Exploitation of Chemical Signaling by Parasitoids: Impact on Host Population Dynamics

Exploitation of Chemical Signaling by Parasitoids: Impacton Host Population Dynamics

Marjolein E. Lof & Maarten De Gee & Marcel Dicke &

Gerrit Gort & Lia Hemerik

Received: 2 November 2012 /Revised: 11 April 2013 /Accepted: 30 April 2013 /Published online: 21 May 2013# Springer Science+Business Media New York 2013

Abstract Chemical information mediates species interac-tions in a wide range of organisms. Yet, the effect of chem-ical information on population dynamics is rarely addressed.We designed a spatio-temporal parasitoid—host model toinvestigate the population dynamics when both the insecthost and the parasitic wasp that attacks it can respond tochemical information. The host species, Drosophilamelanogaster, uses food odors and aggregation pheromoneto find a suitable resource for reproduction. The larvalparasitoid, Leptopilina heterotoma, uses these same odorsto find its hosts. We show that when parasitoids can respondto food odors, this negatively affects fruit fly populationgrowth. However, extra parasitoid responsiveness to aggre-gation pheromone does not affect fruit fly populationgrowth. Our results indicate that the use of the aggregationpheromone by D. melanogaster does not lead to an in-creased risk of parasitism. Moreover, the use of aggregationpheromone by the host enhances its population growth andenables it to persist at higher parasitoid densities.

Keywords Chemotaxis . Aggregation pheromone . Alleeeffect . Competition . Parasitoid-host model . Infochemicals

Introduction

Chemical information plays an important role in the biologyof many species ranging frommicrobes to mammals (Bell andCardé 1984; Dicke and Takken 2006; Kats and Dill 1998;Wyatt 2004). The so-called infochemicals (Dicke and Sabelis1988) provide information on the availability of food ormates,as well as on the presence of competitors or natural enemies.In addition to auditory, tactile, and visual cues, infochemicalsare the most important cues for insects, both over short andlong range distances (Cardé and Miller 2004; Vet and Dicke1992). Another well-known example of an infochemical is thesex pheromone emitted by female moths that attracts conspe-cific males over long distances (Ostränd and Anderbrant2003; Wall and Perry 1987). Once the chemicals are released,they are freely available for every organism in the food web(Bruinsma and Dicke 2008; Dicke and Baldwin 2010;Turlings et al. 1995). Thus, chemical communication amongindividuals of one species potentially can be spied upon by anatural enemy (Fatouros et al. 2005; Hedlund, et al. 1996;Wiskerke et al. 1993; Wyatt 2004). On the other hand, chem-ical information emitted by predators also can potentially beused by prey animals to avoid or escape from predators (Dickeand Grostal 2001; Fraker 2008; Kats and Dill 1998). Moststudies on chemical information focus on the response ofindividuals to chemical cues. However, through changes inthe behavior of individuals, chemical mediation of ecologicalinteractions also can play a significant role at the populationlevel (Vet 1999).

Host-parasitoid interactions can be influenced extensive-ly by the exploitation of chemical information. Chemicalcompounds emitted by the host provide searching parasit-oids with information on where to find their hosts. Parasitoidsoften face a problem known as the reliability-detectabilityproblem (Vet and Dicke 1992). Chemical information emittedby the host is reliable, but usually not well detectable overlong distances, as the host is under strong natural selection tobe inconspicuous to its natural enemies. Chemicals from the

Electronic supplementary material The online version of this article(doi:10.1007/s10886-013-0298-8) contains supplementary material,which is available to authorized users.

M. E. Lof (*) :M. De Gee :G. Gort : L. HemerikBiometris, Department of Mathematical and Statistical Methods,Wageningen University, P.O. Box 100, 6700 AC, Wageningen,The Netherlandse-mail: [email protected]

M. E. Lof :M. DickeLaboratory of Entomology, Wageningen University, P.O. Box8031, 6700 EH, Wageningen, The Netherlands

Present Address:M. E. LofPlant Sciences, Centre for Crop Systems Analysis, Crop and WeedEcology Group, Wageningen University, P.O. Box 430, 6700 AK,Wageningen, The Netherlands

J Chem Ecol (2013) 39:752–763DOI 10.1007/s10886-013-0298-8

host’s habitat often are better detectable over long distances,but this information is not reliable, because their presencedoes not necessarily indicate the presence of the host. Oneway to solve the reliability-detectability dilemma is to exploitchemicals that hosts emit to communicate with conspecifics,such as sex pheromones or aggregation pheromones (Dicke etal. 1994; Fatouros et al. 2008; Hedlund et al. 1996).

Models for host—parasitoid systems have been fruit-ful for many experimental and theoretical investigations,and have provided ample knowledge on parasitoid pop-ulation dynamics (Godfray 1994; Hassell 2000;Wajnberg et al. 2008). Spatial aspects are important inparasitoid-host interactions (Tilman and Kareiva 1997;Turchin 1998). Modeling studies that include spatialeffects have focused mainly on foraging behavior inenvironments where resources are heterogeneously dis-tributed (Bukovinszky et al. 2007; Charnov 1976;Haccou et al. 1991; Wajnberg et al. 2012) or focus ontemporal stability or on spatio-temporal patterns (Hirzelet al. 2007; Ives 1992; Nguyen-Huu, et al. 2006; Pearceet al. 2007; Schofield et al. 2005). There are fewstudies that model the effect of chemical informationon spatio-temporal parasitoid-host dynamics. Pearce etal. (2007), Puente et al. (2008), and Schofield et al.(2002), studied the effects of chemical information;however, in these studies only the parasitoid respondedto chemical information, i.e., herbivore-induced plantvolatiles. They, however, did not include a response ofthe herbivorous host to the chemical information.

In the present study, we modeled how a natural enemythat “eavesdrops” on the chemical communication of itshost may affect the population dynamics of the host. Ourstudy system consisted of Drosophila melanogaster, thecommon fruit fly, and one of its natural enemies, the para-sitoid Leptopilina heterotoma. This is a generalist parasitoidthat attacks the larvae of a variety of Drosophila speciesinhabiting a variety of ephemeral substrates (Janssen et al.1988). Adult female fruit flies emit a volatile aggregationpheromone (Bartelt et al. 1985) that attracts conspecifics andresults in aggregated oviposition by female fruit flies on asuitable resource (Wertheim et al. 2006). Leptopilinaheterotoma parasitoids exploit the aggregation pheromoneof the adult fruit flies to localize their hosts, the larvae of thefruit fly (Wiskerke et al. 1993).

Drosophilid fruit flies tend to aggregate on suitable re-sources. Forming aggregations can benefit individuals inpopulations that are subjected to an Allee effect (a negativeper capita growth rate at small population sizes [(Allee1931) (for review, see Wertheim et al. 2005)]. If the popu-lation density is low, individuals can have difficulties infinding a mate, or in exploiting a resource (Berec et al.2001; Wertheim et al. 2005). For instance, drosophilid fruitflies vector yeast to new substrates making it more suitable

for their offspring by adding extra food to the resource, butalso because yeast competes with fungi that have a negativeeffect on larval survival (Morais et al. 1995; Rohlfs et al.2005; Stamps et al. 2012; Wertheim et al. 2002). Anotherpossible advantage of aggregation can be a diluted risk ofattack by natural enemies at high population densities [theselfish herd theory (Hamilton 1971)]. This can, for instance,be caused by the fact that a predator only is able to attack acertain amount of prey and, thus, more individuals surviveat a high prey density, or that a large group of prey is betterable to defend itself than a small group. Host finding andegg laying take time, which limits the number of hosts aparasitoid can parasitize in a fixed period of time. On theother hand, the formation of aggregations also can havecosts. Individuals within an aggregation often experiencemore severe competition for food and mates than when theyare on their own. A large group of individuals also can bemore conspicuous to natural enemies and more vulnerableto parasites and diseases (Parrish and Edelstein-Keshet1999).

Here, we addressed the positive and negative effectsof using chemical information on the information-emitting species. For this reason, we mainly focusedon the population dynamics of D. melanogaster. Theuse of an aggregation pheromone by D. melanogastermay affect its population dynamics in three ways. First,the use of the aggregation pheromone influences twodensity-dependent effects: the response to the phero-mone promotes the formation of aggregations, whichresults in a reduction in mortality due to the Allee effect(Lof et al. 2009; Rohlfs et al. 2005; Wertheim et al. 2002).Second, the counteracting effect is that with increasing densi-ty, competition increases as well, resulting in larval mortalitydue to (scramble) competition (Hoffmeister and Rohlfs 2001;Rohlfs and Hoffmeister 2003; Wertheim et al. 2002). Thetrade-off between these two effects has been investigated(Lof et al. 2009). The present study focusses on a third aspect:the aggregation pheromone used by D. melanogaster can beexploited by their natural enemies to locate their host or prey(Wertheim et al. 2003). This may increase the mortality due topredation or parasitism. To study the effects of chemicalinformation on the population dynamics of D. melanogasterwhen a parasitoid uses the same infochemicals to find its host,we developed a spatio-temporal model that incorporates odordistribution, the behavioral responses of fruit flies and para-sitoids, and the foraging behavior of the parasitoid. It includeslarval mortality due either to the Allee effect, or to competi-tion, or to parasitism. In our model, both the parasitoid and thehost can respond to chemical information. This is a novelapproach, as other host-parasitoid models only consider che-motaxis for the parasitoid and assume random movement bythe adult host (e.g., Pearce et al. 2006, 2007; Schofield et al.2002, 2005).

J Chem Ecol (2013) 39:752–763 753

Methods and Materials

Description of the Model

We developed a dynamic spatial parasitoid—host modelthat incorporates infochemical concentration, movement ofparasitoids and adults of its hosts (both random and directedtowards the infochemical source), and the interactions be-tween the parasitoid and its larval host. The complete modeland the formulas are found in Supplementary Material S1.Below, we give a brief summary.

Fruit Fly Behavior and Within Generation PopulationDynamics After hibernation (Boulétreau-Merle et al.2003), fruit flies re-colonize an area with suitable breedingsites at the beginning of the breeding season. At the end ofthe breeding season, they leave the breeding area to findshelter. Because the winter acts as a system reset, we fo-cused on the dynamics within one season. For results andbiological conclusions, we looked at population size andpopulation mortalities after a full season.

The development time from freshly oviposited eggs toadult fruit flies is about 19 d (Ashburner et al. 2005). Oursimulation model contained ten non-overlapping fruit flygenerations. At the start of each generation, a fixed numberof resources (e.g., yeast-infected apples) lay randomly in thesimulated orchard. Next, the population dynamics ran for7 d. In the first 3 d, fruit flies dispersed and reproduced; indays 4–7 the parasitoids dispersed, searched for hosts, andparasitized fruit fly larvae (Fig. 1). The larvae that survivedconstituted the next adult fruit fly population 12 d later.

In our model, we considered female fruit flies and femaleparasitoids only. The female fruit fly population was dividedinto three subpopulations according to their current activity.On the resource, the settled flies (with density HS) matedand laid their eggs. When done, they first actively flew awayfrom the resource (leaving flies, with density HL) in arandom direction. After one time step, this gave a donut-shaped distribution. This was simulated in an integro-difference equation with a ring-random or ‘ripple’ dispersalkernel (see online Supplementary Material S2) (Allen et al.2001; Brewster and Allen 1997; Etienne et al. 2002).

Next, they become searching flies (with density HC),flying at random until chemical information directs themtowards a suitable resource (chemotaxis). Bartelt et al.(1985) showed that fruit flies (both females and males)responded more strongly to the combination of food odors(especially yeast odors, with density F) and aggregationpheromone (cis-vaccenyl acetate, with density A) than tofood odors alone, and that they did not respond to theaggregation pheromone alone. We modeled the responseof the searching subpopulations of fruit flies to food odorsand the aggregation pheromone as a composite Monod

function (Monod 1949). When the searching ends in findinga resource, the searching fly may settle on it.

Odor Distribution Recently mated female D. melanogasterdisseminate the aggregation pheromone as a slowly evaporat-ing fluid (Bartelt et al. 1985). Searching fruit flies and para-sitoids can detect this volatile (A) in the air, together with foododors (F). As we assumed that there is no wind, these odorsthus diffuse randomly, i.e., they spread out in all directions atthe same rate. Because odor diffuses in three dimensions,while we modeled in two dimensions, we introduced a lossterm to represent odor molecules that get out of reach of theinsects in the vertical dimension (de Gee et al. 2008).

Parasitoid Behavior and Within-Generation Dynamics Asfrom the fourth simulated day, the larvae that emerged fromfruit fly eggs were amenable to parasitism. From that mo-ment on, we started modeling the dynamics of the femaleparasitoid population. The parasitoids were divided into thesame three activity states as the adult fruit flies.

On the resource, foraging parasitoids (with density PF)search for unparasitized hosts to oviposit in. We modeledparasitism as a Holling type II functional response (Holling1959), which incorporates a saturating maximum parasitismrate. This response is parameterized by the host density, thesearch efficiency of L. heterotoma, and the handling time,i.e., the time between detecting a host and subsequentlyresuming the search for a next host. The rate at whichforaging parasitoids leave a specific resource was estimatedby using host-density dependent patch residence times froma study by Van Lenteren and Bakker (1978). They showedthat the patch residence time of L. heterotoma increases withincreasing numbers of hosts present on the patch.

Leaving parasitoids (with density PL) first actively flyaway from the resource in a random direction, just asfruit flies do (ring-random dispersal, see online Supple-mentary Material S2). In the air, searching parasitoids(with density PC) use chemical information to locate aresource with fruit fly larvae. Leptopilina heterotomahas an innate response to the aggregation pheromoneof D. melanogaster and some other Drosophila species(Hedlund et al. 1996; Wiskerke et al. 1993). Leptopilina.heterotoma also responds more strongly to the combinationof food odors and aggregation pheromone than to food odorsalone (Dicke et al. 1985; Wertheim et al. 2003). We used thesame response function as for fruit flies, although with differ-ent parameter values.

Between-Generation Population Dynamics Apart from par-asitism, scramble competition and an Allee effect influencelocal larval D. melanogaster survival, so larval survival rateis highest at intermediate larval densities. At high densities,competition can cause high mortality. At low densities,

754 J Chem Ecol (2013) 39:752–763

harmful fungi may cause a high mortality of fruit fly larvae(Rohlfs et al. 2005; Rohlfs 2006). In D. melanogaster, adultfruit flies inoculate the resource with yeasts before egg-laying(Morais et al. 1995). This has two beneficial effects: first, itincreases the amount of food for the larvae. Second, by com-peting with disadvantageous fungi, yeast reduces the fungalgrowth.

We assumed that parasitism does not affect the be-havior of the larvae, and that the Allee effect andcompetition affect parasitized and unparasitized larvaeequally. The surviving female larvae constitute the nextadult female generation, and the adults that emerge insitu from their pupa immediately begin to search for asuitable resource. Thus, the next generation adults startby searching.

The development from larvae to adults is not alwayssynchronized between parasitoid and host (Godfray 1994).In our system, the development time for L. heterotoma isapproximately three times longer than for D. melanogaster.Therefore, there is a delay in time of two fruit fly genera-tions before parasitoid larvae develop into adults. In thisstudy, we focused on the costs and benefits of communicat-ing through an aggregation pheromone for the fruit flypopulation dynamics; therefore, we focused on the dynam-ics of the host. Because L. heterotoma is a generalist parasit-oid, attacking more than one host species, it is not likely thatits population dynamics is affected very strongly by that ofD.melanogaster. Therefore, the parasitoid dynamics was notmodeled explicitly. Instead, each generation started with aparasitoid population that either had a fixed size, or a size that

was proportional to the present fruit fly population. In bothcases, the parasitoids started with a random distribution in thearea.

Simulation Set-Up We simulated one summer, consisting often discrete fruit fly generations. Per generation, we simu-lated three dispersal/reproduction days for the fruit fly pop-ulation followed by four dispersal/oviposition days for theparasitoid population, each day consisting of 12 h, dividedin time steps (Δt) of 5 min (Fig. 1). Non-volatile aggrega-tion pheromone is excreted only by fruit flies in the first 3 dof the generation; thereafter, there is only evaporation. Foododors are produced during all seven simulation days. Thenext 12 d that complete a generation do not require spatialsimulations: we just assess the number of larvae that surviveand constitute the next adult fruit fly population. No adultmortality within a generation is incorporated in the model.

We assessed the effect of chemical information usage byboth fruit flies and their parasitoids on the population dy-namics of fruit flies by comparing fruit fly abundance andpersistence for five different combinations of response typesto chemical information: fruit flies that disperse randomly(Dm0) or can use both food odors and aggregation phero-mone (DmFA), combined with parasitoids that disperserandomly (Lh0), use food odors (LhF), or use both foododors and aggregation pheromone (LhFA). These can becombined into six different combinations. However, weassumed that parasitoids cannot do better than fruit flieswith respect to the aggregation pheromone. Therefore, onlythe combinations Dm0-Lh0, Dm0-LhF, DmFA-Lh0, DmFA-

START

Read parameter values.Initialize: fruit fly and parasitoid distribution

next generation

Odor distribution=0

Apply odor loss µ1 dayand redistribute present odor

next day

next time step

Calculate odor evaporation, redistribute evaporated odor and apply odor loss µ5 min

Day ≤ 3 ?yes no

Calculate Sensory index fruit fly population

T iti b t ti it t t f it fli

Dispersal of HCand HL

Calculate Sensory index parasitoid population

T iti b t ti it t t it id

Dispersal of PC and PL

Transition between activity sta es fruit es

Aggregation pheromone secretionand egg laying by HS

ransition e ween activity s a es parasitoid

Ovipositing by PF

More time steps?

More days?

yes

yes

no

no

Calculate larval survivalof parasitism, Allee effect and competition

Death adult population (HL and HS := 0)HC:= surviving female larvae

STOP

More generations?yes

no

Fig. 1 Flow chart of theprocesses in the model. In ourmodel, the time step, Δt, is5 min. We simulated10generations, each generationconsisted of seven simulationdays

J Chem Ecol (2013) 39:752–763 755

LhF, and DmFA-LhFA were simulated. In the simulationswith a fixed number of parasitoids, we set the parasitoidlevel at 0, 300, 500, 700, or 900. In the simulations wherethe number of parasitoids at the beginning of a new gener-ation was a fixed fraction of the number of fruit flies present,the number of parasitoids was 0.125 times that of the fruitflies.

We considered a spatial domain of 90×90 m. Thiswas divided into 512×512 cells with side length of0.1758 m. To mitigate boundary effects, we enlargedthe simulated area to 180×180 m, with the actual do-main in the center. On this enlarged domain, we usedperiodic boundary conditions for the odor distributionand for fruit fly and parasitoid populations. In periodicboundary conditions, the left boundary of the domain isconnected with the right boundary, and the top bound-ary with the bottom boundary. This assures that theinflux of flies and parasitoids into the domain is equalto the outflow. The enlarged area eliminates possibleboundary effects for the odors. In the center of thedomain, we simulated an orchard of 60×60 m withabundant resources (1 apple m−2). To mimic the naturalsituation, for each new generation of flies, 3,600 newapples were randomly allocated. At the beginning of thesimulation, we released 4,000 adult fruit flies randomlydistributed in the orchard. We simulated each combina-tion three times with different realizations of the re-source distributions to verify the consistency of theresults.

To investigate the implications of using chemicalinformation by fruit flies while a natural enemy “spies”upon this information, we kept track of the percentagemortality of fruit fly larvae due to the Allee effect, dueto competition, and due to parasitism for the differentscenarios. With this output, we specifically investigatedwhat mortality factor (“Allee effect”, “competition” or“parasitism”) or combination of mortality factors (“par-asitism and Allee effect” or “parasitism and competi-tion”) caused fruit fly larvae mortality. We alsoinvestigated whether the formation of aggregations byfruit flies in response to the aggregation pheromoneresulted in a diluted risk of parasitism.

The numerical solution of the model equations(see Supplementary Material S2) consists of two steps.First, odor evaporation, odor diffusion, adult fruit flymotion, and parasitoid motion are solved using theintegro-difference (IDE) approach (as in Neubert et al.1995; Powell et al. 1998). Next, the population dynam-ics are simulated.

Statistics For both situations, fixed number of parasitoids(N=500) or fixed fraction of parasitoids (1 parasitoid per8 fruit flies), the development of the female fruit fly

population was studied over the first six generations (duringthe fast population growth). As stated above, each simula-tion started with 4,000 female fruit flies. In each fol-lowing generation, the size of the adult female fruit flypopulation was calculated as half the surviving numberof fruit fly larvae produced by the last generation. Asthe numbers varied over several orders of magnitude,we analyzed the logarithm (base 10) of these populationsizes. We fitted random coefficients models, allowingeach simulation to have its own quadratic regressionover generations, but all starting from the same intercept[log(4,000)]. The five distinguished groups can havedifferent mean regression coefficients for the linear andquadratic terms. The random coefficient model is a typeof linear mixed model. It was fitted using PROCMIXED of the SAS software program (version 9.2).The regression lines were compared pairwise betweengroups with F-tests, and the predicted log(counts) atgeneration 6 were calculated and compared betweenthe groups (t-tests) to quantify population development.As ten pairwise comparisons were done both for the F-tests and t-tests, we used as threshold for the P-value0.05/10=0.005, by applying the Bonferroni method.

In each generation, fractions of fruit flies either sur-vived or died due to the Allee effect, competition, orparasitism. Per generation, the distribution of fruit fliesover the four response categories, which we called aresponse pattern, was studied by using multinomial like-lihood taking into account overdispersion. The responsepatterns were allowed to depend on the five distincttreatments. Next, a comparison of the response patternsamong the five treatment groups was done using ap-proximate F-tests. The F-test statistics were calculatedas the mean change in deviance due to the treatmentgroup, divided by the estimated extra scale parameter.To check treatment effects on the individual responsecategories (Allee, competition, parasitism, and survival),both overall and pairwise comparisons between treat-ments were made per response category by using ap-proximate F-tests. The statistical analysis of the responsepatterns was done by programming in R (version 2.14.2).Separate analyses were done for the cases of fixed numberof parasitoids and fixed percentages of parasitoids.

For the diluted risk of parasitism, we studied thepercentage parasitism per resource. Its relation to larvaldensity was investigated. A diluted risk of parasitismwas shown if the percentage parasitism decreased withlocal larval density. To check whether the ability to usechemical information affected the risk of parasitism, westudied the local percentage parasitism for the five com-binations of infochemical use by fruit flies and L.heterotoma. We also checked whether there were differ-ences in the density-dependent percentages parasitism

756 J Chem Ecol (2013) 39:752–763

between the simulations with a fixed number of parasitoids(500) and a fixed fraction of parasitoids (1 adult parasitoid per8 adult flies).

Results

Infochemical Use Affects Population Growth andEstablishment Figure 2 shows the development of femalefruit fly population sizes over the first six generations (andbeyond). For fixed number of parasitoids and fixed fractionof parasitoids the patterns are comparable. Note, however,that the population growth or decline is more extreme for thefixed number of parasitoids. In an earlier study (Lof et al.2008), we found that the use of infochemicals by fruit flieshad a positive effect on their population growth. In thisstudy, we found that infochemical use by fruit flies stillhad a positive effect even in the presence of a natural enemythat can exploit this information.

For both situations, fixed number and fixed fractionof parasitoids, the total number of fruit flies in theorchard in the first six generations was, according topairwise comparisons of the six regression lines, notsignificantly different between DmFA-LhF and DmFA-LhFA groups (F2,10=0.18, P=0.84 for fixed numbers,and F2,10=0.19, P=0.83 for fixed fraction). All otherpairwise comparisons yielded highly significant differ-ences (P<0.001 for all other comparisons). The totalnumber of fruit flies in the orchard increasesd fastest

when fruit flies responded to chemical information andparasitoids did not (DmFA-Lh0). Both the DmFA-LhFand DmFA-LhFA simulations where parasitoids wereable to use some or all chemical information grewslightly slower. Furthermore, when fruit flies were not ableto use chemical information, the fruit fly populations wenttowards extinction in all replicates when parasitoids were ableto use food odors, and in two out of three replicate simulationswhen parasitoids were unable to use chemical information.The decline was faster for the Dm0-LhF simulations than forthe Dm0-Lh0 simulations.

For a fixed number of parasitoids, the population size grewfastest for group DmFA-Lh0, with predicted log(populationsize) after six generations (± SE) of 4.89±0.03. For the groupswhere parasitoids could use chemical information (DmFA-LhF and DmFA-LhFA) the predicted mean log(size) is 4.50±0.03. If the fruit flies could not use chemical information,then the predicted mean log(size) is 3.34±0.03 for parasitoidsunable to use chemical information (and 3.04±0.03 for para-sitoids able to use chemical information). We conclude thatwhen flies can respond to odors their population size atgeneration 6 is roughly 30 (≈101.5) times larger than whenthey cannot. When parasitoids are able to respond to odors,this results in a 50 % decrease of the fruit fly population (0.5≈10−0.3).

For a fixed fraction of parasitoids, the predicted means(log(size)±SE) at generation six differ in a less extremeway: 4.16±0.03 (for DmFA-Lh0), 4.03±0.03 (for bothDmFA-LhF and DmFA-LhFA), 3.51±0.03 (for Dm0-Lh0),and 3.30±0.03 (for Dm0-LhF). When fruit flies can respond

2 4 6 8 10

2.0

2.5

3.0

3.5

4.0

4.5

5.0

fixed number

generation

log(

popu

latio

n si

ze)

2 4 6 8 10

2.0

2.5

3.0

3.5

4.0

4.5

5.0

bafixed fraction

generation

log(

popu

latio

n si

ze)

Dm0−Lh0Dm0−LhFDmFA−Lh0DmFA−LhFDmFA−LhFA

Fig. 2 The dynamics of the total fruit fly population in the orchard(60×60 m), for all 10 generations. a The simulations were run with afixed total number of 500 parasitoids each generation. b The simula-tions were run with a fixed fraction of parasitoids each generations(Lh:Dm=1:8). The five lines represent the following situations: Dm0-Lh0 ● and black drawn line; Dm0-LhF ○ and black dashed line;

DmFA-Lh0 ■ and grey dashed line; DmFA-LhF □ and grey dottedline; DmFA-LhFA ◊ and drawn grey line. Dm0 Flies that disperserandomly, DmFA Flies that use both food odor and aggregation pher-omone, Lh0 Parasitoids that disperse randomly, LhF Parasitoids thatuse food odors, LhFA Parasitoids that use both food odor and aggre-gation pheromone

J Chem Ecol (2013) 39:752–763 757

to odors, their population size at generation six is roughly 5(≈100.7) times larger than when they cannot. When parasit-oids are able to respond to odors, this results in a 33 %decrease of the fruit fly population (0.67≈10−0.17).

Causes of Mortality and Size of Drosophila Population Inall simulations with a fixed number and fixed fraction ofparasitoids at the start of each generation, we saw the effectof the local parasitoid—host interactions on the numbers inthe orchard as a whole (see Fig. 2). For the fixed fraction ofparasitoid numbers, the Drosophila population stayed be-tween 6 103 and 15 104 individuals, whereas this range wasmuch larger for the simulations with a fixed number ofparasitoids (1 103 to 15 105). Thus, the relative density ofthe host and parasitoid matters. The interaction at the localscale with Allee effect, competition, and parasitism emergesalso in the fates of the total larval population (see online Fig.S3): the Allee effect shows a diminishing impact when thetotal larval population size is between 104 to 3 105 individ-uals (Fig. S3a and e), while above that number the effect ofcompetition increases (Fig. S3b). It is noted, that for the onepopulation where the adult fruit flies did not respond toodors, the mortality due to the Allee effect of their larvae

was lower than when they can. The spatial distribution ofthe adult fruit flies and consequently that of the larvaediffered between those replicates: the colonization was notyet complete. Therefore, local larval densities were higherwhen they could not use odor information. Parasitism isalways present, but it is only an important mortality causefor larval population sizes large enough for the Allee effectto be overcome, and small enough for competition not yet tohave impact (Fig. S3c and g). Thus, as expected, the Dro-sophila larval population size drove the survival and thedistribution over the different mortality causes. Hence, thelarval population size and distribution is affected by theability to use chemical information by the adults. When fruitflies can use chemical information, more fruit flies find aresource, and more eggs are laid. Consequently, the use ofchemical information does affect the initial rate of popula-tion growth or decline as was seen in Fig. 2.

Relative Contributions of Factors for Larval Mortality OverGenerations Figure 3 shows the distribution of the mortalityand survival of the larval population in each generation forthe fixed number of parasitoids (Fig. 3a–e) and the fixedfraction of parasitoids (Fig. 3f–j). In the online Table S4,

1 2 3 4 5 6 7 8 9 10

larv

al m

orta

lity

%

020

4060

8010

0 Dm0−Lh0, fixed numbera b c d e

f g h i j

1 2 3 4 5 6 7 8 9 10

020

4060

8010

0 Dm0−LhF, fixed number

1 2 3 4 5 6 7 8 9 10

020

4060

8010

0 DmFA−Lh0, fixed number

1 2 3 4 5 6 7 8 9 10

020

4060

8010

0 DmFA−LhF, fixed number

1 2 3 4 5 6 7 8 9 10

020

4060

8010

0 DmFA−LhFA, fixed number

1 2 3 4 5 6 7 8 9 10

generation

larv

al m

orta

lity

%

020

4060

8010

0 Dm0−Lh0, fixed fraction

1 2 3 4 5 6 7 8 9 10

generation

020

4060

8010

0 Dm0−LhF, fixed fraction

1 2 3 4 5 6 7 8 9 10

generation

020

4060

8010

0 DmFA−Lh0, fixed fraction

1 2 3 4 5 6 7 8 9 10

generation

020

4060

8010

0 DmFA−LhF, fixed fraction

1 2 3 4 5 6 7 8 9 10

survparonlycompallee

generation

020

4060

8010

0 DmFA−LhFA, fixed fraction

Fig. 3 Local percentage mortality of fruit fly larvae per generation.The simulations were run a–e with a fixed number of 500 parasitoidseach generation and f–j a fixed fraction of parasitoids (Lh:Dm=1:8).The abilities of the fruit flies and parasitoids are (a)+(f) Dm0-Lh0,(b)+(g) Dm0-LhF, (c)+(h) DmFA-Lh0, (d)+(i) DmFA-LhF, and(e)+(j) DmFA-LhFA. The bars depict, per generation, the cumulative

percentage mortality due to Allee effect (black), competition (darkgrey), and parasitism only (light grey) and percentage survival (white).Dm0 Flies that disperse randomly, DmFA Flies that use both food odorand aggregation pheromone, Lh0 Parasitoids that disperse randomly,LhF Parasitoids that use food odors, LhFA Parasitoids that use bothfood odor and aggregation pheromone

758 J Chem Ecol (2013) 39:752–763

first, the results of the overall comparison of the five situa-tions with respect to the distribution of the larvae over thecategories Allee, competition, parasitism, and survival areshown. In almost all generations, the overall test (see S4,exception generation 10 for fixed fraction of parasitoids)indicates a significant difference (P <0.05) between the fivedifferent scenarios for the response to odors for fixed num-ber of parasitoids and fixed fraction of parasitoids. There-fore, we report the results of the pairwise comparisons inonline Table S5. When looking per generation for both thefixed number and fixed fraction of parasitoids, we see thatthese always give rise to different patterns for the Alleeeffect (except the last generation in the fixed fraction simu-lations). Parasitism also is affected by the five differentsituations (except in generation 8 for the fixed number ofparasitoids). Competition is not affected in the fixed fractionsimulations, whereas it is in the last six generations for thefixed number of parasitoids. Survival in the first six gener-ations is significantly different for the five different situa-tions in the fixed number and fixed fraction simulations.After the sixth generation, the overall difference is some-times significant and sometimes not.

The ability of fruit flies to use chemical informationaffects the distribution over the mortality causes of theiroffspring. If fruit flies can respond to chemical information,larval mortality due to the Allee effect decreases in the firstfive generations (Fig. 3c–e and h–j). Moreover, the percent-age mortality due to parasitism increases. Both effects arecaused mainly by the increasing population size. After thefifth generation, larval competition increased so stronglythat it impeded further population growth (Fig. 3c–e) inthe simulations with a fixed number of parasitoids. In thesimulations with a fixed fraction of parasitoids, severe com-petition never took place because the fruit fly populationsize was never large enough for competition to becomeimportant. If fruit flies are not able to respond to chemicalinformation, a small effect in time is observed for percent-age larval mortality due to the Allee effect or competition,compared to fruit flies, responsive to chemical information(Fig. 3a–b) in the simulations with a fixed number of para-sitoids, while the Allee effect seems to be overcome in latergenerations in the simulations with a fixed fraction of par-asitoids (Fig. 3f–g).

Effect of the Parasitoid Number (Fixed Number Simulations) Thenumber of parasitoids in the orchard affects fruit fly persis-tence. When fruit flies cannot use chemical information, thepopulation can persist only at low parasitoid pressure(Table 1). When fruit flies can use chemical information,they can co-exist in the presence of a larger parasitoidpopulation. The presence of more parasitoids delayed thepopulation growth (data not shown). This is caused mainlyby the fruit fly population needing more time to overcome

the Allee effect due to higher parasitism. The use ofinfochemicals by parasitoids delays the population growtheven more. When parasitoids can use chemical information,more parasitoids settle on a resource, and thus, more larvaeare parasitized.

Diluted Risk of Parasitism? For simulations with a fixednumber of parasitoids, we did not find a diluted risk ofparasitism for higher numbers of larvae per resource. Onthe contrary, the percentage parasitism increased with in-creasing numbers of fruit fly larvae on a resource (Fig. 4).The increase in parasitism with number of larvae present ona resource occurred for all simulations. However, when thenumber of parasitoids was fixed, we found more variation inpercentage parasitism after the strong population growth ofthe host (Fig. 4c–d). This indicates that even though there islocally no diluted risk of parasitism, global dilution of riskexists when not enough parasitoids are present. Handling ahost and laying an egg costs time [30 sec for L. heterotoma(Wertheim 2001)]. When many larvae are present, the han-dling time sets the upper limit to how many larvae L.heterotoma can parasitize in a fixed time interval. Whenthe fruit fly population is large, consequently, the totalnumber of eggs deposited on all the resources also will bevery large. Therefore, the parasitoid cannot reach the samehigh percentage parasitism on all resources. When the num-ber of parasitoids is fixed, and the fruit fly population islarge, we found high percentage parasitism only in a part ofthe resources.

In the case where the number of parasitoids is a fixedfraction of the fruit fly population numbers, the variation inpercentage parasitism does not increase when the fruit fly

Table 1 Fruit fly persistence for increasing number of parasitoidspresent, for different combinations of the ability to respond to chemicalinformation by the fruit fly and the parasitoid

Treatment Number of parasitoids

Fruit fly- Parasitoid 0 300 500 700 900

Dm0-Lh0 + + +/− − −

Dm0-LhF + + − − −

DmFA-Lh0 + + + + −

DmFA-LhF + + + +/− −

DmFA-LhFA + + + +/− −

‘+’ denotes fruit fly persistence, ‘−’ denotes fruit fly extinction (num-ber of fruit flies in generation 10 less than in generation 1, thereforeextinction is expected in the next season) and ‘+/−’ denotes fruit flypersistence did not occur in all replicates

Dm0 Flies that disperse randomly, DmFA Flies that use both food odorand aggregation pheromone, Lh0 Parasitoids that disperse randomly,LhF Parasitoids that use food odors, LhFA Parasitoids that use bothfood odor and aggregation pheromone

J Chem Ecol (2013) 39:752–763 759

population reaches its carrying capacity. Per resource, only amaximum number of 70 Drosophila larvae were present. Atthese local densities, almost no variation in percentage par-asitism existed in the simulations with a fixed number ofparasitoids either. In our simulations, at the same larvaldensities for a fixed fraction of parasitoids, the percentageparasitism was 10 to 20 % higher than with a fixed numberof parasitoids. Therefore, we conclude that in the fixedfraction simulations the parasitoids had enough time toexploit all resources efficiently.

Discussion

The response of insects to chemical information has beenstudied in great detail both in the laboratory and in the field(reviewed in Bell and Cardé 1984; Dicke and Baldwin 2010;Dicke and Grostal 2001; Kats and Dill 1998; Fatouros et al.2008; Wertheim et al. 2005). Predators and parasitoids com-monly use chemical cues originating from the host’s habitat(plant volatiles) or cues originating from the host’s chemicalcommunication (for instance sex-pheromones) to locatetheir host (Bruinsma and Dicke 2008; De Boer and Dicke2004; De Moraes et al. 1998; Fatouros et al. 2005; Shiojiri etal. 2001; Turlings et al. 1995). The exploitation of chemicalcues of the host or the host’s habitat by predators or para-sitoids often results in a higher attack or parasitism rate(Wertheim et al. 2003; but see Tentelier and Fauvergue2007). To our knowledge, the consequences of the abilityto use chemical information on population dynamics havenot been addressed in a modeling study when both parasit-oid and host can respond to the same chemical information.

Several modeling studies, however, have addressedparasitoid—host dynamics for systems where only themovement of parasitoids is influenced by chemical informa-tion (Pearce et al. 2007; Puente et al. 2008; Schofield et al.2005).

Effects of the Use of Chemical Information by Host Consistentwith previous studies (Lof et al. 2008, 2009), we found thatthe use of chemical information by fruit flies enhanced theirown population growth rate, even when parasitoids werepresent. In addition, we found that fruit flies that use chem-ical information, i.e., food odors and the aggregation pher-omone, can persist with more parasitoids present than in thesituation where they disperse only randomly. When fruitflies use chemical information, more fruit flies find the re-sources. Therefore, more eggs are laid per resource, fruitflies more easily overcome the Allee effect, and the popu-lation is able to survive higher parasitism pressure.

Effects of Spying Parasitoids In the present study, weshowed that the ability to use chemical information by bothparasitoid and host affects the population dynamics of thehost. In accordance with experimental results of Wertheim etal. (2003) for the same experimental system, we found thatwhen parasitoids exploited chemical information, the aver-age percentage parasitism was higher. As a result, the pop-ulation growth rate of fruit flies is lower, and it takes longerto overcome the Allee effect. This negative effect is evenstronger when fruit flies are not able to use chemical infor-mation and have to find the resources by random dispersal.Protection from predators is viewed as an important selec-tive advantage to being a group member in an aggregation(Parrish and Edelstein-Keshet 1999). In our modeling study,

100 200 300 4000

20

40

60

80

100generation 3

para

sitis

m (

%)

a

100 200 300 4000

20

40

60

80

100generation 7c

100 200 300 4000

20

40

60

80

100generation 5

para

sitis

m (

%)

number of larvae on resource

b

100 200 300 4000

20

40

60

80

100generation 9

number of larvae on resource

d

Fig. 4 Larval mortality due toparasitism for each resourcedepicted for larval density onresource in simulations with afixed number of 500parasitoids; each generation.Depicted are generation a 3, b5, c 7, and d 9 for thesimulations where both fruitflies and parasitoids couldrespond to both food odors andaggregation pheromone(DmFA-LhFA)

760 J Chem Ecol (2013) 39:752–763

we did not find local diluted risk of parasitism at higherlarval densities. Instead, we found that percentage mortalityof larvae due to parasitism increased with larval density on aresource.

Ecological Costs of Chemical Communication Predationand parasitism are important mortality factors for manyinsect herbivores. As usage of chemical information inlocating prey or hosts is especially advantageous whensearching for cryptic prey, predation and parasitism mightbe expected to exert a strong selective pressure on intra-specific communication by chemical information. Neverthe-less, as pheromones are crucial to many aspects of herbivorelife history, radical alterations of these compounds can bedisadvantageous despite their exploitation by predators andparasitoids (but see Raffa et al. 2007).

In Drosophila, larval parasitism is an important mortalityfactor (Allemand et al. 1999; Fleury et al. 2004; Janssen etal. 1988; Wertheim et al. 2003). Since the larval parasitoidL. heterotoma exploits the aggregation pheromone of theadult fruit flies to localize its hosts, we expected that the useof aggregation pheromone by D. melanogaster would in-crease the percentage mortality due to parasitism in thelarvae of D. melanogaster, as was also found in the field(Wertheim et al. 2003). However, in this model study, wedid not find evidence for an ecological cost of the use ofaggregation pheromone by D. melanogaster with respect toincreased risk of parasitism. A possible explanation for thisdiscrepancy is the difference in experimental setup. In oursimulation, we did not compare two types of resources, buttwo types of parasitoids, a ‘wild-type’ parasitoid that canrespond both to food odors and its hosts aggregation pher-omone, and a hypothetical ‘mutant’ parasitoid that can re-spond only to food odors. This simulation potentially canshow the added effect of the response to aggregation pher-omone by the parasitoid, while the field situation shows thepreference for the resource with both food odors and aggre-gation pheromone. This is one of the advantages of model-ing; it can make comparisons possible that are difficult orimpossible to test in real life. Here, we could extract theeffect of the fruit fly aggregation pheromone solely onparasitism rates by comparing parasitism rates when para-sitoids can use both food odors and aggregation pheromone,and parasitism rates when parasitoids can use only foododors or no odors at all, while keeping all other propertiesof the parasitoid’s behavior unchanged. Because the para-sitoid that we studied has an innate response to its host’saggregation pheromone and to the odors produced by itshost’s habitat, this comparison cannot easily be made in afield experiment.

For our study, we were interested in positive and negativeeffects of the use of aggregation pheromone by fruit flies.Therefore, we focused on the dynamics of the host. To make

these dynamics transparent, we excluded possible effectscaused by the dynamics of the parasitoid, by assuming thatthe number of parasitoids present in each generation waseither constant or equal to a fixed proportion of the fruit flypopulation. We expect that the actual dynamics of L.heterotoma lies closest to the fixed proportion of the fruitfly population. Because L. heterotoma is a generalist, it canswitch to other Drosophila species when the density of D.melanogaster is low, or even to an alternative host habitat.Hence, it is not likely that it would go extinct when D.melanogaster is not present. On the other hand, one expectsthat when D. melanogaster is abundant, the density of L.heterotoma would increase (possibly after a time lag).

Our results indicate that parasitism rates where parasit-oids can use only food odors are close to the parasitism rateswhere parasitoids can use both food odors and its host’saggregation pheromone. A possible explanation for thisquestion may be that the patch-leaving rules were estimatedfrom an experiment of Van Lenteren and Bakker (1978) inan environment where hosts were scarce and the host distri-bution was clustered. In our experimental setup, both hosthabitats and hosts were abundant, and because of the ran-dom initial distribution of the fruit flies, almost all appleswere occupied. Finding an apple (by only using food odors)is often sufficient to find hosts, and, due to the high larvaldensities on the apples, parasitoids spend a long time on thepatch before leaving. Searching for an occupied host habitatonly takes a small fraction of the parasitoid’s time. Most ofits time is spent foraging on the apple. To test our assump-tions, it would be interesting to study whether, in abundanceof hosts and host habitats, parasitoids indeed continue toforage with the same decision rules or whether they canadjust their behavior. Furthermore, in our model, parasitoidsdo not distinguish between patches based on the larvaldensity on the patch. When they find a patch, they settleand start foraging. Only the patch residence time is affectedby the number of hosts on the patch. To test whether thisassumption is correct, it would be necessary to test whetherparasitoids avoid patches with very high larval densities,where their offspring are likely to die because of scramblecompetition between the hosts.

In this study, we did not find significant differences in thepercentage mortality of fruit fly larvae and fruit fly popula-tion size between the simulations where parasitoids coulduse only food odors on the one hand, or the simulationswhere they could also exploit the aggregation pheromone ofD. melanogaster on the other hand. However, this does notimply that there is no cost with respect to parasitism whenparasitoids can respond to chemical information. The fruitfly population increases at a slower rate (both for the fixednumber and the fixed fraction of parasitoids) and has a lowercarrying capacity (only at a fixed fraction of parasitoids)when parasitoids can use chemical information as compared

J Chem Ecol (2013) 39:752–763 761

to the simulations where they can search only randomly.This, however, is already the case when parasitoids canrespond only to food odors.

References

Allee WC (1931) Animal aggregations, a study in general sociology.University of Chicago Press, Chicago

Allemand R, Fleury F, Lemaître C, Boulétreau M (1999) Populationdynamics and competitive interactions in two species ofLeptopilina (Hymenoptera : Figitidae) which parasitize Drosophilain the Rhone valley (S-E France). Ann Soc Entomol Fr 35:97–103

Allen JC, Brewster CC, Slone DH (2001) Spatially explicit ecologicalmodels: a spatial convolution approach. Chaos Solitons Fractals12:333–347

Ashburner M, Golic KG, Hawley RS (2005) Drosophila: a laboratoryhandbook. (2nd ed). Cold Spring Harbor Laboratory Press, pp. 162–164

Bartelt RJ, Schaner AM, Jackson LL (1985) Cis-vaccenyl acetate as anaggregation pheromone in Drosophila melanogaster. J Chem Ecol11:747–1756

Bell WJ, Cardé RT (1984) Chemical ecology of insects. Chapman &Hall, London New York

Berec L, Boukal DS, Berec M (2001) Linking the Allee effect, sexualreproduction, and temperature-dependent sex determination viaspatial dynamics. Am Nat 157:217–230

Boulétreau-Merle J, Fouillet P, Varaldi J (2003) Divergent strategies inlow temperature environments for the sibling species Drosophilamelanogaster and D. simulans: overwintering in extension borderareas of France and comparison with African populations. EvolEcol 17:523–548

Brewster CC, Allen JC (1997) Spatiotemporal model for studying insectdynamics in large-scale cropping systems. Environ Entomol26:473–482

Bruinsma M, Dicke M (2008) Herbivore-induced indirect defence: frominduction mechanisms to community ecology. In: Schaller A (ed)Induced plant resistance to herbivory. Springer, Berlin, pp 31–60

Bukovinszky T, Gols R, Hemerik L, Van Lenteren JC, Vet LEM (2007)Time allocation of a parasitoid foraging in heterogeneous vegeta-tion: implications for host-parasitoid interactions. J Anim Ecol76:845–853

Cardé RT, Miller J (2004) Advances in insect chemical ecology.Cambridge University Press, Cambridge

Charnov EL (1976) Optimal foraging: the marginal value theorem.Theor Popul Biol 9:129–136

De Boer JG, Dicke M (2004) Experience with methyl salicylate affectsbehavioural responses of a predatory mite to blends of herbivore-induced plant volatiles. Entomol Exp Appl 110:181–189

de GeeM, LofME, Hemerik L (2008) The effect of chemical informationon the spatial distribution of fruit flies: II parameterization, calibra-tion and sensitivity. Bull Math Biol 70:1850–1868

de Moraes CM, Lewis WJ, Paré PW, Alborn HT, Tumlinson JH (1998)Herbivore-infested plants selectively attract parasitoids. Nature393:570–573

Dicke M, Baldwin IT (2010) The evolutionary context for herbivore-induced plant volatiles: beyond the ‘cry for help’. TIPS 15:167–175

Dicke M, Grostal P (2001) Chemical detection of natural enemies byarthropods: an ecological perspective. Annu Rev Ecol Syst 32:1–23

Dicke M, Sabelis MW (1988) Infochemical terminology: based oncost-benefit analysis rather than origin of compounds? FunctEcol 2:31–139

Dicke M, Takken W (eds) (2006) Chemical ecology: from gene toecosystem. Springer, Dordrecht

Dicke M, Van Lenteren JC, Boskamp GJF, Van Voorst R (1985)Intensification and prolongation of host searching in Leptopilinaheterotoma (Thomson) (Hymenoptera: Eucoilidae) through a kai-romone produced by Drosophila melanogaster. J Chem Ecol11:125–136

Dicke M, Vet LEM, Wiskerke JSC, Stapel O (1994) Parasitoid ofDrosophila larvae solves foraging problem through infochemicaldetour: conditions affecting employment of this strategy. Nor JAgric Sci 16:227–232

Etienne RS, Wertheim B, Hemerik L, Schneider P, Powell JA (2002) Theinteraction between dispersal, the Allee effect and scramble compe-tition affects population dynamics. Ecol Model 148:153–168

Fatouros NE, Huigens ME, Van Loon JJA, Dicke M, Hilker M (2005)Butterfly anti-aphrodisiac lures parasitic wasps. Nature 433:704

Fatouros NE, Dicke M, Mumm R, Meiners T, Hilker M (2008)Foraging behavior of egg parasitoids exploiting chemical infor-mation. Behav Ecol 19:677–689

Fleury F, Ris N, Allemand R, Fouillet P, Carton Y, Boulétreau M(2004) Ecological and genetic interactions in Drosophila-parasit-oids communities: a case study with D. melanogaster, D.simulans and their common Leptopilina parasitoids in south-eastern France. Genetica 120:181–194

Fraker ME (2008) The dynamics of predation risk assessment: re-sponses of anuran larvae to chemical cues of predators. J AnimEcol 77:638–645

Godfray HCJ (1994) Parasitoids: behavioral and evolutionary ecology.Princeton University Press, Princeton

Haccou P, De Vlas SJ, Van Alphen JJM, Visser ME (1991) Informationprocessing by foragers: effects of intra-patch experience on theleaving tendency of Leptopilina heterotoma. J Anim Ecol 60:93–106

Hamilton WD (1971) Geometry for the selfish herd. J Theor Biol31:295–311

Hassell MP (2000) The spatial and temporal dynamics of host—parasitoidinteractions. Oxford University Press, Oxford

Hedlund K, Vet LEM, Dicke M (1996) Generalist and specialistparasitoid strategies of using odours of adult drosophilid flieswhen searching for larval hosts. Oikos 77:390–398

Hirzel AH, Nisbet RM, Murdoch WW (2007) Host-parasitoid spatialdynamics in heterogeneous landscapes. Oikos 116:2082–2096

Hoffmeister TS, Rohlfs M (2001) Aggregative egg distributions maypromote species co-existence—but why do they exist? Evol EcolRes 3:37–50

Holling CS (1959) The components of predation as revealed by a studyof small mammal predation of the European pine sawfly. CanEntomol 91:293–320

Ives AR (1992) Continuous-time models of host-parasitoid interac-tions. Am Nat 140:1–29

Janssen A, Driessen G, de Haan M, Roodbol N (1988) The impact ofparasitoids on natural populations of temperate woodlandDrosophila. Neth J Zool 38:61–73

Kats LB, Dill LM (1998) The scent of death: chemosensory assessmentof predation risk by prey animals. Ecoscience 5:261–394

Lof ME, Etienne RS, Powell J, De Gee M, Hemerik L (2008) Theeffect of chemical information on the spatial distribution of fruitflies: I model results. Bull Math Biol 70:1827–1849

Lof ME, De Gee M, Hemerik L (2009) Odor-mediated aggregationenhances the colonization ability of Drosophila melanogaster. JTheor Biol 258:363–370

Monod J (1949) The growth of bacterial cultures. Annu Rev Microbiol3:371–394

Morais PB, Martins MB, Klaczko LB, Mendonça-Hagler LC, HaglerAN (1995) Yeast succession in the Amazon fruit Parahancorniaamapa as resource partitioning among Drosophila spp. ApplEnviron Microbiol 61:4251–4257

762 J Chem Ecol (2013) 39:752–763

Neubert MG, Kot M, Lewis MA (1995) Dispersal and pattern formationin a discrete-time predator–prey model. Theor Popul Biol 48:7–43

Nguyen-Huu T, Let C, Poggiale JC, Auger P (2006) Effect of move-ment frequency on global host—parasitoid spatial dynamics withunstable local dynamics. Ecol Model 197:290–295

Ostränd F, Anderbrant O (2003) From where are insects recruited? Anew model to interpret catches of attractive traps. Agric ForEntomol 5:163–171

Parrish JK, Edelstein-Keshet L (1999) Complexity, pattern, and evo-lutionary trade-offs in animal aggregation. Science 284:99–101

Pearce IG, Chaplain MAJ, Schofield PG, Anderson ARA, Hubbard SF(2006) Modelling the spatio-temporal dynamics of multi-specieshost-parasitoid interactions: heterogeneous patterns and ecologi-cal implications. J Theor Biol 241:876–886

Pearce IG, Chaplain MAJ, Schofield PG, Anderson ARA, Hubbard SF(2007) Chemotaxis-induced spatio-temporal heterogeneity inmulti-species host-parasitoid systems. J Math Biol 55:365–388

Powell JA, McMillen T, White P (1998) Connecting a chemotacticmodel for mass attack to a rapid integro-difference emulationstrategy. SIAM J Appl Math 59:547–572

Puente M, Magori K, Kennedy GG, Gould F (2008) Impact ofherbivore-induced plant volatiles on parasitoid foraging success:a spatial simulation of the Cotesia rubecula, Pierus rapae, andBrassica oleracea system. J Chem Ecol 34:959–970

Raffa KF, Hobson KR, Lafontaine S, Aukema BH (2007) Can chemicalcommunication be cryptic? Adaptations by herbivores to naturalenemies exploiting prey semiochemistry. Oecologia 153:1009–1019

Rohlfs M (2006) Genetic variation and the role of insect life historytraits in the ability of Drosophila larvae to develop in the presenceof a competing filamentous fungus. Evol Ecol 20:271–289

Rohlfs M, Hoffmeister TS (2003) An evolutionary explanation of theaggregation model of species coexistence. Proc R Soc Lond B270(suppl 1):S33–S35

Rohlfs M, Obmann B, Petersen R (2005) Competition with filamen-tous fungi and its implication for a gregarious lifestyle in insectsliving on ephemeral resources. Ecol Entomol 30:556–563

Schofield PG, Chaplain MAJ, Hubbard SF (2002) Mathematicalmodelling of host-parasitoid systems: effects of chemically medi-ated parasitoid foraging strategies on within- and between-generation spatio-temporal dynamics. J Theor Biol 214:31–47

Schofield PG, Chaplain MAJ, Hubbard SF (2005) Dynamic heteroge-neous spatio-temporal pattern formation in host-parasitoid sys-tems with synchronized generations. J Math Biol 50:559–583

Shiojiri K, Takabayashi J, Yano S, Takafuji A (2001) Infochemicalmediated tritrophic interaction webs on cabbage plants. PopulEcol 43:23–29

Stamps JA, Yang LH, Morales VM, Boundy-Mills KL (2012)Drosophila regulate yeast density and increase yeast communitysimilarity in a natural substrate. PLoS One 7:e42238

Tentelier C, Fauvergue X (2007) Herbivore-induced plant volatiles ascues for habitat assessment by a foraging parasitoid. J Anim Ecol76:1–8

Tilman D, Kareiva P (eds) (1997) Spatial ecology: the role of space inpopulation dynamics and interspecific interactions. PrincetonUniversity Press, Princeton

Turchin P (1998) Quantitative analysis of movement: measuring andmodeling population redistribution in animals and plants. SinauerAssociates Inc., Sunderland

Turlings TCJ, Loughrin JH, McCall PJ, Rose USR, Lewis WJ,Tumlinson JH (1995) How caterpillar-damaged plants protectthemselves by attracting parasitic wasps. Proc Natl Acad Sci US A 92:4169–4174

Van LENTEREN JC, Bakker K (1978) Behavioural aspects of thefunctional responses of a parasitoid (Pseudeucoila bochei Weld)to its host (Drosophila melanogaster). Neth J Zool 28:213–233

Vet LEM (1999) From chemical to population ecology: infochemicaluse in an evolutionary context. J Chem Ecol 25:31–49

Vet LEM, Dicke M (1992) Ecology of infochemical use by naturalenemies in a tritrophic context. Ann Rev Entomol 37:141–172

Wajnberg E, Bernstein E, Van Alphen JJM (eds) (2008) Behavioralecology of insect parasitoids: from theoretical approaches to fieldapplications. Blackwell Publishing Ltd, Oxford

Wajnberg E, Coquillard P, Vet LEM, Hoffmeister T (2012) Optimalresource allocation to survival and reproduction in parasitic waspsforaging in fragmented habitats. PLoS One 7:e38227

Wall C, Perry JN (1987) Range of action of moth sex-attractantsources. Entomol Exp Appl 44:5–14

Wertheim B (2001) Individual risk of parasitism in host aggrega-tions: a behaviour-based model on functional and numericalresponses. Ecology of Drosophila aggregation pheromone: amultitrophic approach. pp. 111–138. PhD Thesis, WageningenUniversity, Wageningen, The Netherlands

Wertheim B, Marchais J, Vet LEM, Dicke M (2002) Allee effect inlarval resource exploitation in Drosophila: an interaction amongdensity of adults, larvae, and micro-organisms. Ecol Entomol27:608–617

Wertheim B, Vet LEM, Dicke M (2003) Increased risk of parasitism asecological costs of using aggregation pheromones: laboratory andfield study of Drosophila - Leptopilina interaction. Oikos100:269–282

Wertheim B, Van Baalen E-J, Dicke M, Vet LEM (2005) Pheromone-mediated aggregation in nonsocial arthropods: An evolutionaryecological perspective. Ann Rev Entomol 50:321–346

Wertheim B, Allemand R, Vet LEM, Dicke M (2006) Effects of aggre-gation pheromone on individual behaviour and food web interac-tions: a field study on Drosophila. Ecol Entomol 31:216–226

Wiskerke JSC, Dicke M, Vet LEM (1993) Larval parasitoid usesaggregation pheromone of adult hosts in foraging behaviour: asolution to the reliability-detectability problem. Oecologia 93:145–148

Wyatt TD (2004) Breaking the code: illicit signallers and receivers ofsemiochemical signals. In: Pheromones and animal behaviour:communication by smell and taste. Cambridge University Press,Cambridge, pp 229–250

J Chem Ecol (2013) 39:752–763 763