Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
Roxina Soler, Department of Multitrophic Interactions, Netherlands Institute of Ecology (NIOO-KNAW), PO Box 40 6666 ZG Heteren, the Netherlands. Tel.: +31 26 4791412; Fax: +31 26 4723227; E-mail: firstname.lastname@example.org
1Plants and insects are part of a complex multitrophic environment, in which they closely interact. However, most of the studies have been focused mainly on bi-tritrophic above-ground subsystems, hindering our understanding of the processes that affect multitrophic interactions in a more realistic framework.
2We studied whether root herbivory by the fly Delia radicum can influence the development of the leaf feeder Pieris brassicae, its parasitoid Cotesia glomerata and its hyperparasitoid Lysibia nana, through changes in primary and secondary plant compounds.
3In the presence of root herbivory, the development time of the leaf herbivore and the parasitoid significantly increased, and the adult size of the parasitoid and the hyperparasitoid were significantly reduced. The effects were stronger at low root fly densities than at high densities.
4Higher glucosinolate (sinigrin) levels were recorded in plants exposed to below-ground herbivory, suggesting that the reduced performance of the above-ground insects was via reduced plant quality. Sinigrin contents were highest in plants exposed to low root fly densities, intermediate in plants exposed to high root fly densities and lowest in plants that were not exposed to root herbivory.
5Our results show, for the first time, that root herbivory via changes in plant quality can reduce the performance of an above-ground multitrophic level food chain. This underlines the importance of integrating a broader range of above- and below-ground organisms to facilitate a better understanding of complex multitrophic interactions and interrelationships.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Most plants employ a range of direct defence strategies to protect them from attacking herbivores. These include the production of morphological structures such as trichomes on the leaf surface, which may impede herbivore colonization or movement, and/or the production of toxic compounds (allelochemicals) that can act as feeding deterrents or alter the development and physiology of herbivores, through reduced rates of growth, smaller size and increased mortality. However, the performance of natural enemies of herbivores may be seriously compromised by toxins consumed by herbivores and sequestered in their tissues. A number of studies have indeed reported that allelochemicals contained in host or prey diet negatively affect the development and survival of their parasitoids or predators (reviewed by Hare 2002).
Studies exploring interactions over several trophic levels have traditionally focused on the above-ground community (e.g. Harvey, Van Dam & Gols 2003). However, it is becoming increasingly clear that plants and insects are part of a complex multitrophic above-ground and below-ground environment, in which they closely interact (van der Putten et al. 2001). Therefore, by focusing exclusively on the above-ground subsystem, our understanding of the processes that affect multitrophic interactions in a more realistic framework has been hindered. Over the past two decades attention has been paid to interactions between above-ground and below-ground insect herbivores sharing the same host plant. It is now acknowledged that herbivorous insects can indirectly interact even when they are spatially or temporally separated from other herbivores associated with the same host plant (Gange & Brown 1989; Masters 1995; Masters & Brown 1997; Gange 2001; Masters, Jones & Rogers 2001).
Thus far, most above–below-ground interaction research has been limited to two trophic levels (Moran & Whitham 1990; Masters & Brown 1992; Salt, Fenwick & Whittaker 1996). However, a recent study reported that densities of above-ground parasitoids may be increased by root feeders (Masters et al. 2001). Studies with other soil organisms, such as earthworms and arbuscular mycorrhizas, have also shown that there is the potential for effects of soil organisms on the third trophic level above-ground (Gange, Brown & Aplin 2003; Wurst & Jones 2003). In the field, parasitoids may also be attacked by hyperparasitoids and predators, which may play an important role by influencing community structure (Brodeur 2000). However, we are not aware of any study that has examined the effects of root herbivores on the fourth above-ground trophic level. Changes in plant quality have been shown to affect hyperparasitoid fitness, via herbivore and parasitoid effects (Harvey et al. 2003). Thus, the effects of root herbivores may trickle up to the fourth trophic level above-ground.
Interactions between above- and below-ground insect herbivores and their antagonists may be mediated by primary (nutrients) and secondary (phytotoxins) plant chemistry. Two hypotheses have been proposed to explain above–below-ground herbivore interactions with opposite predictions for the effects of root feeders on above-ground herbivores (reviewed by Bezemer et al. 2002). The ‘stress response hypothesis’ predicts that root herbivores induce a stress response in the host plant that reduces water and nutrient uptake and leads to the translocation and accumulation of soluble nitrogen (N) and carbohydrates in above-ground plant parts, ultimately enhancing the performance of the foliar feeders (Masters, Brown & Gange 1993). The ‘defence induction hypothesis’, argues that root herbivory induces a defence response in the plant that can lead to the accumulation of secondary plant compounds above-ground, which in turn can reduce the performance of the foliar feeders (Bezemer et al. 2003, 2004a; van Dam et al. 2003). Studies of the effects of root feeding insects on above-ground herbivores have provided evidence for both hypotheses and it appears that both mechanisms are not mutually exclusive. However, as far as we know, no study has thus far examined if changes in primary and/or secondary plant compounds induced by below-ground herbivory can also influence the performance of parasitoids, predators and even higher trophic levels.
This study examines the effects of root herbivory over four trophic levels of an above-ground system, as possibly mediated by changes in primary and secondary plant compounds. Using the wild cruciferous plant Brassica nigra L. (Brassicaceae), we studied the effects of root herbivory by Delia radicum L. (Diptera: Anthomyiidae) larvae on an above-ground system consisting of the herbivore Pieris Brassicae L. (Lepidoptera: Pieridae), the parasitoid Cotesia glomerata L. (Hymenoptera: Braconidae) and its secondary hyperparasitoid, Lysibia nana Gravenhorst (Hymenoptera: Ichneumonidae).
We selected a cruciferous plant species because wild crucifers possess potent inducible direct defences via the production of glucosinolates (GLS) and their breakdown products, and thus provide great potential for the study of multitrophic interactions (van Dam et al. 2003). B. nigra is particularly interesting because GLS production is at least three to five times higher than in related cultivated species, such as Brassica oleracea L. (Harvey et al. 2003). Furthermore, the expression of secondary plant compounds in above-ground shoots of B. nigra is known to be influenced by root herbivory (van Dam, Witjes & Svatos 2004). P. brassicae is a specialist chewing herbivore that only feeds on plants containing GLS. However, despite this specialization, a negative correlation between larval feeding and GLS (sinigrin) levels has been observed in several Pieris species (Olsson & Jonasson 1994; Traw & Dawson 2002a). C. glomerata is a fairly specialized gregarious koinobiont endoparasitoid that parasitizes first to third instar larvae of P. brassicae and related species. L. nana is a solitary idiobiont hyperparasitoid that parasitizes pre-pupae and pupae of C. glomerata and related species in the Microgastrinae.
We specifically addressed the following questions: (1) Is the growth, development and survival of above-ground herbivores, parasitoids and hyperparasitoids affected by below-ground herbivory? (2) Are the effects of root herbivory consistent across the above-ground trophic spectrum [e.g. are the effects on development time and body size (positive or negative)] the same in all of the organisms under investigation? (3) If variation in the performance of above-ground insects is observed in response to the presence or absence of root herbivory, can these results possibly be attributed to changes in the primary and secondary plant compounds of above-ground shoots as affected by below-ground herbivory (thus testing the ‘stress-response’ and ‘defence induction’ hypotheses)? Lastly, we discuss the importance of linking above- and below-ground multitrophic interactions with processes occurring in larger ecological communities.
Materials and methods
P. brassicae (leaf herbivore) and C. glomerata (parasitoid) were obtained from an insect culture maintained at the Laboratory of Entomology of Wageningen University, the Netherlands. P. brassicae and C. glomerata were cultured on Brassica oleracea. L. nana (hyperparasitoid) was obtained from a culture maintained at the Netherlands Institute of Ecology, Heteren, the Netherlands. The hyperparasitoid was cultured exclusively on pre-pupae and pupae of C. glomerata. D. radicum (root herbivore) was obtained from an insect culture of the Swiss Federal Research Station for Fruit-Growing, Wädenswil, Switzerland. Root flies were cultured on the roots of Brassica napus and B. oleracea.
B. nigra seeds were collected from a single B. nigra population in the north-west of Wageningen, the Netherlands. Seeds were surface sterilized and germinated on glass beads. One week after germination seedlings were transplanted into 1·2-L pots filled with soil collected from a restoration area that was abandoned in 1996, at De Mossel in Ede, the Netherlands. The sandy loam soil had been sieved (2 cm) mixed with 25% white sand, and sterilized using gamma radiation (25 KGray) to eliminate all soil organisms. The plants were grown in a greenhouse, at a temperature of 25 ± 1 °C (day) and 22 ± 1 °C (night), with a relative humidity of 70% and with a photoperiod of 16 : 8 h (day/night). Natural daylight was supplemented by 400 W metal halide bulbs (one per 1·5 m2). Plants were watered daily and were supplemented with nutrients once when the plants were 3 weeks old.
Forty-five days after transplanting, D. radicum larvae were introduced into the experimental pots. Depending upon treatment, five (low density) or 20 (high density) newly hatched first instar larvae were carefully placed with a small brush next to the main stem of the plant and observed to ensure that they successfully crawled into the soil. A third set of plants was kept undamaged and served as control. Ten plants were used for each root herbivore density (0, 5 and 20).
It is known that the leaf GLS concentration in Brassica plants increases after some days of feeding by P. brassicae. For B. nigra it takes approximately 7 days to reach the maximum GLS concentration (N.M. van Dam, pers. comm.). To clearly visualize the potential effect of the root herbivore on the above-ground insects, it is necessary to discern it from the effects that P. brassicae could cause. A second set of replicates was set up for this purpose. In this extra set of plants, P. brassicae larvae were periodically moved to a new host plant (every 3–4 days throughout the course of the experiment). For this, a second series of plants was transplanted and inoculated with D. radicum on different days, so that the time between inoculation and exposure to above-ground herbivory was kept constant.
above-ground insect measurements
Thirteen days after root fly inoculation, plants were placed individually into meshed cylindrical cages (height 1 m, diameter 35 cm). Into each cage two unparasitized and two parasitized second instar larvae of P. brassicae were introduced by placing them on the eighth mature leaf of the plant. Parasitism was carried out by individually offering second instars of P. brassicae to single C. glomerata females in plastic vials.
Development of unparasitized P. brassicae larvae was checked daily and fresh pupae were weighed on a Mettler Toledo Microbalance (accuracy ± 1 µg). Larval development time of P. brassicae was determined as the number of days until pupation. To measure the amount of plant tissues consumed by early larval instars of P. brassicae, all feeding damage on each leaf was marked on a transparent sheet and the total area per plant eaten was measured using a leaf-area scanner. The measurements were performed on the plant renewal set of plants, after the first (consumption during days one to four) and second (consumption during days 5–7) plant renewal.
For parasitized P. brassicae, the number of emerged cocoons per clutch was counted. Larval development time was determined as the number of days between parasitism and pre-pupal egression; pupal development time was determined as number of days between pre-pupal egression and adult emergence. From each clutch, five cocoons were randomly selected for hyperparasitism. The remaining C. glomerata cocoons of each clutch were placed, per a given clutch, into a Petri dish (10 cm diameter), and checked daily (at intervals of 2 h) for adult emergence. At emergence, adult size and development time were recorded. Each cocoon presented to L. nana was individually weighed on the microbalance. Cocoons were presented in groups of five with three to five L. nana females in Petri dishes (diameter 10 cm). Upon parasitism, cocoons were immediately removed and transferred individually into plastic vials (2 cm diameter, 5 cm length). Cocoons were checked daily, and at eclosion adult hyperparasitoids were weighed on the microbalance. Egg to adult development time was also recorded.
Foliar GLS concentrations, N, phosphorus (P), potassium (K) and carbon/N (C/N) ratio were measured in an additional set of plants exposed to low and high levels of root herbivory and in undamaged control plants. Plants were treated as described above and the period of root herbivory was also the same.
To determine GLS, for each plant two immature (the second and third leaf) and two mature leaves (the ninth and tenth) were removed from the plant using a razor blade and immediately frozen at −80 °C, then freeze-dried and ground. Foliar GLS was also measured after the plants were exposed to root herbivory and above-ground herbivory (17 days of root herbivory and the last 4 days also above-ground herbivory). GLS was determined using high performance liquid chromatography (HPLC) as described by Van Dam et al. (2004).
To determine the content of foliar N, P and K and C/N ratio the remaining leaves of each plant were harvested, oven-dried at 70 °C, ground and analysed as described in Troelstra et al. (2001). Total foliar and root biomass were also measured.
The statistical analysis of the performance of P. brassicae, C. glomerata and L. nana was based on a model with two factors (root herbivory and plant renewal). As more than one observation was taken from each experimental unit (pot-plant) there are two sources of experimental error, that of subsamples (within each experimental unit) and between experimental units. Within a mixed model framework these two random terms can be specified. The following mixed model describes the experimental set up:
where Yijk is the observed value, µ is the general mean, αi the fixed effect of the plant renewal level i (i = with or without renewal), βj (the fixed effect of root fly density; j = 0, 5, 20 larvae per pot), and αβij the plant renewal by density interaction. The two random terms in the model correspond to the between pot variation dij () and the within pot variation eijk (), both assumed independently and normally distributed with variances and , respectively. Model fit was done by employing restricted maximum likelihood (REML) in Genstat 6 and statistical tests for fixed effects were done using an approximated F-test. Normality, independence and homogeneity of variance were checked by inspection of the residuals after fitting the model.
The effect of root herbivory and above-ground herbivory on GLS and primary plant compound levels in plant tissues were analysed using a fixed model:
where Yijkl is the observed value, µ is the general mean, αi corresponds to the effect of above-ground herbivore i (i = with or without renewal), βj the effect of root fly density (j = 0, 5, 20 larvae per pot), λk the effect of leaf age i (i = 1, 2; young and old, respectively). Finally, all two-way and three-way interaction between the factors were included (αβij, αλik, βλjk, αβλijk). Residuals where assumed independently and normally distributed with variance . Analyses were carried out using STATISTICA 7. Normality, independence and homogeneity of variance were checked by inspection of the residuals after model fit.
Root herbivory significantly affected larval development time of P. brassicae, and the effect differed between root herbivore densities (Table 1; Fig. 1a). Relative to control plants, larval development time was extended by, on average, 0·7 days at low root herbivore densities, but no effect was observed at high densities. Pupal weight was not affected by root herbivory (Table 1; Fig. 1b). Early instar food consumption also did not differ between treatments, neither during the first period (days 1–4, F2,12 = 0·41; P = 0·66) nor the second (days 5–7, F2,12 = 0·80; P = 0·47; data not shown). There was no effect of plant renewal on P. brassicae performance (Table 1).
Table 1. Approximate F-test for the fixed effects from REML analysis of the effect of plant renewal and root herbivory on the larval development time and pupal weight of the herbivore P. brassicae and on the larval development time, cocoon weight, and clutch size of the parasitoid C. glomerata
Larval development time
Larval development time
Plant renewal (1)
Root herbivory (2)
There were highly significant effects of root herbivory on C. glomerata larval development time, cocoon weight and clutch size, and the effects differed between root herbivore densities (Table 1; Fig. 2a–c). Larval development time was significantly longer on plants exposed to the low root herbivore densities compared with undamaged plants and plants exposed to the high densities of root herbivores (Fig. 2a). Independent of root herbivore density, parasitoid cocoon weight was significantly reduced in the presence of root herbivores (Fig. 2b). Secondary parasitoid clutch size was significantly higher on plants exposed to low root herbivore densities compared with undamaged plants or plants exposed to high root herbivore densities (Fig. 2c). There was no significant effect of plant renewal on parasitoid larval development time, cocoon weight or clutch size (Table 1).
Both parasitoid pupal development time and parasitoid adult weight were affected differently by parasitoid sex, plant renewal and root herbivory, resulting in a significant three-way interaction for both variables (Table 2). There was a strong effect of sex on pupal development time, with males developing faster than females (Table 2). Without plant renewal, female pupal development time was not affected by root herbivory, but with plant renewal females had shorter development times on plants exposed to low root fly densities (Fig. 3a). Male adult parasitoid weight was not affected by root herbivory or plant renewal, but in the treatment without plant renewal female weight in plants exposed to low and high root fly densities was lower than in control plants (Fig. 3b). Mortality was not significantly affected by root herbivory (F2,81 = 0·42; P = 0·66; data not shown).
Table 2. Approximate F-test for the fixed effects from REML analysis of the effect of sex, plant renewal and root herbivory on the pupal development time and adult weight of the parasitoid C. glomerata
Pupal development time
Plant renewal (2)
Root herbivory (3)
Independent of sex, L. nana adult weight was significantly affected by root herbivory and plant renewal (Table 3), while egg to adult development time was not affected (Table 3; Fig. 4a). In the no-renewal treatment, in the presence of root herbivores, hyperparasitoid weight was significantly reduced compared with the control treatment (Fig. 4b). Adult hyperparasitoid weight was lower for adults emerging from the renewal treatment than from the no-renewal treatment. Mortality was not significantly affected by root herbivory (F2,55 = 1·01; P = 0·36; data not shown).
Table 3. Approximate F-test for the fixed effects from REML analysis of the effect of sex, plant renewal and root herbivory on the pupal development time, and adult weight of the hyperparasitoid L. nana
Pupal development time
Plant renewal (2)
Root herbivory (3)
There was a strong positive relationship between hyperparasitoid adult weight and parasitoid cocoon weight (host size) (F1,201 = 176·1; P < 0·001) but this relationship was not affected by root herbivory treatment (F2,201 = 0·61; P = 0·54) (Fig. 5).
Plants in all treatments had similar shoot biomass (F2,21 = 0·22; P = 0·80), but root herbivory significantly decreased root biomass (F2,21 = 4·76; P = 0·01). Plants exposed to high root herbivory had significantly lower root biomass compared with undamaged plants, while intermediate root biomass was observed in plants exposed to low root herbivore density (Table 4). Foliar N (F2,24 = 4·88; P = 0·03) and C/N ratio (F2,22 = 8·93; P = 0·005) were significantly affected by root herbivory. N levels were significantly lower in plants exposed to high root herbivore densities compared with control plants and plants exposed to low root herbivory, resulting in significantly higher C/N ratios in plants with high root herbivory (Table 4). No change in P and K content of mustard foliage was recorded.
Table 4. Chemical composition (mean ± SE) in shoot tissue and root and shoot biomass of B. nigra plants exposed to low or high root fly densities and in undamaged control plants. Root herbivory was caused by D. radicum. Within rows, means followed by identical letters are not significantly different (P < 0·05) based on a Tukey test
3·79 ± 0·26b
3·79 ± 0·37b
2·77 ± 0·04a
10·35 ± 0·27a
11·19 ± 0·96a
14·19 ± 0·17b
0·57 ± 0·06
0·60 ± 0·04
0·69 ± 0·12
4·04 ± 0·27
4·33 ± 0·35
4·37 ± 0·33
Root biomass (g)
1·11 ± 0·09a
0·93 ± 0·1ab
0·61 ± 0·1b
Shoot biomass (g)
2·73 ± 0·2
2·61 ± 0·1
2·55 ± 0·1
Foliar GLS content was significantly increased in the presence of root herbivory (Table 5). The changes in total GLS levels were mainly caused by changes in the major GLS, sinigrin, which represented 99·7% of the total GLS contents. In the presence of root herbivory sinigrin content significantly increased in young leaves (Fig. 6a) but not in mature leaves (Fig. 6b). Sinigrin levels were significantly higher at low root herbivore densities than in undamaged plants while intermediate level of sinigrin was observed in plants exposed to high root herbivore density (Fig. 6a). There were also strong effects on sinigrin levels following 4 days of above-ground herbivory (Table 5). In young leaves, sinigrin levels were significantly lower in the presence of above-ground herbivory (Fig. 6a).
Table 5. anova of the effect of above-ground herbivory, root herbivory and leaf age on sinigrin levels of B. nigra plants. Root herbivory was caused by D. radicum larvae and above-ground (AG) herbivory was caused by P. brassicae larvae
AG herbivory (1)
Root herbivory (2)
Leaf age (3)
This experiment is the first to provide evidence that root feeding insects, via the shared host plant, can influence not only above-ground herbivores and parasitoids, but also hyperparasitoids. Our study also reveals that effects of root herbivory do not uniformly influence the performance of all above-ground insects, but that they can mostly bypass one trophic level while negatively affecting others higher up in the above-ground food chain.
Larval development times of P. brassicae and C. glomerata were extended in both cases by about 1 day on plants exposed to root herbivores. The developmental programme of many endoparasitoids is known to be physiologically co-ordinated with biochemical characteristics of the host, including endocrinological factors such as juvenile and prothoracic hormone titres (Beckage 1985; Lawrence 1990). Many koinobiont parasitoids, including C. glomerata, do not begin destructive feeding and thus commence exponential larval growth, until the host enters its final instar (Harvey, Harvey & Thompson 1994). This ensures that sufficient resources are available for parasitoid progeny to complete their development. Consequently, the increase in development time observed in C. glomerata is therefore most probably attributable to a concomitant increase in the development time of its herbivore host, P. brassicae. The increase in development time that was observed in P. brassicae and C. glomerata associated with plants exposed to D. radicum was not, however, carried over to L. nana. Idiobiont parasitoids, such as L. nana, attack hosts that do not feed or grow, and thus represent mostly ‘static’ resources (Mackauer & Sequeira 1993).
Different effects were observed for the response on pupal or adult weight. While pupal weight of P. brassicae was not significantly affected, cocoon weight of C. glomerata and adult weight of L. nana were significantly reduced in the presence of root herbivory. This suggests the transmission of a qualitative effect from the plant via the herbivore, to the parasitoid and the hyperparasitoid. Adult body size in idiobionts is often strongly correlated to the initial size of the host on which they developed (Mackauer & Sequeira 1993), and this is also true for L. nana (Harvey et al. 2003). Thus, the reduction in body size of the hyperparasitoid when developing on C. glomerata cocoons originating from plants exposed to D. radicum is probably attributable to the negative quantitative effects of root herbivory on the size of the primary parasitoid host.
Sinigrin content increased by about 50% in plants exposed to low root fly densities compared with control plants, whereas in plants exposed to high root fly densities levels were intermediate. Other work has shown that root herbivory by Delia floralis can almost double the amount of GLS in shoots of cultivated Brassica species (Birch et al. 1992). Sinigrin is the primary GLS produced by B. nigra plants, and is known to act as a feeding deterrent for many herbivorous insect pests of Brassica and Sinapis species (Olsson & Jonasson 1994; Schoonhoven & Liner 1994). Our results suggest that the reduced performance of above-ground insects on plants previously exposed to root herbivory may have been the indirect consequence of allelochemicals in the host plant, induced by D. radicum below-ground. Our results thus at least partially support the ‘defence-induction hypothesis’.
The amount of root biomass was greatly reduced when plants were exposed to high root fly densities. However, the maximum concentration of secondary plant defence compounds (sinigrin) were recorded on plants exposed to low root fly densities. It is possible that plants exposed to high root fly densities were severely stressed and thus unable to maintain optimal direct defence capabilities, as suggested by a decrease in the N content of their foliage. On the other hand, plants exposed to low root fly densities, which experienced less root damage, retained their vigour and were thus able to allocate defence compounds to the shoots more effectively.
Specialist herbivores of crucifers are thought to have evolved ways to minimize the toxic effects of the GLS (Ratzka et al. 2002). GLS can stimulate feeding of specialist herbivores (e.g. Bodnaryk 1991), but other reports have shown no effect (e.g. Bodnaryk & Palaniswamy 1990) or a negative effect on the feeding and/or development of specialists (e.g. Stowe 1998). Our results indicate that although P. brassicae is a specialist herbivore and is thought to be well adapted to high levels of GLS (e.g. Harvey et al. 2003), indirect interactions with below-ground herbivores can change plant quality (via the increased production of defence compounds to shoots) enough to impact their development and that of its parasitoid and hyperparasitoid.
The plant renewal factor was added with the aim of excluding the possible effects of foliar feeding on the expression of above-ground secondary compounds. In that context, it is important to note that we did not find a significant effect of plant renewal on P. brassicae and C. glomerata development. Moreover, the interaction between renewal and other factors were mostly insignificant with the sole exceptions of the parasitoid pupal development time and adult weight that showed a less straightforward and more complex three-way interaction. The only main effect of renewal was observed on the adult weight of the hyperparasitoid, where wasps were larger in the no-renewal treatment than in the renewal treatment. This suggests that the performance of most above-ground insects was negatively affected irrespective of the renewal treatment under investigation. Several factors may explain why differences were not observed in these experiments. First, it is possible that P. brassicae did not cause additional induction of GLS in the plants. Alternatively, the effects of root herbivory may have been much stronger than above-ground herbivory.
Foliar N levels were also negatively affected by root herbivory, but only in plants exposed to the highest densities of D. radicum. It is well established that N is a limiting factor for insect growth. Increasing the levels of N in plants that initially contained deficiencies in this nutrient has been shown to enhance the growth of many organisms, including insects (Mattson 1980). However, this is unlikely to have been a major factor affecting the development of the above-ground insects in this study, because N levels were very high to begin with, irrespective of treatment, and because insect performance was most negatively affected when associated with plants exposed to low root fly densities that had N levels similar to controls.
In summary, this is the first study to have shown that root herbivores, such as D. radicum, can exert indirect influences over the development of at least three trophic levels of an above-ground system. These effects are driven, in part, by differences in plant quality that are probably the result of changes in levels of secondary plant compounds as influenced by damage in the roots of the host plant. Most importantly, these results have significant implications for our understanding of community-level processes in the above- and below-ground domains. They emphasize the need to acknowledge the role of both above- and below-ground biota in determining the structure and function of ecological communities (Bezemer et al. 2004b; Wardle et al. 2004). By integrating a broader range of above- and below-ground systems, future work will help to facilitate a better understanding of complex multitrophic interactions.
The authors wish to thank Marcos Malosetti for assistance with the statistical analysis, Leo Koopman at Wageningen University for supplying herbivore and parasitoid cultures, Nicole van Dam for constructive comments during the experiment, Leontien Witjes for assistance in maintenance of insect cultures, Gregor Disveld for assistance in maintenance of plants, and Ciska Raaijmakers for assistance in the GLS extractions.