letters to nature

78 NATURE | VOL 409 | 4 JANUARY 2001 | www.nature.com

suf®cient to produce observably subchondritic ratios of Nb/Ta andNb/La in the silicate Earth. More robustly, the similar behaviour ofV and Nb at both low and high pressure means that, if a signi®cantproportion of the Earth's V is in the core, it must be accompanied bya similar fraction of its Nb. At the appropriate oxygen fugacity forsingle-stage core formation in the Earth, Cr is slightly more side-rophile than Nb while Ta is much less siderophile than Nb (Fig. 2).These results are all consistent with signi®cant dissolution of V,Cr and Nb in the core and the completely lithophile behaviour ofTa.

Partition coef®cients for Si and Ga (Fig. 2, Table 2) suggest that Siis not a strong enough siderophile, even at 25 GPa, for the core tocontain 8% Si and that Ga is a stronger siderophile than is requiredto explain its depletion in the silicate Earth1. Temperature effects onDmetal/sil can be large, however17, so the limited temperature range ofour experiments preclude de®nitive conclusions in these cases.

The two competing hypotheses for Nb depletion in the silicateEarth depend only on the nature of the hidden reservoir. Either thecore contains signi®cant fractions of the Earth's V, Cr and Nb, or thedepletions of V and Cr in the mantle are solely due to incompleteaccretion to the Earth. In the latter case the Nb depletion must bedue to a hidden silicate reservoir such as subducted refractoryeclogite5. The second hypothesis would be supported if some HIMUbasalts were found to have superchondritic Nb/Ta and Nb/La(ref. 5), or if V were found to be more volatile in the solar nebulathan Fe. Current data suggest that neither of these conditionsare met7,9. M

Received 27 March; accepted 7 November 2000.

1. McDonough, W. F & Sun, S.-s. The composition of the Earth. Chem. Geol. 120, 223±253 (1995).

2. AlleÁgre, C. J., Poirier, J.-P., Hummler, E. & Hofmann, A. W. The chemical composition of the Earth.

Earth Planet. Sci. Lett. 134, 515±526 (1995).

3. Newsom, H. E. in Global Earth Physics (ed. Ahrens., T. J.) 159±189 (American Geophysical Union

Reference Shelf 1, Washington DC, 1995).

4. Hofmann, A. W. Chemical differentiation of the Earth: the relationship between mantle, continental

crust and oceanic crust. Earth Planet. Sci. Lett. 90, 297±314 (1988).

5. Rudnick, R. L., Barth, M., Horn, I. & McDonough, W. F. Rutile-bearing refractory eclogites: missing

link between continents and depleted mantle. Science 287, 278±281 (2000).

6. Drake, M. J., Newsom, H. E. & Capobianco, C. J. V, Cr and Mn in the Earth, Moon, EPB and SPB and

the origin of the Moon: Experimental studies. Geochim. Cosmochim. Acta 53, 2101±2111 (1989).

7. Wasson, J. T. Meteorites: Their Record of Early Solar-system History (Freeman & Co., New York, 1995).

8. Hofmann, A. W. Mantle geochemistry: the message from oceanic volcanism. Nature 385, 219±228

(1997).

9. Hofmann, A. W & Jochum, K. P. Source characteristics derived from very incompatible trace elements

in Mauna Loa and Mauna Kea basalts, Hawaii Scienti®c Drilling Project. J. Geophys. Res. 101, 11831±

11839 (1996).

10. Robie, R. A., Hemingway, B. S. & Fisher, J. R. Thermodynamic properties of minerals and related

substances at 298.15K and 1 bar (105 Pascals) pressure and at higher temperatures. US Geol. Surv. Bull.

1452 (1978).

11. Li, J. & Agee, C. B. Geochemistry of mantle-core differentiation at high pressure. Nature 381, 686±689

(1996).

12. Righter, K., Drake, M. J. & Yaxley, G. Prediction of siderophile element metal-silicate partition

coef®cients to 20 GPa and 2800 degrees C: The effects of pressure, temperature, oxygen fugacity, and

silicate and metallic melt compositions. Phys. Earth Planet. Int. 100, 115±134 (1997).

13. Righter, K. & Drake, M. J. Effect of water on metal-silicate partitioning of siderophile elements: a high

pressure and temperature terrestrial magma ocean and core formation. Earth Planet. Sci. Lett. 171,

383±399 (1999).

14. Thibault, Y. & Walter, M. J. The in¯uence of pressure and temperature on the metal-silicate partition-

coef®cients of nickel and cobalt in a model-c1 chondrite and implications for metal segregation in a

deep magma ocean. Geochim. Cosmochim. Acta 59, 991±1002 (1995).

15. Kilburn, M. R. & Wood, B. J. Metal-silicate partitioning and the incompatibility of S and Si during

core formation. Earth Planet. Sci. Lett. 152, 139±148 (1997).

16. Kilburn, M. R. Geochemical Constraints on the Formation of the Earth's Core. Thesis, Univ. Bristol

(1999).

17. Gessmann, C. K., Wood, B. J., Rubie, D. C. & Kilburn, M. R. Solubility of silicon in liquid metal at high

pressure: implications for the composition of the Earth's core. Earth Planet. Sci. Lett. (in the press).

18. Rudnick, R. L. Making continental crust. Nature 378, 571±578 (1995).

19. Wood, B. J. Phase transformations and partitioning relations in peridotite under lower mantle

conditions. Earth Planet. Sci. Lett. 174, 341±354 (2000).

Acknowledgements

This work was supported by the NERC. Experiments at Bayreuth were performed withassistance from the EU Large Scale Facility programme. B.J.W. acknowledges a Max Planckresearch award.

Correspondence and requests for materials should be addressed to B.J.W.(e-mail: b.j.wood@bris.ac.uk).

.................................................................Symbiotic fungal endophytescontrol insect host±parasiteinteraction websMarina Omacini*, Enrique J. Chaneton*, Claudio M. Ghersa*& Christine B. MuÈ ller²³

* IFEVA-Departamento de Recursos Naturales y Ambiente, Facultad de

Agronomia, Universidad de Buenos Aires, Av. San Martin 4453, 1417 BuenosAires, Argentina² Department of Biology and NERC Centre for Population Biology,

Imperial College at Silwood Park, Ascot, Berkshire SL5 7PY, UK³ The Zoological Society of London, Institute of Zoology, Regent's Park,London NW1 4RY, UK

..............................................................................................................................................

Symbiotic microorganisms that live intimately associated withterrestrial plants affect both the quantity and quality ofresources1,2, and thus the energy supply to consumer populationsat higher levels in the food chain. Empirical evidence on resourcelimitation of food webs points to primary productivity as a majordeterminant of consumer abundance and trophic structure3±6.Prey quality plays a critical role in community regulation7,8. Plantsinfected by endophytic fungi are known to be chemically pro-tected against herbivore consumption9±11. However, the in¯uenceof this microbe±plant association on multi-trophic interactionsremains largely unexplored. Here we present the effects of fungalendophytes on insect food webs that re¯ect limited energytransfer to consumers as a result of low plant quality, ratherthan low productivity. Herbivore±parasite webs on endophyte-free grasses show enhanced insect abundance at alternate trophiclevels, higher rates of parasitism, and increased dominance by afew trophic links. These results mirror predicted effects ofincreased productivity on food-web dynamics12. Thus `hidden'microbial symbionts can have community-wide impacts on thepattern and strength of resource±consumer interactions.

A central issue in ecology is to understand to what extent popula-tions are limited by resources or natural enemies13±16. There isemerging consensus that the interplay between these forces controlfood-web structure in response to resource enrichment6,17 as well asacross natural gradients of productivity7,12. The potential role ofsymbiotic microorganisms, such as mycorrhizae and endophytes,in the regulation of terrestrial food webs has only recently beingaddressed15. Fungal endophytes in the genus Neotyphodium (Asco-mycetes: Clavicipitaceae) form mutualistic associations with a varietyof grasses9,11. The fungal hyphae grow intercellularly in leaf and stemtissue, causing asymptomatic infections that are transmitted exclu-sively through the seeds of the host plant. Endophytic fungi obtainnutrients from their hosts, whereas infected plants may gain protec-tion from insect herbivores or vertebrate grazers via the toxic ordeterrent effects of alkaloids synthesized by the fungus9,10. No study todate has examined the impact of these fungal symbionts on multi-trophic insect assemblages. We predicted that plant infection withendophytes would alter herbivore abundance and strength ofinteractions at higher trophic levels.

We studied the structure of an aphid±parasite food web, naturallyassembled on monocultures of Lolium multi¯orum (Italian rye-grass) grown from Neotyphodium-infected or endophyte-free seeds.Aphids (Homoptera: Aphididae) are external plant feeders thatform species-rich communities with their hymenopteran para-sitoids. Aphid parasitoids encompass both specialist and polypha-gous parasitic insects, whose larvae develop singly within an aphidhost, eventually killing it. Dead aphids turn into a `mummy' inwhich parasitoid pupation takes place. Parasitoids can be sortedconveniently into trophic levels. `Primary' parasitoids attack aphid

© 2001 Macmillan Magazines Ltd

letters to nature

NATURE | VOL 409 | 4 JANUARY 2001 | www.nature.com 79

nymphal instars. `Secondary' parasitoids attack and consume pri-mary parasitoids, either within a living parasitized aphid (hyper-parasitoids) or a mummi®ed aphid (mummy parasitoids). Mummyparasitoids may also kill hyperparasitoids when they co-occur; theyare therefore the top consumers in the food chain. Interactionsbetween aphids and parasitoids are easy to quantify18 as only oneadult primary or secondary parasitoid emerges from each mummy.When secondary parasitoids are present, they consume the primaryparasitoid before emergence. In this case, secondary parasitoids inthe interaction web are linked to the source aphid18. We used thissystem to examine the impact of fungal endophytes on a food chainof four trophic levels.

The density of aphid herbivores was three times higher onendophyte-free grass monocultures (-E) than on endophyte-infected plots (+E) (Fig. 1a). The two aphid species colonizing theexperiment, Rhopalosiphum padi and Metopolophium festucae, weredifferently affected by the endophyte treatment (Fig. 1a). Removingthe endophyte also enhanced parasitoid activity, resulting in an8-fold increase in total density of parasitized aphids (t18 � 4:27,P , 0:001. This effect was driven by the higher number of R. padimummies on -E plots (t18 � 2:91, P , 0:01; for M. festucae,P � 0:35). More importantly, aphids suffered a proportionallyhigher rate ofparasitism on-E plots than on+Eones (Fig. 1b). Despitethis overall increase in parasitoid pressure, primary parasitoids wererelatively less successful on endophyte-free plots, as a result of adisproportionate increase in secondary parasitism (secondary/-

primary parasitoid ratio: +E � 2:1 versus -E � 6:5, x2 � 4:19,P � 0:041; Fig. 1b). In addition, fungal endophytes in¯uencedindividual parasitoid traits, without affecting aphid mummy size(F1;87 � 0:49). The body size of secondary parasitoids emergingfrom R. padi mummies on -E plots was larger than of those frommummies on +E plots (two-way analysis of covariance, ANCOVA,endophyte effect: F1;41 � 5:56, P � 0:023, endophyte ´ sex:P � 0:45, host size covariate: F1;41 � 3:28, P � 0:08). No effect ofendophyte infection on body size was detected for secondaryparasitoids emerging from M. festucae (P � 0:82). Thus, removingthe limitation imposed by endophytes on herbivore density gener-ated a cascade of effects, with unequal consequences for consumersat upper trophic levels. Whereas herbivores and top parasites

0

10

20

30

40

50

60

70

80

90a

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

+ E – E

b

Endophyte infection

Ap

hid

den

sity

(num

ber

per

plo

t)R

ate

of p

aras

itism

Figure 1 Response of insects to the grass-endophyte association. a, Density response of

the aphids Rhopalosiphum padi (shaded bars) and Metopolophium festucae (empty bars)

to the removal of fungal endophytes. Bars show that the mean (6s.e.m.) total aphid

density differed (t 18 � 2:42, P � 0:026) between plots with (+E) and without (-E)

endophytes. This resulted from a signi®cant difference for R. padi (t 18 � 3:00,

P � 0:008) but not for M. festucae (P � 0:63). b, Total rate of aphid parasitism in the

two endophyte treatments, including the proportion of emerged primary (shaded bars) and

secondary (open bars) parasitoids. The rate of parasitism re¯ects the proportion of all

aphids that were mummi®ed; a randomization test showed that this rate differed between

treatments (P , 0:025).

Rhopalosiphum padi Metopolophium festucae

Aphidius sp. Asaphes sp.

Dendroceruscarpenteri

Dendrocerusaphidum Aphidius

rhopalosiphi

Aphidius ervi

Parasitoid scale: aphid × 1.1

Aphid density: 14.2 per plot (+E)

a

Rhopalosiphum padi Metopolophium festucae

Aphidius sp. Asaphessp.

Dendroceruscarpenteri

Dendrocerusaphidum

Phenoglyphisvillosa

Aphidiusrhopalosiphi

Aphidius ervi

Parasitoid scale: aphid × 0.62

Aphid density: 63.6 per plot (-E)

b

Figure 2 Aphid±parasite interaction webs established on Lolium multi¯orum

monocultures. a, With (+E) and b, without (-E) infection by fungal endophytes. The length

of the horizontal bars represents the population density of two aphid species (bottom level)

and their hymenopteran parasitoid consumers (top level), including primary parasitoids

(black bars), secondary mummy parasitoids (dark grey bars) and a hyperparasitoid (light

grey bar). The basal width of the connections from parasitoids to aphids re¯ects the

proportion of different parasitoid species that emerged from each host species per

treatment. Web diagrams are scaled according to the respective mean aphid densities to

aid visual comparison of treatments.

© 2001 Macmillan Magazines Ltd

letters to nature

80 NATURE | VOL 409 | 4 JANUARY 2001 | www.nature.com

reacted positively to endophyte removal, intermediate parasitesshowed no substantial change in reproductive success, despitetheir increased attack rate. Such `bottom-up cascades' are predictedby food-chain models that stress the interaction between basalresources and dynamical consumer feedbacks in the control ofcommunity structure3,12,15.

Our food web was structured upon two aphid-based trophicchains that were connected at the top level by polyphagous second-ary parasitoids (Fig. 2). The increased dominance of R. padi on -Eplots produced an increase in the median strength of parasiticinteractions (Table 1). Conversely, trophic links between parasitoidsand M. festucae became weaker compared to those based on R. padi(Fig. 2b), which reduced the evenness of pairwise interactionstrengths across the web (Table 1). Endophyte removal enhancedthe complexity of the host±parasite web by increasing the numberof trophic links per species, connectance, and the number of indirectlinks through shared parasitoids (Table 1; Fig. 2). This was partlyexplained by the appearance of a generalist hyperparasitoid on -Eplots (Fig. 2b). Moreover, the addition of a hyperparasitoid specieseffectively generated a longer food chain on endophyte-free grassmonocultures. The observed changes in food-web structure weredriven mainly by the different response of the two aphid species tothe endophyte treatment. Several mechanisms may account for thisheterogeneous reaction within the herbivore trophic level. First, thetwo aphid species may have different sensitivity to endophyteinfection10. Second, R. padi may have a competitive advantageover M. festucae that is ampli®ed in the absence of endophytes.Third, the larger number of parasitoids supported by R. padi on -Eplots could increase the consumer pressure on M. festucae throughparasite-mediated apparent competition19±21. Theoretical modelssuggest that the existence of weak trophic links maintained bygeneralist consumers is important in promoting community per-sistence and stability22.

We initially assumed that endophyte effects on insect consumerswould be mediated by measurable changes in productivity and/orquality of the plant resource. Here, changes in insect performanceoccurred without signi®cant differences in total above-ground plantbiomass between +E (210.7 g) and -E (191.6 g per 0.25 m2) mono-cultures (t8 � 0:04, P � 0:97). In addition, we measured leafnitrogen (N) concentration as an estimate of endophyte effects onplant quality. Against expectation, endophyte removal decreasedleaf N concentration in L. multi¯orum (+E � 1:02 6 0:04 versus-E � 0:89 6 0:01, t8 � 3:47, P , 0:01). This result might indicatethat part of the extra N-infected plants is diverted into alkaloidsynthesis by the fungus, making it inaccessible to insect herbivores10.Together with the well-documented accumulation of chemicaldefences in endophyte-infected grasses9±11, our results suggest thatconsumers were limited by unidenti®ed components of resourcequality, but not by primary productivity.

It is conceivable that the effect of plant endosymbionts on foodwebs will cascade up through various trophic pathways. We found

that the frequency of grass stems attacked by leaf mining insects wasconsistently higher on -E (33.3%) than on +E plots (19.3%)throughout the experiment (repeated-measures analysis of varianceANOVA, endophyte effect: F1;18 � 7:83, P � 0:012; endophyte ´date: F2;36 � 0:16, P � 0:86). Negative impacts of fungal endo-phytes on different phytophagous insects were reported by manyother studies9±11. Moreover, a few laboratory experiments showedthat grass endophytes can affect parasitoid performance23,24. Ourexperiment demonstrates for the ®rst time that endosymbionts ofplants are able to alter species abundance and consumer±resourceinteractions across multiple trophic levels in a ®eld setting.

Microorganisms can greatly affect many terrestrial com-munities25. Research on the ecological role of microbial symbiontshas focused on their impact in plant communities2,26. In particular,endophytes can mediate competitive interactions between plantspecies affecting vegetation diversity and succession27. We haveshown that fungal endophytes control food-web structure by dis-rupting the transfer of energy from plants to upper trophic levels. Asendophytes live concealed within the host plant tissue, their impacton natural communities and biodiversity may easily be overlooked.These inconspicuous mutualistic associations can, however, exert aregulatory force on food-web dynamics that is qualitatively similarto that of primary productivity or nutrient supply5±7,17 in manyecosystems. M

MethodsExperimental design

Forty plots of Italian ryegrass were established on 2 July 1999, using 50 ´ 50 ´ 15 cm woodencontainers laid out in an experimental garden at the College of Agronomy, University ofBuenos Aires. Plots were arranged in a 5 ´ 8 grid, with 30-cm-wide corridors kept short bymowing. The containers were ®lled with soil extracted from a grassland site in the InlandPampa, eastern Argentina. The seeds for the experiment were collected in the same area fromold-®eld communities dominated by L. multi¯orum populations with high levels (85±95%)of endophyte infection. Before the experiment, a batch of seeds was treated with a systemicfungacide (triadimenol: 5 mg per g seed) to obtain endophyte-free plants. Plots wererandomly assigned to one of two treatments, with (+E) or without (-E) endophyteinfection, and were sown with 2 g of L. multi¯orum seed (,700 seeds per plot). These grassmonocultures were maintained by frequent weeding. Levels of endophyte infection werecon®rmed at the end of the experiment by microscopic examination (aniline blue±lacticacid stain) of 30 seeds taken from ten plants per plot. The fungal mycelium was found to beassociated to the aleurone layer of the seed28. 95% of seeds collected from +E plots wereinfected, whereas seeds collected from -E plots contained no fungal endophytes.

Insect and plant sampling

Six months after sowing, at peak above-ground grass biomass, the natural occurrence ofherbivorous insects was recorded in 10 plots of each treatment. Three times between 4 and30 November 1999, the density of aphids and their hymenopteran parasitoids wasestimated by counting the number of living aphids and mummies on 40 grass stemsselected at random within each plot. The assessment of the aphid±parasite interaction webwas based upon counts made on 4 November when aphid and parasitoid abundances wereat their observed maximum. Thus, we report effects on insect densities that most probablyintegrated population responses across several aphid generations, followed by theaggregative response of parasitoids to differences in prey availability. Insect densities werecompared between +E and -E grass plots using t-tests on log-transformed data.Differences in the total rate of parasitism were evaluated through a randomization test,using Resampling Stats version 4.2 (Julian Simon, Resampling Stats, Inc.). The contri-bution of primary and secondary parasitoids to overall parasitoid emergence frommummi®ed aphids was examined with the x2 statistic.

To produce an enhanced description of the parasitoid community, the sample ofmummi®ed aphids obtained during the censuses was supplemented with an extensivecollection of mummies from across all the plots. Parasitoids were reared individually ingelatine capsules under ambient temperature and identi®ed and measured (body size:length in mm) upon emergence. Endophyte-driven effects on parasitoid body size weretested using factorial ANCOVA, with endophyte treatment and parasitoid sex as the maineffects, and mummy size as a covariate.

Plant density was estimated in late November by counting the number of grass stemswithin a 6 ´ 30 cm strip-quadrat in each plot. This measure was used to adjust insectdensities to a common, plot-area basis18. To determine above-ground plant biomass (g dryweight, after 48 h at 72 8C), a grass sample cut to soil level within a 10-cm-diametercylinder was extracted from ®ve replicate plots per treatment at the end of the experiment(December). Plant quality was assessed by measuring total leaf nitrogen concentration in acomposite sample of ®ve plants per plot, using a semi-micro Kjeldahl acid digestion.

Interaction webs

The webs were constructed using mean densities of living and mummi®ed aphids to

Table 1 Effect of fungal endophyte infection on insect host±parasitoidinteraction webs naturally established on Lolium multi¯orum grass mono-cultures

Endophyte

Food-web attributes Present Absent.............................................................................................................................................................................

Total species richness 8 9Linkage density 1.0 1.2Connectance 0.67 (8/12) 0.79 (11/14)Shared parasitism 0.33 (2/6) 0.57 (4/7)Interaction strength

Median 1.18 2.83 P , 0:017Evenness 0.84 0.62 P , 0:025

.............................................................................................................................................................................

Linkage density re¯ects the number of observed aphid±parasitoid links divided by the total numberof species in the web. Connectance refers to the ratio between observed and maximum possiblenumber of trophic links in a web. Shared parasitism is the proportion of parasitoid species attackingboth aphid hosts. For descriptors of interaction strength, P values were derived from rank tests.

© 2001 Macmillan Magazines Ltd

letters to nature

NATURE | VOL 409 | 4 JANUARY 2001 | www.nature.com 81

obtain a quantitative description of herbivore and parasitoid trophic levels at the scale ofplots of 0.25 m2. The strength of each pairwise parasitic interaction within a treatment wasestimated by determining the proportional contribution of each parasitoid species to thetotal number of parasites emerged from a given aphid species in laboratory rearings18.These proportional estimates of parasitoid abundance per host species and treatment weretranslated into absolute measures of interaction strength after multiplying by the meandensity of parasitized aphids (that is, mummy density per host species) recorded in the®eld censuses. We determined levels of linkage density, connectance, and shared parasitismfor the food web representing each separate treatment. The median strength of all pairwiseaphid±parasite interactions was computed for each food web and compared by theMann±Whitney two-sample test. The distribution of parasitic interaction strengthswithin each web was summarized using a standard evenness index29 and was evaluatedstatistically after jackknife resampling of the original data30.

Received 21 September; accepted 31 October 2000.

1. Gehring, C. A. & Whitham, T. G. Interactions between aboveground herbivores and mycorrhyzal

mutualists of plants. Trends Ecol. Evol. 9, 251±255 (1994).

2. van der Heijden, M. G. A. et al. Mycorrhizal fungal diversity determines plant biodiversity, ecosystem

variability and productivity. Nature 396, 69±72 (1998).

3. Wootton, J. T. & Power, M. E. Productivity, consumers, and the structure of a river food chain. Proc.

Natl Acad. Sci. USA 90, 1384±1387 (1993).

4. Kaunzinger, C. M. K. & Morin, P. J. Productivity controls food-chain properties in microbial

communities. Nature 395, 495±497 (1998).

5. Mikola, J. & SetaÈlaÈ, H. Productivity and trophic-level biomasses in a microbial-based solid food web.

Oikos 82, 158±168 (1998).

6. Hulot, F. D., Lacroix, G., Lescher-MoutoueÂ, F. & Loreau, M. Functional diversity governs ecosystem

response to nutrient enrichment. Nature 405, 340±344 (2000).

7. Leibold, M. A., Chase, J. M., Schurin, J. B. & Downing, A. L. Species turnover and the regulation of

trophic structure. Annu. Rev. Ecol. Syst. 28, 467±494 (1997).

8. Abrams, P. A. Effects of increased productivity on the abundance of trophic levels. Am. Nat. 141, 351±

371 (1993).

9. Clay, K. Fungal endophytes of grasses. Annu. Rev. Ecol. Syst. 21, 275±297 (1990).

10. Dahlman, D. L., Eichenseer, H. & Siegel, M. R. in Microbial Mediation of Plant-Herbivore Interactions

(eds Barbosa, P., Krischik, V. A. & Jones, C. G.) 227±252 (John Wiley, New York, 1991).

11. Breen, J. P. Acremonium endophyte interactions with enhanced plant resistance to insects. Annu. Rev.

Entomol. 39, 401±423 (1994).

12. Oksanen, L., Fretwell, S. D., Arruda, J. & NiemelaÈ, P. Exploitation ecosystems in gradients of primary

productivity. Am. Nat. 118, 240±261 (1981).

13. Hairston, N. G., Smith, F. E. & Slobotkin, L. B. Community structure, population control, and

competition. Am. Nat. 94, 421±425 (1960).

14. Price, P. W. et al. Interaction among three trophic levels: in¯uence of plants on interactions between

insect herbivores and natural enemies. Annu. Rev. Ecol. Syst. 11, 41±65 (1980).

15. Hunter, M. D. & Price, P. W. Playing chutes and ladders: heterogeneity and the relative roles of

bottom-up and top-down forces in natural communities. Ecology 73, 724±732 (1992).

16. Osenberg, C. W. & Mittelbach, G. G. in Food Webs: Integration of Patterns and Dynamics (eds Polis, G.

& Winemiller, K. O.) 134±148 (Chapman & Hall, New York, 1996).

17. DeAngelis, D. L. Dynamics of Nutrient Cycling and Food Webs (Chapman & Hall, London, 1992).

18. MuÈller, C. B., Adriaanse, I. C. T., Belshaw, R. & Godfray, H. C. J. The structure of an aphid-parasitoid

community. J. Anim. Ecol. 68, 346±370 (1999).

19. Holt, R. D. Predation, apparent competition and the structure of prey communities. Theor. Pop. Biol.

12, 197±229 (1977).

20. Holt, R. D. & Lawton, J. H. Apparent competition and enemy-free space in insect host-parasitoid

communities. Am. Nat. 142, 623±645 (1993).

21. Chaneton, E. J. & Bonsall, M. B. Enemy-mediated apparent competition: empirical patterns and the

evidence. Oikos 88, 380±394 (2000).

22. McCann, K., Hastings, A. & Huxel, G. Weak trophic interactions and the balance of nature. Nature

395, 794±798 (1998).

23. Barker, G. M. & Addison, P. J. In¯uence of Clavicipitaceous endophyte infection in ryegrass on

development of the parasitoid Microctonus hyperodae Loan (Hymenoptera: Braconidae) in Listronotus

bonariensis (Kuschel) (Coleoptera: Curculionidae). Biol. Contr. 7, 281±287 (1996).

24. Bultman, T. L., Borowicz, K. L., Schneble, R. M., Coudron, T. A. & Bush, L. P. Effect of a fungal

endophyte on the growth and survival of two Euplectrus parasitoids. Oikos 78, 170±176 (1997).

25. Dobson, A. P. & Crawley, M. J. Pathogens and the structure of plant communities. Trends Ecol. Evol. 9,

393±398 (1994).

26. Grime, J. P., MacKey, J. M. L., Hillier, S. H. & Read, D. J. Floristic diversity in a model system using

experimental microcosms. Nature 328, 420±422 (1987).

27. Clay, K. & Holah, J. Fungal endophyte symbiosis and plant diversity in successional ®elds. Science 285,

1742±1744 (1999).

28. Latch, G. C. M., Christensen, M. J. & Hickson, R. E. Endophytes of annual and hybrid ryegrasses. New

Zeal. J. Agr. Res. 31, 57±63 (1988).

29. Alatalo, R. V. Problems in the measurement of evenness in ecology. Oikos 37, 199±204 (1981).

30. Magurran, A. E. Ecological Diversity and its Measurement (Princeton Univ. Press, New Jersey, 1988).

Acknowledgements

We thank A. Austin, M. Bonsall, A. Bourke, C. Godfray, D. Golombek, T. H. Jones,N. MazõÂa, R. Pettifor, S. Power, P. Roset, S. Semple and M. Vila-Aiub for comments on themanuscript; E. Demartin, P. Gundel and M. Rabadan for ®eld assistance; R. Belshaw andF. van Veen for helping with parasitoid identi®cation; and C. Godfray for constructing theweb diagrams. This study was funded by grants from the Agencia Nacional de Promocio nCienti®ca y Tecnolo gica of Argentina and Fundacio n Antorchas.

Correspondence and requests for materials should be addressed to M.O.(e-mail: omacini@ifeva.edu.ar) or C.B.M. (e-mail: christine.mueller@ioz.ac.uk).

.................................................................Evolution of the bilaterianlarval foregutDetlev Arendt*, Ulrich Technau² & Joachim Wittbrodt*

* European Molecular Biology Laboratory, Developmental Biology Programme,Meyerhofstrasse 1, 69012 Heidelberg, Germany² Molecular cell biology, Zoological Institute, Darmstadt University of Technology,

Schnittspahnstrasse 10, 64287 Darmstadt, Germany..............................................................................................................................................

Bilateria are subdivided into Protostomia and Deuterostomia1,2.Indirect development through primary, ciliary larvae occurs inboth of these branches; however, the closing blastopore developsinto mouth and anus in Protostomia and into anus only inDeuterostomia. Because of this important difference in larvalgut ontogeny, the tube-shaped guts in protostome and deuteros-tome primary larvae are thought to have evolved independently2,3.To test this hypothesis, we have analysed the expression ofbrachyury, otx and goosecoid homologues in the polychaete

a bProtostomia Deuterostomia

an

bl

an

bl

sto

at

gut

m

a

m

a

sto

at

gut

Figure 1 Different ontogeny but similar body plans of Protostomia and Deuterostomia

primary larvae as shown by similar expression of brachyury in the ventral developing

foregut and otx in ciliary bands bordering the mouth region. Late gastrula embryos (top)

develop into pelagic, ciliary primary larvae (bottom). a, Polychaeta (Protostomia). The

lateral blastopore lips fuse along the later ventral midline. The blastopore gives rise to

mouth and anus at opposite ends. In the trochophora larva, brachyury (blue) is expressed

in the ventral portion of the stomodaeum and in the proctodaeum, and otx (red) is

expressed in two bands of cells along the preoral prototroch and the postoral metatroch.

b, Enteropneusta (Deuterostomia). The tip of the gastrulation cavity touches the lateral

body wall on the future ventral side, where the mouth later breaks through. The blastopore

gives rise to the anus only. In the early tornaria larva, brachyury (blue) is expressed in the

ventral portion of the stomodaeum and in the proctodaeum7, and otx (red) is expressed in

two upper bands parallel to the preoral ciliated band and in two lower bands parallel to the

postoral ciliated band9. A similar otx pattern is observed in the 30-h auricularia of the sea

cucumber8. a, anus; an, animal pole; at, apical tuft; bl, blastopore; m, mouth; sto,

stomodaeum. (Ciliary larvae adapted from ref. 2).

© 2001 Macmillan Magazines Ltd