Portal de Periódicos da CAPES

The relative importance of resources and natural enemies in determining herbivore abundance: thistles, tephritids and parasitoids

2008; Wiley; Volume: 77; Issue: 5 Linguagem: Inglês

10.1111/j.1365-2656.2008.01406.x

1365-2656

Matthew P Walker, Susan E. Hartley, T. Hefin Jones,

Insect behavior and control techniques

Journal of Animal EcologyVolume 77, Issue 5 p. 1063-1071 Free Access The relative importance of resources and natural enemies in determining herbivore abundance: thistles, tephritids and parasitoids Matthew Walker, Corresponding Author Matthew Walker Present address: UKPopNet, Environment Department, University of York, York Y010 5DD, UK. *Correspondence author. E-mail: mw40@york.ac.ukSearch for more papers by this authorSusan E. Hartley, Susan E. Hartley Present address: School of Life Sciences, University of Sussex, Falmer, Brighton BN1 9QG, UK.Search for more papers by this authorT. Hefin Jones, T. Hefin Jones Present address: Cardiff School of Biosciences, Cardiff University, Cardiff CF10 3TL, UK.Search for more papers by this author Matthew Walker, Corresponding Author Matthew Walker Present address: UKPopNet, Environment Department, University of York, York Y010 5DD, UK. *Correspondence author. E-mail: mw40@york.ac.ukSearch for more papers by this authorSusan E. Hartley, Susan E. Hartley Present address: School of Life Sciences, University of Sussex, Falmer, Brighton BN1 9QG, UK.Search for more papers by this authorT. Hefin Jones, T. Hefin Jones Present address: Cardiff School of Biosciences, Cardiff University, Cardiff CF10 3TL, UK.Search for more papers by this author First published: 13 August 2008 https://doi.org/10.1111/j.1365-2656.2008.01406.xCitations: 26 AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinked InRedditWechat Summary 1 The relative importance of host-plant resources and natural enemies in influencing the abundance of insect herbivores was investigated in potted plant and natural population experiments, using tephritid (Diptera: Tephritidae) flies, their host plant, creeping thistle Cirsium arvense, and their Hymenoptera parasitoids. 2 Experimental manipulation of host-plant quality (i.e. levels of host-plant nutrients) and resource availability (i.e. the number of buds) increased tephritid abundance. There was no evidence that the seed-feeding tephritid fly Xyphosia miliaria preferentially oviposited on fertilized C. arvense. 3 At low thistle densities, X. miliaria showed a constant rate of resource exploitation. At higher thistle densities, a threshold was detected, above which additional buds were not attacked. 4 Parasitism attack was variable across host (tephritid) densities but levels of parasitism were consistently higher on the fertilized thistles. 5 Experimental manipulation of host-plant quality and resource availability (quantity) not only directly affects the tephritid population but also, indirectly, leads to high rates of parasitism. Both chemical and physical characteristics of host plants affect the performance of natural enemies. 6 Both top-down and bottom-up forces act to influence tephritid abundance, with bottom-up influences appearing to be the most important. Introduction The relative importance of resources (bottom-up forces) and natural enemies (top-down forces) in determining the abundance of insect herbivores remains a major focus of debate in population ecology (Hunter & Price 1992; Walker & Jones 2001; Richmond et al. 2004; Singer & Stireman 2005). Despite repeated calls for a synthesis (e.g. Quinn & Dunham 1983; Karban 1989; Polis 1994; Jones, Hassell & Godfray 1997; Bailey & Whitham 2006), it is only relatively recently that quantitative experimental studies have attempted to assess the relative roles of top-down and bottom-up forces in determining insect population densities (e.g. Dyer & Letourneau 1999; Forkner & Hunter 2000; Finke & Denno 2004; Aquilino, Cardinale & Ives 2005). Many ecologists now consider that both factors are important influences (e.g. Denno et al. 2002; Richmond et al. 2004; Aquilino et al. 2005) but, there remains a lack of data to arrive at a consensus about the ways in which plants and natural enemies interact, or what factors increase or reduce the relative strength of top-down and bottom-up forces within particular interactions. Although many studies have demonstrated the importance of natural enemies to herbivore populations (Hassell 2000 and references therein), in most systems both the quantity and nutrient quality of host-plant resources are also likely to be important bottom-up factors (Hartley & Jones 1997). Both latter factors may influence herbivore populations by altering feeding preference and performance (Waring & Cobb 1992). For example, the plant vigour hypothesis (see Price 1991) suggests that herbivores show a feeding preference for vigorous plants, which provide improved food quality and hence improved growth rates. Plant quality and quantity can also indirectly affect top-down factors by altering the effectiveness of herbivore natural enemies. Several mechanisms have been proposed for these tritrophic interactions. For example, it has been suggested that poor food quality leads to increased herbivore movement as they search for acceptable leaves and hence they suffer increased exposure to natural enemies (Bergelson & Lawton 1988). It has also been argued that plant-mediated reduction in larval growth increases susceptibility to mortality from natural enemies (the slow-growth, high-mortality hypothesis: Clancy & Price 1987; Williams 1999). An increasing number of experimental studies have aimed to demonstrate the mechanisms involved in plant-based impacts on the effectiveness of natural enemies. Some studies have directly measured the impacts of the chemical basis of host-plant quality (Cornelissen & Stiling 2006) while others have used surrogates of quality such as plant architecture (Forkner & Hunter 2000; Denno et al. 2002). There is, however, a lack of manipulative studies that simultaneously measure the impacts of plant quality and quantity on the natural enemies of insect herbivores. The close association between tephritid flies (Diptera: Tephritidae) and their host plants, in addition to being well documented (Zwölfer 1988; Straw 1989; Abrahamson & Weis 1997; Headrick & Goeden 1998; Vanbergen et al. 2006), provides an ideal model system within which to investigate the relative role of both bottom-up and top-down factors in determining insect abundance. The quality and quantity of the host resource can easily be altered and measured, and the natural enemies of tephritids are easy to monitor and identify. In this study we use the creeping thistle Cirsium arvense (L.) Scop. (Compositae), its seed-feeding tephritid flies (both galling and non-galling) and their Hymenoptera parasitoids to: (1) determine the role played by both plant resources and natural enemies in influencing the abundance of tephritid populations, and (2) investigate the mechanisms by which resource availability and host-plant quality, both architectural and chemical, affect populations of thistle-attacking tephritids and their parasitoids. In doing this we aim to examine, simultaneously and within one model system, in both potted plant and natural population field experiments, the interacting effects of plant resources and natural enemies in influencing herbivore population abundance. Materials and methods biology Three species of tephritid, Xyphosia miliaria (Schrank), Terellia (= Orellia) ruficauda (F.) and Urophora stylata (F.), develop in the capitula (buds) of C. arvense at the study site at Silwood Park. Flowering occurs between June and August, and young flowerheads are attacked by tephritids during late June to early August. Tephritid larvae feed on thistle achenes (seeds), pappi and/or receptacles; U. stylata forms a woody many-chambered gall. All three species are parasitized by two species of hymenoptera parasitoid: Torymus chloromerus Boheman (Hymenoptera: Torymidae) and Pteromalus (= Habrocytus) elevatus (Walker) (Hymenoptera: Pteromalidae). These two species represent > 98% of the total parasitism recorded. Tephritid larvae overwinter inside buds with both adult tephritids and parasitoids emerging the following June. We restrict our studies in experiments 2 and 3 (see below) to X. miliaria. This was, by far, the most common species attacking C. arvense at Silwood Park. experimental design Three experiments were conducted. In Experiment 1 (Summer 1998) fertilized and unfertilized potted thistle plants were exposed to field populations of tephritids and parasitoids. This experiment aimed to test the impact of increasing host-plant resources on tephritid and parasitoid oviposition behaviour. Experiments 2 and 3 were complementary field experiments carried out in summer 1999. Experiment 2 used potted plants placed in the field to investigate the response of X. miliaria behaviour and performance to variations in the quality and quantity of thistle resources. Experiment 3 focused on natural thistle populations and examined the effects of the quantity of plant resources on higher trophic levels over a naturally occurring range of host-plant densities. Experiment 1 C. arvense plants were grown from seed (John Chambers Seeds, Northants, UK) in trays containing John Innes no. 1 Compost. Germination occurred in heated frames (c. 25 °C) within 3–4 weeks of sowing (27 February 1998). After 8 weeks seedlings were transplanted individually into 520 7·5-litre container pots containing a mix of two-thirds All Purpose Compost (Sinclair Composts, Lincoln, UK) and one-third sharp sand on a 3–4-cm basal layer of fine gravel. Once transplanted the potted plants were randomly assigned to either control or fertilized treatment regimes (260 plants each). Fertilized plants received 12 g nitrogen m−2 (Nitrapill, Kemira Fertilisers, London, UK) dissolved in excess water. This dose produces significant differences in the chemical composition of tissues of potted thistles (Williams, Jones & Hartley 2001). Fertilizer was applied on 1 and 8 May 1998, and subsequently at monthly intervals. Control plants received an equal volume of water only. The date of the first fertilizer application (1 May 1998) was designated Week 1 and all experimental time was measured from this point. Tephritid and parasitoid population densities vary over both time and space (Solbreck & Sillén-Tullberg 1986; Römstock-Völkl 1990; Redfern, Jones & Hassell 1992). To maximize the likelihood of attack of the experimental arrays (see below) by natural tephritid and parasitoid populations, five field sites at Silwood Park with established tephritid populations (Jones, Godfray & Hassell 1996; Williams et al. 2001) were chosen. All sites were separated by at least 20 m. A total of 480 thistles (240 fertilized, 240 control) were moved to the field sites on 20 May 1998 (week 3), with plants being arranged in six arrays of eight pairs of one control and one fertilized plant per field site. Within each array the plants were arranged randomly in pairs in a 4 m × 4 m matrix. The remaining 40 plants (20 control, 20 fertilized) were kept separate but under identical environmental conditions and were used for monitoring plant nutrient changes. Daily field observations of the presence or absence of tephritids on the experimental plants were carried out (20 May–9 September 1998) to determine tephritid activity phenology. Weekly measurements were also taken of the number of open flowerheads per plant. Thistles were harvested in late September 1998. Buds were collected, pooled for each plant, placed in insect emergence chambers (Jones et al. 1996) and kept in an outside insectary until emergence of tephritids and parasitoids occurred in late spring 1999. The number of each insect species emerging was recorded; dissection of thistle buds showed that failure to emerge was negligible (personal observation; Jones et al. 1996). At harvest, plant biomass was partitioned into root, stem and leaf before being oven-dried at 80 °C for 72 h and weighed. At Weeks 13 (maximum tephritid oviposition activity) and 17 (parasitoid oviposition period, inferred from Redfern et al. 1992), 20 randomly selected plants (10 fertilized, 10 control) were taken from the 40 thistles and placed aside for nutrient sampling. For nutrient analysis, dried plant material (25 youngest leaves) were milled and analysed for percentage nitrogen and percentage phosphorus content using a continuous flow calorimetric autoanalyser (Segmented Flow Autoanalyser, Burkard Scientific, Uxbridge, UK) following wet acid digestion (Allen 1989). The nitrogen content was measured as ammonium by a modified Bertholet reaction (Hinds & Lowe 1980; Rowland 1983) and phosphorus was measured as phosphate by the ‘Molybdenum blue’ method (Allen 1989). Phenolic content, a measure of plant chemical defence, was determined using the Folin–Ciocalteu method using tannic acid as a standard, following extraction of dried leaf material in 50% aqueous methanol (Waterman & Mole 1994). Phenolic content was expressed as percentage dry weight (calculated as mg tannic acid equivalents per mg leaf mass × 100%). Experiment 2 As in Experiment 1, C. arvense plants were potted (13 May 1999) into 240 7·5-litre container pots. Thistle plants were randomly assigned to either control or fertilized treatment regimes. Plants were fertilized at concentrations used in experiment 1. Fertilizer was applied on 1, 7 and 15 June 1999, and subsequently at fortnightly intervals. Control plants received an equal volume of water only. Thistles (120 fertilized, 120 control) were positioned at the Gunnes's Hill (Silwood Park) field site on 14 May 1999. Plants were arranged in 15 arrays of eight pairs of one control and one fertilized plant. The pairs were arranged randomly in a 4 m × 4 m matrix. All plants were examined on a weekly basis and the presence of X. miliaria (to allow determination of the timing of tephritid emergence from overwintering larvae) and floral phenology (to determine the timing of first flowering) recorded. Thistles were harvested on 29 July 1999, 4 weeks after the first flowers opened and 6 weeks after the first X. miliaria were observed at the field site. At harvest, open and closed buds were counted and dissected for tephritid larvae. Four plants per array (two fertilized, two control) were partitioned into stem and leaf components for biomass sampling. Samples were oven-dried and weighed as for Experiment 1. Experiment 3 A 20 m × 30 m rabbit-proof fence was erected at Rookery Slope (approximately 400 m away from the Gunnes's Hill site). This area enclosed approximately 6000 creeping thistles. Thistles from 50 randomly chosen 1 m × 1 m quadrats within the fenced area were harvested on 26 July 1999. The number of thistles, and the numbers of open and closed buds per quadrat, were counted before the buds were dissected for X. miliaria larvae. Biomass was partitioned into stem and leaf components, oven-dried as above. statistical analysis In Experiment 1 rabbits grazed 27 experimental plants. As the plant architecture and biomass variables of these plants were affected by an uncontrolled factor, they were removed from the data set. Statistical analysis was performed using GLIM 4 (Royal Statistical Society, London, UK). For chemical data normal error distributions were used; for counts of flowers, emergences and X. miliaria larvae, a Poisson error distribution was used; for percentage parasitism and data of the proportion of buds attacked by X. miliaria a binomial error distribution was applied; and for the biomass data (increasing variance with n) a gamma distribution. Overdispersion in residual plots was determined following Crawley (1993). Results experiment 1 The flight period for tephritids during summer 1998 occurred between Weeks 13 and 16 (mid July to early August), and that of the two main parasitoid species, which oviposit on third instar tephritid larvae, from Week 17 onwards (also see Redfern et al. 1992; Jones et al. 1996). Total plant biomass was higher in fertilized thistles than in control plants. This was primarily due to significant increases in stem biomass; root and leaf biomass were not affected by fertilization. Fertilization also had significant temporal and quantitative effects on thistle inflorescence number. From Week 14 onwards fertilized thistle plants had significantly more open flowerheads at each sampling point while peak flowering occurred on average, 1–2 weeks earlier in control thistles (Fig. 1). Figure 1Open in figure viewerPowerPoint Peak flower opening and number (mean ± SE) of open flowerheads per thistle on control () and fertilized () thistles in Experiment 1. NS = not significant, *P= 0.05, **P = 0.01. Week 1 = week commencing 4 May 1998. At Week 13, by the time tephritid flies started ovipositing, fertilized thistles had a significantly higher nitrogen and phosphorus concentration in their leaves than the control plants. There were no fertilizer treatment effects on percentage dry weight phenolics. By Week 17, at which time parasitoid oviposition had begun, fertilized thistles continued to have higher nitrogen concentrations than control plants, but there was no longer a significant treatment effect on phosphorus. At Week 17, control (unfertilized) thistles had higher phenolic concentrations than fertilized plants. In summary, fertilization resulted in qualitative and temporal changes of thistle quality both in terms of physical (biomass and number of open flowerheads) and chemical (plant nutrient and secondary compound content) parameters. All three species of tephritid (X. miliaria, T. ruficauda and U. stylata) and the two parasitoid species (T. chloromerus and P. elevatus) emerged from the overwintering emergence chambers in July 1999 (Fig. 2). Significantly higher numbers of X. miliaria and U. stylata emerged from fertilized thistles compared with control thistles; T. ruficauda exhibited no significant differences. Higher numbers of T. chloromerus and P. elevatus also emerged from fertilized thistles (Fig. 2). Approximately double the number of tephritids (both unparasitized and parasitized) (anovaχ2 = 19·180; P = 0·001) and four times the number of parasitoids (anovaχ2 = 15·980; P = 0·001) were recorded on fertilized thistles compared with control thistles. Figure 2Open in figure viewerPowerPoint Emergence data (mean number of insects per thistle ± SE) of three species of tephritid fly: Xyphosia miliaria, Terellia ruficauda, and Urophora stylata, and two species of hymenopteran parasitoid: Torymus chloromerus and Pteromalus elevatus for control () and fertilized () thistles in Experiment 1. NS = not significant, **P = 0.01, ***P = 0.001. Analysis of covariance (ancova) found an effect of host (tephritid) density (χ2 = 18·720; P = 0·001) on the number of parasitoids emerging from experimental thistles, regardless of host-plant treatment (χ2 = 0·874; NS). The interaction between host-plant treatment and the density of tephritids hosts was not statistically significant (χ2 = 0·877; NS), although the increased numbers available on fertilized thistles led to different parasitoid responses. On control thistles, the numbers of parasitoid emergences were independent of host density (anovaχ2 = 0·809; NS) (Fig. 3a), while on fertilized thistles greater numbers of hosts resulted in increased numbers of parasitoids (anovaχ2 = 18·010; P = 0·001) (Fig. 3b). Neither host-plant treatment nor host (tephritid) density had any effect on the proportion of hosts (tephritids) parasitized (anovaχ2 = 1·680; NS and χ2 = 2·337; NS, respectively). The interaction between host-plant treatment and the density of tephritid hosts was not statistically significant (ancovaχ2 = 2·042; NS) and a constant number of tephritids were parasitized regardless of host density on both control thistles (anovaχ2 = 0·326; NS) and fertilized thistles (anovaχ2 = 3·714; NS) (Fig. 3c,d). Figure 3Open in figure viewerPowerPoint The effect of host density on the number of parasitoids (m−2) for (a) control and (b) fertilized thistles, and on the proportion of hosts attacked by parasitoids at different host densities (m−2) for (c) control and (d) fertilized thistles in Experiment 1. Control thistles =, fertilized thistles =, broken trendline. The line in (b) is fitted using GLIM and is of the form y= ln(mx+c) where m= 0.795 and c= 1.669. Numbers refer to multiple data points. experiment 2 Fertilization of the potted plants led to significant increases in leaf biomass (F = 17·58, P = 0·01), total biomass (F = 10·25, P = 0·01) and leaf nitrogen (F = 54·42, P = 0·01) (Fig. 4a,b), but had no effect on the mean numbers of buds per thistle (Fig. 4c). Fertilization had no effect on the numbers of X. miliaria larvae (m−2). The number of open buds (i.e. the cumulative number of buds that were available for tephritid oviposition available per plant), however, did have a highly significant effect on the numbers of X. miliaria larvae (χ2 = 58·20; P = 0·001) in both treatments (Fig. 5a). The proportion of open buds attacked by X. miliaria remained independent of bud density and approximately 14% of open buds were attacked over a range of bud densities (0–64 buds m−2) (Fig. 5b). Figure 4Open in figure viewerPowerPoint The effect of host-plant fertilization (in Experiment 2) on the mean (± SE): (a) dry weight biomass (g) of leaf, stem and total vegetation; (b) % of nitrogen, % phosphorus and total phenolic content; and (c) the number of open and closed buds and total bud number in control () and fertilized () thistles. Figure 5Open in figure viewerPowerPoint The effect of host-plant fertilization (in Experiment 2) on: (a) the number of Xyphosia miliaria larvae at different densities of open buds for control and fertilized thistles (regression control y= 0.0839x; fertilized y = 0.0904x); and (b) the proportion of open buds attacked by X. miliaria at different densities of open buds for control (, dashed line) and fertilized (, solid line) thistles. experiment 3 In the natural population experiment, both thistle density (m−2) and leaf biomass (m−2) and were closely correlated with the number of opened buds (m−2) (F = 32·11; P = 0·01 and χ2 = 575·1; P = 0·001, respectively; Fig. 6a,b). As with Experiment 2, the number of open buds had a highly significant effect on the number of X. miliaria larvae found per thistle (χ2 = 70·75; P = 0·001) (Fig. 6c). In contrast to the potted plants in experiment 2, the proportion of open buds attacked by X. miliaria did not remain constant. Proportionally fewer buds were attacked by X. miliaria at high bud densities (χ2 = 24·68; P = 0·001) (Fig. 6d). Figure 6Open in figure viewerPowerPoint (a) The effect of thistle density (m−2) (at natural thistle densities) on the number of open buds (m−2) (regression y= 1.9387x); and (b) the effect of leaf biomass (m−2) on the number of open buds (m−2) in Experiment 3 (regression y= 0.5479x). The effect of the density of open buds (m−2) (at natural thistle densities) on (c) the number of buds attacked by X. miliaria (regression y= 0.0921x) and (d) the proportion of open buds attacked by X. miliaria in Experiment 3 (regression y=−0.0024x+ 0.2568). Discussion direct bottom-up influences Relatively few tritrophic studies have directly measured the chemical basis of host-plant quality, in terms of both nutrients and secondary compounds, at the same time as host-plant quantity. The temporal availability of resources (quantity) determines the window of attack for many phytophagous insects (Feeny 1970; Watt 1987; Weis, Walton & Crego 1988; but see Kerslake & Hartley 1997), with oviposition usually timed to coincide with the maximum availability of resources. Female tephritids show a marked oviposition preference for a specific development stage of their host plant that maximizes subsequent larval performance. For example, Rivero-Lynch & Jones (1993) report that for T. ruficauda ovipositing on C. arvense, thistles are only suitable for oviposition for a single day prior to flowering. This and other studies (e.g. Anderson et al. 1989; Straw 1991; Fondriest & Price 1996) illustrate the importance of the synchronization of tephritid life cycles with the availability of host-plant resources (in this case, open flowerheads). Host-plant chemistry (quality) may provide an alternative mechanism by which bottom-up effects influence populations of tephritids. Nitrogen is commonly a limiting factor for herbivorous insects (White 1984; Hartley & Gardner 1995) and hence foliar nitrogen content is an important determinant of the performance of many phytophagous insects (Feeny 1976; McNeill & Southwood 1978; but see Faeth, Mopper & Simberloff 1981; Waring & Cobb 1992). In thistles, foliar nitrogen content is closely correlated with bud nitrogen content (Williams et al. 2001). The window for tephritid attack is timed to coincide not only with the availability of resources (quantity) but also with periods during which nutrient availability (quality) is at its maximum. The higher nutrient levels recorded in fertilized thistles from week 13 onwards may have resulted in a higher number of tephritid attacks being recorded on these plants. In Experiment 1, almost twice as many tephritids (both parasitized and unparasitized) were found on fertilized thistles compared to control thistles. Fertilization of experimental thistles increased both the number of suitable buds and the duration over which these buds were apparent (sensu Feeny 1976) to tephritids. This suggests a critical role for the phenology of resource availability in determining tephritid abundance. Host-plant fertilization, however, also led to changes in the levels of plant nutrients particularly in nitrogen content, which was almost three times higher during the tephritid oviposition period. Four weeks later during the parasitoid oviposition period, fertilized plants also had lower levels of phenolic compounds, but this was not statistically significant overall. Both resource availability (quantity) and host-plant chemistry (quality) are likely to play a part in the observed increase in tephritid abundance. In the more detailed X. miliaria investigations of Experiments 2 and 3, although fertilizer induced changes in leaf nitrogen and biomass, this had no effect on the number of larvae per thistle or on the proportion of open buds attacked by X. miliaria. Indeed, X. miliaria appeared to attack a constant proportion of open buds in the potted plant experiment, regardless of host-plant quality. In the natural population, the tephritids exhibited an inverse density dependent response to resource availability attacking proportionally fewer buds at higher bud densities. This difference in resource exploitation patterns between the two experiments may be explained by the fact that although the mean open bud densities were roughly equivalent in both studies (11·63 ± 2·20 buds m−2 potted plant; 18·10 ± 3·80 buds m−2 natural), the density of thistles was an order of magnitude different (1 thistle m−2 potted plant; 9·96 ± 1·07 thistles m−2 natural). top-down influences and indirect effects of plant-based factors mediated by top-down effects Parasitism attack was variable across tephritid densities. Similar patterns of parasitism have been observed for other tephritid species attacking thistles (e.g. Redfern et al. 1992; Williams et al. 2001; Vanbergen et al. 2006). Variation in parasitism, be it density-dependent or independent of host, can all be important contributors to population regulation if they involve sufficient heterogeneity in the risk of parasitism between individuals in the host population density (Chesson & Murdoch 1986; Hassell & May 1988). In Experiment 1 levels of parasitism were higher on the fertilized thistles; while tephritid numbers only doubled on fertilized plants the mean number of parasitoids quadrupled. The mechanism for this remains uncertain. It is possible that parasitoids may have simply been better synchronized with the delayed tephritid populations on fertilized plants. On the other hand, the increased levels of foliar nitrogen in fertilized thistles during the parasitoid oviposition period may have substantially changed rates of parasitoid attack. In the present study, tephritid populations were simultaneously determined by top-down effects of parasitism and an opposing bottom-up effect of host-plant quantity and quality, which modifies the host–parasitoid interaction. Although top-down effects play a part in determining population structure it is the bottom-up influences that appear to be the most important factor. Experimental manipulation of host-plant quality and resource availability (quantity) not only directly affects the tephritid population but also, indirectly, leads to high rates of parasitism. The exact mechanism by which this indirect interaction is mediated remains uncertain, but there is a growing body of evidence (e.g. Dyer & Stireman 2003; Gratton & Denno 2003; Harmon et al. 2003; Aquilino et al. 2005) that both chemical and physical characteristics of host plants may affect the performance of natural enemies. Much attention has been focused on so-called indirect defences, when volatile signals released by damaged plants attract predators and parasitoids (De Moraes et al. 1998; Kressler & Baldwin 2001), but there are also examples of other aspects of plant quality and plant architecture impacting on the third trophic level (Cloyd & Sadof 2000; Gingras, Dutilleul & Boivin 2002; Fagundes, Neves & Wilson 2