• Open Access

Soil microbial community structure of range-expanding plant species differs from co-occurring natives


  • Elly Morriën,

    Corresponding author
    • Department of Terrestrial Ecology, Netherlands Institute of Ecology (NIOO-KNAW), 6700 AB Wageningen, The Netherlands
    Search for more papers by this author
  • Wim H. van der Putten

    1. Department of Terrestrial Ecology, Netherlands Institute of Ecology (NIOO-KNAW), 6700 AB Wageningen, The Netherlands
    2. Laboratory of Nematology, Wageningen University and Research Centre, 6700 ES Wageningen, The Netherlands
    Search for more papers by this author

Correspondence author. E-mail: e.morrien@nioo.knaw.nl


  1. Due to global warming and other changes in the environment, many native and exotic plant species show range expansion from lower to higher latitudes. In the new range, the (in)ability of range-expanding plants to establish associations with local soil microbes can have important consequences for plant abundance; however, very little information exists on rhizosphere communities of range-expanding plant species. Here, we examine the rhizosphere microbial community composition of range-expanding plant species in comparison with phylogenetically related species that are native in the invaded range.

  2. We tested the hypothesis that range-expanding plants species would promote fewer shifts in rhizosphere communities than congeneric natives would. In order to test this, soil was collected from the invaded habitat and six range-expanding and nine congeneric natives were planted individually in pots to condition soil microbial communities.

  3. After harvesting, individuals of the same species were planted in conditioned own and control soils to test the legacy effects of soil conditioning on biomass production. The control soils were mixtures of soils conditioned by all other plant species, except congenerics. After 10 weeks of plant growth, we determined the rhizosphere community composition of bacteria, fungi, arbuscular mycorrhizal fungi (AMF) and Fusarium spp.

  4. All groups of microbes were analysed qualitatively using denaturating gradient gel electrophoresis (DGGE). Ergosterol was determined as a quantitative measure of nonarbuscular mycorrhizal fungal biomass, and real-time PCR was applied to detect amounts of Fusarium spp.

  5. Range-expanding plants had less fungal hyphal biomass and lower amounts of Fusarium spp. in the rhizosphere than congenerics. Bacterial community composition was influenced by a combination of soil conditioning and plant origin, whereas fungal communities, AMF and Fusarium spp. were less pronounced in their responses to the experimental treatments.

  6. Synthesis. We conclude that the lack of legacy effects in range-expanding plant species compared with natives may be due to differences in bacterial rhizosphere community composition, or to different quantities of potential pathogenic fungi. If the range-expanding plant species were benefiting more from AMF, effects will not have been due to differences in community composition, but we cannot exclude other options, such as different effectiveness of AMF or other soil biota in the rhizosphere of range-expanding vs. native plant species. The greater accumulation of bacterial and fungal pathogens in the rhizosphere of natives in relation to range expanders might explain the successful establishment of range-expanding plants.


Many native and exotic plant species are expanding their ranges polewards, as a consequence of global climate warming (Walther et al. 2002; Parmesan & Yohe 2003; Pearson & Dawson 2003; Tamis 2005; Hickling et al. 2006). During range expansion, interactions between plants and higher-trophic-level organisms may become disrupted because dispersal rates can differ (van der Putten et al. 2004; van Grunsven et al. 2007; Thuiller, Richardson & Midgley 2007; Menendez et al. 2008). A consequence of unequal dispersal rates is that range-shifting plants can become released from their native enemies (van Grunsven et al. 2007, 2010; Engelkes et al. 2008). As many soil biota are relatively slow dispersers (Berg et al. 2010), soil communities in the new range may not necessarily contain pathogens, decomposers and symbiotic mutualists from the native range. When compared with native plants, enemy release may cause a competitive advantage for exotic plant species in the invaded habitat (Walker et al. 2003; van der Putten, Klironomos & Wardle 2007; Inderjit & van der Putten 2010), whereas nonadapted mutualists and decomposers may cause a disadvantage for plants that have moved away from their home fields (Richardson et al. 2000; Chapman & Koch 2007; Ayres et al. 2009; Pregitzer et al. 2010). The question that we will address in the present study is how the rhizosphere community is influenced by exotic range-expanding plant species in comparison with congeneric plant species that are native in the invaded habitat.

In a previous study, soil communities from the invaded range were less pathogenic to range-expanding plant species than to phylogenetically related species native to the invaded habitat (Engelkes et al. 2008). The lack of legacy effects, also called neutral feedback, between soil microbiota from the invaded range and exotic plants (Reinhart et al. 2003; Callaway et al. 2004; Bennett, Alers-Garcia & Bever 2006; Pringle et al. 2009) may be due to a shifted balance towards the relative absence of pathogens, as well as to the accumulation of mutualistic fungi in the roots and rhizosphere of exotic plants. Such net soil community effects are determined by measuring feedback effects after a phase of conditioning by the same or other plant species (Bever, Westover & Antonovics 1997). Negative feedback between plants and soil communities is a widespread phenomenon in native vegetation, and this so-called plant–soil feedback contributes to plant diversity and coexistence (Bever, Westover & Antonovics 1997; Klironomos 2002; Bever 2003; Petermann et al. 2008).

Exotic plants that have been moved to a different continent can be released from negative soil feedback (Klironomos 2002). This release has been proposed to promote plant invasiveness (Reinhart & Callaway 2006). Studies on above-ground pathogens of exotic plants have shown that recently introduced exotic species have fewer pathogens than exotic plant species with a longer residence time (Mitchell et al. 2010). When the time since invasion increases, soil feedback effects can become increasingly negative (Diez et al. 2010). Based on these studies, we would expect that the rhizosphere of recently introduced and currently range-expanding exotic plants may have a lower abundance of soil pathogens than phylogenetically related natives. Root exudates are the resources for rhizosphere bacteria and fungi, and it has been recognized that the composition of these substrates can differ among plant species (Nelson 1990). Differences in root-derived substrates are believed to explain the plant species-specific rhizosphere bacterial communities that have been observed for different plant species under otherwise similar conditions (Marschner et al. 2001; Kowalchuk et al. 2002; Carney & Matson 2005; Hartmann et al. 2009). Similar growth conditions are important when comparing rhizosphere communities (Berg & Smalla 2009; Eisenhauer et al. 2011) since root exudate composition is also affected by soil conditions such as pH, nutrient limitation, soil moisture and exposure to pathogens (Yang & Crowley 2000). Therefore, such comparisons between range-expanding and native plant species need to be made under otherwise controlled conditions.

Both bacteria and fungi respond to plant community composition and processes in the rhizosphere (Gomes et al. 2003; De Boer, Kowalchuk & Van Veen 2006). We examined differences in soil communities of general bacteria and fungi between soils conditioned by exotics and related natives and compared them to communities obtained from control soils. In addition, we examined specific effects of exotic range expanders and congeneric natives on two taxonomic groups of microbes: one including symbiotic mutualists (arbuscular mycorrhizal fungi; AMF) and the other including the genus Fusarium, which contains a number of economically important plant pathogenic species that can cause substantial damage to crop plants. They are a large genus of fungi widely distributed including relatively abundant members of the rhizosphere microbial community (Booth 1971). Therefore, Fusarium spp. may lend themselves well to intra-genus comparisons involving a larger number of plant species. In addition and in contrast to the host-specific nature of many pathogenic microbes, we included AMF, because many AMF taxa tend to infect a broad range of host plants (Eom, Hartnett & Wilson 2000; Moora et al. 2011).

In order to analyse the microbial community structure across a wider range of plant species, we compared the rhizosphere microbial community structure of six exotic range-expanding plant species with nine phylogenetically related natives. We tested the hypothesis that range-expanding plants species would promote fewer shifts in rhizosphere communities than congeneric natives would. We expected this because one of the possible causes of reduced plant–soil feedback effects of range-expanding plant species is that soil pathogens in the roots and rhizosphere of range-expanding plants will be mostly absent. This was reflected by smaller shifts in the general bacterial and fungal communities after a soil conditioning phase in range expanders when compared with natives.

Our main focus was on the interaction between soil treatments (soil conditioning effect) and plant origin (natives and range expanders). Additionally, this study provides the opportunity to assess the similarity between species within genera and among genera. Rhizosphere communities of own and control soils of range expanders and natives were analysed qualitatively using denaturating gradient gel electrophoresis (DGGE), as this method focuses on the most abundant microbial taxa. We determined ergosterol as a quantitative measure of nonarbuscular mycorrhizal fungal (non-AMF) biomass and applied real-time PCR to Fusarium spp. In line with our hypothesis, we expected to find lower amounts of ergosterol and Fusarium spp. in the rhizosphere of range-expanding plants species than of natives, because range-expanding plant species may benefit less from home-field advantage in decomposition, which might result in less promotion of decomposer fungi. Moreover, when fewer pathogens associate with range expanders, we also expected to observe a lower abundance of non-AMF biomass. We expected the AMF to show an opposite pattern to the Fusarium spp. based on the assumption that AMF are less specific than pathogenic fungi. We also expected AMF diversity to be similar amongst range expanders and natives.

Materials and methods

Plant species selection

We used the National Standard List of the Dutch flora and the updated version of this list (Tamis 2005) to select recently introduced exotic range-expanding plant species and phylogenetically related natives (Agrawal et al. 2005; Funk & Vitousek 2007) all co-occurring in a riverine habitat in eastern Netherlands (51°87′ N, 6°01′ E). Three exotic plant species originated from Eurasia: Artemisia biennis (North-Asia), Centaurea stoebe (Central Europe) and Angelica archangelica (North East-Europe). The other three range-expanding exotics originated from other continents: Bidens frondosa (North-America), Senecio inaquidens (South-Africa) and Solidago gigantea (North-America). We compared the exotic plants with congeneric native plant species (Bidens cernua, Bidens tripartita, Senecio viscosus, Senecio vulgaris, Artemisia vulgaris, Solidago virgaurea, Centaurea cyanus, Centaurea jacea and Angelica sylvestris). In three genera, we included two species in order to test the sensitivity of our analysis to species-specific effects on rhizosphere community composition. All our chosen plant species are known to associate with AMF.

Experimental set-up

In order to determine whether short-term plant species legacy effects already influenced rhizosphere microbial community structure, we began by conditioned the soil (phase I) and did the actual measurements in the soil feedback phase II. This is a common approach for plant–soil feedback experiments (Bever, Westover & Antonovics 1997).

Phase I: soil conditioning

Seventy-five 4-litre pots were filled with a 5:1 mixture of sterilized soil and living inoculum soil. We established five replicate pots of each plant species (six exotics and nine natives, resulting in 75 pots). Each pot received four seedlings to promote soil conditioning, and the experiment was carried out in a greenhouse. After 8 weeks of growth, the plants were harvested and the conditioned soils were used for a second growth phase to test the feedback effect of soil community conditioning.

Phase II: soil feedback

The conditioned soil from every pot in phase I of the growth experiment was split into two halves. One half was placed in a 1.3-litre pot and named ‘own’ soil. The other half was used to create control soils. The control soil of every plant species contained soil conditioned by all other plant species, excluding plants from the same genus. Because all controls shared soil from five genera, we assumed initial soil nutrient conditions to be similar for all control pots. We established five replicates with own and five with control soils (resulting in 150 pots). After week 10, the pots were harvested, and three quarters of the roots were collected to estimate the total amount of root dry biomass. Above-ground biomass and roots were air-dried at 70 °C for 48 h and weighed. Data for the individual plant species are presented in the study by Engelkes et al. (2008). In our study, we made use of one quarter of the soil originating from the experiment presented by Engelkes et al. (2008) to collect root and rhizosphere samples according to Kowalchuk et al. (2002). Briefly, we collected roots from the soil and removed the bulk soil by gently shaking the roots using tweezers. Rhizosphere soil was collected by brushing off the soil particles from the roots. The remaining roots with tightly attached soil particles were cut with scissors and stored in an Eppendorf tube at −80 °C. All equipment used was sterilized in between samples using 98% ethanol.

DNA isolation and PCR-DGGE analyses

We extracted genomic DNA from approximately 0.25 g (wet weight) of the root and attached soil particles using the PowerSoil™ DNA isolation kit (MoBio Laboratories, Carlsbad, CA, USA). Prior to isolation, root samples were ground in a mortar under liquid nitrogen. The composition of the fungal, AMF and Fusarium spp. community was studied using fungal 18S rRNA gene-specific primers. For bacterial composition, we used the 16S rRNA gene–specific primers (Table 1). Details regarding primers, thermocycling regimes and electrophoresis conditions are listed in Table 1, and all PCR was performed with a PTC-200 thermal cycler (MJ-Research, Waltham, MA. Each amplification reaction mixture (25 μL) consisted of 0.5 μL (30 pm) of each primer, 2.5 μL 10 × PCR-buffer, 2.5 μL (2 mm) of the dNTPs mix, 0.40 μL (0.056 U) Expand High Fidelity DNA polymerase (Roche, Mannheim, Germany), 1 μL template DNA (between 15–95 ng depending on the plant species) and 17.60 μL MilliQ (Millipore BV, Etten-Leur, the Netherlands).

Table 1. Primers, PCR and denaturating gradient gel electrophoresis (DGGE) conditions used in this experiment
CommunityPrimersPCR-protocolaDGGE gradientsReference
  1. a

    PCR protocols are given as annealing temperature and number of cycles. The remaining of the procedure is given in the text.

  2. b

    100% denaturant is defined as 40% (v/v) formamide and 7 m urea.



65 °C to 55 °C; 35 cycles

45–65% denaturantbHeuer et al. (1997)


55 °C to 47 °C; 37 cycles

40–55% denaturantbVainio & Hantula (2000)


Followed by FLR3/FLR4


1st 58 °C; 35 cycles

2nd 58 °C; 35 cycles

20–60% denaturantbGollote, van Tuinen & Atkinson (2004)
Fusarium spp.


Followed by Alfie1-gc/Alfie2


1st 50 °C; 29 cycles

2nd 67 °C; 34 cycles

40–60% denaturantbYergeau et al. (2005)

All thermocycling programs were preceded by an initial denaturation step (95 °C for 5 min) and followed by a final elongation step phase (72 °C for 10 min). Each PCR cycle consisted of a denaturation step at 95 °C for 1 min, an annealing step at the specified temperature (Table 1) for 1 min and an elongation step at 72 °C for 1 min. Touchdown protocols started with the highest annealing temperature, which was subsequently lowered by 2 °C after every two cycles until the target annealing temperature was reached. The amplicons of Fusarium spp. were diluted (1:1000) and reamplified in a second PCR round (Table 1). PCR products were examined by standard 1.5% (w/v) agarose 0.5 × TBE gel electrophoresis with ethidium bromide staining to confirm product integrity and estimate yield.

The composition of the fungal and bacterial communities was characterized using denaturating gradient gel electrophoresis (DGGE), using the method of Muyzer, Dewaal & Uitterlinden (1993) as modified by Kowalchuk et al. (2002), except that the linear gradients were as indicated in Table 1. To ensure well-polymerized slots, a 10-mL top gel containing no denaturant was added before polymerization was complete. Denaturating gels were prepared with the gradient former Bio-Rad model 230 (Bio-Rad Laboratories, Veenendaal, the Netherlands) at a speed of 5 mL min−1. A GC-rich sequence, also called GC clamp (indicated as -GC), was attached to one of the primers in the set to prevent complete melting of PCR products during separation in the denaturating gradient gel.

All DGGE analyses were run using a D-Gene system (Bio-Rad Laboratories, Hercules, CA, USA). Total PCR products (25 μL) were applied to the gel, and DGGE was performed in 1 ×  TAE buffer at 60 °C. Electrophoresis was carried out for 10 min at 200 V, after which the voltage was lowered to 70 V for an additional 16 h. In order to compare banding patterns, a reference marker was added in triplicate to each gel. Gels were stained in MilliQ water containing 0.5 mg L−1 ethidium bromide and destained in MilliQ water prior to UV transillumination. Gel images were digitally captured using the ImaGo system (B&L Maarsen, the Netherlands). To facilitate comparative statistical analyses, all gels of the same community were combined into a composite image using Corel PHOTO-PAINT 12 (Corel Corporation, Ottawa, Canada) prior to further analysis. Banding patterns were normalized with respect to standards of known composition as well as samples loaded across multiple gels using the Image Master 1D program (Amersham Biosciences, the Netherlands).

Real-time PCR

Real-time PCR was performed using the ABsolute QPCR SYBR green mix (AbGene, Epsom, UK) on a Rotor-Gene 3000 (Corbett Research, Sydney, Australia) to quantify Fusarium spp. 18S rRNA gene copies. The mixes were made using a CAS-1200 pipetting robot (Corbett Research, Sydney, Australia). Quantification of Fusarium gene copies in the rhizosphere was carried out using the Alfie1 and Alfie2 primers as described in Table 1 except that the forward primer lacked a GC clamp. Standards were made from a pure Fusarium spp. colony from which DNA was extracted using the PowerSoil™ DNA isolation kit (MoBio Laboratories, Carlsbad, CA, USA). DNA amount was quantified on a ND-1000 spectrophotometer (Nanodrop Technologies, Wilmington, DE, USA), and the number of gene copies μL−1 was calculated using the molecular weight of one genome as calculated from sequences deposited in GenBank. DNA extracts from the rhizosphere samples of the plants species used in our experiment were all diluted to 5 ng μL−1 of which 2.5 μL was used for real-time PCR. Using 10-fold increments, the standard concentrations were adjusted from 106 to 101 SSU rRNA gene copies μL−1. R2-values of the standard curve were 0.99 in all of the runs. All samples and standards were assessed in two different runs to confirm the reproducibility of the CT values. The results of the different runs were read out at a threshold of 0.52 where the efficiencies of the different runs were most comparable. All samples with a CT value lower than that of the lowest standard where regarded as undetectable and set to zero. Since we acknowledge that this might artificially increase the differences between the treatments, we also analysed the data by setting all undetectable samples on the value of the lowest standard at 10 copies μL−1.

Ergosterol measurements

We used the alkaline ergosterol extraction described by Bååth (2001) with minor modifications as described by de Ridder-Duine et al. (2006). Ergosterol is the dominating sterol in ascomycetes and basidiomycetes, while the more primitive taxa, except for Mucorales, contain other sterols. Ergosterol is not present in AM fungi (Olsson et al. 2003).

Statistical analyses

We analysed the DGGE gels using binary data, presence or absence of bands, generated by the Image Master 1D program (Amersham Biosciences, the Netherlands). The resulting binary matrices were exported and used in statistical analyses as ‘species’ presence–absence data. Multivariate tests of significance of the effects of soil treatment and plant origin in DGGE patterns were carried out using a principal coordinate analysis (PCoA) using a Jaccard index followed by a distance-based redundancy discriminant analysis (db-RDA) for the analysis of species, genus, soil treatment (own or control soil), plant origin (native or range expander) and interaction effects (Legendre & Anderson 1999). We implemented the experimental design as split plots (soil treatments) within a whole plot (plant species) using 499 random permutations in the Monte Carlo permutation test in canoco 4.5 (Ter Braak & Šmilauer 2002). When we were interested in the effects on the species level (species, genus and plant origin), we permuted whole plots, but not split plots. When we were interested in the effect of the soil treatment, we permuted the split plots, but not whole plots. The interaction between plant origin and soil treatment was analysed regarding species as fixed effects in a Welch test where the same split plots and whole plot were both permuted at random (Welch 1947). Data were visualized as RDA ordination scatter plots. When plotting samples, we made sure to derive the sample scores from the species data, which makes it possible to plot sample scores as variation around the class centroids (tested environmental variable). To test the effects of soil treatment and plant origin, these treatments were coded as dummy variables and used as environmental data in the RDA. In addition to the RDA, we performed a one-way analysis of similarities (ANOSIM) using PAST (PAlaeodontological STatistics version 2.06, Hammer, University of Oslo), which is a nonparametric test of significant differences between groups based on comparing distances between groups with distances within groups (Clarke 1993) using Jaccard as distance measure running 10000 permutations. The distances are converted to ranks. Pairwise ANOSIMs between all pairs of groups (using four groups: native own soil, native control soil, range expander own soil, range expander control soil) are provided as post hoc tests using x = 0.05 as significance cut-off.

To analyse the differences in number of Fusarium spp. copies per microlitre sample between soil treatments, origin and between plant species within origin, we used a nested anova. All variables were considered to be fixed. To analyse the differences in amounts of ergosterol between soil treatment, origin and between plant species within origin, we used a similar nested anova. All variables were considered fixed, including species, because we used all species that were available in our experimental site (Engelkes et al. 2008). In order to test for a relationship between the distance of RDA scores in own and control soil on the first axis and total dry weight biomass difference between own and control soil treatments of the plants, a Spearman rank order correlation was performed with the species as replicate units. All univariate statistical analyses were carried out in statistica 9 (StatSoft, Inc., Tulsa, USA). To improve normality and homogeneity of variances of residuals among groups defined by the statistical models, 18S rRNA gene copy number per microlitre sample and ergosterol values were natural log-transformed (x + 1) prior to analysis.


Profiling of bacteria, fungi, arbuscular mycorrhizal fungi (AMF) and Fusarium spp. communities illustrated that variation could be explained, at least in part, by plant genus (see Appendix S1 in Supporting Information, Fig. 1). However, there was no further effect of plant species (see Appendix S1). For all four groups of microbes the separation on genus level was significant, but the visual separation for general bacteria and fungi was more apparent than for AMF and Fusarium spp. (Fig. 1). In addition to differences among genera, there were significant effects of soil treatment (own vs. control soil) on the community structures of bacteria, fungi and Fusarium spp. (Table 2). However, there were no such differences in the AMF communities between own and control soils (Table 2). The DGGE profiles did not reveal an effect of plant origin (native and range expander) irrespective of the microbial groups examined (Table 2). However, there were significant interactions between soil conditioning and plant origin for bacteria, fungi and AMF, but not for Fusarium spp. (Table 2). The bacterial community in own vs. control soils differed more for natives than for range expanders (Fig. 2). Although less obvious, differences in fungal communities between own and control soils were also greater in natives than in range expanders (Fig. 2). However, for Fusarium spp., range expanders showed greater differences between own vs. control soils than natives, whereas for the AMF, control soils of natives and range expanders showed highly similar profiles but dissimilarity among natives and range expanders own soils was compounded by shifts in opposite directions along RDA axis 1 (Fig. 2). The variation exhibited amongst individual AMF samples in the individual plant species was considerably greater than in the other microbial groups (Fig. 2). There was no correlation between differences in RDA scores of soil microbial community and plant biomass determined in the growth experiment (see Engelkes et al. 2008) when comparing own and control soils: bacteria (n = 15; R2 = 0.30; P = 0.277), fungi (n = 15; R2 = −0.35; P = 0.201), AMF (n = 15; R2 = −0.19; P = 0.499) and Fusarium spp. (n = 15; R2 = −0.17; = 0.541) (data not shown). Thus, these differences in microbial community composition of the most dominant microbes, which are the ones picked up most likely by DGGE, are not indicative for general soil feedback effects.

Table 2. Results of the analyses using RDA and permutation tests showing the significance of the explanatory variables rhizosphere bacteria, fungi, arbuscular mycorrhizal fungi (AMF) and Fusarium spp. as affected by soil (own versus control soil), plant origin (range-expanding versus phylogenetically related native) and their interaction
Explanatory variablesCovariables BacteriaFungiAMFFusarium spp.
  1. Data are centred by species. No standardization by samples was done. Explanatory variables: environmental variables in CANOCO terminology. % expl. 1st axis: percentage of species variability explained by the first ordination axis, a measure of the explanatory power of the variables. r 1st axis: species–environment correlation on the first axis. F ratio: the F ratio statistics for the test on the trace. P: corresponding probability value obtained by the Monte Carlo permutation test using 499 random permutations. O, own soil treatment; C, control soil treatment; RE, range expander; N, native. The asterisk (*) between two terms indicates their interaction.

Soil treatment O, CNoneimage_n/jec12117-gra-0001.png% expl. 1st axis0.
r 1st axis0.4830.5150.4610.455
F ratio1.250.941.001.47
P 0.0040.0060.1040.008
Origin RE, NGenus 1, 2, 3, 4, 5 & 6image_n/jec12117-gra-0001.png% expl. 1st axis1.
r 1st axis0.6580.6840.5330.566
F ratio3.172.541.291.85
P 0.4200.5440.9640.932
Soil treatment * OriginAll species, O, Cimage_n/jec12117-gra-0001.png% expl. 1st axis0.
r 1st axis0.5520.7140.5700.598
F ratio1.751.321.701.82
P 0.0020.0500.0080.260
Figure 1.

Redundancy discriminant analysis ordination plots of denaturating gradient gel electrophoresis (DGGE) banding patterns (left panels) with species plotted as nominal environmental variables in a scatter plot. The first and second RDA axes are plotted on the x- and y-axes, respectively. The right panels show the variation of the individual sample scores around their class centroids (species) from the left panel in a similar RDA ordination scatter plot. Symbols of samples in the right panels equal those of class centroids in the left panels: Angelica (filled circle), Bidens (filled square), Senecio (filled up-triangle), Artemisia (open circle), Solidago (open square) and Centaurea (open up-triangle). Panels show (a) bacteria, (b) fungi, (c) AMF and (d) Fusarium spp..

Figure 2.

Redundancy discriminant analysis centroid scores of the nominal environmental variables on the first RDA axis shown for native and range-expanding plants. The RDA scores of the individual samples projected around their class centroids (soil treatment) on the first RDA axis are plotted as ‘own soil’ (open diamond) and ‘control soil’ (filled diamond) for each native and range-expanding plant species. The distance between own and control soil lines indicates the average difference between their soil communities. Panels show (a) bacteria, (b) fungi, (c) AMF and (d) Fusarium spp..

According to the ANOSIM for the bacteria, the communities within groups (native control soil, native own soil, range expander control soil, range expander own soil [see lines in Fig. 2)] were more similar than between groups (R = 0.0514; P = 0.002). Own soil in native plants differed significantly from own and control soil in range expanders (control soil: = 0.0011; own soil: = 0.0043). Moreover, the rhizosphere bacterial communities of native and range-expanding plants in control soil differed significantly from each other (= 0.0417) (Fig. 2). The fungi (R = −0.018; = 0.875), AMF community (R = 0.0152; = 0.146) and Fusarium spp. (R = −0.001; = 0.485) did not show similarity between the four plant origin and soil groups when using ANOSIM.

Native plant rhizospheres contained more fungal biomass than those of range expanders, based on ergosterol measurement (F1,83 = 31.79; < 0.001) (Fig. 3). There was no difference in fungal biomass between own and control soils either for natives or for range expanders (F1,83 = 2.62; = 0.11) (origin x soil interaction) (Fig. 3).

Figure 3.

Ergosterol (mg.kg−1) in roots and attached rhizosphere soil in native and range-expanding plants after growing in own soil (white bar) and control soil (grey bar). Bars show back-transformed means ± 1 SE of natural log+1 transformed data. Asterisk indicates significant differences.

Real-time PCR data revealed that the number of Fusarium spp. genome copies was significantly lower in the rhizosphere of range expanders than natives (F1,108 = 17.81; < 0.01). This effect remained significant even when all samples that fell below the detection limit were set at 10 copies μL−1 (F1,108 = 5.12; = 0.03) (Fig. 4). However, within plant origin (native or range expander), there were no differences in the number of genome copies between own and control soil, so that the observed biomass differences between own and control soil of the growth experiment cannot be attributed to Fusarium spp. infection.

Figure 4.

Number of Fusarium spp. genome copies per microlitre sample in native and range-expanding plants. Both plant origins have two treatments: own soil (white bar) and control soil (grey bar). Bars show back-transformed means ± 1 SE of natural log+1 transformed data. Asterisk indicates significant differences.


In the plant–soil feedback experiment, native plant species on average had less shoot and root biomass in own than in control soil, whereas phylogenetically related range expanders did not show such reduction. Moreover, the direction of the effect sizes of plant shoot and root biomass between own vs. control soil treatments was remarkably similar at the species level (between natives and range expanders). However, the direction of the effect sizes was different among genera (Engelkes et al. 2008). In the present study, we found that communities of bacteria, fungi, AMF and Fusarium spp. in the rhizosphere of these 15 plant species were influenced by plant genus, but there were no plant species-specific effects and, therefore, no range expander vs. native plant effects within genera. Interestingly, plant species-specific differences in rhizosphere community composition using the same techniques have been reported (Kowalchuk et al. 2002); however, in those studies, there was no particular emphasis on comparisons within genera. Most likely, the differences between plant species of the same genus in our experimental setting were too subtle to be picked up by fingerprinting techniques such as DGGE, whereas inter-genus effects were detectable. Therefore, in future studies on plant community composition as a driver of the microbial rhizosphere community (Marschner et al. 2001; Kowalchuk et al. 2002; Carney & Matson 2005; Hartmann et al. 2009), plant phylogenetic relationships may require more emphasis than has been received hitherto.

In line with our expectations, we found higher amounts of ergosterol in roots and rhizosphere soils of native plants than of range expanders. Although ergosterol measurements cannot distinguish between living and dead fungi (Mille-Lindblom, von Wachenfeldt & Tranvik 2004; Zhao, Lin & Brookes 2005), our data suggest that range expanders had less fungal hyphal biomass in and around their roots than native plants. This might have been caused by relatively fewer pathogens in and around the roots of range expanders. Although ergosterol levels varied between natives and range expanders, there were no systematic responses to soil type for either group. This might be because ergosterol measurements provide no information on species composition or ecology of fungi, so that this biomarker may have originated from saprophytes, beneficials (but not AMF) or pathogens.

Real-time PCR data of Fusarium spp. revealed that the number of Fusarium spp. genome copies was substantially lower in the rhizosphere of range expanders than of native plants. Although this would suggest – in line with our expectations – that range expanders are less exposed to potential pathogenic fungi, there were no differences in the number of genome copies in own soil and control soil within plant origin (native and range expander). If Fusarium spp. had been involved in causing negative plant–soil feedback effects of native plant species, we would have expected a different number of gene copies between own and control soil of native plants. However, it is also possible that due to their potential pathogenic nature, Fusarium spp. were rapidly promoted by native species but not by range expanders and that a preconditioning (e.g. own vs. control) had very little impact on their abundance in relation to the natural predisposition (native vs. range expander) of the plant.

There were significant effects of own vs. control soil on microbial community composition of bacteria, fungi and Fusarium spp., yet there were no such effects observed for AMF communities. This supports the view that AMF have relatively low species specificity, when compared with soil pathogens (Eom, Hartnett & Wilson 2000; Moora et al. 2011), although there are also contrasting views on mycorrhizal specificity (e.g. Klironomos 2003). DGGE did not reveal any effect of plant origin (native and range expander) on any of the microbial groups that were tested with DGGE. However, DGGE revealed significant interactions between soil conditioning and plant origin for bacteria, fungi and AMF due to significant soil treatment differences between origins. This means that taking the soil conditioning treatments into consideration, there are differences between native and range-expanding plant rhizosphere communities. Nevertheless, the interaction effects only explained a relatively small amount of the total variation, a small part of this unexplained variation might be a consequence of using binary data that usually result in a lower multivariate fit (Legendre & Legendre 1998). Since there was no correlation between the distances of the RDA scores in own and control soils in any of the microbial groups (based on DGGE patterns) and the biomass of the plants (measured as biomass differences between own and control soil), we could not show any causal linkages between plant growth reduction and rhizosphere microbial community composition. This would suggest that either there is no pattern, or the effects on the biomass of the plants were caused by rare microbial groups that require more subtle detection techniques than DGGE to show differences, such as high-throughput sequencing techniques.

Analysis of similarities revealed that the bacterial communities were more similar within each of the treatments (native plant species on control or own soil, and range-expanding plant species on control or own soil) than between these treatments, which is consistent with the findings based on the RDA displayed in Fig. 2. The similarity between native plants on own soils differed significantly from similarity between range-expanding plants in both own and control soil. This seems consistent with studies that have reported intraspecific plant–soil feedback effects (Newsham, Fitter & Watkinson 1995; Packer & Clay 2000; Mitchell & Power 2003; Sanon et al. 2009). Studies on plant species–specific soil communities may also be helpful in further unravelling why plant–soil feedback effects may operate more strongly on rare plants than on dominants (Klironomos 2002). The strong effect of native plants on general microbial community composition in their own soils may explain why species can experience a ‘home-field advantage’ in decomposition and nutrient cycling (Gholz et al. 2000; Chapman & Koch 2007; Ayres et al. 2009; Pregitzer et al. 2010; Suding et al. 2013).

The ANOSIM did not reveal treatment differences for fungi, AMF and Fusarium spp. The ANOSIM gave a slightly different outcome than the RDA with these microbial groups. This may be because ANOSIM is a robust nonparametric test that is not able to detect minor differences as RDA does, for example, it was not possible for ANOSIM to take into account the hierarchical structure of the design of the experiment. The patterns in the fungi and AMF might not be robust enough to be picked up by ANOSIM, due to the lower species specificity in AMF. For Fusarium spp., spores might have been absent from the soil inoculum community, since in a large number of samples, no PCR product was recovered. However, mycelium was present, as was revealed by real-time PCR data on Fusarium spp. This makes the analysis of fingerprinting data of Fusarium spp. less reliable, since the translation to binary data inflates small differences in the data set.

We conclude that the more neutral plant–soil feedback of range-expanding plant species when compared with congenerics that are native in the expanded range can be due to several responses in the soil biota. The community composition of soil bacteria showed more similarity between own and control soil for range expanders than for natives. This suggests that natives may accumulate more bacterial pathogen species than range expanders. The native plant species also accumulated more hyphal biomass of Fusarium spp., which is a group of fungi that contains potential pathogens. This would suggest that the negative plant–soil feedback of natives might also be due to enhanced densities of pathogens. Finally, AMF communities did not significantly differ between own and control soils for both range expanders and natives. This does not exclude the possibility of more AMF biomass, or different AMF activities, which could tip the balance from negative to neutral soil feedback in the case of range-expanding plant species. Accumulation of more bacterial and fungal pathogens in the rhizosphere of natives compared with range expanders might explain the relative success of invading plant species and sheds light on the mechanisms by which range-expanding plant species are successful in new habitats. This phenomenon will most probably be a temporal effect that may diminish over time as soils from invaded range may develop increasingly negative feedback effects when time since introduction proceeds (Diez et al. 2010).


We thank Tim Engelkes for the invaluable help in the set-up and development of the experiment; Staatsbosbeheer Regio Oost for allowing permission to collect soil and seeds in the Millingerwaard; Baudewijn Odé, Wil Tamis, Kees Groen, Arjen Biere, Martijn Bezemer, Jeffrey Harvey and the late Ruud van der Meijden for discussions; Wiecher Smant, Miranda Vlag, Lonneke Hensen and Freddy ten Hooven for chemical and molecular analyses; Agata Pijl, Tanja Bakx-Schotman, George Kowalchuk, Etienne Yergeau, Petr Šmilauer, Jan Lepš and Cajo ter Braak for advices during molecular data interpretation; and George Kowalchuk, Emilia Hannula and two anonymous reviewers for their helpful comments and suggestions on a previous version of the manuscript. This study was funded by ALW-VICI grant number 865.05.002 to WvdP. Publication 5455 Netherlands Institute of Ecology (NI00-KNAW).