Intraspecific differences in plant chemotype determine the structure of arthropod food webs


It is becoming increasingly appreciated that the structure and functioning of ecological food webs are controlled by the nature and level of plant chemicals. It is hypothesized that intraspecific variation in plant chemical resistance, in which individuals of a host-plant population exhibit genetic differences in their chemical contents (called ‘plant chemotypes’), may be an important determinant of variation in food web structure and functioning. We evaluated this hypothesis using field assessments and plant chemical assays in the tansy plant Tanacetum vulgare L. (Asteraceae). We examined food webs in which chemotypes of tansy plants are the resource for two specialized aphids, their predators and mutualistic ants. The density of the ant-tended aphid Metopeurum fuscoviride was significantly higher on particular chemotypes (borneol) than others. Clear chemotype preferences between predators were also detected. Aphid specialist seven-spotted ladybird beetles (Coccinella septempunctata) were more often found on camphor plants, while significantly higher numbers of the polyphagous nursery web spider (Pisaura mirabilis) were observed on borneol plants. The analysis of plant chemotype effects on the arthropod community clearly demonstrates a range of possible outcomes between plant-aphid-predator networks. The findings help to offer a deeper insight into how one important factor—plant chemical content—influences which species coexist within a food web on a particular host plant and the nature of their trophic linkages.


Conceptions of what determines the trophic structure and functioning of ecological food webs increasingly recognize that nutrients, plant defence, herbivory and predation play interdependent roles (Van der Putten et al. 2001; Poehlman et al. 2008; Schmitz 2008; Mooney et al. 2010; Burghardt and Schmitz 2015). Such understanding derives largely from studies examining interspecific variation in plant-herbivore-predator interactions (Hare 1992; Poehlman et al. 2008; Schmitz 2010; Mooney et al. 2010; Burghardt and Schmitz 2015). There is also growing appreciation that genotypic and phenotypic differences within species can be important in explaining trophic structure and functioning (Johnson and Agrawal 2005; Crutsinger et al. 2006; Johnson 2008; Whitham et al. 2012; Barbour et al. 2015). But, in many cases, the mechanism determining such variation remains elusive (Barbour et al. 2015).

A candidate mechanism in food chains involving arthropod-plant interactions is intraspecific trait variation in the nature and concentration of plant chemical defences (Linhart et al. 2005; Gols et al. 2008; Johnson and Agrawal 2005; Johnson 2008; Poehlman et al. 2008; Burghardt and Schmitz 2015). At an interspecific level, plant defensive chemicals are known to influence the trophic structure of food chains by changing how herbivores mediate bottom-up effects of plants on predators and top-down effects of predators on plants, or through changes in plant traits, such as release of volatile attractants, that alter the direct effect of plants on predators (Hare 1992; Roth et al. 1997; Harvey et al. 2003, 2007; Baldwin 2006; Gols et al. 2008; Dicke and Baldwin 2010). It therefore stands to reason that if intraspecific differences in chemical defence mediate the magnitude of interspecific variation, then we may improve our understanding of changes in food chain structure through explicit consideration of intraspecific variation in defence expression.

To this end, we report on an analysis of intraspecific variation in chemical defence of the herb tansy, Tanacetum vulgare, and its implications for associated arthropod food web structure. Tansy exhibits substantial intraspecific variation in essential oil content, in particular monoterpenes and sesquiterpenes that are emitted as volatiles. The chemicals are distributed throughout each plant but are predominantly present in leaves and young stems (Holopainen 1989; Keskitalo et al. 1998; Rohloff et al. 2004). Intraspecific differences in chemical form and concentration have been described in terms of plant chemotype, where a chemotype is based on the dominant compound in the chemical profile (Holopainen 1989; Wolf et al. 2012). Breeding experiments as well as studies using molecular markers have shown that the chemical oil produced by a particular tansy chemotype has a genetic basis (Holopainen et al. 1987a, b; Holopainen 1989; Keskitalo et al. 1998).

Tansy chemotype could determine arthropod community structure. It is well known that predatory arthropods and parasitoids use volatiles emitted by plants as signals to locate plants carrying their prey or host, which in turn can determine food web structure through predator top-down control of herbivores (Baldwin 2006; Soler et al. 2007; Bruinsma et al. 2009). To date, more than 25 species from various taxa of predaceous arthropods (including those studied in the present work), parasitic nematodes, and insectivorous birds (Mantyla et al. 2004; Rasmann et al. 2005; Runyon et al. 2006; Baldwin 2006; Soler et al. 2007; Bruinsma et al. 2009; Benedek et al. 2015) are known to be attracted to plant volatiles and potentially contribute to plant indirect resistance (Mumm and Dicke 2010; Dicke and Baldwin 2010). Furthermore, variability in individual expression of these volatiles may explain variability in the insect food web composition associated with different varieties of a plant (Bukovinszky et al. 2008).

We tested whether chemical content of individual tansy chemotypes does indeed influence associated food web structure under field conditions. Using systematic field surveys, we examined relations between arthropod species and tansy chemical contents. We asked the following questions:

  1. 1.

    Does essential oil composition of tansy vary between plants?

  2. 2.

    Are there differences in arthropod abundances across different plant chemotypes?

  3. 3.

    Are different characteristics of these food webs (species composition and food web linkages) affected by tansy chemical content?

Materials and methods

Study system

Tansy is a perennial herbaceous composite from Europe and Asia that preferentially grows in disturbed, well-drained, poor soils (Keskitalo et al. 1998). It often forms isolated patches alongside river valleys, railway tracks and on wastelands. Single plants comprise a ‘genetically identical’ genet with up to 50 flowering ramets (shoots) (but usually far fewer). Genets propagate clonally underground via stolons.

Two dominant and specialized aphid species feed on tansy: Macrosiphoniella tanacetaria (Kaltenbach) and Metopeurum fuscoviride Stroyan. M. tanacetaria is not attended by ants and forms small colonies mainly on the top of ramets (shoots), while M. fuscoviride is an obligatory ant-tended aphid that feeds in more compact and often large colonies on leaves and near the apex of ramets. The black garden ant, Lasius niger (L.) commonly tends M. fuscoviride aphids (Flatt and Weisser 2000). Many predators attack both aphid species, the most important being the aphid specialist ladybird beetles, e.g. the seven-spotted ladybird (Coccinella septempunctata); predation from the generalist nursery web spiders (Pisaura mirabilis) and minute pirate bugs (Orius spp.) can also be observed (Benedek et al. 2015). The attending ants help to defend the aphids from the predators (Mehrparvar et al. 2013; Benedek et al. 2015).

Field assessment of arthropods on tansy

The 2-year (2011 and 2012) field study was conducted in three sites along a 110-km transect in Transylvania, Europe. Each site was between 2 and 3 km long and the sites were separated by 50 km (Supplementary material, Fig. 1). Site 1 contained several thousand tansy plants while sites 2 and 3 each contained several hundred plants. Sites were not isolated from each other (principal railway-linked sites), therefore exchange of tansy material, i.e. pollen exchange between sites, was possible. There were no obvious environmental and climate differences between sites—both soil and climatic conditions were similar. The soil was well drained, poor clay in all sites. The climate was temperate specific for the sub-Carpathian region, with abundant precipitation in spring and fall, a relatively warm summer, and low to very low temperatures (to −20 °C) in winter.

In total 100 plants were randomly selected for the study (50 plants in site 1, 25 plants in sites 2 and 3). The distance between the selected plant individuals within each site was 100–110 m to ensure that separate genotypes were selected. All plants were labelled individually at the start of the assessment, and the same plants were examined during the entire study. In both years, weekly censuses of the marked plants were carried out, between 7 and 11 a.m. Sampling began in May when the aphids first appeared and continued until the end of September when aphids had disappeared from the plants. Aphids and other arthropods observed on the marked plants were counted. M. tanacetaria and M. fuscoviride were counted as individuals, except when the colony size of M. fuscoviride was large. In this case, all aphids on a 5-cm section of the shoot within the aphid colony were selected and all aphids counted. Total aphid number on a ramet was then estimated by multiplying that count with total colony length (in centimetres) along the ramet (e.g. 80 aphids in a 5-cm2 area, colony length 15 cm, total 240 aphids). The aphid species were distinguished by their body colour: M. tanacetaria is green and M. fuscoviride is black. After counting the aphids, a 10-min visual scan was made of all aphid colonies on a plant. We counted all ant workers and all aphid predators that were observed attending the aphid colony or were less than 1 cm away from an aphid.

We quantify food web linkages by noting ants that were actively tending aphids, and any interactions between ants and a predator, between a predator and an aphid, and between predators.

Volatile extraction from plants and quantitative analysis of volatiles

In June of year 2, live plant material (100 g; fresh leaves and stems) was collected from unoccupied ramets of each marked plant and stored at −20 °C. These samples were used to measure and compare, using gas chromatography, the main spectrum of volatile compounds in each individual plant. Standards of pure volatile compounds (camphor, borneol, cineol, piperiton and α-thujone and β-thujone) were obtained from the Roth Laboratory (Canada) to create chromatograms for comparison.

Plant material from an individual plant was placed into a round-bottom Clevenger micro-distillation flask containing 50 ml water. Flasks were heated at 250 W for 4 h. A 6-ml aliquot of water and essential oil was then extracted with n-hexane (4 ml) using a mixer-settler method (András et al. 2009). Samples were vortex mixed (IKA Vortex Genius 3) for 30 min. The resulting emulsion was separated by centrifuging (Hettich Universal 32) at 1200 r.p.m. for 5 min. The separated organic phase was stored in 4-ml glass vials over anhydrous sodium sulphate at 4 °C until it was subjected to gas chromatographic analysis. Extracted essential oils were analysed using a Varian CP 3380 gas chromatography coupled to a flame ionization detector [CP-Sil 88 (100 m × 0.25 mm) silica capillary column]. The system was operated at hydrogen gas pressure of 235 kPa, with 2 μl in the injection probe at a temperature program of 270 °C. The starting temperature varied from 50 to 210 °C with a 5 °C min−1 gradient, and then a constant temperature was applied during the 55-min analysis for each sample (András et al. 2009; Kapás et al. 2011).

The volatile components (the percentage values based on compound quantity from each plants) were identified by comparing their retention times with the standard chromatogram (camphor, borneol, cineol, piperiton, α-thujone and β-thujone). Individual plants were identified to chemotype based on their volatile compound quantity. Plants were designated as pure chemotypes when a single chemical made up 90 % or more of the total quantity of compounds detected e.g. ‘thujone pure’. Hybrid chemotypes were designated as those plants in which two volatiles together made up more than 90 % of the bouquet i.e. ‘thujone hybrid’. In all these cases, the proportion of the dominant volatile was at least three times higher than the next most abundant volatile compound. When more than two compounds made up >90 %, the mixed chemotype was designated by the dominant compound present, e.g. ‘thujone mixed’.

Field data analyses

We quantified food web structure in terms of species composition and abundance. Our analysis began by testing for inter-annual differences in abundance of each arthropod species by constructing principal response curves (Brink and Braak 1999), which account for population dynamics through the season. Mean number of arthropods per individual plant per sampling date was considered. This analysis revealed that M. fuscoviride, ants and predators were consistently present together from 22 June until 22 August in the first year, and from 20 June until 18 August in the second year. Therefore only the nine weekly sampling dates between these start and end dates were considered for further data analyses. We tested for a year effect by comparing the mean number of arthropods on each plant using a repeated measures analysis on weekly sampling dates between years. No significant temporal difference was detected (multivariate ANOVA; MANOVA). Therefore we pooled data for both years in all subsequent analyses.

We used two generalized linear modelling approaches with quasi-Poisson error distributions to analyse the data further. The first model tested for the effect of discrete chemotype (considered pure or hybrid as defined above) on arthropod abundance, using each species group as a response variable, and site, plant chemotype and arthropod species groups as explanatory variables. The second model tested for the effect of plant chemotype profile (considering all volatiles according to their relative proportion in a tansy plant) on different attributes of the food web. This was done by first subjecting data on the relative amount of each chemical per plant (%) to a principal components analysis (PCA) to identify the proportion of variation in each PCA axis (1–3) that was explained by each chemical (α-thujone, β-thujone, camphor, borneol, cineol and piperiton) (Fig. 1). We then used the average counts of each invertebrate grouping as response variables, and used site, PCA axis 1 (PCA1), PCA2 and PCA3 scores, each chemical and species group as independent variables. PCA covariance analyses were run using community analysis package 4 (Pisces Conservation). All other statistics were run in R Studio version 0.97.314 using R version 3.0.1 (R Core Team 2013).

Fig. 1

Principal components analysis (PCA) covariance plot of chemotype profiles by the most dominant chemotypes

Food web construction and its parameters

We tested pure and hybrid chemotypes to determine the direction and strength of correlation between M. fuscoviride aphids and ants, M. fuscoviride aphids and predators, ants and predators and between predators (i.e. spiders and Orius, spiders and ladybirds). We assumed that a significant negative correlation reveals a strong predatory effect; a significant positive correlation reveals mutualism.

All observed interactions during the nine sampling dates/year when all arthropods were present were grouped according to the plant chemotypes on which they were observed. These data included 334 observed interactions between ants and M. fuscoviride aphids, 13 ant-M. tanacetaria aphid interactions, 218 predator-M. fuscoviride aphid interactions, 16 predator-M. tanacetaria aphid interactions, 49 ant-predator and 20 predator–predator interactions. Based on these observed interactions, and estimated correlations between arthropod species, we constructed food web networks using CoSBiLab software (Jordán et al. 2012). We calculated the following indices for each chemotype-based food web:

  1. 1.

    Web degree, which represents linkage density (links per species) that considers the number of species and number of links between them. Both metrics can be simply derived by counting the species that interact and the number of links (interactions) between them. If, for example ladybird predation on M. fuscoviride aphids was observed several times, this was considered as one interaction. This parameter indicates the complexity of the web. This is the most local index of food web topology (Barabási and Albert 1999; Dunne et al. 2002).

  2. 2.

    Bottom-up (K bu,i ) and top-down (K td,i ) effects are indices which emphasize vertical interactions and indicate a strong interference between trophic groups (Jordán et al. 2012). Large top-down effects require weak interference, while large bottom-up effects require both weak interference and strong prey dependence (Dunne et al. 2002). These indices can be calculated as: 

    $$K_{bu,i} \; = \;\sum\limits_{c = 1}^{n} {\frac{1}{{d_{c} (1\; + \;K_{bc} )}}}$$

    For node i, i.e. species i, K bu,I quantifies the bottom-up effect of species I where n is the number of predators eating species i, d c is the number of prey of its cth predator and K bc is the bottom-up keystone index of the cth predator.

    $$K_{td,i} \; = \;\sum\limits_{e = 1}^{m} {\frac{1}{{f_{e} (1 + \;K_{te} )}}}$$

    K td,I quantifies the top-down effect of species i. Here, m is the number of prey eaten by species i, f e is the number of predators of its eth prey and K te is the top-down keystone index of the eth prey. K bc (bottom-up) and K te (top-down) keystone indices are computed as the sum of the dominant (key) predators/plants, respectively, and dominant herbivore/plant averaged over the other species from the food web.

The calculated values for the food web indices were tested for normality of errors and homogeneity of variance. Bottom-up and top-down indices were considered as separate variables for each individual plant per sampling date. Effects of chemotype on food web bottom-up and top-down indices for each sampling period were tested using repeated-measures MANOVA.

Model fit with observed field interactions was compared using χ2-tests on the differences between the covariance matrices, and by the root mean square error of approximation. The statistical analyses were performed in R version 2.8.0. and the package bipartite for food web analyses and SEM (R Core Team 2013).


Classification of tansy by essential oils, arthropod densities on different tansy chemotypes

Ninety-five of the 100 plants examined could be assigned to either a pure or hybrid chemotype. Fifty plants were pure chemotypes. Of these, 25 were camphor (camphor pure), seven borneol (borneol pure) and 18 thujone (thujone pure) chemotypes. Another 44 plants were considered hybrid chemotypes of which 15 were dominated by camphor (camphor hybrid), four dominated by borneol (borneol hybrid), 25 dominated by thujone (thujone hybrid). Only one plant was assigned as mixed thujone (thujone mix) (Supplementary material, Fig. 1). When thujone was detected in both pure and hybrid plants, this was β-thujone. We used only these 95 plants for further analyses because our confidence in assigning them to chemotype was high, which was not the case for the five plants that were excluded from further analysis.

All three sites contained multiple chemotypes, although the borneol chemotype did not occur at site 2. Sites 1 and 2 were dominated by camphor and thujone chemotypes and site 3 was dominated by borneol and thujone chemotypes (Fig. 1). We detected a significant effect of site and chemotype on most arthropod abundances, except those of Orius (Table 1). For example, significantly more M. fuscoviride aphids were found on borneol plants than on other chemotypes throughout the nine census dates in both years, and abundances of M. tanacetaria aphids were lower on thujone plants (Table 1). This aphid species inhabited the top of the plants in small colonies where predation by ants and all predators was observed. There was a significant positive effect of the borneol hybrid chemotype on the number of ants observed (Table 1). Three predators were often observed on plants: the seven-spotted ladybird (C. septempunctata), the nursery web spider P. mirabilis and several species of minute pirate bugs (Orius spp.). The numbers of the seven-spotted ladybird was lower on thujone pure chemotypes, and more nursery web spiders were observed on borneol pure and borneol hybrid plants than on other chemotypes (Table 1). The abundance of M. fuscoviride aphids was positively associated with ants and spiders, but negatively associated with the other aphid species and Orius (Table 1). M. tanacetaria was negatively associated with M. fuscoviride and positively associated with spider abundance. Ant abundance was positively associated with M. fuscoviride and negatively associated with M. tanacetaria and Orius. Ladybird abundance was only influenced by M. fuscoviride abundance. Therefore, both aphid species and Orius had positive associations with spiders, while negative associations between ants and spiders and Orius were detected (Table 1).

Table 1 The effect of discrete chemotype grouping on different aspects of the community, using each species group as a response variable, and site, discrete chemotype and species groups as explanatory variables with interactions between species and chemotype

PCA showed that variation in chemical profile among individual plants is primarily explained by the relative amount of camphor and β-thujone in the plants and following this, the amount of borneol then α-thujone (Fig. 1; Table 2). Consistent with the discrete chemotype analysis, borneol had a significant positive effect on M. fuscoviride density (PCA2; Fig. 2a; Table 3). The abundance of ladybirds was to some extent positively influenced by β-thujone and camphor (PCA1) concentrations and negatively influenced by α-thujone (PCA3). The abundance of spiders was also influenced by PCA1 and PCA2, explained by a higher abundance of spiders on plants containing more β-thujone or borneol and less camphor (Fig. 2b; Table 3).

Table 2 Variance component analysis on principal components analysis (PCA) scores, showing the relevant chemicals associated with each axis
Fig. 2

Effect of plant chemotype a % borneol on Metopeurum fuscoviride (ME) aphid number and b chemotype profile on spider number, with PCA axis 2 (black solid line and solid circles) and PCA axis 1 (red solid triangles and dashed line) (Color figure online)

Table 3 The effect of plant chemotype profile (continuous using PCA) on different aspects of the community, using each species group as a response variable, and site, the first three PCA axes scores and species groups as explanatory variables with interactions between species and PCA axes

Interactions, food webs and metrics

There were high positive associations for all cases between ants and M. fuscoviride aphids (camphor pure r = 0.72, p < 0.05; thujone pure r = 0.59, p < 0.05; borneol pure r = 0.60, p < 0.05; camphor hybrid r = 0.65, p < 0.05; thujone hybrid r = 0.61, p < 0.01; borneol hybrid r = −0.74, p < 0.01). Negative associations were detected between M. fuscoviride and the seven-spotted ladybird beetles on camphor pure (r = −0.42, p < 0.05), camphor hybrid (r = −0.87, p < 0.01) and thujone hybrid plants (r = −0.61, p < 0.01). The nursery web spider density was negatively correlated with the density/abundance of M. fuscoviride only on borneol plants (borneol pure r = −0.67, p < 0.01; borneol hybrid r = −0.47, p < 0.05). A negative association between the number of ladybirds and spiders (r = −0.406, p < 0.001) was observed.

Analyses revealed that different food webs exist on different tansy chemotypes. Dominance of ladybirds on camphor and of nursery web spiders on borneol chemotypes was clearly observed, while no predator dominance occurred on thujone plants (Fig. 3a–c). Path analyses revealed that significant trophic relations exist between the arthropod species in all tansy chemotypes. Our model predicted relations between ladybirds and M. tanacetaria on borneol plants, but statistics revealed no significant effects (Fig. 3b). Ants protected M. fuscoviride colonies on all pure and hybrid chemotypes. Intra-guild predation was observed only on borneol plants, where nursery web spiders preyed upon Orius and ladybirds (Fig. 3b). There was no observed ant predation of M. tanacetaria aphids on borneol (Fig. 3b).

Fig. 3

Food web on tansy pure camphor pure (a), borneol pure (b), thujone pure plants (c). Bold arrow represents mutualistic relations between Lasius niger and M. fuscoviride aphids. Hybrid chemotypes had the same food webs as the appropriate dominant plants, thus they are not repeated again in figures. Numbers represent χ2-values for significant path coefficients. *p < 0.05, **p < 0.01, ***p < 0.001

The number of species used for food web analyses was seven for all pure and hybrid chemotypes. The number of links between species did not differ considerably. There were 13 links for camphor and borneol plants (both pure and hybrid) and 12 for thujone plants (both pure and hybrid). No significant differences of bottom-up effects were detected between pure chemotypes, or between hybrid chemotypes (Table 4). There was, however, a significantly higher top-down influence on borneol pure and borneol hybrid plants, while no differences in top-down effects on camphor and thujone plants were observed (Table 4).

Table 4 Food web parameters on tansy pure and hybrid chemotypes


This study demonstrates that variation in chemical defence exists within the tansy species T. vulgare, with the plants in our study area dominated by β-thujone, camphor and borneol chemotypes. Moreover, there is considerable spatial variation in the composition of different chemotypes (i.e. β-thujone and camphor, β-thujone and borneol, camphor and borneol) across sites (Supplementary material, Fig. 1). A previous study has also shown that genetic differences also exist between M. fuscoviride aphids on different tansy chemotypes (Benedek et al. 2015).

The ant-tended aphid M. fuscoviride was dominant in borneol and borneol-hybrid plants. Ants had clear protective effects on M. fuscoviride aphids on all plants irrespective of their chemotype. Consequently, M. fuscoviride predators were uniformly influenced by ants on all plants. Previous studies suggested that the population growth and colony persistence of M. fuscoviride in the presence of ants are much greater than in the absence of ants (Stadler and Dixon 2008). M. tanacetaria aphid colonies were relatively unstable and disappeared rapidly from all plants. As M. tanacetaria is not ant-tended it may be preyed on by both ants and predators, therefore high predator pressure may lead to its rapid extinction from plants. Species-species interactions revealed that ant abundance was positively correlated with M. fuscoviride and negatively influenced by spiders (also described by Mestre et al. 2014) and Orius.

The variation in the associations of ants, aphids and predators within food webs is complex, but some of that complexity can be explained by tansy chemotypes supporting different food webs according to their chemical compositions. Food web analyses revealed that the aphid specialist ladybirds were significantly and predominantly associated with camphor pure and camphor hybrid, and negatively associated with thujone hybrid (if the chemical was α-thujone) plants. Significantly higher numbers of the polyphagous nursery web spider were observed on borneol pure and borneol hybrid plants. While predator variation by chemotypes could mean spatial dispersion and separation due to interspecific competition (Begon et al. 2006), our study offers a new, alternative mechanism that plant volatile chemical composition mediates animal community composition. The predators may occupy different plants to partition prey resources based on plant chemical composition rather than by classic resource-partitioning mechanisms. Therefore, the differences in M. fuscoviride aphid densities and their higher abundance on borneol plants most probably are due to differences in predator densities. In particular, different predators may be attracted by different volatile compounds, i.e. the attraction of spiders to borneol. Further, the association of ants with borneol plants could either be a result of the higher number of M. fuscoviride aphids or, alternatively, a direct effect of plant chemotype on ants, which then tend the M. fuscoviride aphids more efficiently on borneol plants. Experiments assessing the preference of ants and aphids for different plant chemotypes are required to determine the direction of this effect. Since there is still some confounding effect of site and chemotype, despite the use of appropriate statistics, it cannot be excluded that geographical differences among the three sites might contribute to differences in arthropod abundances. However, our previous research where tansy chemotypes were replanted in a common field also shows the same effects of borneol on spiders (Benedek et al. 2015). Thus, higher M. fuscoviride abundances on borneol plants may also be the consequence of a lower predation rate by the generalist nursery web spiders, which is also suggested by our modelling approaches.

The attraction of predators to specific volatiles is also supported by the results of the present study. Separately assessing individual tansy plants, e.g. camphor plants found in sites dominated by thujone (site 2) and borneol plants dominated by camphor (site 1) (Supplementary material, Fig. 1) reveals similar patterns [spiders related to borneol and ladybirds to camphor and β-thujone (PCA1)]. Both pure and hybrid borneol plants are, nevertheless, certainly capable of attracting higher numbers of spiders, and the same effect of camphor and β-thujone plants on ladybird numbers was also observed (PCA1). Previous studies on thyme demonstrated a similar role of chemotype, where the negative effects of spider and coccinellid predators on aphids varied by plant chemical content (Linhart et al. 2005). Aphid (Aphis serpylli) densities on thyme plants containing carvacrol, geraniol and thymol monoterpenes were reduced by arthropod predators. Aphids on plants with linalol, however, were not affected by predators, perhaps because predators showed chemotype-specific behaviour and avoided linalol-containing plants (Linhart et al. 2005). Host plants may also have trait-mediated effects on predators, such that particular chemical signals from plants can indicate higher prey quality and abundance that may have fitness effects on predators [i.e. increase fitness by feeding on prey on a particular tansy chemotype (Huang et al. 2010; Snoeren et al. 2010; Pierre et al. 2011)]. The higher M. fuscoviride density on borneol plants probably resulted from intraguild predator interactions that weakened top-down effects; specifically here it appears that the generalist nursery web spider may have preyed upon other predators (ladybird beetles and Orius) (Table 4).

The degree of top-down control can also be modulated if predators like ladybirds and spiders are differentially attracted by tansy volatile signals to locate plants carrying their prey. By partitioning host plants based on their volatiles some predators (like spiders and lady beetles in this study) can avoid intra-guild predation on tansy. Such resource partition by predators may be a consequence of variability in tansy inducible resistance causing variability in the predator species’ preferred prey being associated with particular varieties of a plant. Or, some predators like the generalist nursery web spider may be less affected by plant chemicals than specialist ladybirds with high attraction to camphor, as suggested by our food web modelling.

In conclusion, our study demonstrates a range of possible community-level outcomes between plant-aphid-predator networks that can be influenced by plant chemical content. Our work points to hypotheses that explain the mechanisms underlying species’ presence and mediating interactions (intra-guild effects, resource partitioning based on chemotype) that are not considered in classic analyses of species interactions. These hypotheses should help to motivate more experimental work that will resolve these mechanisms through the manipulation of chemotype abundances within local sites.


  1. András CsD, Kapás Á, Bartha Z, Salamon R, Cs Csajági, Székely G, Sz Salamon, György É (2009) Volatile extraction from fennel (Foeniculum vulgare Mill.). J Oil Soap Cosm 58:66–72

    Google Scholar 

  2. Baldwin IT (2006) Volatile signaling in plant–plant interactions: ‘talking trees’ in the genomics era. Science 311:812–815

    CAS  Article  PubMed  Google Scholar 

  3. Barabási AL, Albert R (1999) Emergence of scaling in random networks. Science 286:509–512

    Article  PubMed  Google Scholar 

  4. Barbour MA, Rodrigues-Cabal MA, Wu ET, Julkunen-Tiitto R, Ritland CE, Miscampbell AE, Jules ES, Crutsinger GM (2015) Multiple plant traits shape the genetic basis of herbivore community assembly. Funct Ecol. doi:10.1111/1365-2435.12409 (in press)

    Google Scholar 

  5. Begon M, Townsend CR, Harper JL (2006) Ecology: from individuals to ecosystems, 4th edn. Blackwell, Oxford

    Google Scholar 

  6. Benedek K, Bálint J, Salamon VR, Kovács E, Ábrahám B, Cs Fazakas, Loxdale HD, Balog A (2015) Tansy plant (Tanacetum vulgare L.) chemotype determines aphid genotype and its associated predator system. Biol J Linn Soc 114:709–719

    Article  Google Scholar 

  7. Brink PJV, Braak CJFT (1999) Principal response curves: analysis of time-dependent multivariate responses of biological community to stress. Toxicol Chem 18(2):138–148

    Article  Google Scholar 

  8. Bruinsma M, Posthumus MA, Mumm R, Mueller MJ, van Loon JJ, Dicke M (2009) Jasmonic acid-induced volatiles of Brassica oleracea attract parasitoids: effects of time and dose, and comparison with induction by herbivores. J Exp Bot 60:2575–2587

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  9. Bukovinszky T, van Veen FF, Jongema Y, Dicke M (2008) Direct and indirect effects of resource quality on food web structure. Science 319:804–807

    CAS  Article  PubMed  Google Scholar 

  10. Burghardt KT, Schmitz OJ (2015) Influence of plant defenses and nutrients on trophic control of ecosystems. Ch. 9. In: Hanley T, La Pierre K (eds) Trophic ecology: bottom-up and top-down interactions across aquatic and terrestrial systems. Cambridge University Press, Cambridge

    Google Scholar 

  11. Crutsinger GM, Collins MD, Fordyce JA, Gompert Z, Nice CC, Sanders NJ (2006) Plant genotypic diversity predicts community structure and governs an ecosystem process. Science 313:966–968

    CAS  Article  PubMed  Google Scholar 

  12. Dicke M, Baldwin IT (2010) The evolutionary context for herbivore-induced plant volatiles: beyond the ‘cry for help’. Trends Plant Sci 15(3):167–175

    CAS  Article  PubMed  Google Scholar 

  13. Dunne JA, Williams RJ, Martinez ND (2002) Food-web structure and network theory: the role of connectance and size. Proc Natl Acad Sci USA 99:12917–12922

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  14. Flatt T, Weisser WW (2000) The effects of mutualistic ants on aphid life history traits. Ecology 81:3522–3529

    Article  Google Scholar 

  15. Gols R, Wagenaar R, Bukovinszky T, van Dam NM, Dicke Bullock JM, Harwey JA (2008) Genetic variation in defence chemistry in wild cabbage affects herbivores and their endoparasitoids. Ecology 89(6):1616–1626

    Article  PubMed  Google Scholar 

  16. Hare JD (1992) Effects of plant variation on herbivore-natural enemy interactions. In: Fritz RS, Simms EL (eds) Plant resistance to herbivores and pathogens: ecology, evolution, and genetics. University of Chicago Press, Chicago, pp 278–300

    Google Scholar 

  17. Harvey JA, van Dam NM, Gols R (2003) Interactions over four trophic levels: food plant quality affects development of a hyperparasitoid as mediated through a herbivore and its primary parasitoid. J Anim Ecol 72:520–531

    Article  Google Scholar 

  18. Harvey JA, van Dam NM, Witjes LMA, Soler R, Gols R (2007) Effects of dietary nicotine on the development of an insect herbivore, its parasitoid and secondary hyperparasitoid over four trophic levels. Ecol Entomol 32:15–23

    Article  Google Scholar 

  19. Holopainen M (1989) A study on the essential oil of tansy (Tanacetum vulgare L.). University of Helsinki, Finland

    Google Scholar 

  20. Holopainen M, Hiltunen R, Lokki J, Forsen K, Schantz MV (1987a) Model for the genetic-control of thujone, sabinene and umbellulone in Tansy (Tanacetum vulgare L). Hereditas 106:205–208

    CAS  Article  Google Scholar 

  21. Holopainen M, Hiltunen R, von Schantz M (1987b) A study on tansy chemotypes. Planta Med 53:284–287

    CAS  Article  PubMed  Google Scholar 

  22. Huang M, Abel C, Sohrabi R, Petri J, Haupt I, Cosimano J, Gershenzon J, Tholl D (2010) Variation of herbivore-induced volatile terpenes among Arabidopsis ecotypes depends on allelic differences and sub cellular targeting of two terpene synthases, TPS02 and TPS03. Plant Physiol 153:1293–1310

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  23. Johnson MTJ (2008) Bottom-up effects of plant genotype on aphids, ants and predators. Ecology 89:145–154

    Article  PubMed  Google Scholar 

  24. Johnson MTJ, Agrawal AA (2005) Plant genotype and environmental interact to shape a diverse arthropod community on evening primrose (Oenothera biennis). Ecology 86:874–885

    Article  Google Scholar 

  25. Jordán F, Gjata N, Mei S, Yule CM (2012) Simulating food web dynamics along a gradient: quantifying human influence. PLoS One 7(7):e40280. doi:10.1371/journal.pone.0040280

    Article  PubMed  PubMed Central  Google Scholar 

  26. Kapás Á, András CsD, Dobre TG, Vass E, Székely G, Stroescu M, Lányi Sz, Ábrahám BÉ (2011) kinetics of essential oil separation from fennel by microwave hydro distillation (MWHD). UPB Sci Bull Series B 73(4):127–130

    Google Scholar 

  27. Keskitalo M, Linden A, Valkonen J (1998) Genetic and morphological diversity of Finnish tansy (Tanacetum vulgare L., Asteraceae). Theor Appl Genet 96:1141–1150

    Article  Google Scholar 

  28. Linhart YB, Keefover-Ring K, Mooney KA, Breland B, Thompson JD (2005) A chemical polymorphism in a multitrophic setting: thyme monoterpene composition and food web structure. Am Nat 166:517–529

    Article  PubMed  Google Scholar 

  29. Mantyla E, Tero K, Erkki H (2004) Attraction of willow warblers to sawfly damaged mountain birches: novel function of inducible plant defences? Ecol Lett 7:915–918

    Article  Google Scholar 

  30. Mehrparvar M, Mahdavi Arab N, Weisser WW (2013) Diet-mediated effects of specialized tansy aphids on survival and development of their predators: is there any benefit of dietary mixing? Biol Control 65:142–146

    Article  Google Scholar 

  31. Mestre L, Bucher R, Entling MH (2014) Trait-mediated effects between predators: ant chemical cues induce spider dispersal. J Zool 293(2):119–125

    Article  Google Scholar 

  32. Mooney KA, Halitschke R, Kessler A, Agrawal AA (2010) Evolutionary trade-offs in plants mediate strength of trophic cascades. Science 327:1642–1644

    CAS  Article  PubMed  Google Scholar 

  33. Mumm R, Dick M (2010) Variation in natural plant products and the attraction of bodyguards for indirect plant defence. Can J Zool 88(7):628–667

    CAS  Article  Google Scholar 

  34. Pierre PS, Jansen JJ, Hordijk CA, van Dam NM, Cortesero AM, Dugravot S (2011) Differences in volatile profiles of turnip plants subjected to single and dual herbivory above- and belowground. J Chem Ecol 37:1–10

    Article  Google Scholar 

  35. Poelman EH, van Loon JJA, Dicke M (2008) Consequences of variation in plant defense for biodiversity at higher trophic levels. Trends Plant Sci 13:534–541

    CAS  Article  PubMed  Google Scholar 

  36. R Core Team (2013) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna. ISBN 3-900051-07-0. Accessed 7 June 2014

  37. Rasmann S, Tobias G, Köllner J, Degenhardt IH, Stefan T, Ulrich K, Jonathan G, Ted CJ (2005) Recruitment of entomopathogenic nematodes by insect-damaged maize roots. Nature 434:732–737

    CAS  Article  PubMed  Google Scholar 

  38. Rohloff J, Mordal R, Dragland S (2004) Chemotypical variation of tansy (Tanacetum vulgare L.) from 40 different locations in Norway. J Agric Food Chem 52:1742–1748

    CAS  Article  PubMed  Google Scholar 

  39. Roth S, Knorr C, Lindroth RL (1997) Dietary phenolics affects performance of the gypsy moth (Lepidoptera: Lymantriidae) and its parasitoid Cotesia melanoscela (Hymenoptera: Braconidae). Environ Entomol 26:668–671

    CAS  Article  Google Scholar 

  40. Runyon JB, Mescher MC, De Moraes CM (2006) Volatile chemical cues guide host location and host selection by parasitic plants. Science 313:1964–1967

    CAS  Article  PubMed  Google Scholar 

  41. Schmitz OJ (2008) Herbivory from individuals to ecosystems. Annu Rev Ecol Evol Syst 39:133–152

    Article  Google Scholar 

  42. Schmitz OJ (2010) Resolving ecosystem complexity. Monographs in population biology. Princeton University Press, Princeton

    Google Scholar 

  43. Snoeren TAL, Kappers IF, Broekgaarden C, Mumm R, Dicke M, Bouwmeester HJ (2010) Natural variation in herbivore-induced volatiles in Arabidopsis thaliana. J Exp Bot 61:3041–3056

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  44. Soler R, Harvey JA, Kamp AFD, Vet LEM, Van der Putten WH, Van Dam NM, Stuefer JF, Gols R, Hordijk CA, Bezemeret TM (2007) Root herbivores influence the behaviour of an aboveground parasitoid through changes in plant-volatile signals. Oikos 116:367–376

    CAS  Article  Google Scholar 

  45. Stadler B, Dixon AFG (2008) Mutualism: ants and their insect partners. Cambridge University Press, New York

    Google Scholar 

  46. Van der Putten WH, Vet LEM, Harvey JA, Wäckers FL (2001) Linking above- and belowground multitrophic interactions of plants, herbivores, pathogens, and their antagonists. Trees 16:547–554

    Google Scholar 

  47. Whitham TG, Gehring CA, Lamit LJ, Wojtowicz T, Evans LM, Keith AR, Smith DS (2012) Community specificity: life and afterlife effects of genes. Trends Plant Sci 17:271–281

    CAS  Article  PubMed  Google Scholar 

  48. Wolf VC, Gassmann A, Clasen BM, Smith AG, Muller C (2012) Genetic and chemical variation of Tanacetum vulgare in plants of native and invasive origin. Biol Control 61:240–245

    CAS  Article  Google Scholar 

Download references


This research was founded by the Institute of Research Programs of the Sapientia University, grant no. 1/13/05.01.2012. The authors declare no conflict of interest.

Author contribution statement

W. W. W. and A. B. conceived the experiments. A. B., K. B. and J. B. designed the experiments. A. B., K. B., J. B., R. V. S. and M. M. performed the experiments. S. E. Z., W. W. W., O. J. S. and A. B. analysed the data. A. B., W. W. W. and O. J. S. wrote the manuscript. S. E. Z. provided editorial advice.

Author information



Corresponding authors

Correspondence to Klára Benedek or Adalbert Balog.

Additional information

Communicated by Nina Farwig.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary material 1 (JPEG 168 kb)

Supplementary material 2 (DOCX 180 kb)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Bálint, J., Zytynska, S.E., Salamon, R.V. et al. Intraspecific differences in plant chemotype determine the structure of arthropod food webs. Oecologia 180, 797–807 (2016).

Download citation


  • Aphids Ants
  • Bottom-up effects
  • Interactions
  • Predation
  • Top-down effects