compartments in a marine food web associated with phylogeny, … · 2014-05-08 · the food web we...

L E TTERCompartments in a marine food web associated withphylogeny, body mass, and habitat structure

Enrico L. Rezende,* Eva M.

Albert, Miguel A. Fortuna and

Jordi Bascompte

Integrative Ecology Group,

Estacion Biologica de Donana,

CSIC, Americo Vespucio s ⁄n,E-41092 Sevilla, Spain

*Correspondence and present

address: Departament de

Genetica i de Microbiologia,

Facultat de Biociencies, Edifici

Cn, Universitat Autonoma, de

Barcelona, E-08193 Bellaterra

(Barcelona), Spain. E-mail:

[email protected]

AbstractA long-standing question in community ecology is whether food webs are organized in

compartments, where species within the same compartment interact frequently among

themselves, but show fewer interactions with species from other compartments. Finding

evidence for this community organization is important since compartmentalization may

strongly affect food web robustness to perturbation. However, few studies have found

unequivocal evidence of compartments, and none has quantified the suite of

mechanisms generating such a structure. Here, we combine computational tools from

the physics of complex networks with phylogenetic statistical methods to show that a

large marine food web is organized in compartments, and that body size, phylogeny, and

spatial structure are jointly associated with such a compartmentalized structure. Sharks

account for the majority of predatory interactions within their compartments.

Phylogenetically closely related shark species tend to occupy different compartments

and have divergent trophic levels, suggesting that competition may play an important

role structuring some of these compartments. Current overfishing of sharks has the

potential to change the structural properties, which might eventually affect the stability

of the food web.

KeywordsCompartments, complex networks, food web, phylogenetic analyses, sharks, trophic

level.

Ecology Letters (2009) 12: 779–788

I N TRODUCT ION

Food webs are representations of who eats whom inecological communities. Understanding how these webs arestructured, and which factors underlie such a structure, aremajor goals in ecology with far-reaching consequences forthe conservation of species-rich communities. A long-heldtenet in ecology has been to determine whether food websare structured in compartments, i.e. subgroups of highlyinteracting species with little interaction between thesubgroups (Paine 1963, 1980; Cohen 1978; Pimm & Lawton1980; Sugihara 1983; Raffaelli & Hall 1992; Krause et al.2003; Melian & Bascompte 2004). This search was spurredby the belief that compartmentalization greatly affects foodweb stability (May 1972; Pimm 1979; Pimm & Lawton 1980;Solow et al. 1999). For example, a compartmentalizedstructure may reduce the propagation of a perturbationthrough the entire food web (Melian & Bascompte 2002).This is particularly relevant for marine ecosystems becausethey are currently under intense fishing pressures (Pauly

et al. 1998; Myers & Worm 2003), and have the potential forcommunity-wide cascading effects (Carpenter & Kitchell1993; Bascompte et al. 2005; Mumby et al. 2007; Myers et al.2007).

In spite of all this interest, however, there are fewunequivocal examples of compartmentalization in foodwebs (e.g. Girvan & Newman 2002; Krause et al. 2003; seeOlesen et al. 2007 for plant-animal mutualistic interactions),and even less mechanistic explanations for such compart-ments. Habitat boundaries were first suggested as the mostlikely explanation of compartments (Pimm & Lawton 1980),and more recent papers have found evidence for such afactor (Girvan & Newman 2002; Krause et al. 2003). Bodysize (Paine 1963; Cohen et al. 1993, 2003; Williams &Martinez 2000; Emmerson & Raffaelli 2004; Stouffer et al.2005; Petchey et al. 2008) and evolutionary history (Cattinet al. 2004) have also been adduced to be an importantexplanation for food web structure. Therefore, new studiesare necessary to disentangle the role of these factors inshaping the architecture of food webs. These studies should

Ecology Letters, (2009) 12: 779–788 doi: 10.1111/j.1461-0248.2009.01327.x

! 2009 Blackwell Publishing Ltd/CNRS

also consider the role of phylogeny as a determinant ofcommunity structure (Webb et al. 2002), because phenotypicattributes such as body mass, relative abundance, andtrophic level may be partly determined by species evolu-tionary history (Blomberg et al. 2003; Garland et al. 2005).

Here, we (i) detect compartments in a real, large marinefood web; (ii) test if body size, habitat structure, andphylogeny are associated to such a compartmentalizedstructure; and (iii) analyse the disproportionate importanceof sharks in shaping these compartments. To do so, wecombine recently developed computational tools to analysecomplex networks (Newman & Girvan 2004; Guimera &Amaral 2005) with phylogenetic statistical tools (Blomberget al. 2003; Garland et al. 2005). Our results suggest that pastevolutionary history of the species involved can be veryimportant in explaining current community patterns (Cattinet al. 2004; Rezende et al. 2007).

MATER IA L S AND METHODS

The food web

We analysed a Caribbean marine food web depicting a totalof 3313 trophic interactions between 249 species ⁄ trophicgroups (Opitz 1996; Bascompte et al. 2005). This food webencompasses all benthic and pelagic communities from thesurface to a 100-m depth in an area of c. 1000 km2, andincludes detritus, 3 primary producer groups, 35 invertebratetaxa, 208 fish species, sea turtles and sea birds (see Dataset S1for the food web with its constituent species and interac-tions). We also used additional information on speciespresence ⁄ absence across five habitat types, which provides acrude estimate of spatial structure in this marine ecosystem(Melian et al. 2005): mangrove ⁄ estuaries, coral reefs, sea-grass beds ⁄ algal mats, sand, and off-shore reefs. Details onhow the food web was assembled, on habitat characteristicsand on the data strengths and limitations (particularly inrelation to the analysis of gut contents), are discussedelsewhere (Bascompte et al. 2005; Melian et al. 2005).

Analyses of network structure were performed with thetotality of the food web. Employing information on speciesbiomass, body mass, metabolic rates, and diet composition(Bascompte et al. 2005), we calculated the strength of apredator–prey interaction as:

!Q=B"j #DC ij ; !1"

where (Q ⁄B)j is the number of times an age-structuredpopulation of predator j consumes its own weight per day,and DCij is the proportion of prey i in the diet of predatorj estimated from stomach content analysis. Thus, ourresulting matrix describes the absolute magnitude of biomassflowing from each prey to each predator per unit time

(Berlow et al. 2004). Note that the measure in eqn 1 isequivalent to the one used by Bascompte et al. (2005), butwithout dividing by prey biomass. This is because here we areinterested in a property of the predator species, namely thebiomass flowing from a prey to its predator, while Bascompteet al. (2005) focused in the per capita effects of predators ontheir prey, which is directly dependent on the prey biomass.

Network structure and trophic levels

We first addressed if the food web was significantlysubdivided into different compartments in relation torandom expectation. We used a module-finding algorithm(Newman & Girvan 2004) combined with a simulatedannealing optimization approach (Guimera & Amaral 2005)to detect food web compartments. Specifically, we haveused the simplest generalization to weighted networks of themodularity implemented in the Guimera and Amaral!salgorithm (Guimera & Amaral 2005; Guimera et al. 2007).Basically, the program employs a heuristic procedure to findan optimal solution for the maximization of a functioncalled modularity (Newman & Girvan 2004). For weightednetworks, where wij represents the weight of the interactionbetween predator j and prey i, the modularity is given by (seeGuimera et al. 2007):

MW !P" $XNM

s$ 1

w ins

W

! "% wall

s

2W

! "2" #

; !2"

where W $P

i&j wij is the sum of the weights of allpredator–prey interactions across the entire network, w in

s isthe sum of the weights of the links wij within compartment s,and wall

s $P

i2sP

j wij is the sum of the weights of theinteractions involving species i within module s with allother species.

This function maximizes the weights of links betweennodes belonging to the same compartment and minimizesthe weight of links between nodes belonging to differentcompartments. A network with high modularity is such thatthe density of links (and their weights) inside compartmentsis significantly higher than the random expectation for sucha density. Because values of interaction strength spanned sixorders of magnitude, we ranked interaction strengths from 1to 6 according to their order of magnitude (note that theseranks reflect the flux of biomass from prey to predators ona log-scale). From these analyses, we obtained the valueof modularity for the food web, the number and identity ofcompartments detected by the algorithm, and the role ofeach species in the web. We estimated how well connectedeach species is to the other species inside the compartment(z-score or standardized within-compartment degree, z), andhow well the interactions of a species are distributed amongdifferent compartments (participation coefficient, PC, vary-

780 E. L. Rezende et al. Letter


ing between 0 and 1). The role of each species in the webcan be described with these two indices. For instance,species with high z and low PC are well connected onlywithin their compartments; species with low z and high PCact primarily as connectors between compartments; andspecies with both high z and PC are network hubs, i.e. wellconnected both within and across compartments (details inGuimera & Amaral 2005).

We also calculated the trophic level of each species ⁄ taxonin the food web, employing the positional index d (Bersieret al. 2002). This index takes into account the amount ofbiomass that comes to a species from its prey, and howmuch biomass flows from this species to its predators,varying between 0 and 1: basal species that do not predateany other species on the web have a d = 0, top predatorshave a d = 1, and intermediate species that act as both preyand predators have a d between 0 and 1. Because we wereinterested primarily on the interactions between fish species,we excluded other taxa on the food web prior tocalculations of d.

The phylogeny

We assembled a fish phylogeny (Fig. S1) encompassing 116species employing multiple sources from the literature (Textin Supporting Information). This corresponds to 46.6% of allnodes in the web or 55.8% of the fish species. These speciesare involved in 2307 trophic links, corresponding to 69.6% ofthe total number of interactions. Visual inspection of the dataand preliminary analyses suggested that this subset of specieswas an accurate representation of the food web (no biaseswere observed when comparing results obtained with theentire dataset against analyses including only species withavailable phylogenetic information). Our composite phy-logeny was based primarily on the supertree of Mank et al.(2005), whereas phylogenetic relations between species atlower levels were based on differentmolecular studies (Text inSupporting Information). Conflicting branching patternswere resolved conservatively and kept as soft polytomies.Major groups that could not be confidently placed in the treewere not included (e.g. Serranidae, Pomacanthidae, Poma-centridae, and Sciaenidae). Divergence times were notavailable for all clades, hence we employed Ornstein–Uhlenbeck (OU) transformations to obtain adequate branchlengths for phylogenetic analyses (Blomberg et al. 2003). Weselected the branch lengths that would best fit the distributionof body mass in our sample, because the distribution of bodymass often matches a simple stochastic model of characterevolution (Blomberg et al. 2003). Starting from three differentarbitrary branch lengths with different levels of hierarchy(Nee, Grafen, and Pagel in PDTREE;Garland et al. 1992), wefirst obtained the best fitting tree for each case employingmaximum likelihood methods and subsequently selected the

branch lengths that best explained the covariance of bodymass across species (i.e. lowest Mean squared error (MSE);Blomberg et al. 2003).

Statistical analyses

Analyses encompassed the following procedures: (i) discri-minant analyses to assess which ecological and phenotypicvariables could significantly discriminate different compart-ments; (ii) phylogenetic analyses to determine whether thesevariables were partly explained by phylogeny; (iii) ANCOVA

testing for prey selection among predators of differentcompartments; (iv) randomization to determine if compe-tition between shark species from different compartments islower than that expected from random sampling.

Discriminant function analysesDiscriminant analyses were performed separately for eachvariable (five 0–1 habitat categories coding forabsence ⁄presence of each species in each of the five habitattypes, trophic level, log-transformed body mass and relativeabundance), to diagnose which phenotypic attributes couldunderlie the compartmentalization of the food web.

Phylogenetic effectsWe employed two complementary approaches to estimatehow phylogeny affected trophic level, body mass, andhabitat (the relevant traits associated with food webstructure as judged by discriminant analyses; see Results).First, we compared goodness of fit of ordinary least-squareregression models (OLS) with phylogenetically generalizedregression models (PGLS). Whereas OLS assumes nohierarchical structure in the data, the correlations betweenresiduals in the PGLS models vary depending on the degreeof relatedness between species, accounting for phylogeneticeffects (Garland et al. 2005). We compared the Akaike!sinformation criterion (AIC) values across models asindicators of the models! goodness of fit, using thesmaller-is-better formulation (as a rule of thumb, modelswhose AIC is £ 2 units larger than the best fit model alsohave substantial support, whereas those with modelsresulting in AIC values > 10 units larger have virtually nosupport). Regression models included trophic level as thedependent variable and body mass and habitat as theindependent variables (comparisons between sharks andother fish species were also performed by including a 0–1variable in these models), and were performed with theMatlab program Regressionv2.m (Lavin et al. 2008).

Second, we employed randomization tests to determinewhether the phenotypic data had significant phylogeneticsignal (i.e. the tendency of closely related species to havesimilar phenotypes; Blomberg et al. 2003; Garland et al.2005). These analyses were primarily performed to estimate

Letter Correlates of food web compartments 781


if phylogenetic effects in food web structure exist (see alsoIves & Godfray 2006; Rezende et al. 2007). Subsequentcomparisons did not control for phylogeny because thestructure of the food web is inherently determined by theidentity of the species in the community.

Prey selectionTo determine whether predators from different compart-ments select prey of contrasting sizes, we performed anANCOVA comparing mass of prey consumed by differentpredators weighting by the strength of each interaction. Thesecomparisons encompassed 205 fish species (invertebrateswere not included because estimates of size were notavailable), and pairwise differences between compartmentswere estimated with a posteriori Tukey test. All prey wereincluded in this analysis (including those belonging to othercompartments), and interactions were weighted employingthe rank from 1 to 6 used to estimate compartmentalization.This analysis is more refined and has a higher statistical powerthan the discriminant analysis described before because itdiscriminates between predators and prey within eachcompartment. As a complementary analysis, we employed alogistic regression to test for size effects at the interactionlevel. Specifically, we assessed whether the relation betweenpredator and prey sizes differs across predatory interactionswithin compartments vs. interactions between compart-ments. Significance of the predator by prey size interactionin the logistic regression was tested employing the Waldcoefficient, which follows a chi-squared distribution.

Competition between shark speciesSubsequently, we explored if competition between sharkspecies was associated with the compartmentalized structureof the food web. Competition was estimated employing theJaccard index of similarity from qualitative interactionmatrices. The similarity between two predator species, i andj, was defined as S (i, j) = a ⁄ (a + b + c), where a, b, and crepresent the number of shared prey species, the number ofprey exclusive to predator i, and the number exclusive topredator j, respectively. This resulted in a matrix of pairwisecompetitive interactions between species within a singlecompartment (31 interactions) or involving species indifferent compartments (74 interactions), with their respec-tive average competitions, S1 and S2. We tested if thisclassification was able to discriminate between regions of highand low competition with a randomization procedure: sharkswere shuffled across compartments 1000 times to generate arandom distribution of S1. The null hypothesis was rejected ifS1 in the real matrix was higher than 95% of the randomiza-tions (a = 0.05). This would suggest that competition issignificantly higher between species in the same compartmentand lower between species belonging to different compart-ments than expected from random sampling.

RESUL T S

Network structure

The Caribbean food web presented a significantly compart-mentalized structure. Its average modularity level was0.212 ± 0.001 SD (n = 100 replicates of the simulatedannealing algorithm). Hence, the number of predator–preyinteractions within each of the assigned compartments isroughly 21% higher than that expected from a network withthe same compartments but random interactions betweenspecies. To assess the significance of this compartmental-ized structure controlling for each species! number ofinteractions, we compared the previous modularity levelwith that corresponding to 100 randomizations of the foodweb preserving the number of interactions per species. Themodularity of this population of randomizations was0.1672 ± 0.002 SD, significantly lower than for the realfood web (P < 0.01).

Our subsequent analyses of the Caribbean food web arebased on the output of a randomly chosen replicate of thesimulated annealing algorithm. This replicate indicated adivision of the food web into five distinct compartments(colour-coded in Fig. 1). To assess if this replicate was arepresentative sample for subsequent analyses and toquantify the consistency of results across replicates, we firstchecked how the number of compartments varied acrossreplicates. There was no variation across the 100 replicates,i.e. simulated annealing always ended up in detecting fivecompartments. Then, we quantified how conserved thedistribution of species within compartments was acrossreplicates. We calculated, for a given pair of speciesobserved in the same compartment in the working replicate,how often that particular pair of species was also foundwithin the same compartment in the remaining 99 replicates.Coincident results represented 78% of the cases. Hence,nearly 4 out of 5 pairs of species fell consistently within thesame compartments across replicates of the simulatedannealing algorithm running on the real food web. Amongsharks, results were even more robust, with coincidentresults corresponding to 83% of the cases. Of the 15 sharkspecies in the food web, only 2 species were not consistentlyassigned to the same compartment across replicates. Whenthese species are removed, coincident results increase to99.6% of the times. This is a strong indication of therobustness of our classification of the food web incompartments, particularly with regards to shark species.

Determinants of compartmentalization

Discriminant analyses suggested that body mass, trophiclevel, and habitat (0–1 coding for species inhabiting in-shoreor off-shore) can significantly discriminate between com-partments (F4,200 > 4.37, P £ 0.002 in all cases). However,



only trophic level remained significant when these threefactors were included in the model, suggesting thatdifferences between compartments in trophic level, bodysize, and habitat fell within a single discriminant axis.Comparisons of discriminant canonical functions estimatedat group means suggested that trophic level and habitat candiscriminate the purple compartment from the remainingones (Fig. 2). Habitat can also discriminate the redcompartment, but canonical functions cannot be calculatedin this case because all species from this compartmentinhabit in-shore (Fig. 2).

These discriminant analyses did not discern betweenpredators and prey within each compartment, hence we also

analysed whether predators from different compartmentsselected prey of different sizes. Results from logisticregressions showed that the interaction between predatorand prey body mass significantly explains whether predation

Figure 1 Compartmentalized structure of the Caribbean food web.(a) The entire food web. Nodes of different colours representspecies belonging to different compartments, whereas each linkrepresents a predator–prey interaction. Squares, circles, andtriangles indicate non-fish, bony fish, and shark species, respec-tively. (b) Schematic diagram of the compartmentalized structureof the food web. Each node represents a compartment and arrowsindicate the flow of biomass from the prey to predator within(loops) and between compartments. The size of each node isproportional to the number of species in that compartment. Thethickness of the arrows indicates the fraction of the interactionsbetween these two compartments in relation to the total for thefood web.

Figure 2 Effects of habitat and body size across compartments. (a)Relative frequency of species belonging to different compartmentsin shore vs. off-shore habitats. (b) Range of prey size consumed bypredators from different compartments. There was a cleardichotomy between compartments where predators eat large prey(L) and compartments where predators consume smaller prey (S ):pairwise comparisons were statistically significant across (Tukey!sP < 0.001 in four pairwise comparisons) but not within (P > 0.14in two pairwise comparisons) these groups. Results of comparisonswith the red group were not included in the figure for clarity; itencompassed just a few predatory interactions and only differedsignificantly from the purple compartment (P = 0.033). (c)Trophic level of species in each compartment, ranging between 0(species with no fish prey) and 1 (species with no fish predators).The median (line within the box), 25th and 75th percentiles (box),10th and 90th percentiles (error bars), and outliers are shown inboxplots (b) and (c).



occurs within or between compartments (v2 = 5.4, 1 d.f.,P = 0.02), and supported that predatory interactionsbetween species of similar body size occur significantlymore often within compartments. This suggests thatpredation on different prey sizes contributes to thecompartmentalization of the food web. Accordingly,predators of different compartments consumed prey ofsignificantly different size, according to a regular ANOVA

weighed by interaction strength (F4,1948 = 23.9, P <0.0001). A subsequent Tukey test showed that pairwisedifferences between compartments were significant in someinstances (P < 0.001), and separated two compartmentswhere predators consume larger prey (green and purple)from two other compartments where predators eat smallerprey (yellow and blue; Fig. 2). A similar analysis showed thatprey size differences were more dramatic across sharks(yellow, green, and purple compartments). Prey consumedby sharks from different compartments differed significantlyin size (F2,1298 = 68.3, P < 0.0001), and pairwise compar-isons were always statistically significant (Tukey!sP < 0.0001 in all cases). In summary, these analyses suggestthat the compartmentalized structure of the food web isassociated with interspecific differences in trophic level,habitat, and body size.

Phylogenetic effects

The association between body size and habitat with trophiclevel was significant in most OLS and PGLS regressions(Table 1), hence these variables are correlated as suggestedby the discriminant analysis. Comparisons between differentmodels suggested that phylogenetic effects can be veryimportant in our dataset because PGLS models performedconsiderably better than OLS regressions (differencesbetween values of the AIC from OLS models and theirPGLS counterpart were always higher than 50; Table 1).Interestingly, habitat effects were significant or borderedsignificance in OLS but not in PGLS models (Table 1),

suggesting that habitat distribution is also phylogeneticallystructured. Randomization analyses support this conclusion:habitat, body mass, relative abundance, and trophic levelexhibited significant phylogenetic signal (P < 0.001 in allcases; Table S1).

Phylogenetic effects were pervasive in this dataset butcontrastingly different between sharks and bony fishes.Whereas compartments encompassed closely phylogeneti-cally related species of bony fishes (P < 0.05 in three of thefive compartments), the opposite pattern was observed insharks (Fig. 3). Trophic levels among sharks showedsignificant phylogenetic overdispersion (i.e. closely relatedspecies tended to differ more than expected at random;Blomberg et al. 2003; Helmus et al. 2007) when analyseswere performed for this group alone (P = 0.023; Fig. 3).That is, closely related shark species tended to be segregatedin different compartments so that sharks within eachcompartment were less related to each other than expectedif they were picked at random. Interestingly, randomizationresults suggested that competitive interactions were signifi-cantly lower than expected by chance when they involvedspecies in different compartments (P < 0.008; Fig. 4),suggesting that compartmentalization decreases competitionbetween shark species.

Sharks also differed from other fishes in several aspects.For instance, sharks were significantly larger and occupiedhigher trophic levels than other fishes (F1,114 = 10.80,P = 0.001 and F1,114 = 4.84, P = 0.030, respectively).Although body mass was a significant predictor of trophiclevel (Table 1), the difference in trophic level betweensharks and other fishes remained significant after controllingfor habitat and body mass (F1,112 = 2.85, one-tailedP = 0.047; Table 1). Furthermore, their number of preda-tory interactions was higher than those of other fish speciesafter controlling for body mass (F1,113 = 45.0, P < 0.001).Because sharks are opportunistic generalist feeders, it ispossible that their prominent role in the food web partlystems from characteristic foraging mode and feeding

Table 1 Comparison between different ordinary (OLS) and phylogenetic regression models (PGLS) testing for the effects of habitat, bodymass, and taxonomic affiliation (bony fishes vs. sharks) on trophic levels

Model Regression AIC Habitat Mass Taxa

Habitat OLS 131.18 F1,114 = 10.01**PGLS 36.97 F1,114 = 0.01

Habitat + mass OLS 96.57 F1,113 = 3.15 F1,113 = 41.94***PGLS 28.63 F1,113 = 0.37 F1,113 = 10.53**

Habitat + mass + taxa OLS 86.27 F1,112 = 2.32 F1,112 = 12.36*** F1,112 = 12.52***PGLS" 27.71 F1,112 = 0.40 F1,112 = 7.09** F1,112 = 2.85

**Two-tailed P < 0.01, ***P < 0.001."Model with best fit according to smaller-is-better AIC criterion.AIC, Akaike information criterion.



strategies. In our dataset, sharks predated upon species witha significantly broader range of sizes than other fish speciesof similar size, according to PGLS controlling for body size(F1,41 = 4.0, one-tailed P = 0.026; species with trophiclevel = 0 were excluded from this analysis).

Sharks account for a disproportionally large fraction ofthe predations in the food web (Fig. 3), hence we alsoanalysed their contribution to food web structure. The roleof each species in the network can be described by acombination of its standardized within-compartment degree(z), and its PC. While z measures how well connected onespecies is to another in its compartment, PC indicates howwell distributed the interactions of a species are amongdifferent compartments (see Methods). Whereas the PC didnot differ significantly between sharks and bony fishes(mean 0.65 ± 0.01 SE across all species; ANCOVA,F1,114 = 0.01, P = 0.90), the standardized within-compart-ment degree was significantly higher in sharks (ANCOVA,F1,114 = 192.2, P < 0.001; Fig. 4). Neither PC nor z was

significantly associated with body mass (P > 0.65 in bothcases). Therefore, the importance of sharks for networkstructure stems primarily from their higher total number ofinteractions.

D I SCUSS ION

Although the existence of compartments and their role forfood web stability have been debated for more than threedecades, there are very few unequivocal examples ofcompartmentalization (e.g. Girvan & Newman 2002; Krauseet al. 2003). More importantly, despite early suggestion of thepotential role of habitat boundaries (Pimm & Lawton 1980;Girvan &Newman 2002; Krause et al. 2003), almost no studyhas clearly identified the mechanisms behind a compartmen-talized structure. In this paper, not only do we describe

Figure 3 Fish phylogeny employed for phylogenetic statisticalanalyses, showing also species-specific ecological traits andcompartment identity. Histograms show how frequently eachspecies is a prey (left) and a predator (right) in the food web,estimated as the total number of interactions weighed by theirstrength (see Materials and methods).

Figure 4 (a) Role of each species within the food web. Eachsymbol describes the participation coefficient and the within-compartment degree of each fish species. Symbol size isproportional to trophic level. Species are colour-coded accordingto the compartment to which they belong. Triangles andcircles represent sharks and bony fishes, respectively. (b) Meancompetition between shark species across and within compart-ments. Randomization analyses show that pairwise competition issignificantly lower between species belonging to differentcompartments.



compartments in a largemarine foodweb, but also assess theirassociation with several ecological and evolutionary variables.Specifically, here we provide the first empirical evidence thatbody size mediating predator–prey associations (Paine 1963;Cohen et al. 1993, 2003; Emmerson & Raffaelli 2004) can beassociated to a compartmentalized food web. This isparticularly remarkable considering that we worked withspecies averages, hence effects were strong enough to bedetected despite considerable measurement noise (e.g. intra-specific variation in size during development should affectspecies! potential preys and predators). Additional factorssuch as diet specialization and spatial structure also contributeto this compartmentalized organization.

This food web has a comparatively high number ofspecies and interactions, a high resolution (the bulk of nodescorrespond to taxonomic species), and quantitative infor-mation. However, some caveats about this dataset should bemade explicit, such as a high variability in diets and biomassestimates, an unequal level of resolution, and potentialeffects of anthropogenic impact. The first issue arises fromthe limitation of the samples of stomach contents. While thestomach contents of all fish species in the food web wereobtained from the exhaustive analysis by Randall (1967), thenumber of sampled individuals per species was highlyvariable, with 54 species represented only by 10 or lessindividuals. Still, Randall!s (1967) attempts to minimizebiases are considerable, focusing only on stomach contentsto control for differences in digestibility across prey itemsand making an effort to collect individuals across allhabitats. Therefore, despite the small sample size in a fewcases, the samples were as representative as possible.

The second potential limitation of this food web is itsunequal level of resolution across all taxa. While fishes areall resolved to the species level, there is some degree oflumping for zooplankton, microfauna, sponges, octopuses,echiuroids, and other invertebrates. It is very difficult toassess to what extent results may be affected by this biasedlevel of resolution (except that the food web would becomelarger and more complex). Nonetheless, it is worth notingthat most of the invertebrates fall within two compartments(red and blue; Fig. 1). The remaining network, composedprimarily of fishes with a more homogeneous level ofresolution, also presents a compartmentalized structure,which apparently results from the feeding choices of sharks.

With regards to anthropogenic impact, the whole Carib-bean ecosystem has been largely overfished in the last decadeswith profound implications for its structure and functioning(Jackson 1997). It is very difficult to assess to what extent thestructure reported here reflects the pristine organization ofthe community or a transient stage along a destructiveprocess, and future work is necessary in this context (e.g.Randall!s dataset has been assembled over four decades ago,hence more up-to-date records may help elucidating the

community-wide effects of continuous overfishing). Insummary, this study employs one of the most comprehensivedatasets to analyse the evolutionary and ecological correlatesunderlying food web structure, with the caveat that potentiallimitations associated with this dataset should always be keptinmind (for a detailed account of the strengths and limitationsof this dataset, see the SupportingMaterial in Bascompte et al.2005).

Results suggest that phylogenetic effects are importantdeterminants of community structure, in part because bodysize and habitat were significantly associated with evolu-tionary history. Phylogenetic analyses also point towards theunique role of some groups of species in the food web. Ouranalyses show that a few shark species account for asignificantly large fraction of the predatory interactionswithin compartments (Figs 3 and 4). Although body sizepartly accounts for the role of sharks as top predators andorganizers of compartments, differences in foraging modeprobably account for the size-independent differences introphic levels and number of predatory interactions betweensharks and other fishes. To our knowledge, this is the firsttime that phylogeny is shown to correlate with food webstructure. This complements the dominant concept of bodysize as a driver of food web structure and imposescontrolling for phylogeny before any test on the role ofbody size. Our results support recent work emphasizing thatwhile a model of optimal foraging and allometric consid-erations predicts 65% of trophic links, an accurate predic-tion of food web structure will require incorporation oftraits other than body size (Petchey et al. 2008). Our paperbuilds on that by identifying additional factors such as pastevolutionary history and habitat structure.

Previous studies have suggested that compartmentalizationmay increase network stability (May 1972; Krause et al. 2003;Teng & McCann 2004; however, see Pimm 1979). Besides itsrelation to stability sensu strictu, it would be interesting toaddress how a compartmentalized structure could reducecompetition between closely related species and thus facilitatecommunity biodiversity, as seems to be the case with sharks(Paine 1980; Duffy et al. 2007). Significant phylogeneticoverdispersion among sharks suggests that closely relatedspecies tend to occupy different compartments, emphasizinghow phylogenetic information can shed light on the evolu-tionary basis of community structure (see also Webb et al.2002). Accordingly, competition was significantly higherbetween shark species in the same compartment. Althoughthis result, together with the observed phylogenetic overdi-spersion, suggests that niche displacement between toppredators might partly explain the compartmentalized struc-ture, it is unclear if these patterns have resulted from ongoingcompetition or reflect competitive exclusion that hasoccurred in the past (or the interaction between both).Regardless of the ultimate determinants of these patterns, it is



clear that closely related shark species tend to occupy differentcompartments, and this pattern might partly explain thecoexistence of these species in the community.

To sum up, our results suggest that the Caribbean foodweb is organized in compartments that correlate withhabitat and dietary preferences (prey size). They alsoillustrate how network and phylogenetic analyses can shedlight on the ecological and evolutionary mechanismsunderlying community structure (Helmus et al. 2007),providing circumstantial evidence that competition andniche displacement at higher trophic levels might beassociated with the compartmentalized structure of thefood web. These observations give rise to testable predic-tions that may provide important insights on the relationbetween species composition, species richness, and com-munity stability. For instance, future studies may addresswhether modularity across food webs is positively relatedwith interspecific variation in body size or phylogeneticdiversity, and determine how consistently competition andniche displacement between predators high in the foodchain can be associated with compartmentalization. From aconservation perspective, our results highlight the impor-tance of sharks for the compartmentalized organization,hence potentially for the stability and biodiversity of theCaribbean food web. Selective fishing of predators can havecascading effects across the entire ecosystem (Daskalov et al.2007) and this might be particularly true for sharks in thiscommunity (see also Bascompte et al. 2005). Our findingsadd to these previous results by showing that the overfishingof a few species of sharks may jeopardize the structuralproperties of the entire food web and its derived stability. Inthis context, it is worth noting that the role of sharks in thecommunity is partly determined by their evolutionaryhistory, and not a mere reflection of their larger body sizes.

ACKNOWLEDGEMENTS

The authors thank C.J. Melian, H.G.A. de Alencar, M.R.Helmus, M. Kondoh, R. Paine, P. Buston, D. Stouffer, andthree anonymous referees for their help in different stagesof this work. E.L. Rezende is a Ramon y Cajal fellow(MCyT, Spain) and E.M. Albert is a postdoctoral fellow(Excellence Grant, Junta de Andalucıa). This work wasfunded by the European Heads of Research Councils, theEuropean Science Foundation, and the EC Sixth Frame-work Programme through a EURYI (European YoungInvestigator) Award to J. Bascompte.

RE F ERENCES

Bascompte, J., Melian, C.J. & Sala, E. (2005). Interaction strengthcombinations and the overfishing of a marine food web. Proc.Natl. Acad. Sci. USA, 102, 5443–5447.

Berlow, E.L., Neutel, A.-N., Cohen, J.E., De Ruiter, P.C., Man,B.E., Emmerson, M. et al. (2004). Interaction strengths in foodwebs: issues and opportunities. J. Anim. Ecol., 73, 585–598.

Bersier, L.F., Banasek-Richter, C. & Cattin, M.F. (2002). Quanti-tative descriptors of food-web matrices. Ecology, 83, 2394–2407.

Blomberg, S.P., Garland, T. Jr & Ives, A.R. (2003). Testing forphylogenetic signal in comparative data: behavioral traits aremore labile. Evolution, 57, 717–745.

Carpenter, S.R. & Kitchell, J.F. (1993). The Trophic Cascade in LakeEcosystems. Cambridge University Press, New York.

Cattin, M.-F., Bersier, L.-F., Banasek-Richter, C., Baltensperger, R.& Gabriel, J.-P. (2004). Phylogenetic constraints and adaptationexplain food-web structure. Nature, 427, 835–839.

Cohen, J.E. (1978). Food Webs and Niche Space. Princeton UniversityPress, Princeton, USA.

Cohen, J., Pimm, S., Yodzis, P. & Saldana, J. (1993). Body sizes ofanimal predators and animal prey in food webs. J. Anim. Ecol.,62, 67–78.

Cohen, J.E., Jonsson, T. & Carpenter, S.R. (2003). Ecologicalcommunity description using the food web, species abundance,and body size. Proc. Natl. Acad. Sci. USA, 100, 1781–1786.

Daskalov, G.M., Grishin, A.N., Rodionov, S. & Mihneva, V.(2007). Trophic cascades triggered by overfishing reveal possiblemechanisms of ecosystem regime shifts. Proc. Natl. Acad. Sci.USA, 104, 10518–10523.

Duffy, J.E., Cardinale, B.J., France, K.E., McIntyre, P.B., Thebault,E. & Loreau, M. (2007). The functional role of biodiversity inecosystems: incorporating trophic complexity. Ecol. Lett., 10,522–538.

Emmerson, M.C. & Raffaelli, D. (2004). Predator–prey body size,interaction strength and the stability of a real food web. J. Anim.Ecol., 73, 399–409.

Garland, T., Harvey, P.H. & Ives, A.R. (1992). Procedures for theanalysis of comparative data using independent contrasts. Syst.Biol., 41, 18–32.

Garland, T. Jr, Bennett, A.F. & Rezende, E.L. (2005). Phylogeneticapproaches in comparative physiology. J. Exp. Biol., 208, 3015–3035.

Girvan, M. & Newman, M.E.J. (2002). Community structure insocial and biological networks. Proc. Natl. Acad. Sci. USA, 99,7821–7826.

Guimera, R. & Amaral, L.A.N. (2005). Functional cartography ofcomplex metabolic networks. Nature, 433, 895–900.

Guimera, R., Sales-Pardo, M. & Amaral, L.A.N. (2007). Moduleidentification in bipartite and directed networks. Phys. Rev. E, 76,036102.

Helmus, M.R., Savage, K., Biebel, M.W., Maxted, J.T. & Ives, A.R.(2007). Separating the determinants of phylogenetic communitystructure. Ecol. Lett., 10, 917–925.

Ives, A.R. & Godfray, H.C. (2006). Phylogenetic analysis of trophicassociations. Am. Nat., 168, E1–E14.

Jackson, J.B.C. (1997). Reefs since Columbus. Coral Reefs, 16, S23–S32.

Krause, A.E., Frank, K.A., Mason, D.M., Ulanowicz, R.E. &Taylor, W.W. (2003). Compartments revealed in food-webstructure. Nature, 426, 282–285.

Lavin, S.R., Karasov, W.H., Ives, A.R., Middleton, K.M. &Garland, T. (2008). Morphometrics of the avian small intestine,compared with non-flying mammals: a phylogenetic approach.Physiol. Biochem. Zool., 81, 526–550.



Mank, J.E., Promislow, D.E.L. & Avise, J.C. (2005). Phylogeneticperspectives in the evolution of parental care in ray-finnedfishes. Evolution, 59, 1570–1578.

May, R. (1972). Will a large complex system be stable? Nature, 238,413–414.

Melian, C.J. & Bascompte, J. (2002). Complex networks: two waysto be robust? Ecol. Lett., 5, 705–708.

Melian, C.J. & Bascompte, J. (2004). Food web cohesion. Ecology,85, 352–358.

Melian, C.J., Bascompte, J. & Jordano, P. (2005). Spatial structureand dynamics in a marine food web. In: Aquatic Food Webs (edsBelgrano, A., Scharler, U., Dunne, J. & Ulanowicz, R.). OxfordUniversity Press, Oxford, pp. 19–24.

Mumby, P.J., Hastings, A. & Edwards, H.J. (2007). Thresholds andthe resilience of Caribbean coral reefs. Nature, 450, 98–101.

Myers, R.A. & Worm, B. (2003). Rapid worldwide depletion ofpredatory fish communities. Nature, 423, 280–283.

Myers, R.A., Baum, J.K., Shepherd, T.D., Powers, S.P. & Peterson,C.H. (2007). Cascading effects of the loss of apex predatorysharks from a coastal ocean. Science, 35, 1846–1850.

Newman, M.E.J. & Girvan, M. (2004). Finding and evaluatingcommunity structure in networks. Phys. Rev. E, 69, 026113.

Olesen, J.M., Bascompte, J., Dupont, Y.L. & Jordano, P. (2007).The modularity of pollination networks. Proc. Natl. Acad. Sci.USA, 104, 19891–19896.

Opitz, S. (1996). Trophic interactions in Caribbean coral reefs. Interna-tional Center for Living Aquatic Resources Management Tech-nical Reports 43, International Center for Living AquaticResources Management, Makati City, Philippines.

Paine, R.T. (1963). Trophic relationships of 8 sympatric predatorygastropods. Ecology, 44, 63–73.

Paine, R. (1980). Food webs: linkage, interaction strength andcommunity infrastructure, the third Tansley lecture. J. Anim.Ecol., 49, 667–685.

Pauly, D., Christensen, V., Dalsgaard, J., Froese, R.M. & Torres,F. (1998). Fishing down marine food webs. Science, 279,860–863.

Petchey, O.L., Beckerman, A.P., Riede, J.O. & Warren, P.H.(2008). Size, foraging, and food web structure. Proc. Natl. Acad.Sci. USA, 105, 4191–4196.

Pimm, S.L. (1979). The structure of food webs. Theor. Popul. Biol.,16, 144–158.

Pimm, S.L. & Lawton, J.H. (1980). Are food webs divided intocompartments? J. Anim. Ecol., 49, 879–898.

Raffaelli, D. & Hall, S. (1992). Compartments and predation in anestuarine food web. J. Anim. Ecol., 61, 551–560.

Randall, J.E. (1967). Food habits of reef fishes of the West Indies.Stud. Trop. Oceanogr., 5, 665–847.

Rezende, E.L., Lavabre, J.E., Guimaraes, P.R. Jr, Jordano, P. &Bascompte, J. (2007). Non-random coextinctions in phyloge-netically structured mutualistic networks. Nature, 448, 925–928.

Solow, A.R., Costello, C. & Beet, A. (1999). On an early result onstability and complexity. Am. Nat., 154, 587–588.

Stouffer, D.B., Camacho, J., Guimera, R., Ng, C.A. & Amaral,L.A.N. (2005). Quantitative patterns in the structure of modeland empirical food webs. Ecology, 86, 1301–1311.

Sugihara, G. (1983). Holes in niche space: a derived assembly ruleand its relation to intervality. In: Current Trends in Food WebTheory, Report 5983 (eds De Angelis, D.L., Post, W. & G.Sugihara). Oak Ridge National Laboratory, Oak Ridge, Ten-nessee, USA, pp. 25–35.

Teng, J. & McCann, K.S. (2004). Dynamics of compartmented andreticulate food webs in relation to energetic flows. Am. Nat., 164,85–100.

Webb, C.O., Ackerly, D., McPeek, M. & Donoghue, M. (2002).Phylogenies and community ecology. Ann. Rev. Ecol. Syst., 33,475–505.

Williams, R.J. & Martinez, N.D. (2000). Simple rules yield complexfood webs. Nature, 409, 180–183.

SUPPORT ING IN FORMAT ION

Additional Supporting Information may be found in theonline version of this article:

Dataset S1 Excel file describing food web.Figure S1 Phylogeny: Phylogenetic hypothesis employed inthis study.Table S1 Phylogenetic signal: Testing for the presence ofphylogenetic signal in species phenotypic attributes, employ-ing the MatLab program PHYSIG.M (14).

Please note: Wiley-Blackwell are not responsible for thecontent or functionality of any supporting materials suppliedby the authors. Any queries (other than missing material)should be directed to the corresponding author for the article.

Editor, Marcel HolyoakManuscript received 12 January 2009First decision made 11 February 2009Manuscript accepted 17 April 2009



compartments in a marine food web associated with phylogeny, … · 2014-05-08 · the food web we...

Documents