Plant-mediated effects of soil invertebrates and summer drought on above-ground multitrophic interactions


Correspondence author. E-mail:


1. We conducted a large-scale field study to determine how the interactive effects of earthworms (Aporrectodea caliginosa) and summer drought affected plant communities containing barley (Hordeum vulgare), shepherd’s purse (Capsella bursa-pastoris) and common groundsel (Senecio vulgaris), and how such effects then influenced populations of the aphid Rhopalosiphum padi and its parasitoid, Aphidius ervi.

2. Total biomass of plant communities decreased by around 25% with summer drought, but was increased by the presence of earthworms by over 11%. The effects of drought and earthworms on H. vulgare differed in monocultures and mixed plant communities. In monocultures, earthworms had the biggest impact, increasing plant biomass, but drought had the biggest effect in multi-species communities, causing a decrease in biomass.

3. Drought had an extremely negative impact on S. vulgaris shoot and root biomass, but this was ameliorated in the roots when earthworms were present. Capsella bursa-pastoris was not significantly affected by drought or earthworms. Drought caused a significant increase in shoot nitrogen concentrations in all plants, including H. vulgare (from 19.2 to 23.8 mg g−1).

4. In total, 35 234 aphids were recorded, of which 3936 were parasitized. Drought conditions reduced aphid abundance by over 50%. The interaction was moderated by earthworms, which caused further declines in R. padi populations under drought conditions, most notably in monocultures.

5. The plant-mediated effects of drought and earthworms (negative and positive, respectively) on R. padi had cascading effects on the parasitoid, Aphidius ervi, which declined in abundance with reduced numbers of R. padi. In addition, drought had a negative impact on A. ervi abundance beyond its impacts on aphid density, suggesting reduced prey quality as well as quantity.

6.Synthesis. This study demonstrated the effect of predicted climate change (i.e. reduced summer rainfall) on plant-mediated interactions between earthworms and above-ground multitrophic groups. These effects were seen to differ between monocultures and multi-species plant communities, suggesting that changes in above-ground–below-ground linkages in response to drought may influence plant communities in the future.


The occurrence of extended periods of summer drought is predicted to increase in central and Southern Europe according to current climate change models (Christensen et al. 2007). Changed abiotic conditions such as the occurrence droughts can disrupt interactions between plants and invertebrates, thereby altering terrestrial food webs (Emmerson et al. 2004; Staley et al. 2007). Drought can result in an increase in the concentration of foliar nitrogen, which has led to the suggestion that it benefits population growth in insect herbivores (White 1984; Mattson & Haack 1987). However, a more recent meta-analysis found that the majority of phloem-feeding invertebrates responds negatively to host plant water stress, which may be due to a loss of turgor pressure under drought that makes phloem sap less available (Huberty & Denno 2004). Drought can also change the balance of carbon and nitrogen in the soil, potentially resulting in an increase in soil carbon concentration (Garten, Classen & Norby 2009).

Detritivores have the potential to affect above-ground food webs through plant-mediated interactions, as they can cause large increases in plant growth and performance due to enhanced nutrient availability in the soil. For example, the biomass of Lolium perenne and Plantago lanceolata was increased by 50% when earthworms were added (Wurst & Jones 2003). In 79% of studies earthworms increased shoot biomass (Scheu 2003). There are several possible mechanisms for these effects, including mineralization of organic matter, stimulation of soil microbes or symbionts, control of pests and parasites, and changes to soil porosity (Scheu 2003).

Earthworms can increase the amount and quality of food resources available to foliar herbivores and, consequently, their parasitoids (Bezemer et al. 2005). For example, the fecundity of the aphid Myzus persicae feeding on Cardamine hirsuta and the concentration of nitrogen in the shoots were increased in the presence of earthworms (Wurst & Jones 2003). Poveda et al. (2005) compared the effects of decomposers and root herbivores on Sinapis arvensis growth and seed production. Decomposers increased plant and seed biomass, while root herbivores reduced plant biomass and the number of seeds produced per plant (Poveda et al. 2005). The abundance of aphids was increased by decomposers, and the number of aphid parasitoids increased at the same rate as aphid populations (Poveda et al. 2005). Although shoot biomass is usually increased by earthworms in well-watered plants, there was no effect of earthworms on biomass in plants subjected to drought (Blouin, Lavelle & Laffray 2007). Earthworms can reduce photosynthesis (assimilation and transpiration) in drought conditions as they may increase soil drainage and thereby exacerbate the effects of drought (Blouin, Lavelle & Laffray 2007).

In addition to interacting with individual foliar herbivores and parasitoid species, decomposers have the potential to alter interactions between plant species, thereby potentially affecting succession and plant diversity. Earthworms increase the growth of grass species in particular, while legumes barely respond to their presence (Eisenhauer et al. 2009; Laossi et al. 2009). These studies suggest that soil invertebrates are capable of altering the composition of plant communities, specifically the competitive balance between different functional groups. If such shifts in competitive balance occur, plant assemblages of different composition, ranging from monocultures to multi-species communities, may be differentially affected by the presence of detritivores and changes in abiotic soil conditions.

The effects of climate change on interactions between soil invertebrates, plants and above-ground invertebrates have not been widely addressed (Van der Putten et al. 2009). The number of potential interactions is overwhelming, so Van der Putten et al. (2004) suggest that plants are assigned to functional groups to determine above-ground–below-ground interaction characteristics, as a way to facilitate the search for general patterns in the effects of climate change on these interactions. The lack of empirical information in this area was the impetus for this study, one which uses a relatively simple community as a study system incorporating two plant functional groups (grasses and forbs) and three invertebrate functional groups (a decomposer, a herbivore and its natural enemy).

This large-scale field study addressed how both drought and soil decomposers affected plant communities and the abundance of herbivores and their natural enemies above ground. We specifically aimed to test whether the effects of summer drought on plants, and the abundances of aphids and their natural enemies, were modified by the presence of decomposers (the earthworm Aporrectodea caliginosa). Our study system used plant communities containing barley (Hordeum vulgare), shepherd’s purse (Capsella bursa-pastoris) and common groundsel (Senecio vulgaris). Hordeum vulgare often has a relatively simple trophic community associated with it (Hawes et al. 2009), so the impacts of both abiotic (drought) and biotic (earthworms) soil parameters above ground can be detected and elucidated. The bird–cherry oat aphid (Rhopalosiphum padi) L. (Homoptera: Aphididae) is normally one of the dominant aphid species present (Blackman & Eastop 2000), which often promotes populations of generalist parasitic wasps (Aphidius spp.) (Haliday, Hymenoptera: Aphidiinae) (Powell 1982).

We hypothesized that (i) drought will negatively affect plant growth, resulting in reduced aphid densities and, consequently, lower parasitoid abundance; (ii) earthworms will positively affect plant growth and thereby promote populations of aphids and parasitoids on well-watered plants, but will intensify the negative effects of drought stress; (iii) the presence of earthworms will have smaller impacts on the abundance of aphids and their parasitoids in a multi-species plant community compared with a grass monoculture.

Materials and methods

Experimental Procedure

The field experiment was conducted on a south-facing field site at Scottish Crop Research Institute (56°27′20″ N, 3°4′17″ W) in Dundee, UK, during August 2009. The site was surrounded by a wire mesh fence to prevent mammalian herbivory. Approximately 1.7 tonnes of soil (eutric cambisol) was excavated, air-dried and sieved to < 4 mm to remove larger debris and macro-invertebrates. Twenty-five rain shelters, similar to those described by Yahdjian & Sala (2002), were erected at the site. These comprised four vertical struts that supported a corrugated Perspex sheet (90 × 90 cm) at a 15° angle (to disperse rainfall), which was positioned 90 cm above ground at the lowest point. Each shelter was enclosed within an insect-proof mesh tent (1.2 × 1.2 m2) which also reduced incidental rainfall from the sides of the shelter. Shelters contained four 15-L plant pots containing 13 kg of soil that had been lined with thick nylon mesh to prevent escape of earthworms. On 1 July 2009, 12 plants (c. 14 days old) were transferred to each pot; two pots contained a monoculture of H. vulgare (cv. Optic) and two pots were multi-species communities comprising eight H. vulgare, two C. bursa-pastoris (ecotype 367, SCRI, UK) and two S. vulgaris (91792, Herbiseed, Twyford, UK). The two forb species were inter-planted amongst the H. vulgare seedlings. Twelve shelters were randomly assigned as drought treatments and the remaining 13 were assigned as ambient treatments. Based on the 10-year average weekly rainfall during July–August at the site (18.63 mm), drought treatments received 50% less rainwater than ambient treated shelters. This is in line with projections for the UK by 2080 based on current circulation models (UKCP09, 2010). Rainfall was collected at the site and delivered to the pots twice per week with dosage being calculated on the basis of pot surface area; 60 mL week−1 for drought treatments and 120 mL week−1 for ambient treatments. After 17 days, adult earthworms (A. caliginosa), c. 80 mm long, were excavated from the field site (eutric cambisol soil) using mustard extraction (see Gunn 1992 for details), and assessed for good health according to Fründ et al. (2010). Four worms were added to one of the monoculture pots and four to the multi-species pots. The entire front of the mesh tents was opened to allow ingress of flying and walking invertebrates. Soil water content was measured every 3–4 days using profile probes and a moisture meter (Delta-T Devices, Cambridge, UK).

Plants were harvested 6 weeks after being transferred to pots. Aphids and mummies (parasitized aphids) were manually removed from all plants using an electric pooter. Shoot and roots were excavated to measure fresh mass. Plant material was frozen in liquid nitrogen and freeze-dried prior to being re-weighed and analysed for carbon and nitrogen concentrations. Carbon and nitrogen concentrations were determined by flash combustion and chromatographic separation of c. 1.5 mg of ground and homogenized shoot material, which was calibrated against a standard (C26H26N2O2S) using an elemental combustion system (Costech Instruments, Milan, Italy). Full details of analysis in Johnson et al. (2010).

Statistical Analysis

Split-plot analysis of variance (anova), with shelter and pot fitted as block terms, were used to analyse whether plant and aphid responses were affected by (i) drought and (ii) earthworm presence. For H. vulgare responses, a further factor of (iii) plant community (monoculture or multi-species) was initially included in the model, but was removed for subsequent analysis due to non-significance. Interaction terms were included in all analyses. Hordeum vulgare root mass and insect abundance were log+1-transformed prior to analysis. The number of parasitized aphids was analysed with an analysis of covariance (ancova), in which aphid abundance was included as the covariate. All factors (i–iii) and interactions between factors were included in analyses of aphid and parasitoid abundance. All tests were conducted in Genstat (version 12, VSN International, UK).


The mean daily temperature ranged from 11 °C to 20 °C during the course of the field experiment, with temperatures under the rain shelters being approximately 1.0 °C higher than ambient. The average soil water content in the top 5 cm was 23.9% under ambient irrigation and 9.6% under drought conditions. Other than by drought treatments (i.e. irrigation regime), soil water content was not affected by earthworm or plant composition treatments.

Plant Responses

In total, 1165 plants were harvested at the end of the experiment; only 2.9% of plants died during the experiment. The total biomass of plant communities (shoots and roots) was significantly reduced under drought conditions and marginally (= 0.05) increased by the presence of earthworms (Fig. 1). Total plant biomass was higher in multi-species plant communities, but there was a significant interaction between plant community composition and earthworm presence, whereby earthworms would cause an increase in plant biomass in monocultures, but not in multi-species communities (Fig. 1).

Figure 1.

 The effects of drought and earthworms on total plant biomass (mean dry mass per pot ± SE) in (a) monoculture and (b) multi-species plant communities. Ambient (inline image) and drought (inline image) conditions shown with (E+) and without (E−) the presence of earthworms. N = 13 for ambient irrigation and N = 12 for drought treatments. Statistically significant results (< 0.05) highlighted in bold.

When the responses of individual plant species were considered, drought had a statistically significant negative effect on H. vulgare dry mass (F1,23 = 4.96, = 0.036), which was largely due to reductions in shoot biomass (F1,23 = 5.54, = 0.027) rather than root biomass (F1,23 = 1.25, = 0.276). The presence of earthworms enhanced plant growth (F1,69 = 4.96, = 0.036), again in the shoots (F1,69 = 4.00, = 0.049) rather than the roots (F1,69 = 1.85, = 0.178). When responses were analysed separately for the two types of community, H. vulgare responded to drought and earthworm presence differently depending on whether they were grown in monoculture or multi-species communities (Fig. 2). When H. vulgare was grown in monocultures, earthworm presence significantly increased total plant biomass (Fig. 2a) and shoot biomass (Fig. 2b), but not root biomass. In contrast, when H. vulgare was grown in multi-species communities, then drought became the more significant influence, with statistically significant reductions in total (Fig. 2c) and shoot biomass (Fig. 2d) being observed, but not so in the roots.

Figure 2.

 The effects of drought and earthworms on Hordeum vulgare biomass (mean dry mass per plant ± SE) when grown in (a–b) single and (c–d) multi-species plant communities. Statistically significant differences (highlighted in bold) were seen and total (a and c) and shoot (b and d) biomass, but not root biomass. Shaded bars represent drought treatments, and the presence or absence of earthworms is indicated with E+ and E−, respectively. Sample size as Fig. 1. Statistically significant results (< 0.05) highlighted in bold.

The two forb species responded to the experimental treatments differently. In particular, S. vulgaris biomass was severely reduced by drought (Fig. 3) in both the shoots (Fig. 3b) and the roots (Fig. 3c). Earthworms lessened the detrimental effects of drought, however, with interactive effects of drought and earthworms being evident in the shoots (Fig. 3b) and roots (Fig. 3c). In contrast, C. bursa-pastoris biomass was not significantly affected by either earthworm presence or drought. In terms of community composition, there was no significant effect of drought (F1,23 = 0.21, = 0.648), earthworms (F1,23 = 0.62, = 0.441) or drought × earthworm effects (F1,23 = 1.07, = 0.311) on the relative biomass of H. vulgare and the forb species.

Figure 3.

 The effects of drought and earthworms on Senecio vulgaris biomass (mean dry mass per plant ± SE) according to (a) total biomass, (b) shoot biomass and (c) root biomass. Shaded bars represent drought treatments, and the presence or absence of earthworms is indicated with E+ and E−, respectively. Sample size as Fig. 1. Statistically significant results (< 0.05) highlighted in bold.

Drought caused large increases in shoot nitrogen concentrations across all three plant species (Fig. 4a), but had no effect on shoot carbon concentrations, except in C. bursa-pastoris where it caused a slight decrease (Fig. 4b). Nitrogen and carbon concentrations in H. vulgare shoots was similar, regardless of whether plants were grown in single or multi-species communities. Earthworms had no effect on shoot nitrogen or carbon concentrations.

Figure 4.

 The effects of drought on shoot concentrations (mean mg g−1 ± SE) of (a) nitrogen and (b) carbon in the three plant species subjected to ambient (open bars) and drought (shaded bars) rainfall patterns. Sample size for ambient and drought treated plants, respectively: Hordeum vulgare N = 52, 48; Senecio vulgaris and Capsella bursa-pastoris N = 13, 12. Statistically significant results (< 0.05) highlighted in bold.

Herbivore–Parasitoid Responses

In total, 35 234 aphids were recorded, of which 3936 were parasitized. Negligible numbers of aphids and mummies were found on the two forb species, and all of the aphids on H. vulgare were identified as R. padi. A subset of 20 mummies were reared out and the parasitoid identified as Aphidius ervi (Powell 1982).

Drought had a highly significant negative impact on aphid density above ground (Fig. 5a) in both monocultures and multi-species plant communities. There was a significant interactive effect of drought and earthworm presence, whereby the effects of drought on aphid abundance were exacerbated when earthworms were present, especially in the monocultures (Fig. 5a). When aphid responses were scaled for plant mass (i.e. aphids per H. vulgare dry mass), the interaction between the effects of drought and earthworms remained statistically significant (F1,69 = 7.07, = 0.010), but in addition earthworms had a negative effect on aphid abundance overall (F1,69 = 5.27, = 0.025), suggesting that earthworms may be reducing plant suitability for aphids.

Figure 5.

 Abundance of (a) Rhopalosiphum padi (b) Aphidius ervi on Hordeum vulgare plants (mean number of aphids per plant ± SE) subjected to drought (inline image) and ambient (inline image) rainfall with (E+) and without (E−) earthworms. N = 26 for ambient treatments, N = 24 for drought treatments. Statistically significant results (< 0.05) highlighted in bold.

The size of A. ervi wasp populations can be directly inferred from the number of mummies (i.e. parasitized bodies of aphids) (Omacini et al. 2001). Aphidius ervi abundance was closely correlated (rs = 0.345, < 0.001) with aphid abundance when it was included as a covariate in the analysis (F1,68 = 9.07, < 0.001), indicative of a positive relationship between prey density and parasitoid density (Fig. 5b). In addition, drought had a statistically significant negative effect on the abundance of A. ervi (Fig. 5b), over and above the negative effects of drought on aphid density. The proportion of aphids parasitized (i.e. parasitism rate) was not significantly affected by any of the treatments, although it was generally lower under drought conditions (F1,23 = 3.20, = 0.078).


This study has illustrated how projected climate change, in this case reduced summer rainfall, can interact with soil decomposers to affect both plants and herbivore–parasitoid populations above ground, producing a series of interactions between different components of the above- and below-ground communities, summarized in Fig. 6. Reduced summer rainfall in this case was 50% of ambient, which is line with projections for the UK by 2080 based on current circulation models (UKCP09, 2010).

Figure 6.

 Schematic summary of the major above-ground–below-ground interactions operating in multi-species plant communities and monocultures. (1) Earthworms positively affected Hordeum vulgare shoot growth in monocultures, but not multi-species communities. (2) earthworms ameliorated the negative effects of drought on Senecio vulgaris roots. (3) Drought reduced both root and shoot growth of S. vulgaris and also (4) root growth of H. vulgare in multi-species communities. In terms of trophic effects, (5) earthworms caused decreases in Rhopalosiphum padi abundance (scaled for plant size). (6) Drought had a negative impact on aphid density in both single and multi-species plant communities, which was (7) exacerbated by earthworm presence when plants were suffering from drought. (8) Reductions in aphid density reduced the abundance of Aphidius ervi, which was (9) further negatively affected by drought.

Plant Responses

Soil-dwelling decomposers, such as earthworms, often promote plant growth through several mechanisms, including mineralization of organic matter, stimulation of soil microbes and changes to soil porosity (Scheu 2003). In this study, we also found positive effects of A. caliginosa on growth of H. vulgare in monocultures (Fig. 6, mechanism 1), but these effects were absent when H. vulgare was part of a mixed plant community. The fact that the beneficial effects of soil invertebrates on H. vulgare are negated by the presence of competing plant species is interesting and suggests that, in this study system at least, grasses are less successful in capturing any resources released by the activities of earthworms when forbs are present. Aporrectodea caliginosa also had positive effects by ameliorating (Fig. 6, mechanism 2) the negative effects of drought (Fig. 6, mechanism 3) on S. vulgaris growth, but did not lessen the negative effects of drought (Fig. 6, mechanism 4) on H. vulgare in mixed plant communities. In terms of plant species, S. vulgaris was far more affected by drought than C. bursa-pastoris, with the former suffering a reduction in both root and shoot biomass under drought conditions and the latter showing no response. This may be due to differences in root architecture and rooting depth between the two species. Senecio vulgaris has a herringbone rooting structure with an average depth of 12 cm (Bernston & Woodward 1992), while C. bursa-pastoris has a tap-root with a deeper rooting depth of 33 cm (Aksoy, Dixon & Hale 1998), which may enable it to access water under drought conditions more effectively than S. vulgaris. Earthworms ameliorated the impacts of drought on S. vulgaris, contrary to our hypothesis, but had no such positive effect on H. vulgare. This may simply reflect the fact that S. vulgaris is the plant most severely affected by drought and thus benefits most from any impacts of earthworms, which have the potential to reduce drought stress, such as an increase in soil nitrogen availability (Scheu 2003).

Herbivore Responses

Studies have demonstrated how earthworms can influence herbivore (especially aphid) populations through plant-mediated mechanisms, with examples of positive (e.g. Wurst & Jones 2003; Poveda et al. 2005), neutral (e.g. Bonkowski et al. 2001) and negative (e.g. Ke & Scheu 2008) effects being reported. In the present system, A. caliginosa showed some tendency to promote populations of R. padi on H. vulgare under ambient rainfall conditions, but this was not significant and, once effects of plant size were taken into account, earthworm impacts on R. padi were negative (Fig. 6, mechanism 5), particularly under drought conditions in monocultures. Drought caused a marked decline in R. padi populations (Fig. 6, mechanism 6), and this was exacerbated by earthworms (Fig. 6, mechanism 7). Drought often has very negative effects on insect herbivores, particularly sap-sucking insects such as aphids (Huberty & Denno 2004), although it can sometimes make plants more susceptible to insect attack by reducing resistance and increasing plant nutritional quality (White 1984; Koricheva, Larsson & Haukioja 1998). Drought caused significant increases in shoot nitrogen in H. vulgare in the current study, which has also been widely reported for many other plants under water stress (Mattson 1980; White 1984) and often reflects increases in free amino acids arising from protein degradation and reduced protein synthesis (Brodbeck & Strong 1987). Increased concentrations of shoot nitrogen did, however, not have any beneficial effects on R. padi, which may indicate that increased levels of nitrogen were either not accessible or compositionally unsuitable, or both. In particular, drought is likely to reduce turgor pressure and increase phloem sap viscosity, which would make it more difficult for sap-sucking insects to access (Raven 1983). Moreover, drought often induces non-essential amino acids (e.g. proline) rather than essential dietary amino acids (Brodbeck & Strong 1987), which may be of less benefit to the aphid. Hale et al. (2003) reported that R. padi performance was reduced on water-stressed plants (albeit non-significantly for H. vulgare) and concluded that while amino acid concentrations increased, lower sap ingestion rates on drought stressed plants resulted in reduced performance overall.

The mechanism by which earthworms exacerbated the adverse effects of drought on aphids remains unclear. Earthworms can reduce assimilation, transpiration and stomatal conductance of Graminaceae species under drought conditions, which may be due to a reduced soil water storage capacity when earthworms are present (Blouin, Lavelle & Laffray 2007). Although plant biomass was not reduced by earthworms in the drought treatment, it is possible that plant quality for the aphids was affected by reduced assimilation and stomatal conductance.

Parasitoid Responses

Several studies have demonstrated that plant-mediated effects of earthworms on aphid abundance can have cascading effects on aphid parasitoids, often mediated through changes in prey density (Wurst & Jones 2003; Bezemer et al. 2005). Here we also found that changes in R. padi density were positively correlated with changes in the abundance of A. ervi (Fig. 6, mechanism 8), which further illustrates the importance of linkages between soil biota and above-ground multitrophic interactions. In this system, drought also had an additional indirect negative impact of A. ervi abundance beyond the effects of drought on R. padi numbers (Fig. 6, mechanism 9). This suggests that drought might be negatively affecting prey quality as well as quantity, whereas existing studies that consider tri-trophic interactions between above- and below-ground organisms (e.g. Poveda et al. 2005), have only identified correlations between herbivore and parasitoid densities. This highlights how climate perturbation may cause many nonlinear responses in above-ground–below-ground interactions.


This study illustrates that under summer drought conditions the outcome of interactions between detritivores, plants and above-ground foliar herbivores can be changed markedly. Further work would be needed to determine if this is a general pattern or whether climate change effects differ with the system under study. Very recent studies (Erb et al., 2010) have also illustrated the importance of water relations in mediating such above-ground–below-ground interactions. Given the different responses we observed between just three plant species examined here, it seems likely that effects will be contingent on a range of factors including plant functional group, as Van der Putten et al. (2004) suggest, whether species are growing alone or in competition with other species and the sequence with which plants are challenged (Erb & Robert, 2010).

The importance of investigating the effects of soil invertebrates in the context of multi-species plant communities is clearly demonstrated by this study. The effects of earthworms and drought on plant biomass differed between the H. vulgare monoculture and the multi-species plant community. Previous studies addressing interactions between soil and above-ground invertebrates have only used one plant species, or several plant species each grown in separate pots. In contrast, this study demonstrated how the impacts of soil conditions on above-ground food differed between monocultures and multi-species communities. The study of interactions between plants, soil invertebrates and above-ground invertebrates has been an active research area for about 20 years. However, thus far the effects of climate change have not been incorporated into many studies of these interactions (Van der Putten et al. 2009), nor have any studies addressed the interacting effects of climatic factors and soil invertebrates on above-ground trophic groups in mixed plant communities. This study shows that the use of multi-species experimental systems (discussed by Van Dam & Heil, 2010) is critical to understanding how climate change may affect plant-mediated interactions between above- and below-ground communities in the future.


The authors would like to thank Lewis Fenton, Sarah Reid and Phil Dyer for their assistance with fieldwork, and in particular Steven Gellatly for his monumental patience in counting invertebrates. Thanks also to Naomi Ewald who performed the chemical analysis and to Clare Morton for assistance with the graphics. We would also like to thank Pete Iannetta for providing the C. bursa-pastoris seed and Cathy Hawes for useful discussions. S.N.J. acknowledges funding from the Scottish Government’s Rural and Environment Research and Analysis Directorate Workpackages 1.7 and 1.3 for this research.