IDEAPlants protect their roots by alertingthe enemies of grubsRob W.H.M. van Tol,1AntonT.C. van der Sommen,2MariI.C. Boff,3,4Jan van Bezooijen,2Maurice W. Sabelis5andPeter H. Smits31Applied Plant Research,Nursery Stock Research Unit,P.O. Box 118, 2770 AC Boskoop,The NetherlandsE-mail:[email protected], Departmentof Nematology, P.O. Box 8123,6700 ES Wageningen,The Netherlands3Plant Research International,P.O. Box 16, 6700 AAWageningen, The Netherlands4Current address: Universidadedo Estado de Santa Catarina ±Centro de CieÃnciasAgroveterinaÂrias, Av. Luis deCamo~es, 2090, 88520-000, Lages,Santa Catarina, Brazil5Institute for Biodiversity andEcosystem Dynamics, P.O. Box94084, 1090 GB Amsterdam,The NetherlandsAbstractPlant roots in the soil are under attack from many soil organisms. Although manyecologists are aware of the presence and importance of natural enemies in the soil thatprotect the plants from herbivores, the existence and nature of tritrophic interactions arepoorly understood. So far, attention has focused on how plants protect their above-ground parts against herbivorous arthropods, either directly or indirectly (i.e. by gettinghelp from the herbivore's enemies). This article is the ®rst in showing that indirect plantdefences also operate underground. We show that the roots of a coniferous plant (Thujaoccidentalis) release chemicals upon attack by weevil larvae (Otiorhynchus sulcatus) and thatthese chemicals thereby attract parasitic nematodes (Heterorhabditis megidis).KeywordsSynomones, semiochemicals, roots, herbivory, entomopathogenic nematodes, vine weevil,Otiorhynchus sulcatus, Heterorhabditis megidis, Thuja occidentalis.Ecology Letters (2001) 4: 292±294Plants protect themselves against insect herbivores, eitherdirectly, e.g. using toxins, or indirectly by promoting theeffectiveness of the herbivore's enemies. Indirect plantdefences may help explain why insect herbivores aregenerally predator-controlled and therefore why ``the Worldis green'' (Sabelis et al. 1999). The evidence for this modeof defence comes from studies on above-ground plantparts (stems and leaves) providing shelter, food or SOS-signals to the herbivore's enemies (Dicke et al. 1990;Turlings et al. 1995; Takabayashi & Dicke 1996; Sabeliset al. 1999). However, roots are a vital, yet vulnerable partof the plant and the role of soil-dwelling natural enemies insuppressing populations of root-feeding insects in naturalecosystems has been shown (Strong et al. 1996, 1999).Thus, one may wonder whether plants actively protect theirroots by attracting these natural enemies. We tested thishypothesis by studying the interaction between coniferroots (Thuja occidentalis), root-feeding vine weevil larvae(Otiorhynchus sulcatus) and entomopathogenic nematodes(Heterorhabditis megidis). Olfactometry revealed that rootsdamaged by weevil larvae release exudates that attractparasitic nematodes. These chemicals therefore function asan SOS, signalling the presence of herbivores to theirnatural enemies.Little is known of the searching behaviour of entomo-pathogenic nematodes in the soil, let alone the role ofchemical communication. Entomopathogenic nematodesare known to be attracted to undamaged plant roots (Bird &Bird 1986; Choo et al. 1989; Lei et al. 1992; Wang & Gaugler1998), to their insect hosts (Schmidt & All 1978; Gaugleret al. 1980; Lei et al. 1992; Lewis et al. 1993) and theassociated cues (e.g. CO2and faeces) (Schmidt & All 1978;Ecology Letters, (2001) 4: 292±294Ó2001 Blackwell Science Ltd/CNRSGaugler et al. 1980; Grewal et al. 1993; Lewis et al. 1993), butonly one study (Wang & Gaugler 1998) suggests that cuesfrom intact and wounded grass roots in¯uence host ®ndingability of the nematodes. The possibility that plant rootsrelease SOS signals upon being eaten by insects has beenlargely ignored, even though there is a wealth of evidencefor such herbivore-induced signals from leaves (Dicke et al.1990; Turlings et al. 1995; Takabayashi & Dicke 1996;Sabelis et al. 1999). We hypothesized that this is alsopossible in the soil environment where plant roots are underinsect attack and would be able to maintain their function bythe aid of the herbivore's natural enemies.To test the olfactory behaviour of the nematodes we useda Y-shaped tube, comprising ®ve short tubes (two per arm ±7 cm long, 3.5 cm diam. ± and one at the base ± 5.5 cmlong, 4.5 cm diam.), each ®lled with silver-sand (moisturecontent 10% w/w) and closed using nylon gauze to isolateinsects and roots, yet allow nematodes to pass through (Boffet al. 2001). The short tube on top of each Y-arm wasdisconnected and incubated for four days with one of thefollowing odour-emitting objects in silver-sand: six weevillarvae, undamaged, mechanically damaged (cutting 10% ofroot tips) or weevil-damaged (with or without six larvae)roots of one intact thuja plant with its above-ground partssealed off from the tube. After incubation, the tube partswere connected again on top of the Y-arms and the Y-tubewas positioned vertically, arms up, by clamping it to a standin a climate room (15 °C; L : D  16 : 8 h). One daythereafter, nematodes (900 in 0.5 mL tap water) werereleased in a pipette inserted up the middle of the base tube,i.e. 16 cm from the top of the Y tube. One day later, theirnumbers in either arm were counted after sand-extractionwith an independently estimated 90±95% ef®ciency (Boffet al. 2001). Each experiment was replicated at least fourtimes with fresh odour sources, fresh sand and new batchesof nematodes. For the choice tests between weevil-damagedand mechanically damaged or undamaged thuja roots, we®rst rinsed the roots with tap water, placed them in waterfor one day and ®nally replanted the thujas in fresh silver-sand in the Y-tubes. Thus, odours from insects and theirfaeces are absent during these tests and the observedresponse of the nematodes must be attributed to odoursreleased from the plant.We found that nematodes were more attracted to weevillarvae alone and to undamaged Thuja roots alone when cleanair was the alternative, and that they were more attracted toweevil-infested roots than to larvae alone or roots alone(Table 1). The nematodes also moved to odours fromweevil-damaged roots, freed of weevil larvae prior to theexperiment, instead of to odours from undamaged ormechanically damaged roots (Table 1).The attractive plant odour probably does not travel by airto the nematode's sensory organs. GC-MS analysis