biodiversity and ecosystem functioning in evolving food...

on July 17, 2018http://rstb.royalsocietypublishing.org/Downloaded from

rstb.royalsocietypublishing.org

ResearchCite this article: Allhoff KT, Drossel B. 2016Biodiversity and ecosystem functioning

in evolving food webs. Phil. Trans. R. Soc. B

371: 20150281.http://dx.doi.org/10.1098/rstb.2015.0281

Accepted: 8 February 2016

One contribution of 17 to a theme issue

Biodiversity and ecosystem functioning

in dynamic landscapes.

Subject Areas:evolution, computational biology, ecology,

environmental science, systems biology,

theoretical biology

Keywords:food web evolution models, ecosystem

services, global change, evolutionary

emergence, community assembly,

bioenergetics approach

Author for correspondence:K. T. Allhoff

e-mail: [email protected]

& 2016 The Author(s) Published by the Royal Society. All rights reserved.

Electronic supplementary material is available

at http://dx.doi.org/10.1098/rstb.2015.0281 or

via http://rstb.royalsocietypublishing.org.

Biodiversity and ecosystem functioningin evolving food webs

K. T. Allhoff1,2 and B. Drossel1

1Institute for Condensed Matter Physics, Technische Universitat Darmstadt, Darmstadt, Germany2Institute of Ecology and Environmental Sciences, Universite Pierre et Marie Curie, Paris, France

KTA, 0000-0003-0164-7618

We use computer simulations in order to study the interplay between biodi-versity and ecosystem functioning (BEF) during both the formation andthe ongoing evolution of large food webs. A species in our model is character-ized by its own body mass, its preferred prey body mass and the width of itspotential prey body mass spectrum. On an ecological time scale, populationdynamics determines which species are viable and which ones go extinct.On an evolutionary time scale, new species emerge as modifications of existingones. The network structure thus emerges and evolves in a self-organizedmanner. We analyse the relation between functional diversity and five com-munity level measures of ecosystem functioning. These are the metabolicloss of the predator community, the total biomasses of the basal and the pred-ator community, and the consumption rates on the basal community andwithin the predator community. Clear BEF relations are observed during theinitial build-up of the networks, or when parameters are varied, causingbottom-up or top-down effects. However, ecosystem functioning measuresfluctuate only very little during long-term evolution under constant environ-mental conditions, despite changes in functional diversity. This resultsupports the hypothesis that trophic cascades are weaker in more complexfood webs.

1. IntroductionDuring the past decades, the relation between biodiversity and ecosystem func-tioning (BEF) has been intensely investigated (for reviews, see [15]). TheseBEF studies are motivated by the need to understand the mechanisms that med-iate the functioning of diverse ecosystems and to predict the consequences ofrapid changes in biodiversity owing to current extinction events [6,7]. Duffyet al. [2] emphasized the importance of taking into account processes that occurboth within and among trophic levels, because trophic processes between levelsaffect ecosystem functioning as much as facilitation and competition withintrophic levels. The authors point out that many earlier BEF studies focusedinstead on rather simple systems, as for example on a single trophic level ofrandomly assembled species, but not on complex multi-trophic communitieswith a coevolutionary history. While these studies provided first insights intoBEF relations, they are far from providing a complete picture. An overview ofnew approaches dealing with multi-trophic and non-equilibrium biodiversity,with larger spatial or temporal scales and with different types of ecosystems,can be found in the introductory chapter of this theme issue [8].

Here, we follow the suggestion of Loreau [9] that evolutionary food webmodels provide an excellent tool to study BEF-related questions. Such modelsinclude evolutionary changes in species composition in addition to populationdynamics [1012]. The network structure is not static, but evolves in a coevolu-tionary manner via the dynamical interplay between population dynamics andthe introduction of new species or morphs. It is thus possible to investigate thetime-dependent behaviour of the functioning of large food webs, both duringthe initial build-up of a network and during the ongoing species turnoveron larger time scales. Such a long-term perspective may lead to surprises thatsignificantly differ from short-term experiments, as shown by Reich et al. [13].

http://crossmark.crossref.org/dialog/?doi=10.1098/rstb.2015.0281&domain=pdf&date_stamp=2016-04-25mailto:[email protected]://dx.doi.org/10.1098/rstb.2015.0281http://dx.doi.org/10.1098/rstb.2015.0281http://rstb.royalsocietypublishing.orghttp://rstb.royalsocietypublishing.orghttp://orcid.org/http://orcid.org/0000-0003-0164-7618http://rstb.royalsocietypublishing.org/

2s3

log10( f3)log10(m)

FCR

Figp

XCC

R

3

1 2

0log10(m3)log10(m2)log10(m1)log10(m0 = 1) = 0

Figure 1. Model illustration using four species. Species 3 (black triangle) is characterized by its body mass m3, the centre of its feeding range f3 and the width of itsfeeding range s3. The Gaussian function (black curve) describes its attack rate kernel N3j on potential prey species. Here, species 3 feeds on species 2 and 1 (greytriangles) with a high-respiration low attack rate. Species 1 and 2 are consumers of the external resource, represented as species 0 with a body mass m0 1 (whitetriangle). Also illustrated is the corresponding network graph with the five measures of ecosystem functioning. After [19].

rstb.royalsocietypublishing.orgPhil.Trans.R.Soc.B

371:20150281

2


Well-known examples of evolutionary food web modelsare the webworld model [14,15] and the matching model[16,17]. An individual-based approach was recently takenby Takahashi et al. [18], who found abrupt communitytransitions and cyclic evolutionary dynamics in complexfood webs. These three models use abstract trait vectors tocharacterize the ecological niche of a species. By contrast,the model by Allhoff et al. [19] used here is based on bodymasses. It is related to the model by Loeuille & Loreau [20],and its later modifications [2123]. The species in ourmodel differ concerning their body masses (as in [20]), butalso concerning their preferred prey body masses and thewidths of their potential prey body mass spectrum. Thisreflects different possible feeding strategies and results inmore realistic and less static food web structures [19].

We investigate the relationship between the functionaldiversity of the evolving networks and five community levelmeasures of ecosystem functioning. These are the metabolicloss of the predator community, the total biomasses of thebasal and the predator community, and the consumptionrates on the basal community and within the predator commu-nity. Theoretical [2426] and empirical [27] food web studiessuggest a large variety of different BEF relations, owing tobottom-up or top-down effects. Other studies suggest a satur-ation of BEF relations owing to the dampening of trophiccascades in complex and diverse communities [2833]. Atthe beginning of our simulations, when the networks are stillrelatively small, each species addition or extinction causesmajor changes in the network structure and hence in the abilityof the consumer guild to exploit the resource. We thereforeexpect all measures of ecosystem functioning (except forresource biomass) to be positively correlated with biodiversity.However, we expect the ecosystem functioning to saturateduring the ongoing fluctuations in the network structurelong after this initial build-up, when a complex multi-trophiccommunity has formed, where the function performed by aspecies that goes extinct can be retained by others.

We also analyse the impact of two model parameters on theBEF relations, which are both well known to respond to promi-nent drivers of global change: the respiration and mortality rateincreases owing to an increased temperature in a climatechange scenario [34,35], and the carrying capacity may eitherincrease owing to nutrient enrichment or decrease with anincreasing temperature [36]. Both parameters are assumed tohave a significant impact on the resulting network structures.We expect that an increased carrying capacity leads tobottom-up effects that enable the emergence of a more diverseconsumer community, whereas an increased respiration andmortality rate leads to biomass loss in the consumer guild

and hence to a release of the resource owing to decreasedtop-down control.

2. ModelThe model includes fast ecological processes (populationdynamics) which determine for a given species compositionthe population sizes, biomass flows, and whether a species isviable in the environment created by the other species. Addition-ally, slow evolutionary processes (speciation or invasion events)add new species to the system that are similar to existing ones,leading to an ever-changing network structure. A species i ischaracterized by its body mass, mi, the centre of its feedingrange, fi and the width of its feeding range, si. These traits deter-mine the feeding interactions in the community and thereby thepopulation dynamics, as illustrated in figure 1.

(a) Population dynamicsThe population dynamics follows the multi-species generaliz-ation of the bioenergetics approach by Yodzis & Innes [37,38].The rates of change of the biomass densities Bi of the populationsare given by

_B0 G0B0 X

jconsumersg j0Bj 2:1

for the external resource (species 0 with body mass m0 1);and by

_Bi X

jresourcesejgijBi

Xjconsumers

g jiBj xiBi 2:2

for consumer species. The coefficient G0 r(1 2 B0/K)describes the logistic growth of the external resource, with agrowth rate r and a carrying capacity K. The time scale of thesystem is defined by setting r 1. The efficiency ej is prey-dependent and equals either 0.45 for feeding on the resourceor 0.85 for preying on other consumer species. gij is the mass-specific rate with which species i consumes species j, andxi x0m0:25i is a combined, mass-specific rate that describesis losses owing to respiration and mortality. The carryingcapacity K (in the same units as biomass density) and the con-stant of the respiration and mortality rate x0 (in units of kg

0.25

per year) are model parameters that are varied in this study.The mass-specific consumption rate is described via a

BeddingtondeAngelis functional response [39]:

gij 1

mi

aijBj1

Pkres: hiaikBk

Plcomp: cilBl

: 2:3

http://rstb.royalsocietypublishing.org/


371:2015028

3


The per capita rate of successful attacks of predator i on prey j,aij, is based on a Gaussian feeding kernel Nij,

aij m0:75i Nij m0:75i1

siffiffiffiffiffiffi2pp exp

log10 fi log10 mj2

2s2i

" #:

2:4

The parameter hi 0:398m0:75i in equation (2.3) is thehandling time of species i for one unit of prey biomass, andcil quantifies interference competition among predators iand l. It depends on their similarity, as measured by theoverlap Iil

NijNljd(log10mj) of their feeding kernels, via

cil cfoodIilIii

for i = l: 2:5

We assume that interference competition is higher within aspecies than between different species, e.g. owing to territorialor mating behaviour. We therefore introduce an intraspecificcompetition parameter cintra and set cii cfood cintra. The influ-ence of these competition parameters has been discussed in aprevious article [19]. Here, we use fixed values: cfood 0.8and cintra 0.6.
1
(b) Speciation eventsEach simulation starts with a single ancestor specieswith body mass m1 100 and optimal feeding parametersf1 1 and s1 1. The initial biomass densities are B0 K 100 for the resource and B1 m1e 2 102 for the ances-tor species. The parameter e 2 104 is the extinctionthreshold, i.e. the minimum population density requiredto survive.

A speciation event occurs every 104 time units. Then, eachspecies with a population size below the extinction thresholdis removed from the system, and one of the remaining species(but not the external resource) is chosen randomly as parentspecies i for a mutant species j. The logarithm of themutants body mass, log10(mj), is chosen randomly from theinterval [log10(0.5 mi), log10(2 mi)], meaning that the bodymasses of parent and mutant species differ at most by afactor of 2. The mutants initial biomass density is set toBj mje and is taken from the parent species. The logarithmof the mutants feeding centre, log10fj, is drawn randomlyfrom the interval [log10(mj) 2 3.5), (log10(mj) 2 0.5)], meaningthat the preferred prey body mass is 31000 times smallerthan the consumers body mass, consistent with the resultsfrom Brose et al. [40]. The width of the feeding range sj isdrawn randomly from the interval [0.5, 1.5]. Several vari-ations of these rules, most of them with only minor impactson the resulting network structures, were discussed in aprevious article [19].

3. MethodsTwo species with similar feeding traits (e.g. species 1 and 2 infigure 1) have a similar function in the food web and are poten-tially redundant: one of them can retain their function when theother one goes extinct. Large extinction events may thus changethe systems diversity (measured as the number of species S)with little impact on its functioning. To account for this, weuse the following measure of functional diversity (taken fromthe work of Schneider et al. (Schneider FD, Brose U, Rall BC,Guill C. 2014 The dynamics of predator diversity and ecosystem

functioning in complex food webs, unpublished data)):

FD 11

max( N1j, N2j, . . . , NSjdlog10mj: 3:1

It represents the area below the envelope of all Gaussian feedingkernels Nij, with 1 i S. Note that the area below each feedingkernel is normalized to 1, so that two species with little (much)overlap in their feeding kernels have a functional diversityclose to 2 (1). Consequently, the little network illustrated infigure 1 has a functional diversity slightly above 2. FD is thus ameasure of complementarity in the feeding preferences androughly corresponds to the number of trophic levels. Alternativedefinitions of functional diversities can be found in [41].A measure of link overlap that additionally includes the overlapin predator links and that is applicable to discrete trophic layerswas introduced by Poisot et al. [26].

We analyse the relationship between FD and five measuresof ecosystem functioning:

(1) The total biomass density of all consumer species,C

PSi1 Bi:

(2) The total biomass density of the resource species R B0.(3) The total energetic loss of the system owing to the respiration

and mortality of the consumers,XC

PSi1 xiBi:

(4) The total consumption rate on the resource,FCR

PSi1 0:45 gi0Bi:

(5) The intraguild consumption rate,Figp

PSi1

PSj1 0:85 gijBi

:

We analyse the development of these measures during the initialbuild-up of the networks and during the ongoing species turnoverlater on. For each value of the respiration and mortality rate, x00.3, 0.5, 0.7, 0.9, we performed 60 simulations (with a fixed value ofthe carrying capacity K 100). Simulations with identical parametervalues differ concerning the set of random numbers and concerningtheir runtime (Tend 5 108, 1 108, 2.5 107, 1 107, 5 106,1 106). For comparison, the generation time of the initial ances-tor species with body size m1 100 is of the order of 1/x11000.25/0.314 10 time units. The measures of ecosystem functioningare evaluated after every single, 10th or 50th mutation event, depen-dent on the runtime. In addition to these 240 simulations, weperformed another 40 simulations with a runtime of Tend 5 . 108and a fixed value of x0 0.3, but with different values of the carryingcapacity, K 50, 100, 150, 200. K and x0 both respond to promi-nent drivers of global change, as highlighted in Introduction. Theirvariation thus reflects different environmental conditions.

4. Results(a) Time seriesExamples of the initial build-up (columns 2 and 4) and of thelong-term behaviour (columns 1 and 3) of the evolving net-works are shown in figure 2. After a short period of strongdiversification, we observe a fairly layer-like structure. With alow respiration and mortality rate, x0 0.3, we obtain networkswith approximately three body mass clusters around 1, 2.5 and4. Note that the species in one body mass cluster can differ intheir feeding preferences and hence belong to different trophiclevels. Higher respiration and mortality rates represent anincreased biomass loss, so that the emergence of higher-levelspecies is hampered. With x0 0.9, only two body massclusters can emerge. The ongoing species turnover is due tonewly emerging mutants that are better adapted to availableprey species or experience less predation pressure, and thereforedisplace other species with similar feeding preferences.


543210

5.0

4.0

3.0

2.0

0

1

2

3

4

5

6

0

0.001

0 1.5 108 3 108

time

biom

ass

flow

biom

ass

func

tiona

l div

ersi

tyno

. spe

cies

trop

hic

posi

tion

log(

body

mas

s)

4.5 108

FCR

x0 = 0.3 x0 = 0.3 x0 = 0.9 x0 = 0.9

FigpXC

0 3 105 6 105

time9 105 0 1.5 108 3 108

time4.5 108 0 3 105 6 105

time9 105

0 1.5 108 3 108 4.5 108 0 3 105 6 105 9 105 0 1.5 108 3 108 4.5 108 0 3 105 6 105 9 105

0 1.5 108 3 108 4.5 108 0 3 105 6 105 9 105 0 1.5 108 3 108 4.5 108 0 3 105 6 105 9 105

0 1.5 108 3 108 4.5 108 0 3 105 6 105 9 105 0 1.5 108 3 108 4.5 108 0 3 105 6 105 9 105

0 1.5 108 3 108 4.5 108 0 3 105 6 105 9 105 0 1.5 108 3 108 4.5 108 0 3 105 6 105 9 105

0 1.5 108 3 108 4.5 108 0 3 105 6 105 9 105 0 1.5 108 3 108 4.5 108 0 3 105 6 105 9 105

0.01

0.1

1

10

2040

resourceconsumers

6080

100

510152025

totaltr. Pos. >2.530

3540

5.5

4.5

3.5

2.5

Figure 2. Four exemplary simulation runs with different run times and different values of the respiration and mortality rate x0. Shown is the evolution of bodymasses and flow-based trophic positions, and the time series of the measures that are defined in the Methods section. A detailed analysis of various networkproperties can be found in [19].


371:20150281

4


Comparing the two long-term simulations, we find thatboth parameter sets lead to food webs with a similarnumber of species. Their network structure and the measuresof ecosystem functioning, however, differ significantly. Thisconfirms the expectation that functional diversity is a betterpredictor of ecosystem functioning than the pure number ofspecies. For comparison, the same analysis as in the follow-ing, but with the species number instead of the functionaldiversity, is presented in the electronic supplementarymaterial.

Functional diversity (figure 2, line 4) increases at the begin-ning of the simulations. Later, it fluctuates around a constantvalue in case of x0 0.3 or it shows a step-like behaviour incase of x0 0.9. This step-like behaviour represents a changing

number of trophic levels: the species with body masses around3 occurring in the middle of the simulation run feed on theexternal resource, so that both body mass clusters form asingle trophic level. The two body mass clusters that emergeat the end of the simulation represent two distinct trophiclayers, as also reflected in the increased values of functionaldiversity and intraguild predation.

In general, the measures of ecosystem functioning (lines 5and 6) remain surprisingly stable after the initial build-up,despite the ongoing changes in the trophic structure. Notethat at population equilibria we find

0 X

_Bi FCR XC 1 ej

ejFigp: 4:1


1

10

met

abol

ic lo

ss

1.0 1.5 2.0 2.5

functional diversity

3.0 3.5 4.0 4.5 1.0 1.5 2.0 2.5


3.0 3.5 4.0 4.5 1.0 1.5 2.0 2.5


3.0 3.5 4.0 4.5

200

406080

100

cons

umer

bio

mas

s

707580859095

100

reso

urce

bio

mas

s

1

10

pred

atio

n on

res

ourc

e

104103102101

1

10

intr

agui

ld p

reda

tion

10

0

20

30

40

50no

. spe

cies

x0 = 0.3 x0 = 0.5 x0 = 0.7 x0 = 0.9

1.0 1.5 2.0 2.5


3.0 3.50

1 1082 1083 1084 1085 108

4.0 4.5

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

Figure 3. The relationship between biodiversity and ecosystem functioning during the evolutionary history of the food webs with different values of the respiration andmortality rate x0. The carrying capacity is set to K 100. Different colours indicate different times: black represents data from networks shortly after the simulation start,whereas light blue represents data from fully developed networks after 1.5 108 time units. (Online version in colour.)


371:20150281

5


The last term (describing efficiency losses owing to intraguildpredation) is by far the smallest, so that the predation on theresource FCR and the total metabolic loss XC are of the sameorder of magnitude.

(b) Biodiversity and ecosystem functioning duringthe initial build-up of the networks

Figure 3 shows BEF relations during the evolutionary history ofthe networks, with early and later stages coded in black or lightblue. For a low respiration and mortality rate, (x0 0.3, column1), the number of species and functional diversity increaseduring the initial period of diversification. The more diversethe consumer guild becomes, the more biomass is accumu-lated, which can be observed as an increasing value of thetotal consumer biomass C. This leads to a decreasing resource

biomass R, owing to an increased predation pressure FCR onthe resource. We also observe an increasing intraguild preda-tion Figp, owing to the emergence of higher trophic levels,and an increasing total metabolic loss XC. If we focus on thedata points long after the initial build-up (in light blue), wefind that the measures of ecosystem functioning remain sur-prisingly stable, even though functional diversity fluctuatesbetween 2.5 and 4.5, representing networks with differentnumbers of trophic levels.

The comparison of the different columns reveals thestrong impact of the respiration and mortality rate x0. Thedata clouds shift to the left with increasing values of x0,reflecting flatter network structures with fewer trophiclevels and hence lower values of functional diversity. How-ever, all datasets result in networks of similar speciesnumbers, again highlighting the importance of functional


50

10

1

10

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

101

102

103

104

1

10

1

40

30

20

10

0

100

80

60

40

20

0

10510095908580

functional diversity functional diversity

7570

no. s

peci

es

pred

atio

n on

res

ourc

ein

trag

uild

pre

datio

nm

etab

olic

loss

cons

umer

bio

mas

sre

sour

ce b

iom

ass

Figure 4. The relationship between biodiversity and ecosystem functioning during the ongoing species turnover after the initial build-up of the emerging food webs.Only those data points that emerged from the simulations with the longest runtime, tend 5 108, are shown, so that the initial build-up plays only a minor role.The colours represent four different values of the respiration and mortality rate: red (in the background) x0 0.3, green x0 0.5, blue x0 0.7, pink (in theforeground) x0 0.9. The carrying capacity is set to K 100. (Online version in colour.)


371:20150281

6


diversity. In the last column, we observe two clusters of datapoints that reflect the existence or absence of a second trophiclevel above the resource, as explained in 4a.

(c) The influence of environmental conditions onbiodiversity and ecosystem functioning relations

The different colours in figure 4 represent different values ofthe respiration and mortality rate x0. In the first panel, weobserve again that the resulting networks differ strongly intheir functional diversity, but not so much in species number.Looking at the other panels, we observe the same qualitativetrends as in figure 3 with data from the initial build-up. Thisis due to a top-down effect: increasing values of x0 lead toless biomass flow into higher trophic levels and hence to flatternetwork structures with lower functional diversities and smal-ler amounts of consumer biomass, and therefore to a reducedpredation pressure on the resource.

However, this is not a universal pattern. Figure 5shows data that were generated with different values of thecarrying capacity K. Here, we observe a bottom-up effect:higher values of the carrying capacity correspond to moreresource biomass and therefore to a larger amount of availableenergy for the system. This enables the emergence of largernetworks (in terms of both number of species and functionaldiversity) and explains the increase in consumer biomassand in the rates of biomass flow. A similar (but weaker)

effect can be obtained by increasing the growth rate r of theresource, as shown in the electronic supplementary material.

5. DiscussionAs Loreau [9] highlighted, evolutionary food web models canprovide major contributions to the BEF debate. Communitiesgenerated with such models emerge from individual levelprocesses and share a coevolutionary history, instead ofbeing randomly put together or suffering from random, arti-ficial extinction events. In our study, we analyse BEF relations(i) during the initial build-up of our networks and (ii) duringongoing changes in the species composition after that. Theformer reveals that an increasing functional diversity(which corresponds to an increasing trophic complexity inthe consumer guild) correlates with an increasing total consu-mer biomass, a decreasing resource biomass, and increasingbiomass flows into, within and out of the consumer guild(figure 3). This can be explained with a top-down argument:the larger the network, the more biomass can be accumulatedand the higher is the predation pressure on the resource.A similar pattern has been found in a meta-analysis of exper-imental studies [42]. The authors show that the average effectof a decreasing species richness is to reduce the abundance orbiomass of the focal trophic group, leading to less completedepletion of resources used by that group.


60

10

1

10

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

101

1

10

1

40

50

30

20

10

0

250

200

150

100

50

0

200

150

100

50

functional diversity functional diversity

0

no. s

peci

es

pred

atio

n on

res

ourc

ein

trag

uild

pre

datio

nm

etab

olic

loss

cons

umer

bio

mas

sre

sour

ce b

iom

ass

Figure 5. As in figure 4, but with different values of the carrying capacity: red (in the background) K 200, green K 150, blue K 100, pink (in theforeground) K 50. The respiration and mortality rate is set to x0 0.3. (Online version in colour.)


371:20150281

7


Our results are also consistent with a study by Schneideret al. (Schneider FD, Brose U, Rall BC, Guill C. 2014 Thedynamics of predator diversity and ecosystem functioningin complex food webs, unpublished data), who investigatedthe same measures of ecosystem functioning in a non-evolvingfood web model. The authors reject a long-established hypo-thesis, which suggests the release of the basal communityfrom feeding pressure with growing functional diversityowing to an increased intraguild predation within the consumercommunity [2,3,30,43]. Schneider et al. showed that such anincrease of functional diversity indeed leads to an increasedintraguild predation in the consumer community, but not toan increase of the total biomass of the basal community. Adiverse predator community might simultaneously be moreexploitative but less efficient than a species-poor community.

However, the described BEF relations are not valid over thewhole simulation time. Our measures of ecosystem functioningbecome almost constant after the initial build-up of the net-works, although functional diversity continues to varysignificantly, reflecting ongoing changes in the trophic struc-ture (figure 3). A saturation of ecosystem functioning isknown from leaf breakdown by stream fungi [44] and fromother single trophic layer systems. Saturation was also pre-dicted by the early hypothesis of the redundancy model [45].It is nevertheless surprising in our model context, becausethere is theoretical evidence that ecosystem properties greatlydepend on the functional biodiversity and in particular onthe trophic structure [2426]. In addition, empirical studiessuggest diverse BEF relations based on bottom-up or top-down effects in multi-trophic communities. For example,

Gamfeldt et al. [27] studied a controlled marine microbialsystem and found that increasing consumer richness leads toreduced prey and increased consumer biomass, whereas anincreased prey richness leads to enhanced energy transferinto higher trophic levels, and thus to increased biomasses ofconsumers and prey.

One possible explanation for the discrepancy betweenthese diverse results and our observed saturation might bethe use of different approaches to generate communities withdifferent biodiversities. Random extinction events, as com-monly used, might be too simple: Fung et al. [46] analysed amarine food web under harvesting and showed that therelation between total biomass production and the proportionof remaining fish species strongly depends on the algorithm ofspecies deletion. The need to study more realistic scenarios ofextinction was also highlighted in the reviews by Duffy et al.[2] and Cardinale et al. [3]. But no matter what algorithm isused, most studies assume that species loss is irreversible, with-out taking into account that empty niches might be refilled viaevolution or immigration. In our model, many extinctionevents take place because one existing species is replacedby a similar but slightly better-adapted new species. In thisparticular case, intact functioning is indeed no surprise.

However, we observe intact functioning even if occasion-ally a whole trophic level disappears, supporting thehypothesis that trophic cascades are weaker in more complexfood webs [2830]. Duffy predicted that the addition of anew top predator to a diverse, multi-trophic system influencesspecies abundances in the trophic level directly below, but thisinfluence does not necessarily cascade down to even lower



371:20150281

8


levels (see fig. 4 in [31]), because formerly rare predator-resist-ant species might be released from competition, so that theirfeeding compensates for loss of formerly dominant speciesthat are now eaten by the new predator. This hypothesis wassupported in a meta-analysis by Schmitz et al. [32]. Finke &Denno [33] analysed a coastal marsh community and foundthat adding more and more predators can further dampentrophic cascades owing to increased intraguild predation.These processes can only occur in communities that are suffi-ciently complex and diverse. We assume that this is givenafter, but not during, the initial build-up of our networks.

Another reason why we do not find such strong anddiverse BEF relations as observed in previous studies couldbe the fact that ecosystems in different study sites experienceand adapt to different environmental conditions [47]. This isequivalent to combining in our model simulations performedwith different respiration and mortality rates or carryingcapacities. Both parameters are well known to respond toenvironmental conditions [3436].

By varying the respiration and mortality rate x0, we foundthe same BEF relations as during the initial build-up of the net-works (figure 4). Again, this can be explained with a top-downargument: higher values of the respiration rate increase the bio-mass loss of the system, which directly leads to a decreasedtotal amount of consumer biomass, which then leads to andecreased predation pressure on the resource. Moreover, thedecreased consumer biomass hampers the emergence ofhigher-level species, which goes hand in hand with lowervalues of functional diversity. These trends become moreobvious when considering functional diversity instead of thenumber of species, as has been observed already by manyother researchers (see [48] and references therein).

In contrast, varying the carrying capacity leads to positiveBEF relations both for the resource biomass and for the

consumer biomass (figure 5). This can be explained with abottom-up argument: an increased value of the carryingcapacity K directly corresponds to an increased resource bio-mass and therefore to a larger amount of energy that isavailable to the consumer guild. Thus, larger networks witha higher functional diversity and with a larger amount of con-sumer biomass can emerge. A similar bottom-up effect wasfound in a dataset from the Cedar Creek Ecosystem ScienceReserve (see [5] for more information about this studysystem): Haddad et al. showed that the species richnessand abundance in higher trophic levels (predatory and parasi-toid anthropods) is strongly and positively related to thespecies richness in lower trophic levels (plants) [49]. Inaddition, Scherber et al. reported that plant diversity hasstrong bottom-up effects in multi-trophic networks [50].

To conclude, within a single food web model, we foundinstances of positive, negative or no correlations between func-tional diversity and ecosystem functioning measures. Thediversity of our results highlights the fact that there is noglobal answer to the question of how biodiversity interactswith ecosystem functioning, because different mechanisms areinvolved and each of them affects BEF relations in different ways.

Authors contributions. Both authors designed the study and analysed theresults. K.T.A. performed the simulations and wrote the manuscriptdraft. Both authors reviewed the manuscript.Competing interests. We declare we have no competing interests.Funding. This work was supported by the DFG research unit 1748(contract no. Dr300/13) and by the ANR ARSENIC project(ANR-14-CE02-0012).Acknowledgements. We thank Florian Schneider, Christian Guill, BjornRall, Elisa Thebault and Nicolas Loeuille for useful discussionsand/or comments on early versions of the manuscript.

References

1. Hooper DU et al. 2005 Effects of biodiversity on ecosystemfunctioning: a consensus of current knowledge. Ecol.Monogr. 75, 3 35. (doi:10.1890/04-0922)

2. Duffy JE, Cardinale BJ, France KE, McIntyre PB,Thebault E, Loreau M. 2007 The functional role ofbiodiversity in ecosystems: incorporating trophiccomplexity. Ecol. Lett. 10, 522 538. (doi:10.1111/j.1461-0248.2007.01037.x)

3. Cardinale B, Duffy E, Srivastava D, Loreau M,Thomas M, Emmerson M. 2009 Towards a food webperspective on biodiversity and ecosystemfunctioning. Studies 60, 20. (doi:10.1093/acprof:oso/9780199547951.003.0008)

4. Reiss J, Bridle JR, Montoya JM, Woodward G. 2009Emerging horizons in biodiversity and ecosystemfunctioning research. Trends Ecol. Evol. 24,505 514. (doi:10.1016/j.tree.2009.03.018)

5. Tilman D, Isbell F, Cowles JM. 2014 Biodiversity andecosystem functioning. Annu. Rev. Ecol. Evol. Syst.45, 471 493. (doi:10.1146/annurev-ecolsys-120213-091917)

6. Barnosky AD et al. 2011 Has the earths sixth massextinction already arrived? Nature 471, 51 57.(doi:10.1038/nature09678)

7. Mooney H et al. 2009 Biodiversity, climatechange, and ecosystem services. Curr. Opin.Environ. Sustain. 1, 46 54. (doi:10.1016/j.cosust.2009.07.006)

8. Brose U, Hillebrand H. 2016 Biodiversity andecosystem functioning in dynamic landscapes. Phil.Trans. R. Soc. B 371, 20150267. (doi:10.1098/rstb.2015.0267)

9. Loreau M. 2010 Linking biodiversity and ecosystems:towards a unifying ecological theory. Phil.Trans. R. Soc. B 365, 49 60. (doi:10.1098/rstb.2009.0155)

10. Drossel B, McKane AJ. 2005 Modelling food webs.In Handbook of graphs and networks: from thegenome to the internet, ch. 10. Weinheim, Germany:Wiley-VCH.

11. Fussmann GF, Loreau M, Abrams PA. 2007Eco-evolutionary dynamics of communities andecosystems. Funct. Ecol. 21, 465 477. (doi:10.1111/j.1365-2435.2007.01275.x)

12. Brannstrom A, Johansson J, Loeuille N, Kristensen N,Troost TA, Lambers RHR, Dieckmann U. 2012Modelling the ecology and evolution ofcommunities: a review of past achievements,

current efforts, and future promises. Evol. Ecol. Res.14, 601 625.

13. Reich PB, Tilman D, Isbell F, Mueller K, Hobbie SE,Flynn DFB, Eisenhauer N. 2012 Impacts ofbiodiversity loss escalate through time asredundancy fades. Science 336, 589 592.(doi:10.1126/science.1217909)

14. Drossel B, Higgs PG, McKane AJ. 2001 The influenceof predator prey population dynamics on the long-term evolution of food web structure. J. Theor. Biol.208, 91 107. (doi:10.1006/jtbi.2000.2203)

15. Drossel B, McKane AJ, Quince C. 2004 The impactof nonlinear functional responses on the long-term evolution of food web structure. J. Theor.Biol. 229, 539 548. (doi:10.1016/j.jtbi.2004.04.033)

16. Rossberg AG, Matsuda H, Amemiya T, Itoh K. 2006Food webs: experts consuming families of experts.J. Theor. Biol. 241, 552 563. (doi:10.1016/j.jtbi.2005.12.021)

17. Rossberg AG, Ishii R, Amemiya T, Itoh K. 2008 Thetop-down mechanism for body-mass abundancescaling. Ecology 89, 567 580. (doi:10.1890/07-0124.1)

http://dx.doi.org/10.1890/04-0922http://dx.doi.org/10.1111/j.1461-0248.2007.01037.xhttp://dx.doi.org/10.1111/j.1461-0248.2007.01037.xhttp://dx.doi.org/10.1093/acprof:oso/9780199547951.003.0008http://dx.doi.org/10.1093/acprof:oso/9780199547951.003.0008http://dx.doi.org/10.1016/j.tree.2009.03.018http://dx.doi.org/10.1146/annurev-ecolsys-120213-091917http://dx.doi.org/10.1146/annurev-ecolsys-120213-091917http://dx.doi.org/10.1038/nature09678http://dx.doi.org/10.1016/j.cosust.2009.07.006http://dx.doi.org/10.1016/j.cosust.2009.07.006http://dx.doi.org/10.1098/rstb.2015.0267http://dx.doi.org/10.1098/rstb.2015.0267http://dx.doi.org/10.1098/rstb.2009.0155http://dx.doi.org/10.1098/rstb.2009.0155http://dx.doi.org/10.1111/j.1365-2435.2007.01275.xhttp://dx.doi.org/10.1111/j.1365-2435.2007.01275.xhttp://dx.doi.org/10.1126/science.1217909http://dx.doi.org/10.1006/jtbi.2000.2203http://dx.doi.org/10.1016/j.jtbi.2004.04.033http://dx.doi.org/10.1016/j.jtbi.2004.04.033http://dx.doi.org/10.1016/j.jtbi.2005.12.021http://dx.doi.org/10.1016/j.jtbi.2005.12.021http://dx.doi.org/10.1890/07-0124.1http://dx.doi.org/10.1890/07-0124.1http://rstb.royalsocietypublishing.org/


371:20150281

9


18. Takahashi D, Brannstrom A, Mazzucco R, YamauchiA, Dieckmann U. 2013 Abrupt communitytransitions and cyclic evolutionary dynamics incomplex food webs. J. Theor. Biol. 337, 181 189.(doi:10.1016/j.jtbi.2013.08.003)

19. Allhoff KT, Ritterskamp D, Rall BC, Drossel B, Guill C.2015 Evolutionary food web model based on bodymasses gives realistic networks with permanentspecies turnover. Nat. Sci. Rep. 5, 10955. (doi:10.1038/srep10955)

20. Loeuille N, Loreau M. 2005 Evolutionary emergenceof size-structured food webs. Proc. Natl Acad.Sci. USA 102, 5761 5766. (doi:10.1073/pnas.0408424102)

21. Brannstrom A, Loeuille N, Loreau M, Dieckmann U.2011 Emergence and maintenance of biodiversity inan evolutionary food-web model. Theor. Ecol. 4,467 478. (doi:10.1007/s12080-010-0089-6)

22. Ingram T, Harmon LJ, Shurin JB. 2009 Nicheevolution, trophic structure, and species turnover inmodel food webs. Am. Nat. 174, 56 67. (doi:10.1086/599301)

23. Allhoff KT, Drossel B. 2013 When do evolutionaryfood web models generate complex networks?J. Theor. Biol. 334, 122 129. (doi:10.1016/j.jtbi.2013.06.008)

24. Thebault E, Loreau M. 2003 Food-web constraintson biodiversity ecosystem functioningrelationships. Proc. Natl Acad. Sci. USA 100, 14949 14 954. (doi:10.1073/pnas.2434847100)

25. Thebault E, Huber V, Loreau M. 2007 Cascadingextinctions and ecosystem functioning: contrastingeffects of diversity depending on food webstructure. Oikos 116, 163 173. (doi:10.1111/j.2006.0030-1299.15007.x)

26. Poisot T, Mouquet N, Gravel D. 2013 Trophiccomplementarity drives the biodiversity ecosystemfunctioning relationship in food webs. Ecol. Lett. 16,853 861. (doi:10.1111/ele.12118)

27. Gamfeldt L, Hillebrand H, Jonsson PR. 2005 Speciesrichness changes across two trophic levelssimultaneously affect prey and consumer biomass.Ecol. Lett. 8, 696 703. (doi:10.1111/j.1461-0248.2005.00765.x)

28. Berlow EL, Dunne JA, Martinez ND, Stark PB,Williams RJ, Brose U. 2009 Simple prediction ofinteraction strengths in complex food webs. Proc.

Natl Acad. Sci. USA 106, 187 191. (doi:10.1073/pnas.0806823106)

29. Shurin JB, Borer ET, Seabloom EW, Anderson K,Blanchette CA, Broitman B, Cooper SD, Halpern BS.2002 A cross-ecosystem comparison of the strengthof trophic cascades. Ecol. Lett. 5, 785 791. (doi:10.1046/j.1461-0248.2002.00381.x)

30. Strong DR. 1992 Are trophic cascades all wet?Differentiation and donor-control in specioseecosystems. Ecology 73, 747 754. (doi:10.2307/1940154)

31. Duffy JE. 2002 Biodiversity and ecosystem function:the consumer connection. Oikos 99, 201 219.(doi:10.1034/j.1600-0706.2002.990201.x)

32. Schmitz OJ, Hamback PA, Beckerman AP. 2000Trophic cascades in terrestrial systems: a review ofthe effects of carnivore removals on plants. Am. Nat.155, 141 153. (doi:10.1086/303311)

33. Finke DL, Denno RF. 2004 Predator diversitydampens trophic cascades. Nature 429, 407 410.(doi:10.1038/nature02554)

34. Brown JH, Gillooly JF, Allen AP, Savage VM, WestGB. 2004 Toward a metabolic theory of ecology.Ecology 85, 1771 1789. (doi:10.1890/03-9000)

35. Rall BC, Vuvic-Pestic O, Ehnes RB, Emmerson M,Brose U. 2010 Temperature, predator preyinteraction strength and population stability. Glob.Change Biol. 16, 2145 2157. (doi:10.1111/j.1365-2486.2009.02124.x)

36. Binzer A, Guill C, Brose U, Rall BC. 2012 Thedynamics of food chains under climate change andnutrient enrichment. Phil. Trans. R. Soc. B 367,2935 2944. (doi:10.1098/rstb.2012.0230)

37. Yodzis P, Innes S. 1992 Body size and consumer resource dynamics. Am. Nat. 139, 1151 1175.(doi:10.1086/285380)

38. Brose U, Williams RJ, Martinez ND. 2006 Allometricscaling enhances stability in complex food webs.Ecol. Lett. 9, 1228 1236. (doi:10.1111/j.1461-0248.2006.00978.x)

39. Skalski GT, Gilliam JF. 2001 Functional responseswith predator interference: viable alternatives to theHolling type II model. Ecology 82, 3083 3092.(doi:10.1890/0012-9658(2001)082[3083:FRWPIV]2.0.CO;2)

40. Brose U et al. 2006 Consumer resource body sizerelationships in natural food webs. Ecology 87,

2411 2417. (doi:10.1890/0012-9658(2006)87[2411:CBRINF]2.0.CO;2)

41. Petchey OL, OGorman EJ, Flynn DFB. 2009A functional guide to functional diversity measures. InBiodiversity, ecosystem functioning, and humanwellbeing: an ecological and economic perspective (edsS Naeem, DE Bunker, A Hector, M Loreau, C Perrings),pp. 49 60. Oxford, UK: Oxford University Press.

42. Cardinale BJ, Srivastava DS, Duffy JE, Wright JP,Downing AL, Sankaran M, Jouseau C. 2006 Effects ofbiodiversity on the functioning of trophic groupsand ecosystems. Nature 443, 989 992. (doi:10.1038/nature05202)

43. Finke DL, Denno RF. 2005 Predator diversity and thefunctioning of ecosystems: the role of intraguildpredation in dampening trophic cascades. Ecol. Lett. 8,1299 1306. (doi:10.1111/j.1461-0248.2005.00832.x)

44. Dang CK, Chauvet E, Gessner MO. 2005 Magnitudeand variability of process rates in fungal diversity litter decomposition relationships. Ecol. Lett. 8,1129 1137. (doi:10.1111/j.1461-0248.2005.00815.x)

45. Naeem S, Loreau M, Inchausti P. 2002 Biodiversityand ecosystem functioning: the emergence of asynthetic ecological framework. In Biodiversity andecosystem functioning: synthesis and perspectives(eds M Loreau, S Naeem, P Inchausti), pp. 3 11.Oxford, UK: Oxford University Press.

46. Fung T, Farnsworth KD, Reid DG, Rossberg AG. 2015Impact of biodiversity loss on production in complexmarine food webs mitigated by prey-release. Nat.Commun. 6, 6657. (doi:10.1038/ncomms7657)

47. Partel M, Laanisto L, Zobel M. 2007 Contrastingplant productivity diversity relationships acrosslatitude: the role of evolutionary history. Ecology88, 1091 1097. (doi:10.1890/06-0997)

48. Hooper DU et al. 2002 Species diversity, functionaldiversity and ecosystem functioning. Biodivers.Ecosyst. Funct. Synth. Perspect. 17, 195 208.

49. Haddad NM, Crutsinger GM, Gross K, Haarstad J,Knops JMH, Tilman D. 2009 Plant species lossdecreases arthropod diversity and shifts trophicstructure. Ecol. Lett. 12, 1029 1039. (doi:10.1111/j.1461-0248.2009.01356.x)

50. Scherber C et al. 2010 Bottom-up effects of plantdiversity on multitrophic interactions in abiodiversity experiment. Nature 468, 553 556.(doi:10.1038/nature09492)

http://dx.doi.org/10.1016/j.jtbi.2013.08.003http://dx.doi.org/10.1038/srep10955http://dx.doi.org/10.1038/srep10955http://dx.doi.org/10.1073/pnas.0408424102http://dx.doi.org/10.1073/pnas.0408424102http://dx.doi.org/10.1007/s12080-010-0089-6http://dx.doi.org/10.1086/599301http://dx.doi.org/10.1086/599301http://dx.doi.org/10.1016/j.jtbi.2013.06.008http://dx.doi.org/10.1016/j.jtbi.2013.06.008http://dx.doi.org/10.1073/pnas.2434847100http://dx.doi.org/10.1111/j.2006.0030-1299.15007.xhttp://dx.doi.org/10.1111/j.2006.0030-1299.15007.xhttp://dx.doi.org/10.1111/ele.12118http://dx.doi.org/10.1111/j.1461-0248.2005.00765.xhttp://dx.doi.org/10.1111/j.1461-0248.2005.00765.xhttp://dx.doi.org/10.1073/pnas.0806823106http://dx.doi.org/10.1073/pnas.0806823106http://dx.doi.org/10.1046/j.1461-0248.2002.00381.xhttp://dx.doi.org/10.1046/j.1461-0248.2002.00381.xhttp://dx.doi.org/10.2307/1940154http://dx.doi.org/10.2307/1940154http://dx.doi.org/10.1034/j.1600-0706.2002.990201.xhttp://dx.doi.org/10.1086/303311http://dx.doi.org/10.1038/nature02554http://dx.doi.org/10.1890/03-9000http://dx.doi.org/10.1111/j.1365-2486.2009.02124.xhttp://dx.doi.org/10.1111/j.1365-2486.2009.02124.xhttp://dx.doi.org/10.1098/rstb.2012.0230http://dx.doi.org/10.1086/285380http://dx.doi.org/10.1111/j.1461-0248.2006.00978.xhttp://dx.doi.org/10.1111/j.1461-0248.2006.00978.xhttp://dx.doi.org/10.1890/0012-9658(2001)082[3083:FRWPIV]2.0.CO;2http://dx.doi.org/10.1890/0012-9658(2001)082[3083:FRWPIV]2.0.CO;2http://dx.doi.org/10.1890/0012-9658(2006)87[2411:CBRINF]2.0.CO;2http://dx.doi.org/10.1890/0012-9658(2006)87[2411:CBRINF]2.0.CO;2http://dx.doi.org/10.1038/nature05202http://dx.doi.org/10.1038/nature05202http://dx.doi.org/10.1111/j.1461-0248.2005.00832.xhttp://dx.doi.org/10.1111/j.1461-0248.2005.00815.xhttp://dx.doi.org/10.1038/ncomms7657http://dx.doi.org/10.1890/06-0997http://dx.doi.org/10.1111/j.1461-0248.2009.01356.xhttp://dx.doi.org/10.1111/j.1461-0248.2009.01356.xhttp://dx.doi.org/10.1038/nature09492http://rstb.royalsocietypublishing.org/

Biodiversity and ecosystem functioning in evolving food websIntroductionModelPopulation dynamicsSpeciation events

MethodsResultsTime seriesBiodiversity and ecosystem functioning during the initial build-up of the networksThe influence of environmental conditions on biodiversity and ecosystem functioning relations

DiscussionAuthors contributionsCompeting interestsFundingAcknowledgementsReferences

biodiversity and ecosystem functioning in evolving food...

Documents