The role of body mass in diet contiguity and food-web structure.

Abstract

1. The idea that species occupy distinct niches is a fundamental concept in ecology. Classically, the niche was described as an n-dimensional hypervolume where each dimension represents a biotic or abiotic characteristic. More recently, it has been hypothesised that a single dimension may be sufficient to explain the system-level organization of trophic interactions observed between species in a community. 2. Here, we test the hypothesis that species body mass is that single dimension. Specifically, we determine how the intervality of food webs ordered by body size compares to that of randomly ordered food webs. We also extend this analysis beyond the community level to the effect of body mass in explaining the diets of individual species. 3. We conclude that body mass significantly explains the ordering of species and the contiguity of diets in empirical communities. 4. At the species-specific level, we find that the degree to which body mass is a significant explanatory variable depends strongly on the phylogenetic history, suggesting that other evolutionarily conserved traits partly account for species' roles in the food web. 5. Our investigation of the role of body mass in food webs thus helps us to better understand the important features of community food-web structure and the evolutionary forces that have led us to the communities we observe.

Full text

Page 1

Theroleofbodymassindietcontiguityandfood-web
structure
DanielB.Stouffer1*,EnricoL.Rezende2andLuı´sA.NunesAmaral3,4,5
1IntegrativeEcologyGroup,Estacio ´nBiolo ´gicadeDon ˜ana-CSIC,c/Ame ´ricoVespucios/n,41092Sevilla,Spain;
2DepartamentdeGene `ticaideMicrobiologia,FacultatdeBiocie `ncies,EdificiCn,UniversitatAuto `nomadeBarcelona,
08193Bellaterra,Spain;3HowardHughesMedicalInstitute,NorthwesternUniversity,Evanston,IL60208,USA;
4DepartmentofChemicalandBiologicalEngineering,NorthwesternUniversity,Evanston,IL60208,USA;and
5Northwestern InstituteonComplexSystems(NICO),NorthwesternUniversity,Evanston,IL60208,USA
Summary
1. Theideathat species occupydistinctnichesisafundamentalconceptinecology.Classically,the
niche was described as an n-dimensional hypervolume where each dimension represents a biotic or
abiotic characteristic. More recently,ithasbeenhypothesised that a singledimensionmaybe suffi-
cient to explain the system-level organization of trophic interactions observed between species in a
community.
2. Here, we test the hypothesis that species body mass is that single dimension. Specifically, we
determine how the intervality of food webs ordered by body size compares to that of randomly
ordered food webs. We also extend this analysis beyond the community level to the effect of body
massinexplainingthedietsofindividualspecies.
3. We conclude that body mass significantly explains the ordering of species and the contiguity of
dietsinempiricalcommunities.
4. At the species-specific level, we find that the degree to which body mass is a significant explana-
tory variable depends strongly on the phylogenetic history, suggesting that other evolutionarily
conservedtraitspartlyaccountforspecies’rolesinthefoodweb.
5. Our investigation of the role of body mass in food webs thus helps us to better understand the
important features of community food-web structure and the evolutionary forces that have led us
tothecommunitiesweobserve.
Key-words: complexnetworks,foodwebs,intervality,nichedimension,speciesphylogenetics
Introduction
Food webs are a description of who eats whom in an ecosys-
tem (Cohen, Briand & Newman 1990; Pimm 2002; Pascual &
Dunne2006).Recently,multiplestudieshavehelpedtoquan-
tify food-web structure by revealing a number of statistical
regularities within the data (Camacho, Guimera ` & Amaral
2002a,b; Dunne, Williams & Martinez 2002; Cattin et al.
2004; Stouffer et al. 2005; Beckerman, Petchey & Warren
2006;Stouffer,Camacho&Amaral2006;Camacho,Stouffer
& Amaral 2007; Stouffer et al. 2007; Allesina, Alonso &
Pascual2008;Petcheyet al.2008).Instrumentalintheunder-
standing of these statistical patterns has been the develop-
ment of models that attempt to explain their origin (Cohen &
Newman 1985; Williams& Martinez 2000; Cattinet al.2004;
Loeuille & Loreau 2005; Stouffer et al. 2005, 2006; Allesina
et al. 2008; Petchey et al. 2008; Williams & Martinez 2008).
‘Niche space’ is a fundamental concept in these models
(Cohen1978).
Hutchinson (1957) originally defined niche space as an ‘n-
dimensional hypervolume’ where each dimension accounts
for adistinctbioticorabioticcharacteristic.Aspecies’roleor
position within its community is thus conditioned by a set of
n factors acting upon it. To determine a species’ niche, one
must quantify all n factors and determine those which are
most relevant. More recently, the ecological niche has been
reinterpreted as the minimum set of species attributes that
explains some ecological phenomena (Cohen 1978; Warren
&Lawton1987).Thislatterformulationofthenichehaspro-
ven an integral assumption of current static models of food-
webstructure(Cohen&Newman1985;Williams&Martinez
2000;Stoufferet al.2005,2006;Allesinaet al.2008;Williams
& Martinez 2008) and provides a testable criterion – some-
thing that the ‘n-dimensional hypervolume’ does not – to
study which and how many variables may describe species’
nichesinanecosystem.
*Correspondenceauthor.E-mail:stouffer@ebd.csic.es
JournalofAnimalEcology2011,80,632–639doi:10.1111/j.1365-2656.2011.01812.x
?2011TheAuthors.JournalofAnimalEcology?2011BritishEcologicalSociety

Page 2

One case of particular interest is the set of variables neces-
sary to explain the organization of the interactions within
foodwebs.Cohen&Newman(1985)hypothesizedthatinter-
mediate species and top predators in an ecosystem can be
ranked based upon a single characteristic, i.e. assigned an
ordered set of ‘niche-values’ (Stouffer et al. 2005). This
concept of a trophic ordering is very closely tied to that of
‘food-web intervality’ (Cohen 1977, 1978; MacDonald 1979;
Sugihara 1982, 1984; Fig. 1). According to the original defi-
nition, a food web is interval if its interactions are con-
strained such that diets can be represented as contiguous
segments in the same single-dimensional set-up as the trophic
ordering (Cohen 1977). It has been observed that empirical
food webs are not strictly interval (Williams & Martinez
2000). Nevertheless, it was recently demonstrated that they
exhibit a strong tendency toward intervality (Stouffer et al.
2006) and toward empirically observed diet contiguity
(Williams & Martinez 2000; Stouffer et al. 2006; Allesina
et al. 2008; Williams & Martinez 2008; Fig. 2). Remarkably,
this is true for food webs from a variety of environments.
Intriguingly, a number of robust, empirically observed pat-
terns in food-web structure arise as a direct consequence of
diet contiguity (Williams & Martinez 2008); these include the
over/under-representationoffood-webmotifs(Stoufferet al.
2007) and food-web compartmentalization (Guimera ` et al.
2010). These structural patterns in turn mediate the response
of an ecosystem to threats, such as extinctions or invasive
species(Srinivasanet al.2007;Romanuket al.2009).
Determining which variables account for both trophic
ordering and diet contiguity is therefore of paramount
importance to understand food-web structure and dynamics.
The identity of a true empirical proxy, however, has
remained elusive, although multiple alternative hypotheses
have been suggested (Neubert et al. 2000; Layman et al.
2005; Jonsson, Cohen & Carpenter 2005; Woodward et al.
2005;Allesinaet al.2008).Inparticular, anumberofauthors
have suggested that species’ mass or body size provides the
most suitable mapping of the species in a food web along a
singledimension(Warren&Lawton1987;Cohen1989;Law-
ton 1989; Cohen et al. 1993; Neubert et al. 2000; Woodward
& Hildrew 2002; Cohen, Jonsson & Carpenter 2003; Brose
et al.2006;Barneset al.2010;Zooket al.2011).
Here, we examine the role of species mass in explaining
empirically observed diet contiguity. We choose to examine
the explanatory power of body size not because of conve-
nience (Berlow, Brose & Martinez 2008), but because of its
long-recognized importance. Hutchinson proposed that
body sizes could directly explain the coexistence of different
species within an ecosystem (Hutchinson 1959). Body size is
correlated with many descriptors of species ecology (Whitt-
field 2004; Berlow et al. 2008; Bersier & Kehrli 2008), such as
species abundances (Cohen et al. 2003; Jonsson et al. 2005),
and even leads to finely structured communities at an intra-
specific level (Buston & Cant 2006; Kohda et al. 2008). Body
size has also been shown to be an important indicator of
mammals’ vulnerability to extinction (Davidson et al. 2009),
for example. Moreover, there is a rich history of the inter-
actions between body size and ecological networks (Wood-
ward et al. 2005). The strength of its explanatory power is
fortunateas‘measuringbodysize providesa relativelysimple
means of encapsulating and condensing a large amount of
biologicalinformation’(Woodwardet al.2005).
Our manuscript is organized as follows. First, we test
whether species mass not only provides species a hierarchy
12
3
2
1
Predator species
34
13
24
Prey species
6
5
4
Predator species
1234
Prey species
6
5
4
3
2
1
(a)
Perfectly interval food webs
(b)
(c)
Non-interval food web
Fig. 1.Visualizingfood-webintervality.Potential prey species(circles, orientedhorizontally)are placed inasingledimension called theresource
axis.For everypredator(squares,orientedvertically),a lineisplaced abovethepreythatitconsumes.(a),The foodweb isintervalbecausethere
exists an ordering of the prey species, O ¼ f1;2;3;4g, for which all predators have contiguous diets. (b), This food web is also interval because
the ordering, O ¼ f1;3;2;4g, corresponds to contiguous diets. (c), The food web consisting of the predator species from both (a) and (b) is not
intervalbecausethereisnowaytoreorderthepreyspeciessuchthatalldietsarecontiguous.Notethe‘gaps’inthedietsofpredators4and6.
Bodymassandfood-webstructure 633
?2011TheAuthors.JournalofAnimalEcology?2011BritishEcologicalSociety,JournalofAnimalEcology,80,632–639

Page 3

but can also significantly explain diet contiguity. Second,
we present a novel means to quantify the degree of diet conti-
guity of individual species within a food web. This measure
allows us to then test whether species mass can significantly
explain individual diet contiguity. Finally, we use tools from
community phylogenetics to investigate how intervality var-
iesacrossthespeciesfoundwithinempiricalfoodwebs.
Materialsandmethods
For a food web with S species, there are S!/2 possible orderings of
species Ok¼ s1s2...sS?1sS. The diet contiguity of a particular order-
ing can be computed with a suitable cost function (Stouffer et al.
2006).Anexampleofsuchafunctionis
GðOkÞ ¼
X
S
i¼1
X
ci
j¼1
gijjk:
eqn 1
Here, ciis the number of gaps in the diet of species i, and gij|kis the
number of species in the j-th gap in the diet of species i for the order-
ingOk.
Given an empirical food web F, there exists a specific ordering
OmðFÞ determined by the set of masses {m1,m2,…,mS)1,mS} of the
‘typical’individualofeachoftheSspecies.Wewilladopttheconven-
tion here of ranking species in order of increasing mass, that is, the
species with the smallest mass is assigned rank 1, with the next small-
est mass rank 2, and so on up to the species with the largest mass
which is assigned rank S. We compare the number of gaps
Gm¼ GðOmÞ to the number of gaps~G for a random permutation of
species1.Wequantifythiscomparisonwiththez-score
zm¼h~Gi ? Gm
r~G
eqn 2
whereh~Gi and r~Garetheaverageandstandarddeviationofthenum-
ber of gaps across the ensemble of random permutations, respec-
tively. In our analysis, we consider an ensemble of 10 000 random
permutations. Note that, because of the inverse relationship between
thenumberofgapsinafoodwebanditsintervality,wehavereversed
the traditional order of the numerator in eqn 2. Values zm< )1Æ96
therefore imply that diets in a food webordered bymass are less con-
tiguous than expected at random, values zm> 1Æ96 imply that they
are more contiguous, and values )1Æ96 < zm< 1Æ96 imply that they
areconsistentwiththerandomnullhypothesis.
By following this methodology, the viability of any alternative
hypothesis (Neubert et al. 2000; Jonsson et al. 2005; Layman et al.
2005; Woodward et al. 2005; Allesina et al. 2008) for an empirical
analoguetospecies’nichevaluecanbedirectlygaugedwithitssignifi-
cance. Previous studies of intervality have instead searched for glob-
ally optimum permutations or orderings (Stouffer et al. 2006;
Mouillot, Krasnov & Poulin 2008). While such studies can describe
the existence of a significant pattern of contiguity, focusing only on
the best possible ordering could interfere with the ability to detect
important and significant patterns by mistakenly regarding them as
‘not good enough.’ In this study, by concentrating on the ability of a
specific variable – here, body mass – to explain contiguity, we can
quantify how individual species deviate from the hypothesized pat-
tern. That is, we can measure how the diets of individual species con-
tribute to the intervality of the community as a whole. We can also
directly assess the degree to which species’ masses determine the diet
ofeachspecies.
Recallthateveryspeciesicontributes
di¼
X
ci
j¼1
gijjk;
eqn 3
gaps from their diet to the overall number of gaps G. The value diis
thus the absolute intervality of species i measured as the number of
gapsintheirdiet.
Just as with community intervality, we can compare these species-
specific contributions to their equivalent in the ensemble of random
permutations of species. This means that, in addition to understand-
ing how intervality of the community compares to the random null
hypothesis,we canquantifythe degree to whichindividual speciesdo
aswell.We firstcalculatedi,thenumberofgapsinthedietofspeciesi
when species are ordered by their mass. As before, we compare the
numberofgapsditothenumberofgaps~diforarandompermutation
of species (10 000 permutations). We again quantify this comparison
withthez-score
0 2000
Number of gaps,
4000 60008000
0
0·001
0·002
Probability density
0
1020 30
z-score for maximum intervality
0
10
20
30
z-score when ordered by mass, zm
(a)(b)
Fig. 2. Comparison of intervality for mass-ordered and randomly ordered food webs. (a), The probability of observing a particular number of
gapsfortheCaricaieLakesfoodweb.Thehistogramshowsthedistributionofgapsforrandomorderingsofspecies,thesolidblacklineindicates
the empirical value Gmobtained using the ordering provided by the empirical masses, and the dashed line indicates the best-estimate minimum
number of gaps^G for the empirical food web. We find zm¼ 7Æ89 corresponding to P < 0Æ001 while the z-score for^G is 24Æ90. (b), We compare
the z-score for intervality according to species mass with the same but compared to maximum diet contiguity for the 15 empirical food webs
studied.Differencesbetweenthetwocanberegardedasthefractionofpotentialcontiguitythatisunexplainedbybodymass.
1Note that due to the resolution of the empirical data, two or more empirical
species could be recorded to have the same mass. Should this occur, we report
the largest value Gmencountered over multiple realizations where the equiva-
lentspeciesarepermutedrandomly.
634 D.B.Stouffer,E.L.Rezende&L.A.N.Amaral
?2011TheAuthors.JournalofAnimalEcology?2011BritishEcologicalSociety,JournalofAnimalEcology,80,632–639

Page 4

zi¼h~dii ? di
r~di
eqn 4
where h~dii and r~diare the average and standard deviation, respec-
tively, of the number of gaps in the diet of species i across the ensem-
ble of random permutations. The value zi measures the relative
intervality of species i and relative explanatory power of body mass
onthedietofspeciesi.
We estimate the role of evolutionary history on diet contiguity by
measuringthephylogeneticsignal–i.e.thetendencyofcloserelatives
to resemble each other – on species’ mass mi, number of prey ni
(degree of generalization or specialization) and relative intervality zi.
In each community, we first construct the phylogenetic tree using
species’ taxonomic classifications and branch lengths that best fit
the observed distribution of body mass (Appendix S2). With this
tree, we analyse whether species attributes show significant phylo-
genetic signal by employing a randomization procedure in which
species’ attributes are shuffled across the phylogeny, destroying any
signalthatmayhavebeenpresent(Blomberg,Garland&Ives2003).
Results
We examine 15 empirical food webs for which both trophic
interaction data and species masses have been tabulated (see
Appendix S1 for original references): Benguela, Broadstone
Stream,ScotchBroom,Capinteria,CaribbeanReef,Caricaie
Lakes, Coachella Valley, EcoWEB41, EcoWEB60, Grass-
lands, Mill Stream, Sierra Lakes, Skipwith Pond, Tuesday
LakeandYthanEstuary.
We first obtain the set of orderings fOmg for the 15 empiri-
cal food webs and compare their properties with those of
random orderings (see Methods). Here, we find that the
number of gaps Gm is consistent with the random null
hypothesis for only two of the 15 food webs (Table 1). This
implies that species’ masses have significant explanatory
power not only for a one-dimensional niche space but also
forcontiguityofdiets.Moreover,wefindthatthe rejectionof
the random hypothesis is independent of the food web’s size
(Kruskal–Wallis, P ¼ 0Æ54) and number of links (Kruskal–
Wallis,P ¼ 0Æ45).
We also find that the ability of species’ masses to explain
food-webintervality anddietary contiguityappears to be lar-
gely independent of environment (Chase 2000), something
which has also been observed for other food-web properties
(Dunne, Williams & Martinez 2004; Stouffer et al. 2005;
Stouffer et al. 2007; Dunne et al. 2008; Williams & Martinez
2008). The differences are not significant if we classify the
webs as freshwater, marine and terrestrial (Kruskal–Wallis,
P ¼ 0Æ10). If, on the other hand, we classify food webs more
coarsely as aquatic and terrestrial, we find that aquatic food
webs are significantly more interval (Kruskal–Wallis, P ¼
0Æ037), indicating a more important role of size structuring
across aquatic systems. The two food webs, Broadstone
Stream and Grasslands, for which we cannot reject the
random null hypothesis are from freshwater and terrestrial
ecosystems,respectively.
If we ‘zoom in’ to the level of individual species, we reach
similar conclusions as for whole food webs. Indeed, the diets
of a majority of species are more contiguous than would be
expected at random (Fig. 3 and Methods). This implies that
species mass has explanatory power not just at the food-web
level but also at the level of individual species’ diets. Not sur-
prisingly, we find that as the intervality of a food web
increases so does the diet contiguity of its constituent species.
Nevertheless, and despite the general agreement of species’
diets with orderings based on species mass, we do observe
deviations from this pattern (Fig. 3). We next ask whether
there is a common thread that helps us understand why mass
has strong explanatory power for some species but lacks
explanatorypowerforothers.
Specifically, we ask whether the degree to which species
mass explains species-level intervality is related to species’
evolutionary history, that is,tothe species’ phylogeneticrela-
tionship. We quantify the phylogenetic relationship between
species via their individual taxonomic classification. Note
that detailed taxonomic information is only available for
eight of the 15 empirical food webs we have examined up to
this point: Broadstone Stream, Scotch Broom, Caribbean
Reef, Caricaie Lakes, Grasslands, Mill Stream, Skipwith
PondandTuesdayLake.Therefore,allphylogeneticanalyses
andconclusionsarerestrictedtothissubsetofwebs.
First,weexaminewhetherspeciesthatarephylogenetically
related species tend to have similar body mass. We indeed
find a significant phylogenetic signal on species’ mass for
every food web under consideration, that is, closely related
species tend to have similar body masses (Appendix S2).
Note,however,thatsimilarityinbodymassinnowayimplies
similarity of the species’ diets or how their diets fit into the
overallpatternofcommunityintervality.
Table 1. Comparison of orderings based on species’ masses with
random species permutations. Using eqn 1, we compute the number
of gaps in all diets in the empirical food web Gmgiven the ordering
according to the masses of the individual species. We perform the
same computation for an ensemble of random species permutations
and from this calculate the z-score that measures the number of
standard deviations away from the expected value under the random
null hypothesis. The more positive the value, the more species’
masses can account for empirically observed diet contiguity. For
only two empirical food webs, Broadstone Stream and Grasslands,
wouldwerejectthishypothesisata95%confidencelevel
FoodwebGm
h~Gi
r~G
z-score
Benguela
BroadstoneStream
Broom
Capinteria
CaribbeanReef
CaricaieLakes
Coachella
EcoWEB41
EcoWEB60
Grasslands
MillStream
SierraLakes
SkipwithPond
TuesdayLake
Ythan
102
76
326
702
5425
5825
154
19
146
545
516
42
939
344
1124
283
90
770
1472
7656
8087
235
108
289
569
2134
162
1122
1587
2783
39
10
127
147
377
286
26
20
45
70
224
14
68
164
259
4Æ56
1Æ44
3Æ40
5Æ24
5Æ92
7Æ89
3Æ14
4Æ58
3Æ21
0Æ23
7Æ21
8Æ76
2Æ70
7Æ49
6Æ38
Bodymassandfood-webstructure635
?2011TheAuthors.JournalofAnimalEcology?2011BritishEcologicalSociety,JournalofAnimalEcology,80,632–639

Page 5

To quantify the role of phylogeny on species diets, we
examined indetailits relationshipwithtwo additionalspecies
attributes: (i) species’ number of prey niand (ii) species’ rela-
tive intervality zi. We summarize the results of these analyses
inTable 2.Notably,wefindfrequent,significantcorrelations
between phylogenetic similarity and each of these species-
level attributes across the different food webs (Fig. 4). In
particular, we find a lack of phylogenetic signal on relative
intervality in only three food webs, Broadstone Stream and
Grasslands – the two webs for which we observed no relation
between species mass and food-web intervality – and Skip-
with Pond. The correlations observed in the remainder of the
food webs imply that,in these ecosystems, species’ evolution-
ary historyhelpstoexplaintheir roleswithin a foodweb– i.e.
specialist or generalist predator – and also the manner in
whichtheirdietfitswithinthelargerorganizationofthecom-
munity.
To better understand these results, we examine whether
species that have large relative intervality tend to have large
bodymassesaswell,orviceversa.Wealsotakeintoconsider-
ation species’ number of prey, species’ number of predators
and species’ total number of interactions (Appendix S3). We
find no evidence from which to conclude that species whose
dietsarestronglyintervalalsotendtohavelargebodymasses
(or small),tend tobegeneralizedpredators(or specialized)or
tend to have manyinteractions(or few). Overall,ouranalysis
supports the idea that species need not be the most generalist
predator or be found ‘high up’ in the trophic hierarchy for
body mass to have played an important role in the evolution
of their diet. We did find, however, that species with
non-interval diets tend to have many predators (P ¼ 0Æ034),
indicating a possible link between vulnerability and diet
contiguity.
Discussion
SincethecascademodelofCohen&Newman (1985),mecha-
nistic food-web models have relied upon a conceptual ‘niche-
value’ to provide a species ordering. Because of a lack of
strong evidence, the underlying, but often unstated, assump-
tion was that body mass provided the best proxy for these
models’ niche-values (Cohen et al. 1993; Neubert et al.
2000). Our results strongly support the hypothesis that spe-
cies’ masses explain both the ordering of trophic links (War-
ren & Lawton 1987; Cohen 1989; Lawton 1989; Cohen et al.
1993, 2003; Neubert et al. 2000) and empirically observed
diet contiguity (Williams & Martinez 2000; Stouffer et al.
2006;Allesinaet al.2008;Williams&Martinez2008).
In our examinations at the species-specific level, we again
find strong evidence that body mass can explain individual
species’ diet contiguity. The degree to which this takes place,
however, is significantly modulated by the phylogenetic his-
tory of the community. These results imply that there is a
phylogenetic component to how each species fits within the
larger organization of their community. Importantly, our
results are also consistent with previous studies that have
documentedtheimportanceofphylogenetichistoryinunder-
standing additional aspects of food-web structure (Cattin
et al.2004;Rezendeet al.2009).
010
Relative intervality of food web
–5
0
5
10
15
Relative intervality of species
–4–2
Relative intervality of species
2468
02468 10
0·0
0·1
0·2
0·3
Probability density
(a)
(b)
Fig. 3. Comparison of food-web and species-level intervality. (a), The degree of food-web intervality, as measured by the z-score, compared to
the degree of intervality of its constituent species, also measured by their z-score. All measurements quantify intervality when the species are
ordered by their mass. As expected, species-level intervality increases significantly with increasing food-web intervality. However, we observe
that the intervality of species in a food web can vary substantially. The arrow shown indicates points which fall outside of the visible area. (b),
Theprobabilitydistributionofspecies-levelrelativeintervalityintheCaricaieLakesfoodweb.Amajorityofallpredatorspeciesaresignificantly
interval(62%haverelativeintervalityzi> 1Æ96).
Table 2. Phylogenetic signal on species attributes. For the eight webs
for which we have detailed taxonomic information, we tabulate the
number of species S and the p-values associated with phylogenetic
signal on species body mass mi, species number of prey niand species
relative intervality zi. We observe strong evidence for a phylogenetic
signal on each of these species-specific attributes across the various
food webs studied. The phylogenetic signal on species number of
prey implies that phylogenetically similar species tend to exhibit
similar degrees of specialization. The signal on relative intervality
implies that body mass plays a similar role in the organization of the
dietsofphylogeneticallysimilarspecies
FoodwebSP-value,mi
P-value,ni
P-value,zi
BroadstoneStream
Broom
Caribbean
CaricaieLakes
Grasslands
MillStream
SkipwithPond
TuesdayLake
28
68
200
149
65
76
71
71
<0Æ001
<0Æ001
<0Æ001
<0Æ001
<0Æ001
<0Æ001
<0Æ001
<0Æ001
0Æ201
<0Æ001
0Æ036
0Æ039
0Æ227
0Æ002
0Æ402
<0Æ001
0Æ332
<0Æ001
0Æ011
0Æ016
0Æ906
0Æ012
0Æ999
<0Æ001
636D.B.Stouffer,E.L.Rezende&L.A.N.Amaral
?2011TheAuthors.JournalofAnimalEcology?2011BritishEcologicalSociety,JournalofAnimalEcology,80,632–639

Page 6

Wefindthat foodwebs arehighly size-structured, inagree-
ment with earlier results (Beckerman et al. 2006; Petchey
et al. 2008). Furthermore, our analyses provide insight into
therolethatbodymasshasplayedintheevolutionaryhistory
of the individual species under consideration. Phylogenetic
effects on diet contiguity suggest that other factors, apart
from body mass, likely account for species’ contributions to
the food-web structure. Indeed, this is in close agreement
withrecentanalyses,whichsuggestthatinteractionsinafood
web are predicted with increased accuracy by models that
includetwolatent traitsin additiontobody mass(Rohr et al.
2010). Interestingly, these variables, that describe species
foraging intensity and vulnerability, also appear to show
significantphylogeneticsignal.
The pervasiveness of phylogenetic signal across variables
and food webs provides unequivocal evidence that closely
related species often have similar niches in the community.
Even though phylogenetic signal is expected from shared
ancestryandverysimple evolutionarymodels,itisalsopossi-
ble that the structure of the food web – and the selective pres-
sures inherent to size-structured interactions – contributes to
this pattern, resulting in phylogenetic niche conservatism due
to selection (Losos 2008). Importantly, these alternatives are
by no means mutually exclusive, and the interplay between
phylogenetic and ecological factors in shaping species’ niches
maybestrikinglydifferentacrosstaxaortrophiclevelswithin
asinglefoodweb(Rezendeet al.2009).
It is widely acknowledged that some of the species making
up empirical food webs result from the aggregation of indi-
viduals across size and ontogeny (Werner & Gilliam 1984;
Martinez 1991; Solow & Beet 1998; Rudolf 2007). For exam-
ple, it has previously been observed that taxonomic aggrega-
tion in Broadstone Stream – one of the food webs for which
we cannot reject the random null hypothesis – has created a
food web that no longer reflects the size of the individuals
that actually interact (Woodward & Warren 2007; Wood-
ward et al. 2010). We therefore find the general nature of the
patterns that we observe to be all the more intriguing. Were
food webs more highly resolved, to the level of individuals
and not just individual species, we would expect the signal to
be even stronger. Similarly, we have focused on food webs
thatare largelyfree ofparasitic interactions,despitetheirrec-
ognized importance (Lafferty, Dobson & Kuris 2006; Laffer-
ty et al. 2008; Beckerman & Petchey 2009). Here, we have
treatedallantagonistic interactions– predatory or parasitic –
as equivalent when parasites or parasitoids are present (e.g.
Broom and Grasslands). It will be interesting to see how this
assumption holds up for future data sets with parasitism
incorporatedingreaterquantity.
In the present manuscript, we provide statistical methods
andresultsthat allow us to linkspecies body mass to the con-
cepts of intervality and diet contiguity. This does not, how-
ever, imply that these factors represent sufficient or necessary
ingredients in any model that aims to explain empirical food-
webstructure(Cohen&Newman1985;Williams&Martinez
2000; Stouffer et al. 2006; Allesina et al. 2008; Williams &
Martinez2008).Nevertheless,anysuchmodelwilllikelyneed
to incorporate the observed, phylogenetically related, non
uniform variation across species. The challenge remaining,
then, will be to definitively link the results presented here to a
mechanisticmodeloffood-webstructure.
Acknowledgements
WethankS.Allesina,J.Bascompte,P.M.Buston,M.A.Fortuna, R.Guimera ` ,
R.D. Malmgren,P.McMullen, M.Sales-Pardo,G. Woodward and A.E. Zook
0
25
50
75
0
5
10
(a)
(b)
Number of prey
(c)
Relative intervality
Species
Fig. 4.Phylogeneticsignalandspecies’diet contiguity.(a),ThephylogenetictreefortheCaricaieLakesfoodweb,builtusingspecies’taxonomic
classifications.(b),Thenumberofpreyofeachspeciesorderedaccordingtotheirphylogeneticsimilarity.(c),Therelativeintervalityofeachspe-
ciesorderedaccordingtotheirphylogeneticsimilarity.Tofacilitatevisualclarity,thecolourofthebarsisproportionaltothevaluefromlowval-
ues (light green) to high values (dark green). We observe a significant phylogenetic signal for both number of prey (P ¼ 0Æ039) and relative
intervality (P ¼ 0Æ016). Visually, this means that species that are closer together in the phylogeny have a greater than expected probability of
havingsimilarnumbersofpreyandsimilarrelativeintervalities.
Bodymassandfood-webstructure 637
?2011TheAuthors.JournalofAnimalEcology?2011BritishEcologicalSociety,JournalofAnimalEcology,80,632–639

Page 7

for stimulating discussions and helpful suggestions. D.B.S. acknowledges a
CSIC JAE Postdoctoral Fellowship. E.L.R. is a Ramo ´ n y Cajal fellow of the
MICINN, Spain. All figures were generated with PyGrace (http://pygrace.
sourceforge.net).
References
Allesina, S., Alonso, D. & Pascual, M. (2008) A general model for food web
structure.Science,320,658–661.
Barnes, C., Maxwell, D., Reuman, D.C. & Jennings, S. (2010) Global patterns
in predator-prey size relationships reveal size dependency of trophic transfer
efficiency.Ecology,91,222–232.
Beckerman, A.P. & Petchey, O.L. (2009) Infectious food webs. The Journal of
AnimalEcology,78,493–496.
Beckerman, A.P., Petchey, O.L. & Warren, P.H. (2006) Foraging biology pre-
dicts food web complexity. Proceedings of the National Academy of Sciences
oftheUnitedStatesofAmerica,103,13745–13749.
Berlow, E.L., Brose, U. & Martinez, N.D. (2008) The ‘Goldilocks factor’ in
food webs. Proceedings of the National Academy of Sciences of the United
StatesofAmerica,105,4079–4080.
Bersier, L.F. & Kehrli, P. (2008) The signature of phylogenetic constraints on
food-webstructure.EcologicalComplexity,5,132–139.
Blomberg, S.P., Garland, Jr., T. & Ives, A.R. (2003) Testing for phylogenetic
signal in comparative data: behavioral traits are more labile. Evolution, 57,
717–745.
Brose, U., Jonsson, T., Berlow, E.L., Warren, P., Banasek-Richter, C., Bersier,
L.F., Blanchard, J.L., Brey, T., Carpenter, S.R., Blandenier, M.F.C., Cush-
ing, L., Dawah, H.A., Dell, T., Edwards, F., Harper-Smith, S., Jacob, U.,
Ledger, M.E., Martinez, N.D., Memmott, J., Mintenbeck, K., Pinnegar,
J.K., Rall, B.C., Rayner, T.S., Reuman, D.C., Ruess, L., Ulrich, W.,
Williams, R.J., Woodward, G. & Cohen, J.E. (2006) Consumer-resource
body-sizerelationshipsinnaturalfoodwebs.Ecology,87,2411–2417.
Buston,P.M.&Cant, M.A.(2006)Anewperspectiveonsizehierarchiesinnat-
ure:patterns,causes,andconsequences.Oecologia,149,362–372.
Camacho, J., Guimera ` , R. & Amaral, L.A.N. (2002a) Analytical solution of a
modelfor complexfood webs. Physical Review. E, Statistical,Nonlinear,and
SoftMatterPhysics,65,art.no.030901.
Camacho, J., Guimera ` , R. & Amaral, L.A.N. (2002b) Robust patterns in food
webstructure.PhysicalReviewLetters,88,art.no.228102.
Camacho, J., Stouffer, D.B. & Amaral, L.A.N. (2007) Quantitative analysis of
the local structure of food webs. Journal of Theoretical Biology, 246, 260–
268.
Cattin, M.F., Bersier, L.F., Banas˘ek-Richter, C., Baltensperger, R. & Gabriel,
J.P. (2004) Phylogenetic constraints and adaptation explain food-web struc-
ture.Nature,427,835–839.
Chase, J.M. (2000) Are there real differences among aquatic and terrestrial
foodwebs?TrendsinEcology&Evolution,15,408–412.
Cohen, J.E. (1977) Food webs and the dimensionality of trophic niche space.
Proceedings of the National Academy of Sciences of the United States of
America,74,4533–4536.
Cohen, J.E. (1978) Food Webs and Niche Space. Princeton University Press,
Princeton,NJ.
Cohen, J.E. (ed.) (1989) Ecologists’ Co-Operative Web Bank, version 1.0.
RockefellerUniversity,NewYork,NY.
Cohen, J.E., Briand, F. & Newman, C.M. (1990) Community Food Webs: Data
andTheory.Springer-Verlag,Berlin.
Cohen, J.E., Jonsson, T. & Carpenter, S.R. (2003) Ecological community
description using the food web, species abundance, and body size. Proceed-
ings of the National Academy of Sciences of the United States of America,
100,1781–1786.
Cohen, J.E. & Newman, C.M. (1985) A stochastic theory of community food
webs I. Models and aggregated data. Proceedings of the Royal Society of
London.SeriesB,BiologicalSciences,224,421–448.
Cohen, J.E., Pimm, S.L., Yodzis, P. & Saldan ˜ a, J. (1993) Body sizes of animal
predators and animal prey in food webs. The Journal of Animal Ecology, 62,
67–78.
Davidson, A.D., Hamilton, M.J., Boyer, A.G., Brown, J.H. & Ceballos, G.
(2009) Multiple ecological pathways to extinction in mammals. Proceedings
of the National Academy of Sciences of the United States of America, 106,
10702–10705.
Dunne, J.A., Williams, R.J. & Martinez, N.D. (2002) Food-web structure
and network theory: the role of connectance and size. Proceedings of the
National Academy of Sciences of the United States of America, 99, 12917–
12922.
Dunne, J.A., Williams, R.J. & Martinez, N.D. (2004) Network structure and
robustness of marine food webs. Marine Ecology Progress Series, 273, 291–
302.
Dunne, J.A., Williams, R.J., Martinez, N.D., Wood, R.A. & Erwin, D.H.
(2008) Compilation and network analyses of cambrian food webs. PLoS
Biology,6,e102.
Guimera ` ,R.,Stouffer,D.B.,Sales-Pardo,M.,Leicht,E.,Newman,M.&Ama-
ral, L. (2010) Origin of compartmentalization in food webs. Ecology, 91,
2941–2951.
Hutchinson, G.E. (1957) Concluding remarks. Cold Spring Harbor Symposia
onQuantitativeBiology,22,415–427.
Hutchinson, G.E. (1959) Homage to Santa Rosalia or why are there so many
kindsofanimals?TheAmericanNaturalist,93,145–159.
Jonsson, T., Cohen, J.E. & Carpenter, S.R. (2005) Food webs, body size, and
speciesabundanceinecologicalcommunitydescription.AdvancesinEcologi-
calResearch,36,1–84.
Kohda, M., Shibata, J.Y., Awata, S., Gomagano, D., Takeyama, T., Hori, M.
& Heg,D.(2008)Nichedifferentiation depends onbody sizeina cichlidfish:
amodelsystem ofacommunitystructured according tosize regularities.The
JournalofAnimalEcology,77,859–868.
Lafferty, K.D., Allesina, S., Arim, M., Briggs, C.J., Leo, G.D., Dobson, A.P.,
Dunne, J.A., Johnson, P.T.J., Kuris, A.M., Marcogliese, D.J., Martinez,
N.D., Memmott, J., Marquet, P.A., McLaughlin, J.P., Mordecai, E.A.,
Pascual, M., Poulin, R. & Thieltges, D.W. (2008) Parasites in food webs: the
ultimatemissinglinks.EcologyLetters,11,533–546.
Lafferty, K.D., Dobson, A.P. & Kuris, A.M. (2006) Parasites dominate food
web links. Proceedings of the National Academy of Sciences of the United
StatesofAmerica,103,11211–11216.
Lawton, J.H. (1989) Food webs. Ecological Concepts (ed. J. Cherrett), pp. 43–
78.BlackwellScientific,Oxford,UK.
Layman, C.A., Winemiller,K.O., Arrington,D.A. & Jepsen,D.B. (2005) Body
size and trophic position in a diverse tropical food web. Ecology, 86, 2530–
2535.
Loeuille, N. & Loreau, M. (2005) Evolutionary emergence of size-structured
food webs. Proceedings of the National Academy of Sciences of the United
StatesofAmerica,102,5761–5766.
Losos,J.B.(2008)Phylogeneticnicheconservatism,phylogeneticsignalandthe
relationship between phylogenetic relatedness and ecological similarity
amongspecies.EcologyLetters,11,995–1003.
MacDonald, N.(1979)Simpleaspectsoffoodwebcomplexity.JournalofTheo-
reticalBiology,80,577–588.
Martinez, N.D.(1991)Artifactsorattributes?EffectsofresolutionontheLittle
RockLakefoodweb.EcologicalMonographs,61,367–392.
Mouillot, D., Krasnov, B.R. & Poulin, R. (2008) High intervality explained by
phylogeneticconstraintsinhost-parasitewebs.Ecology,89,2043–2051.
Neubert, M.G., Blumenshine, S.C., Duplisea, D.E., Jonsson, T. & Rashleigh,
B. (2000) Body size and food web structure: testing the equiprobability
assumptionofthecascademodel.Oecologia,123,241–251.
Pascual,M.& Dunne, J.A.(eds) (2006) Ecological Networks:Linking Structure
toDynamicsinFoodWebs.OxfordUniversityPress,Oxford,UK.
Petchey,O.L.,Beckerman,A.P.,Riede,J.O.&Warren,P.H.(2008)Size,forag-
ing,andfoodwebstructure.ProceedingsoftheNationalAcademyofSciences
oftheUnitedStatesofAmerica,105,4191–4196.
Pimm, S.L. (2002) Food Webs, 1st edn. University of Chicago Press, Chicago,
IL,USA.
Rezende, E.L., Albert, E.M., Fortuna, M.A. & Bascompte, J. (2009) Compart-
mentsinamarinefoodwebassociatedwithphylogeny,bodymass,andhabi-
tatstructure.EcologyLetters,12,779–788.
Rohr, R.P., Scherer, H., Kehrli, P., Mazza, C. & Bersier, L.F. (2010) Modeling
food webs: exploring unexplainedstructure using latent traits.The American
Naturalist,176,170–177.
Romanuk,T.N.,Zhou,Y.,Brose,U.,Berlow,E.L.,Williams,R.J.&Martinez,
N.D. (2009) Predicting invasion success in complex ecological networks.
Philosophical Transactions of the Royal Society of London. Series B, Biologi-
calSciences,364,1743–1754.
Rudolf, V.H.W. (2007) Consequences of stage-structured predators: cannibal-
ism,behavioraleffects,andtrophiccascades.Ecology,88,2991–3003.
Solow,A.R.&Beet,A.R.(1998)Onlumpingspecies infood webs.Ecology, 79,
2013–2018.
Srinivasan, U.T., Dunne, J.A., Harte, J. & Martinez, N.D. (2007) Response
of complex food webs to realistic extinction sequences. Ecology, 88, 671–
682.
Stouffer, D.B., Camacho, J. & Amaral, L.A.N. (2006) A robust measure of
food web intervality. Proceedings of the National Academy of Sciences of the
UnitedStatesofAmerica,103,19015–19020.
638D.B.Stouffer,E.L.Rezende&L.A.N.Amaral
?2011TheAuthors.JournalofAnimalEcology?2011BritishEcologicalSociety,JournalofAnimalEcology,80,632–639

Page 8

Stouffer, D.B., Camacho, J., Guimera ` , R., Ng, C.A. & Amaral, L.A.N. (2005)
Quantitative patterns in the structure of model and empirical food webs.
Ecology,86,1301–1311.
Stouffer, D.B., Camacho, J., Jiang, W. & Amaral, L.A.N. (2007) Evidence
for the existence of a robust pattern of prey selection in food webs. Pro-
ceedings of the Royal Society of London. Series B, Biological Sciences, 274,
1931–1940.
Sugihara, G. (1982) Niche Hierarchy: Structure, Organization, and Assembly
in Natural Communities. Ph.D. thesis, Princeton University, Princeton,
N.J.
Sugihara, G. (1984) Graph theory, homology, and food webs. Population Biol-
ogy (ed. S.A. Levin), pp. 83–101. American Mathematical Society, Provi-
dence,RI,vol.30ofProceedingsofSymposiainAppliedMathematics.
Warren,P.H. & Lawton, J.H.(1987) Invertebrate predator-prey bodysize rela-
tionships: an explanation for upper triangular food webs and patterns in
foodwebstructure?Oecologia,74,231–235.
Werner, E.E. & Gilliam, J.F. (1984) The ontogenetic niche and species interac-
tions in size structured populations. Annual Review of Ecology, Evolution,
andSystematics,15,393–425.
Whittfield,J.(2004)Ecology’sbig,hotidea.PLoSBiology,2,e440.
Williams, R.J. & Martinez, N.D. (2000) Simple rules yield complex food webs.
Nature,404,180–183.
Williams, R.J. & Martinez, N.D. (2008)Success and its limits amongstructural
modelsofcomplexfoodwebs.TheJournalofAnimalEcology,77,512–519.
Woodward, G., Blanchard, J., Lauridsen, R.B., Edwards, F.K., Jones, J.I.,
Figueroa, D., Warren, P.H. & Petchey, O.L. (2010) Individual-based food
webs: species identity, body size and sampling effects. Integrative Ecology:
From Molecules to Ecosystems (ed. G. Woodward), pp. 211–266. Academic
Press,London,UK.vol.43ofAdvancesinEcologicalResearch.
Woodward, G., Ebenman, B., Emmerson, M., Montoya, J.M., Olesen, J.M.,
Valido, A. & Warren, P.H. (2005) Body size in ecological networks. Trends
inEcology&Evolution,20,402–409.
Woodward, G. & Hildrew, A.G. (2002) Body-size determinants of niche
overlap and intraguild predation within a complex food web. The Journal of
AnimalEcology,71,1063–1074.
Woodward, G. & Warren, P.H. (2007) Body size and predatory interactions in
freshwaters: scaling from individuals to communities. In: Body Size: The
Structure and Function of Aquatic Ecosystems (eds A.G. Hildrew, D. Raf-
faelli & R. Edmonds-Brown), pp. 98–117. Cambridge University Press,
Cambridge,UK.
Zook, A.E., Eklof, A., Jacob, U. & Allesina, S. (2011) Food webs: ordering
species according to body size yields high degree of intervality. Journal of
TheoreticalBiology,271,106–113.
Received21September2010;accepted19January2011
HandlingEditor:AndrewBeckerman
SupportingInformation
Additional Supporting Information may be found in the online ver-
sionofthisarticle.
AppendixS1.Empiricalfoodwebs.
AppendixS2.Phylogeneticanalyses.
AppendixS3.Comparisonbetweenspeciesattributes.
As a service to our authors and readers, this journal provides sup-
porting information supplied by the authors. Such materials may be
re-organized for online delivery, but are not copy-edited or typeset.
Technical support issues arising from supporting information (other
thanmissingfiles)shouldbeaddressedtotheauthors.
Bodymassandfood-webstructure639
?2011TheAuthors.JournalofAnimalEcology?2011BritishEcologicalSociety,JournalofAnimalEcology,80,632–639