1. Our aim was to explore plant–soil feedback in mixed grassland communities and its significance for plant productivity and community composition relative to abiotic factors of soil type and fertility.
2. We carried out a 4-year, field-based mesocosm experiment to determine the relative effects of soil type, historic management intensity and soil conditioning by a wide range of plant species of mesotrophic grassland on the productivity and evenness of subsequent mixed communities.
3. The study consisted of an initial soil conditioning phase, whereby soil from two locations each with two levels of management intensity was conditioned with monocultures of nine grassland species, and a subsequent feedback phase, where mixed communities of the nine species were grown in conditioned soil to determine relative effects of experimental factors on the productivity and evenness of mixed communities and individual plant species performance.
4. In the conditioning phase of the experiment, individual plant species differentially influenced soil microbial communities and nutrient availability. However, these biotic effects were much less important as drivers of soil microbial properties and nutrient availability than were abiotic factors of soil type and fertility.
5. Significant feedback effects of conditioning were detected during the second phase of the study in terms of individual plant growth in mixed communities. These feedback effects were generally independent of soil type or fertility, and were consistently negative in nature. In most cases, individual plant species performed less well in mixed communities planted in soil that had previously supported their own species.
6.Synthesis. These findings suggest that despite soil abiotic factors acting as major drivers of soil microbial communities and nutrient availability, biotic interactions in the form of negative feedback play a significant role in regulating individual plant performance in mixed grassland communities across a range of soil conditions.
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.
In recent years, there has been a growing awareness among ecologists of the importance of plant–soil feedback as a driver of plant community dynamics, especially in the context of plant succession and invasion (Bever, Westover & Antonovics 1997; Klironomos 2002; Van Der Putten 2003; Van der Heijden, Bardgett & Van Straalen 2008), and ecosystem processes such as nitrogen and carbon cycling (Manning et al. 2006; Van der Heijden, Bardgett & Van Straalen 2008; Bardgett, DeDeyn & Ostle 2009). By altering the physical, chemical and biological nature of their soil environment, individual plants have the ability to influence their performance relative to their competitors, ultimately leading to changes in plant community composition and diversity. A plant species may influence its associated soil biological community, or other abiotic soil properties, in one of two broad ways, leading to either a positive or negative feedback: a given plant species may alter its soil environment in a way that increases its own growth rate relative to that of other plant species, resulting in a positive feedback; or, a plant species might alter soil in a way that decreases its own growth rate relative to that of others, resulting in a negative feedback (Bever, Westover & Antonovics 1997). Positive plant–soil feedback involves enhanced soil nutrient availability via stimulation of soil microbes involved in mineralization processes or promotion of mycorrhizal fungi that enhance plant nutrient uptake (Klironomos 2002). In contrast, negative feedback involves the accumulation of parasites, pathogens and herbivores of roots (Bever, Westover & Antonovics 1997; Klironomos 2002), which can remove carbon and nutrients from plant tissue and reduce root uptake capacity in a species-specific manner, resulting in qualitative differences in plant community composition (Van Der Putten 2003; Wardle et al. 2004). Plant–soil feedback describes the net effect of these co-occurring events, namely positive and negative effects, since they are not occurring in isolation.
Most studies on plant–soil interactions demonstrate either plant effects on soil properties or effects of soil biological properties on plant growth (Wedin & Tilman 1990; Wardle et al. 2004), while studies that attempt to trace the feedback effects are relatively under-represented (Ehrenfeld, Ravit & Elgersma 2005). Moreover, most studies that have explored feedback effects have looked at plant performance in one or two species mixtures in soil conditioned by con-specifics and heterospecifics (Van der Putten, Van Dijk & Peters 1993; Bever 1994; Klironomos 2002). As a result, little is known about how soil conditioning affects plant performance and community dynamics in mixed plant communities (Ehrenfeld, Ravit & Elgersma 2005; Bezemer et al. 2006). The need for further studies of plant–soil feedback in the context of mixed communities was highlighted by Kardol, Bezemer & Van der Putten (2006), who grew mixed communities on soils of different successional stages and showed that temporal variations in plant–soil interactions contributed profoundly to plant community structure and development. Also, in a glasshouse study, Manning et al. (2008) showed that a range of grassland species generated differences in the soil environment that were sufficiently marked to affect the performance of individual plant species in mixed plant communities and hence alter community structure.
Little is also known about the generality of plant–soil feedbacks and their importance in shaping plant communities relative to soil conditions, and whether the incidence and outcome of plant–soil feedback for community properties varies across soils (Ehrenfeld, Ravit & Elgersma 2005; Bezemer et al. 2006; Casper et al. 2008). Previous studies have shown that effects of individual plant species on soil microbial communities can vary with soil fertility (Bardgett et al. 1999; Innes, Hobbs & Bardgett 2004), and it is well-established that the nature of interactions between plants and soil microorganisms depend on soil conditions (Marschner, Crowley & Yang 2004; De Deyn et al. 2009). In terms of plant–soil feedback, however, the picture is less clear. For example, in a glasshouse study Manning et al. (2008) showed that soil N enrichment modifies the outcome of plant–soil feedback relationships in mixed annual plant communities, albeit subtly, whereas Casper et al. (2008) found that the incidence of plant–soil feedback among four grass species was independent of soil type. Given this uncertainty, there is a need for more studies to test how plant–soil feedback varies with soil conditions, including variation in soil fertility.
The overarching aim of this study was to explore the significance for plant productivity and community composition of plant–soil feedback in mixed communities of temperate grassland species relative to abiotic factors such as soil type and fertility, resulting from historic management intensity. We hypothesized that plant–soil feedbacks will act as a significant driver of plant productivity and community structure in mixed grassland communities due to occurrence of positive or negative feedback, and that these effects will be strong enough to override those of soil type and differences in soil fertility resulting from historic management. This was tested in a large-scale mesocosm experiment, established in the field and run over four growing seasons, which consisted of: (i) an initial soil conditioning phase over two growing seasons, whereby soil taken from two locations each with two levels of management intensity (fertilized and unfertilized) was conditioned with monocultures of nine different grassland species covering three functional groups (grasses, herbs and legumes); and (ii) a subsequent feedback phase of two growing seasons, where mixed communities of the same nine species were grown in conditioned soil to determine relative effects of the three experimental factors (soil type, historic management intensity and conditioning by individual species) on the productivity and evenness of mixed communities. As noted by Casper et al. (2008), most previous studies on plant–soil feedback have been carried out under glasshouse conditions, often using soil inocula collected from soils conditioned by different plant species over relatively short time-scales (Bever 1994; Klironomos 2002; Bezemer et al. 2006). Here, we tested the role of plant–soil feedback under more realistic conditions, using field soils collected from different grassland sites and time-scales that permitted longer term effects of soil conditioning on plant communities to develop.
Materials and methods
Vegetation treatments were established in two contrasting soil types taken from permanent grassland sites at two different locations in northern England, namely a clay loam taken from Allendale, Northumberland, UK (54°54′ N, 2°15′ W), and a free-draining sandy alluvial soil taken from Nafferton, Northumberland, UK (54°1′ N, 0°23′ W) (see Table S1 in Supporting Information for site characteristics). At both locations, soils were taken from adjacent fields on the same soil type and topography, but which had been subject to different long-term management regimes, namely intensively managed fertilized grassland and adjacent unimproved grassland with no known history of artificial fertilizer application. All grasslands were grazed throughout the growing season by dairy cattle and occasionally sheep, although grazing pressures will have been historically greater in the intensively managed than unimproved grasslands, and this will have contributed to initial differences in soil biological communities and nutrient availability (Bardgett et al. 2001). Hereafter, we refer to these grasslands as improved and unimproved, respectively. At both locations, soils of improved grassland had greater concentrations of nitrate, ammonium and dissolved organic nitrogen (DON) than unimproved grassland soils, indicative of greater nutrient availability, although total soil C and N did not differ at the start of the experiment (see Table S1). All soils were collected during April 2004 and then transported back to Lancaster University field station where the experiment was established.
Phase 1: conditioning phase
Mesocosms were constructed using square-sectioned pots, with a capacity of 42 L (Bardgett et al. 2006). A 10-cm layer of carboniferous limestone chippings was placed in the base of each pot. In May 2004, pots were filled with unsieved soil (c. 30 L) and left for 1 month to remove species within the seed bank. Then, seeds of nine individual plant species (three grasses, three herbs and three legumes) were sown to create monocultures of each species, namely, Anthoxanthum odoratum, Festuca rubra, Lolium perenne, Achillea millefolium, Plantago lanceolata, Ranunculus repens, Lotus corniculatus, Trifolium pratense and Trifolium repens. The range of species and functional groups reflects those that are known to be successful in restoration schemes (Pywell et al. 2003) and occur across a wide range of British mesotrophic grassland types (Rodwell 1992). They were also selected to cover a spectrum of different growth strategies ranging from the fast-growing species L. perenne and T. repens, which are typically associated with low-diversity, high-fertility grassland (Warren 2000), to the slower-growing species A. odoratum, T. pratense and R. repens, which are common in high-diversity, low-fertility grassland (Smith et al. 2003, 2008) (Table S2).
Seeds were sown into replicate (n = 4) pots containing the two soil types each with two different management intensities, i.e. improved and unimproved. The experimental design yielded 144 pots including nine species in two soil types with two levels of management history. Treatments were arranged in a randomized block design with four replicates of each treatment. All pots were placed on weed matting and surrounded by rabbit-proof fencing to prevent damage from herbivores. Plants were allowed to grow over two growing seasons, and shoot material was harvested in September 2005 to measure above-ground biomass. In December 2005, soils were sampled by taking five individual, randomly positioned cores per pot (1.5 cm diameter × 7 cm depth), which were bulked to form a composite sample to test for conditioning effects of plants on soil microbial and nutrient measures. Soil was passed through a 5-mm sieve and stored at 5 °C before analysis.
In this phase, we tested how conditioning by individual plant species in phase 1 affected plant community dynamics and the performance of the conditioning species relative to other species in mixed communities. In April 2006, as much tissue as possible of the conditioning plant species, including roots from the soil surface, was removed from the pots with minimal disturbance. Into each pot, a mixed community of all nine species was then planted as rooted seedlings (see Table S2), including the conditioner species. Plants were assigned a random location in a grid of 6 × 6 positions, with each replicate having a different configuration, following the design of De Deyn et al. (2003). Each pot, therefore, contained four randomly assigned individuals of each of the nine species. Plants were left to grow for the following two growing seasons, with above-ground plant biomass being harvested in September 2006 and July 2007. Here, we used data from July 2007 when above-ground plant material was sorted to individual plant species and air-dried prior to weighing to enable effects of soil conditioning, soil type and historical management intensity on community evenness and the performance of the conditioner species to be detected.
Data were analysed with statistix v8.1 (Analytical Software, Tallahasse, FL, USA). For the first phase of the experiment, the effect of conditioning species on above-ground biomass of monocultures, soil microbial properties and nutrient availability was analysed using a three-way anova followed by Tukey’s post hoc test. The independent variables were conditioning species, historical management intensity (improved and unimproved) and soil type (clay loam or free-draining alluvial); block was included in the model as a random effect. The dependent variables were nitrate, ammonium, DON, microbial biomass C and N, microbial respiration, total PLFA, total bacterial PLFA, total fungal PLFA and fungal : bacterial PLFA. For the second phase of the experiment, feedback effects on individual and total above-ground plant biomass and community evenness (Pielou’s species evenness index ‘J’) were analysed using a three-way ancova followed by Tukey’s post hoc test. The independent values were as above, and DIN and rate of N mineralization were included as covariables. Dependent variables were normalized prior to analysis using log10 transformation for all plant species biomass data (A. odoratum, F. rubra, L. perenne, A. millefolium, P. lanceolata, R. repens, L. corniculatus, T. pratense and T. repens). Data on total above-ground biomass and community evenness did not require transformation. To calculate plant species performance in home (i.e. species growing in soil conditioned by the same species) versus away (i.e. species growing in soil conditioned by any of the other eight species tested) soil, the following calculation was carried out: ((biomass in away soil − biomass in home soil)/biomass in home soil)/100.
Phase 1: soil conditioning
Shoot biomass was significantly, and almost equally, affected by historic management intensity and conditioning plant species (F1,105 = 69.93, P <0.0001 and F8,105 = 68.27, P <0.0001, respectively). Plants grown in soil collected from improved grassland had a significantly greater biomass than those grown in unimproved soil across both soil types (Tables 1 and S3), confirming the greater fertility of the former. In terms of plant species differences, P. lanceolata had significantly greater biomass than all other species across all soils, whereas R. repens had a significantly lower biomass than all other species (Tables 1 and S3). Above-ground plant biomass also differed between the two soil types (F1,105 = 45.37, P <0.0001), being greater across all species in the clay loam soil than the alluvial soil (Tables 1 and S3).
Table 1. Summary of a three-way anova looking at the effects of conditioning species, management intensity and soil type on above-ground biomass of monocultures and soil nutrient and microbial measures (phase 1 – conditioning effects). Bold values represent significant responses at the P ≤ 0.05 level
Conditioning species × management intensity
Conitioning species × soil type
Management intensity × soil type
Conditioning species × management intensity × soil type
Microbial biomass C
Microbial biomass N
Total bacterial PLFA
Total fungal PLFA
Fungal : bacteria PLFA
Soil nutrient availability
For most species tested, greater nutrient availability was found in the clay loam soil, but when P. lanceolata was planted, no difference in ammonium between the soil types was detected (Fig. 1a) (F1,104 = 5.56, P <0.0001 for the conditioning species × soil type interaction). Similarly, when A. odoratum and A.millefolium were grown, no difference in nitrate between the clay loam and the free-draining alluvial soil was detected (Fig. 1b) (F1,104 = 4.41, P =0.0001 for the conditioning species × soil type interaction). A conditioning species × management intensity interaction was also detected for ammonium (F1,104 = 2.14, P =0.0384), which was greater in improved than unimproved soil when planted with P. lanceolata and T. pratense (Fig. 1c).
Overall, soil type was the most important factor affecting soil nutrient availability. Soil concentrations of ammonium (F1,104 = 191.77, P <0.0001), nitrate (F1,104 = 72.85, P ≤ 0.0001), DON (F1,103 = 9.37, P =0.0028) and the potential rate of N mineralization (F1,105 = 6.64, P =0.0113) were all significantly greater in the clay loam than the sandy alluvial soil (Tables 1 and S4). Significant differences in soil nutrient availability were also detected in the soils of different management intensities. As expected, soil ammonium (F1,104 = 24.60, P < 0.0001), nitrate (F1,104 = 9.15, P =0.0031) and DON (F1,103 = 8.77, P =0.0038) were all significantly greater in the improved than the unimproved soil (Tables 1 and S4). Conditioning species had less of an impact on these soil properties, although soils conditioned with the legumes T. pratense and T. repens had significantly greater concentrations of ammonium (F1,104 = 8.93, P <0.0001) and nitrate (F1,104 = 18.15, P <0.0001), and rates of N mineralization (F1,105 = 3.76, P =0.0060) than many of the other species tested (Table 1, Fig. 1a,b,d).
Soil microbial properties
Soil type was the most important factor affecting the majority of soil microbial properties tested. Microbial biomass C (F1,105 = 68.62, P <0.0001) and N (F1,105 = 127.76, P < 0.0001) and the total abundance of PLFA (F1,105 = 163.71, P <0.0001) were all significantly greater in the clay loam soil than the sandy alluvial soil (Tables 1 and S4). Consistent with this, the abundance of fungal and bacterial signature PLFA’s (F1,105 = 9.29, P =0.0029 and F1,105 = 239.98, P < 0.0001, respectively) were greater in the clay loam than the sandy alluvial soil, whereas the ratio of fungal : bacterial PLFA (F1,105 = 8.40, P =0.0046) was greater in the sandy alluvial soil than the clay loam (Tables 1 and S4).
Consistent differences in soil microbial properties were also detected between the soils of different management intensity. For example, when data were integrated across soil types and plant species, microbial biomass C (F1,105 = 6.12. P = 0.0149) and N (F1,105 = 32.71, P <0.0001), microbial respiration (F1,105 = 6.75, P =0.0107), total PLFA (F1,105 = 9.72, P =0.0024) and the total abundance of bacterial specific PLFA’s (F1,105 = 11.53, P =0.0010) were all significantly greater in improved than unimproved soil (Tables 1 and S4). While conditioning species had a lesser effect on these soil properties, significant differences between plant species were detected (Table 1, Fig. 2a–d). Soil conditioned by A. millefolium had significantly greater microbial biomass C (F1,105 = 3.91, P =0.0005), microbial respiration (F1,105 = 5.74, P <0.0001), total fungal PLFA (F1,105 = 3.45, P =0.0015) and ratio of fungal : bacterial PLFA (F1,105 = 4.04, P = 0.0003) than many of the other species. For example, soils conditioned by A. odoratum, R. repens and T. repens had consistently lower values for microbial measures than those conditioned with A. millefolium (Fig. 2a–d). No significant interactions were detected between conditioning species, management intensity or soil type for any of the soil microbial measures tested.
Phase 2: feedback effects on plant biomass and evenness
Community above-ground biomass and evenness
Management intensity was the most important factor affecting total above-ground plant biomass after the feedback phase of the experiment (F1,78 = 12.55, P =0.0007, Table 2); when data were integrated over all conditioning species and soil types, total above-ground biomass was significantly greater in the improved soil than the unimproved (means of 129 and 108 g, respectively). Soil type also affected this measure, in that significantly greater total above-ground biomass was detected in clay loam than alluvial soils conditioned with A. odoratum, F. rubra, L. perenne, P. lanceolata, R. repens, L. corniculatus and T. repens (F8,78 = 2.49, P =0.0183 for the conditioning species × soil type interaction; Table 2). However, there was no significant difference in above-ground biomass between clay loam and sandy alluvial soils when soils were conditioned with A. millefolium or T. pratense (Fig. 3a).
Table 2. Summary of a three-way ancova looking at the effects of conditioning species, management intensity and soil type on total above-ground biomass, community evenness and above-ground biomass of nine individual plant species grown in a mixed community (phase 2 – feedback effects). Rates of nitrogen mineralization and dissolved inorganic nitrogen (DIN) were included in the model as covariates. Bold values represent significant responses at the P ≤ 0.05 level.
Conditioning species × management intensity
Conitioning species × soil type
Management intensity × soil type
Conditioning species × management intensity × soil type
Rate of N mineralization
Total above-ground biomass
Plant community evenness was most significantly affected by the covariable DIN (F1,78 = 7.68, P =0.007), and a significant negative relationship between plant community evenness and soil DIN was detected (F1,142 = 14.58, P =0.0003, r = −0.11, Pearson’s correlation).When soils were conditioned with A. odoratum, P. lanceolata and T. repens, community evenness was significantly greater in the unimproved soils. However, there was no significant difference in community evenness between the two management intensities for all other species tested (Fig. 3b) (F8,78 = 2.93, P =0.0065 for the conditioning species × management intensity interaction).
Individual plant species performance
Plant biomass within a mixed community was significantly affected by soil type for three of the nine species tested: A. odoratum, L. perenne and P. lanceolata (Table 2). Above-ground biomass of each of these species was significantly greater in improved over unimproved free-draining alluvial soil; however, in the clay loam soil, biomass was not significantly different in improved than unimproved soil (Fig. 4a–c; F1,77 = 13.87, P =0.0004; F1,77 = 9.16, P =0.0034; and F1,77 = 4.91, P =0.0296 for the management intensity × soil type interaction, respectively). Significant effects of historic management intensity were also detected for A. millefolium (F1,77 = 7.51, P =0.0076); this species had significantly greater biomass in a mixed community when grown in soil from improved than unimproved grassland (6 and 4 g, respectively), however, soil type did not affect this measure.
Conditioning by individual plant species had significant effects on subsequent growth of A. odoratum, L. perenne, A. millefolium, P. lanceolata, L. cornuliculatus, T. pratense and T. repens in mixed communities (Table 2, Fig. 5a–i). Anthoxanthum odoratum produced less above-ground biomass (i.e. performed less well) in soils conditioned by itself than all other species studied, except F. rubra, whereas L. perenne produced less shoot biomass in home soil than in soil conditioned by any other species (Fig. 5a,c). Achillea millefolium grew significantly better in terms of biomass when grown in soils conditioned by A. odoratum than by itself (Fig. 5d), while P. lanceolata grew less well in its own soil relative to that conditioned by L. perenne, A. millefolium and T. pratense (Fig. 5e). Lotus corniculatus grew significantly less well when grown in soil conditioned by itself and by A. millefolium and T. pratense, than in soil conditioned by all other species tested (Fig. 5g), whereas T. pratense grew significantly less well in soil conditioned by itself than that conditioned by F. rubra, L. perenne, A. millefolium and P. lanceolata (Fig. 5h). Finally, T. repens grew significantly less well in soil conditioned by itself and T. pratense, than all other species tested (Fig. 5i).
Shoot biomass of F. rubra was significantly greater in improved than unimproved soil when conditioned by itself (F8,75 = 2.60, P =0.0146 for the conditioning species × management intensity interaction; Table 2). However, when soil was conditioned with L. perenne, biomass of this species was significantly greater in unimproved than improved soil (Fig. 6a). A similar interaction was detected for P. lanceolata (F8,77 = 2.71, P =0.0111 for the conditioning species × management intensity interaction; Table 2), which produced more shoot biomass in improved than unimproved soil when soil was first conditioned with A. odoratum or L. perenne. However, there was no significant difference in biomass of P. lanceolata between improved and unimproved soils when soil was conditioned with any of the other species tested (Fig. 6b). Another significant interaction was also detected for P. lanceolata (F8,77 = 2.64, P =0.0131 for the conditioning species × soil type interaction; Table 2), which produced more shoot biomass in the free-draining alluvial soil than the clay loam soil when conditioned by T. pratense. However, when soil had been conditioned with either P. lanceolata or L. corniculatus, the biomass of this species was significantly greater in the clay loam over the alluvial soil (Fig. 6c).
Dissolved inorganic nitrogen, a covariable included in the model, significantly affected the biomass of F. rubra to a greater extent than any of the other variables tested (F1,75 = 20.79, P <0.0001); a significant positive relationship was detected between these two factors (F1,140 = 4.32, P <0.0001, r = 0.34).
Individual plant species exerted strong effects on soil microbial communities and nutrient availability during the first phase of this study. Although the existence of plant species effects on grassland soil biological properties is well known (Bardgett, Wardle & Yeates 1998; Grayston et al. 1998; Wardle et al. 1999; Porazinska et al. 2003; Innes, Hobbs & Bardgett 2004; De Deyn et al. 2009), we found that responses to individual plant species were of much less importance than were the effects of soil abiotic conditions (i.e. soil type and fertility). Despite this, and consistent with our hypothesis, we found significant plant–soil feedback effects on individual plant growth in mixed communities during the second phase of the study. Moreover, these feedback effects were generally consistent across contrasting soil conditions and were always negative in nature, i.e. individual plant species consistently grew less well in mixed communities planted in soil that had previously supported their own species across both soil types and management regimes The importance of negative feedback in regulating individual plant performance and competitive interactions is well-established and mostly attributed to a build-up of species-specific pathogens (Bever 1994; Klironomos 2002; Bezemer et al. 2005; Kardol, Bezemer & Van der Putten 2006). Our findings extend this knowledge by showing that negative feedback consistently affects individual plant performance in mixed plant communities growing on soils of markedly different texture and fertility, and that these responses occur despite the overriding importance of abiotic factors in shaping microbial communities and nutrient availability in grassland soil.
Soil conditioning phase
A key finding of the first phase of the experiment was that soil abiotic factors, namely soil type and variation in fertility resulting from historic management, were the main determinants of below-ground properties For example, we found that microbial biomass (C and N) and the abundance of PLFAs synthesized by bacteria and fungi were consistently greater in the clay loam than sandy alluvial soil, whereas the ratio of fungal-to-bacterial phospholipid fatty acids, a measure of the relative abundance of these microbial groups (Bardgett, Hobbs & Frostegård 1996; Frostegård & Bååth 1996), was greater in sandy than the clay loam soil. These differences reflect the powerful role that inherent differences in soil physical properties of these soils, including texture, pH and moisture status, play as drivers of soil community abundance and structure (Bardgett 2005) This is in agreement with the work of Fierer & Jackson (2006) and Fierer et al. (2009), who found that diversity and richness of soil bacterial communities differed by ecosystem type, and that these differences were explained mostly by differences in soil pH. However, they also probably reflect differences in nutrient availability (inorganic N and DON) and plant production, which were both significantly greater in the clay loam than the sandy alluvial soil. Indeed, several studies report a lower ratio of fungal-to-bacteria biomass in higher fertility grassland soils, indicating a greater role for the bacterial-energy channel in these grasslands (Bardgett & McAlister 1999; Bardgett et al. 2001; De Vries et al. 2006). Moreover, across both soil types, microbial abundance was greater in the improved than the unimproved soils, again suggesting that nutrient availability and plant production, which were both greater in improved soil, are a key determinant of soil microbial community structure in grassland. These findings indicate, therefore, that variation in soil abiotic conditions, including differences in texture, moisture content, pH and nutrient availability, act as the primary drivers of below-ground properties in grassland (Harrison & Bardgett 2004; Risch & Frank 2006).
Despite the above, conditioning effects of different plant species on measures of soil N availability and microbial community structure were detected. Unsurprisingly, the legumes T. pratense and T. repens increased rates of N mineralization and the availability of inorganic N in soil relative to the grass species, especially A. odoratum and F. rubra, and also the herb A. millefolium. The two legumes also significantly affected microbial community structure, causing a significant reduction in the soil fungal-to-bacterial PLFA ratio, indicative of a shift towards a bacterial-based energy channel that is typically associated with rapid rates of nutrient cycling (Bardgett et al. 2006; Van der Heijden, Bardgett & Van Straalen 2008). The herb A. millefolium also had marked effects on the soil microbial community and nutrient cycling, again demonstrating the potential for individual species to modify the size and structure of the soil microbial community. While relatively rare, we also detected significant interactions between plant species and other experimental factors for some measures, indicating that, in some situations, plant species effects on soil properties and nutrient cycling vary with soil conditions. For instance, the positive effect of the legumes on soil N availability, measured as soil nitrate and ammonium concentration, was stronger in the more fertile clay loam than the infertile sandy soil. Collectively, these findings indicate that although plant species can differentially influence soil microbial properties and nutrient cycling, these effects are of less importance than soil abiotic factors in driving these below-ground properties.
Plant–soil feedback effects
At the end of the feedback phase, historic management intensity was identified as the most important factor affecting above-ground biomass of mixed communities, being greater in more fertile improved than infertile unimproved soil. Despite treatment responses of plant production in the feedback phase, community evenness, a measure of the relative abundance of component species, was largely unaffected by conditioning species, management intensity or soil type. This may have been due to the design of the experiment in that all communities were initially sown to the same evenness, and potentially the experiment did not continue long enough for this to be significantly affected by the factors studied here. However, community evenness did differ when soil was conditioned with the species A. odoratum, P. lanceolata and T. repens; when soil had been conditioned by these species during phase 1, the subsequent mixed plant communities were of greater evenness in unimproved than in improved soil. We found the predominant factor affecting community evenness was the availability of inorganic N in soil, and as expected, evenness was generally greater in soils with lower concentrations of DIN.
Although community-level responses to plant conditioning were not detected, the biomass of most of the individual plant species tested was significantly affected by the conditioning species previously grown in the soil. Performance in away versus home soil clearly showed that for each of the seven plant species where significant responses to conditioning were identified, a negative feedback was detected, i.e. when grown in a mixed community, these seven species (especially the grasses A. odoratum and L. perenne) produced less biomass when the soil had been conditioned by themselves relative to when conditioned by one of the other species. Importantly, we also found that these negative feedback responses, which are most likely due to the accumulation of species-specific pathogens (Van der Putten, Van Dijk & Peters 1993; Bever 1994; Klironomos 2002), were predominantly consistent across soils and management histories. These data indicate that while soil microbial communities, nutrient availability and overall plant community biomass are most strongly affected by soil abiotic factors, biotic factors associated with conditioning species and negative feedback play a more important role in affecting individual plant performance in mixed grassland communities.
Little is known about the extent that plant–soil feedback effects are modulated by soil conditions, and the few studies that have been carried out on this topic yield contrasting results (Bezemer et al. 2005; Casper et al. 2008). Moreover, most, but not all, studies that have explored feedback effects have looked at plant performance in one or two species mixtures over relatively short time-scales and responses from glasshouse experiments (Van der Putten, Van Dijk & Peters 1993; Bever 1994; Klironomos 2002). Our findings therefore advance understanding by showing that plant–soil feedback influences the performance of a wide range of individual plant species in mixed grassland communities over relatively long time-scales, and that these responses are generally consistent across a wide range of soil conditions and are mostly negative in nature. These findings support the notion that negative feedback is a general phenomenon in mixed species grassland communities, despite the overriding role that soil abiotic factors play in shaping microbial production, nutrient cycling and plant production in temperate grassland. In drawing these conclusions, it is important to note that our studies were performed under relatively artificial conditions in mesocosms. Therefore, field studies, which include other controls on plant community structure, are required to fully understand the significance of plant–soil feedback processes in shaping below-ground and above-ground communities in grassland.
This work was funded by a grant (DIGFOR) awarded to R.D.B. by the UK Department for Environment, Food and Rural Affairs (DEFRA), as part of their species-rich grassland research programme. We thank Richard Brand-Hardy and Val Brown for their support of our work in this programme. We are also grateful to Roger Smith, Helen Quirk, Eva Tregidgo, Dan Wright, Gerlinde De Deyn, Emma Tukey and Phil Hobbs for technical assistance, and two anonymous referees and the Handling Editor for helpful comments on the manuscript.