Hemisphere. Global Ecology and Biogeography,
17, 274–283.
Sanderson, E.W., Jaiteh, M., Levy, M.A., Redford, K.H.,
Wannebo, A.V. & Woolmer, G. (2002) The human footprint
and the last of the wild. BioScience, 52, 891–904.
Sechrest, W. (2003) Global diversity, endemism, and conserva-
tion of mammals. PhD Thesis, University of Virginia,
Charlottesville.
Smith, F.A., Lyons, S.K., Ernest, S.K.M., Jones, K.E., Kaufman,
D.M., Dayan, T., Marquet, P.A., Brown, J.H. & Haskell, J.P.
(2003) Body mass of late Quaternary mammals. Ecology, 84,
3403.
Smith, F.A., Brown, J.H., Haskell, J.P., Lyons, S.K., Alroy, J.,
Charnov, E.L., Dayan, T., Enquist, B.J., Ernest, S.K., Hadly,
E.A., Jones, K.E., Kaufman, D.M., Marquet, P.A., Maurer,
B.A., Niklas, K.J., Porter, W.P., Tiffney, B. & Willig, M.R.
(2004) Similarity of mammalian body size across the taxo-
nomic hierarchy and across space time. The American
Naturalist, 163, 672–691.
Van Valkenburgh, B. (1999) Major patterns in the history of
carnivorous mammals. Annual Review of Earth and Plane-
tary Sciences, 27, 463–493.
Van Valkenburgh, B. (2007) De ´ja ` vu: the evolution of feeding
morphologies in the Carnivora. Integrative and Comparative
Biology, 47, 147–163.
Van Valkenburgh, B., Wang, X. & Damuth, J. (2004) Cope’s
rule, hypercarnivory, and extinction in North American
canids. Science, 306, 101–103.
Global patterns in Carnivora body size
Journal of Biogeography 36, 2222–2236
ª 2009 Blackwell Publishing Ltd
2235

Page 15

de Vivo, M. & Carmignotto, A.P. (2004) Holocene vegetation
change and the mammal faunas of South America and
Africa. Journal of Biogeography, 31, 943–957.
Webster, A.J., Gittleman, J.L. & Purvis, A. (2004) The life-
history legacy of evolutionary body size change in carni-
vores. Journal of Evolutionary Biology, 17, 396–407.
Werdelin, L. & Lewis, M.E. (2005) Plio-Pleistocene Carnivora
of eastern Africa: species richness and turnover patterns.
Zoological Journal of the Linnean Society, 144, 121–144.
Westoby, M., Leishman, M.R. & Lord, J.M. (1995) On mis-
interpreting the ‘phylogenetic correction’. Journal of Ecology,
83, 531–534.
Wiens, J.J. & Donoghue, M.J. (2004) Historical biogeography,
ecology and species richness. Trends in Ecology and Evolu-
tion, 19, 639–644.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the
online version of this article:
Appendix S1 List of the 209 species of Carnivora analysed in
this study, with values of their log-transformed mean body
mass.
Appendix S2 Results from principal components analysis of
environmental variables (Table S1) and details of phylogenetic
eigenvector regression (PVR) of body size against 21 eigen-
vectors extracted from phylogenetic distances among the 209
species (Table S2).
Please note: Wiley-Blackwell is not responsible for the
content or functionality of any supporting materials supplied
by the authors. Any queries (other than missing material)
should be directed to the corresponding author for the
article.
BIOSKETCH
Jose ´ Alexandre Felizola Diniz-Filho is Professor of Animal
Ecology and Evolution in the Departamento de Biologia Geral
of the Universidade Federal de Goia ´s, Brazil, and is a 1A
Productivity Researcher of the Brazilian Council for Scientific
Development and Technology (CNPq). His main research
interests are the evolutionary aspects of macroecological
theory and the application of spatial statistical methods in
macroecology, evolutionary biology, population biology and
conservation. He is currently an editor of Global Ecology and
Biogeography.
Author contributions: J.A.F.D.-F., M.A.R. and B. H. conceived
the ideas; M.C., M.A. R. and J.H. processed and organized the
data; L.M.B., M.A.O.-T. and J.C.N. conducted different steps of
the analyses; and J.A.F.D.-F. led the writing, which received
significant inputs from all other authors.
Editor: Brett Riddle
J. A. F. Diniz-Filho et al.
2236
Journal of Biogeography 36, 2222–2236
ª 2009 Blackwell Publishing Ltd