No consistent legacy effects of invasion by giant goldenrod (Solidago gigantea) via soil biota on native plant growth

Abstract

Aims

Changes in soil microbial communities after occupation by invasive alien plants can represent legacy effects of invasion that may limit recolonization and establishment of native plant species in soils previously occupied by the invader. In this study, for three sites in southern Germany, we investigated whether invasion by giant goldenrod (Solidago gigantea) leads to changes in soil biota that result in reduced growth of native plants compared with neighbouring uninvaded soils.

Methods

We grew four native plant species as a community and treated those plants with soil solutions from invaded or uninvaded soils that were sterilized, or live, with live solutions containing different fractions of the soil biota using a decreasing sieve mesh-size approach. We measured aboveground biomass of the plants in the communities after a 10-week growth period.

Main Findings

Across all three sites and regardless of invasion, communities treated with <20 µm soil biota or sterilized soil solutions had significantly greater biomass than communities treated with the complete soil biota solution. This indicates that soil biota >20 µm are more pathogenic to the native plants than smaller organisms in these soils. Across all three sites, there was only a non-significant tendency for the native community biomass to differ among soil solution types, depending on whether or not the soil was invaded. Only one site showed significant differences in community biomass among soil solution types, depending on whether or not the soil was invaded; community biomass was significantly lower when treated with the complete soil biota solution than with soil biota <20 µm or sterilized soil solutions, but only for the invaded soil. Our findings suggest that efforts to restore native communities on soils previously invaded by Solidago gigantea are unlikely to be hindered by changes in soil microbial community composition as a result of previous invasion.

INTRODUCTION

Changes in belowground processes are increasingly invoked to explain why alien plant species become invasive and out-compete native species in the introduced range (Maron et al. 2014; Reinhart and Callaway 2006; Suding et al. 2013). Such changes could have long-lasting effects, and could even persist after the invasive species has been eradicated. However, relatively few studies have considered legacy effects, such as the impacts of changes in soil biota after invasion on native plant species establishment and growth (Dickie et al. 2014).

While invasive plant species might be released from specific soil-borne natural enemies in the introduced range, they may still accumulate generalist soil pathogens that have multiple plant hosts, with resulting negative effects on native plant species, e.g. pathogenic nematode accumulation under invasive Marram grass (Eppinga et al. 2006; Knevel et al. 2004) and Fusarium fungus accumulation under invasive Chromolaena odorata plants (Mangla et al. 2008). If the invasive plant is relatively tolerant to these enemies, the latter could then more negatively affect native species, leading to their exclusion (Mangla et al. 2008). Eppinga et al. (2006) described this as the ‘accumulation of local pathogens’ hypothesis. As well as representing a mechanism of invasion, such accumulation of pathogenic soil biota could prevent or reduce establishment of native plant species in soils formerly occupied by the invader. Moreover, such legacy effects could vary depending on species identity and the specific soil micro-organisms that have changed in abundance in invaded soils. This could then have consequences for native species composition in re-establishing communities. Potential legacy effects caused by changes in soil biota may have implications for efforts to restore native vegetation after invader removal.

Solidago gigantea (Asteraceae) is a tall perennial herb native to North America and is now invasive in much of Central Europe, often forming near-monospecific stands in a range of habitats (Weber 1998; Weber and Jakobs 2005). The species can alter ecosystem properties across habitats it has invaded (Scharfy et al. 2009). Solidago gigantea is known to increase soil C and P concentrations, suggesting greater rates of microbial mineralization in invaded soils (Chapuis-Lardy et al. 2006; Koutika et al. 2007), Soil pH can also be lower in invaded compared to uninvaded soils (Herr et al. 2007). Solidago gigantea is also known to benefit from association with arbuscular mycorrhizal fungi (Kytoviita et al. 2003). Furthermore, the species is on the blacklist of invasive species prohibited from sale and planting in Switzerland and is actively managed in Europe (Hartmann et al. 1995; Weber and Jakobs 2005). While efforts to restore previously invaded sites in Europe include mowing, mulching and reseeding of native species (Hartmann et al. 1995), it is not known whether changes in soil biota and in different fractions of soil biota (i.e. bacteria, pathogenic fungi, nematodes and mycorrhizal fungi), represent an important legacy effect of invasion. For example restoration efforts could be impeded through post-invasion changes in soil biota that have a net negative effect on growth of re-establishing native plant species.

In this study, we assessed the potential for legacy effects of soil invaded by the giant goldenrod, S. gigantea, on a community of four native plant species via changes in soil biota, using soils obtained from invaded sites in southern Germany. We used a decreasing sieve-mesh size approach, in order to separate out different fractions of the soil biota into solutions acting as inocula. As fungal and bacterial abundances are still relatively high at mesh sizes <40 µm (Wagg et al. 2014) and these are two important groups containing plant pathogens, we specifically made the following hypotheses

• 1) Native plant species will grow less well on inoculates from soils obtained from vegetation invaded by S. gigantea than from neighbouring uninvaded vegetation.

• 2) If there is such a legacy effect, it will be driven by soil pathogens in the fungal and bacterial fractions of the soil community (those <40 µm).

METHODS

Soil collection and preparation of soil inocula

Soils were collected from three different sites in the vicinity of Constance, southern Germany, during the last week of September 2013 (Table 1). At each site, soil was collected from vegetation invaded by S. gigantea and neighbouring uninvaded vegetation. We obtained 12 soil cores (10cm depth, 5cm diameter) from the invaded and uninvaded patches of vegetation at each site. Due to the irregularity of the patch shapes and sizes, in order to obtain a representative sample of soil, we took soil cores by entering one end of the patch, haphazardly choosing a direction of movement and then we took a core after walking 5–10 m in that direction. We then proceeded to the next sample from that position, in a similar manner. All soil cores for a patch (uninvaded or invaded) at a site were bulked together (amounting to ~2kg of soil) and sealed in plastic bags. Care was taken to wash and clean the soil-corer with 70% ethanol in order to prevent contamination between sampling of invaded- and uninvaded-patch soils within a site and also between sites. Soils per invaded/uninvaded patch were then homogenized by passing through a 5 mm-mesh sieve in a lab. Again, care was taken to avoid contamination between soils by washing with 60°C water and then wiping with 70% ethanol. The soil samples were then kept in a refrigerator at 4°C for no longer than 2 days, before preparation of the soil inocula. In order to understand how differences in soil biota effects among sites might correspond to differences in S. gigantea density, we revisited the sites in June 2015 and counted the number of stems in ten 50cm × 50cm quadrats per site, which were haphazardly placed with the constraint of a minimum 5 m distance between quadrats.

Using similar methods to Klironomos (2002) and Callaway et al. (2011), we created four soil inocula from each soil sample, differing in the fraction of soil biota they contained, one day before planting the seedlings into experimental pots (see below). Previous studies have shown that the diversity of soil communities decreases and composition changes with decreasing mesh size (Callaway et al. 2011; Wagg et al. 2014). Both mycorrhizal fungi and nematodes tend to be reduced in abundance below mesh sizes of 40–50 µm (Callaway et al. 2011; Wagg et al. 2014), while fungi are reduced with a mesh size <25 µm and bacteria form the predominant group. At a mesh size of 250 µm, fungi, bacteria, nematodes and mycorrhizal fungi have abundances close to those of communities passed through a 5000 µm sieve (Wagg et al. 2014) and mesh sizes <100 µm exclude meso- and macrofauna but include microfauna (Bradford et al. 2002).

For each soil sample, we first weighed out 1.2kg of soil and suspended it into 12L of distilled water in a screw-top plastic bottle (100g of soil per litre of water). This gave six soil suspensions in total. Each suspension was then thoroughly mixed by rolling the closed bottles for 5min each. The bottles were then left for 20min to allow soil particulate matter to fall out of suspension. The resulting solution from each soil sample was then passed through a series of fine-mesh sieves (Target Ltd, Glasgow, UK). First, all of the solution (avoiding the particulate sludge at the bottom) was passed through a 5 mm-mesh sieve, to catch coarse plant material, which was placed directly over a 200 μm-mesh sieve. The 200 μm sieve should have allowed most soil biota to pass through into the collected solution (Wagg et al. 2014; bacteria, mycorrhizal fungal spores and hyphae, nematodes and fungal pathogens and saprobes; hereafter referred to as the ‘all-microbe’ solution). One third of the resulting all-microbe solution was held back for use in the experiment, and the remaining two-thirds was then passed through a 40 μm-mesh sieve, which should have largely excluded mycorrhizal fungi (particularly spores) and nematodes, and allowing bacterial and fungal pathogens/saprobes through into the collected solution (hereafter the ‘<40 µm’ treatment). One half of this solution was held back for use in the experiment, while the remaining half was passed through a final 20 μm-mesh sieve, with the aim of allowing mainly bacteria through into the collected solution (hereafter the ‘<20 µm’ solution). These methods ensured similar concentrations of soil biota of different types in each collected solution from a given soil sample.

A fourth soil solution for each soil sample was obtained from 300g of soil sterilized in a steam autoclave (at 121°C for 40min), added to 3L of distilled water and mixed in the same way as the non-sterilized soils. Then, the resulting suspension was passed through the same series of mesh sieves (from 200 to 20 μm), but we only used the final suspension. The sterilized solution was produced in order to check for potentially confounding effects of differences in soil nutrient concentration between invaded and uninvaded soils. In summary, we produced four soil solutions for each invaded and uninvaded patch of vegetation per site, giving a total of 24 soil solutions. These soil solutions were produced within 24h and were kept refrigerated (at 4°C) until the following day.

Native test species

In order to assess the effects of soil biota from soils invaded and not invaded by S. gigantea, we chose four common, native plant species to grow in a simulated ‘community’: Epilobium hirsutum, Hypericum perforatum, Lythrum salicaria and Phleum pratense. These test species are all known to occur at the three sites, either in standing vegetation or in seed banks (Kundel et al. 2014). Thus they are suitable native phytometer species for testing effects of soil biota from invaded soils on early, first-season plant growth. We collected seeds of these species from plants growing in sites other than those that were used for soil collection, earlier in 2012. We bulked together the seeds from different parent plants and sowed out 400 seeds into two shallow trays (200 per tray) containing a 1:1 mixture of washed sand and vermiculite, in mid-September 2014. These seeds were allowed to germinate and grow under lamps for 12 days (temperature = 20°C/15°C 12h each; light level = 150 µmol m⁻²s⁻¹,12h light/dark; relative humidity = 90%).

Experimental set-up and measurements

The seedlings of the four native species were planted out into pots, containing a 1:1 mixture of sand and vermiculite, in a square formation (on the 27th of September 2013). The planting position of each species was haphazardly chosen. The experiment involved a total of 120 pots (square, plastic, 5cm × 5cm × 10cm), representing 3 sites × 2 soil origins [invaded and uninvaded] × 4 soil solutions × 5 replicates. Before planting with the native plant seedlings, each pot was treated with a soil solution. For each of the three sites, we applied 100ml of one of the four soil solutions from the invaded or the uninvaded soils to each of five replicate pots (giving a total of 40 pots per site and 20 pots per uninvaded or invaded soil per site), by pouring the solutions onto the surface of the substrate. We initially placed each pot on a circular saucer, to allow absorption of soil solutions that ran out of the bottom of the pots and to prevent contamination of neighbouring pots with different soil solutions.

We placed the pots on a table in a greenhouse and pots were assigned to random positions with the restriction that pots belonging to the same treatment combination were not clustered. The positions of the pots were re-randomized midway through the experiment in order to minimize position effects. After planting, the saucers of every pot were upturned, to prevent pots from standing in water, and to maintain separation of soil biota among pots. For the first 10 days after planting, plants were sprayed every second day with distilled water to reduce transplant stress. After 10 days, we watered each pot with 100ml of distilled water three times per week. The plants were grown for a total of 10 weeks (from the transplanting date). Extra lighting (125 µmol m⁻² s⁻¹) was provided between 06.00 and 20.00h for the first 8 weeks and then between 07.00 and 21.00h for the last 2 weeks. The temperature was kept at 24°C during day time (~14h) and at 16°C during night time (~10h). To avoid nutrient limitation, we supplied each pot with 200ml of a 50% strength Hoagland’s solution once every 2 weeks, starting from the fourth week of the experiment. At the end of the experiment, aboveground biomass of the plants was harvested separately per species, dried at 70°C for 72h, and weighed. Total community biomass was calculated from the individual species values.

Analysis

We analysed the number of S. gigantea stems per quadrat in relation to site using a quasipoisson generalized linear model (due to overdispersion) followed by a post hoc test of pairwise comparisons, using the ‘glht’ function in the R package ‘multcomp’ (Hothorn et al. 2014). We analysed the total aboveground biomass of plants in each pot using a linear mixed-effects model in the R package ‘lme4’ (Bates et al. 2014), with a two-way interaction between the fixed effects ‘soil origin’ (uninvaded or invaded) and ‘treatment’ (all-microbe, <40 µm, <20 µm and sterilized soil solutions), and site added as a random effect. We assessed the significance of the interaction and main effects using an analysis of variance (ANOVA) from the package ‘lmerTest’ (Kuznetsova et al. 2014). In order to assess the effects of soil solutions and invasion status of soil per site, we used ANOVA with a two-way interaction between ‘soil origin’ and ‘treatment’, for each individual site. To assess the effects of soil biota and soil invasion status for each native species separately, we used linear mixed-effects models again, but this time including the total biomass of the other non-target species in each pot as a covariate, centred on the mean and scaled by the standard deviation. We included the biomass as a covariate, to account for competitive effects of other species in the community; we expected a negative relationship between biomass of other community species and the target species. Significant effects of origin and treatment after accounting for community biomass would indicate additional effects of these factors in addition to indirect effects on the rest of the community. As with total biomass of all four species, significance of the covariate, the interaction between ‘soil origin’ and ‘treatment’ and their main effects were assessed using ANOVA from the R package ‘lmerTest’.

Finally, when a significant interaction was found, we used a priori defined Tukey’s pairwise comparisons, to test when (i) invaded soil solutions resulted in significantly different biomass for native species compared with uninvaded soil solutions (comparing invaded to uninvaded for each level of the soil solution treatments; ‘all-microbe,’ ‘<40 µm’, ‘<20 µm’ or ‘sterile’ solutions), (ii) ‘all-microbe’ solutions in either invaded or uninvaded soils significantly differed from ‘<40 µm’, ‘<20 µm’ and ‘sterile’ solutions. This yielded a total of 10 pairwise comparisons of means for a significant interaction. Finally, when only the main effect of treatment was significant, we made pairwise comparisons of all combinations of soil solution pairs, averaging over the uninvaded and invaded soil levels. Post hoc pairwise comparisons were made using the package ‘multcomp’ (Hothorn et al. 2014). All analyses were done in R version 3.1.1 (R Core Team 2014).

RESULTS

Differences in S. gigantea stem density among invaded sites

Sites significantly differed in the number S. gigantea stems per quadrat (F_2,27 = 16.086, P < 0.001), with highest numbers of stems in site A (means and standard errors shown in Table 1). Both sites A and B had significantly higher numbers of stems on average than site C (Site A–C [log] mean difference = 0.590±0.109, P < 0.001; Site B–C [log] mean difference = 0.327±0.114, P = 0.012). Site A also had higher stem numbers on average than site B ([log] mean difference = 0.263±0.098, P = 0.020).

Effects of soil origin and soil solution treatment on community biomass

Across all three sites, there was no significant overall effect of soil origin, while there was a significant amount of variation explained by the solution treatment overall (Table 2; Fig. 1a). Tukey’s post hoc comparisons revealed that across uninvaded and invaded soils, community biomass was significantly greater when treated with the <20 µm solutions (mean difference = 0.483±0.142g, P = 0.004) and the sterile solutions (mean difference = 0.436±0.142g, P = 0.011) than when treated with the all-microbe solutions. Biomass of communities treated with the <40 µm solutions was greater on average than for communities treated with all-microbe solutions, but the difference was marginally non-significant (mean difference = 0.338±0.142g, P = 0.080).

There was a marginally non-significant interaction between the effects of soil solutions and soil origin on the combined aboveground biomass of the four native species (Table 2). Biomass tended to be lower for plants receiving the all-microbe solution from invaded soils compared with uninvaded soils (Fig. 1a). The effects of soil solutions significantly differed according to soil origin for site B only (Table 2; Figs 1b–d). For invaded soils, community biomass was significantly greater when treated with the <20 µm solution (mean difference = 1.358±0.375g, P = 0.009) and with sterile solution (mean difference = 1.174±0.375g, P = 0.031) than with the all-microbe solution. The difference in biomass between the <40 µm and all-microbe solutions was not significant (mean difference = 0.740±0.375g, P = 0.346). There were no significant differences between the all-microbe solution and other solutions from uninvaded soils, or between invaded and uninvaded soils for each solution type (P > 0.05). Whilst biomass was lower in site B with the all-microbe solution from invaded compared with non-invaded soils (Fig. 1c), this difference was not significant (−0.962±0.375g, P = 0.113).

Effects of soil origin and soil solution treatment on biomass of individual species

Biomass of all four individual species varied significantly with the biomass of the remaining three species (Table 3). Individual species’ biomasses decreased with increasing biomass of the remaining species (slope estimates: Epilobium = −0.041±0.018; Phleum = −0.231±0.036; Lythrum = −0.127±0.055), except for Hypericum, which showed an increase in biomass with increasing total biomass of the other species (0.122±0.060). After accounting for the biomass of the other three species, the effects of soil solutions on individual species only marginally and non-significantly differed according to soil origin for Epilobium hirsutum and Hypericum perforatum (Table 3). Biomass of Epilobium and Phleum varied significantly according to solution type (Table 3). Epilobium biomass (Fig. 2a) was significantly greater with the <20 µm solutions compared with the all-microbe solutions (mean difference = 0.406±0.140g, P = 0.020), and was greater with sterile than all-microbe solutions, but not significantly so (mean difference = 0.342±0.142g, P = 0.075; Fig 2a). The biomass of Phleum was significantly greater with the <20 µm solutions (mean difference = 0.309±0.100g, P = 0.011) and with sterile solutions (mean difference = 0.331±0.099g, P = 0.005), than with the all-microbe solutions (Fig. 2d). Differences in Hypericum biomass under different solution types tended to depend on soil origin, with lower biomass under treatment with sterilized soil solution from invaded than uninvaded soils (Fig. 2b), which corresponded with greater biomass of Epilobium under the invaded than uninvaded sterilized soil solution (Fig. 2a). However, the interaction explaining Hypericum biomass was barely significant (P = 0.054; Table 3). Without the biomass covariate included, only P. pratense showed a significant effect of soil solution type (online supplementary Table S1). Thus, colinearity between the factors soil origin and treatment, and biomass of other community species was not apparent, as high colinearity would have resulted in more significant effects of the two factors after removal of the biomass covariate. Rather, the biomass covariate explained a large amount of variation that soil origin and solution treatment could not.

DISCUSSION

Invasive plant species are acknowledged to have potential impacts on native plant communities, including legacy effects via changes in soil biota and allelopathy, which might influence establishment of native species in soil previously occupied by the invader. We showed in this study, that there was a significant effect of different soil biota fractions overall on native community biomass, regardless of soil origin. Differences in legacy effects between S. gigantea-invaded and non-invaded soils via soil biota changes were weak, and at best site-specific, with only a trend of reduced native plant growth when treated with the full suite of soil biota or with fractions >20 µm from Solidago-invaded areas (Fig. 1). We only found significantly reduced growth of native species when treated with soil biota from Solidago-invaded soils at one of the three sites. Overall, our study suggests that soil biota >20 µm in size have a net antagonistic effect on the first-season growth of native species, regardless of whether the soil was invaded by S. gigantea or not.

The results of our study offer little evidence that accumulation of generalist pathogens in soils invaded by S. gigantea represents a negative legacy effect on colonizing native species. Scharfy et al. (2010) also found that simulated native communities did not respond differently to being grown in soils from Solidago-invaded areas compared to native community soils. In our study, the effects of soil biota from Solidago-invaded soils are variable among sites probably because sites differ in the initial soil microbial communities occurring prior to invasion, and this could be a consequence of native plant species composition and environmental conditions. Sites A and B were wet meadows with similar plant species composition, while site C was a grassland/scrub community directly bordering mixed forest (Table 1; Kundel et al. 2014). However, only site B showed a significant effect of soil invasion status, depending on the soil solution used. It is not clear why the effect was not present in all sites; differences in Solidago density among sites did not correspond with differences in soil biota effects. It could be that site B has been invaded for a longer time, and as a consequence had more time to build up legacy effects. Initial vegetation and environmental differences, and thus soil differences, may lead to variation in accumulation of specific groups of soil microorganisms. Unless the species has differing root traits and root biochemistry at each site, differences in the initial soil community state mean that any changes in composition after invasion are unlikely to be in one particular direction, and subsequent legacy effects on native establishing species may therefore be idiosyncratic.

Although there was no consistent biotic soil legacy effect of invasion by Solidago across sites, overall, the effect of ‘all-microbe’ solutions was negative compared with <20 µm and sterilized solutions, but not significantly so compared with the <40 µm solution. This result indicates that, in general, the larger soil-microbe component has a net negative effect on plant growth in contrast with previous studies (Callaway et al. 2011; Klironomos, 2002). Our results suggest that soil organisms in the 20–200 µm size range are the most pathogenic, compared with the <20 µm microbes (Fig. 1). The >20 µm fractions may contain more fungi, including mycorrhizal fungi (Wagg et al. 2014), but it is also possible that nematodes were excluded from the <20 µm solution (as in Callaway et al. 2011 and Wagg et al. 2014). However, we cannot confirm for certain which specific group of soil organisms >20 µm was responsible for net negative plant growth, compared with sterilized or <20 µm soil solutions.

The reduced biomass of communities treated with the ‘all-microbe’ solution compared with sterilized/<20 µm solutions was driven by the two largest biomass-producers, Epilobium hirsutum and Phleum pratense (accounting for 90% of biomass per pot on average), as they had significantly lower biomass under the ‘all-microbe’ solution than the sterilized or <20 µm solution. The other two native plant species (Lythrum and Hypericum) had much smaller average biomass, and only Hypericum showed a marginally non-significant interaction between soil origin and solution treatment. Competition from Epilobium and Phleum may have precluded any effect of treatment and soil origin being realized for Lythrum and Hypericum. Further work might consider how these soil biota effects on individual native species affect their competitive ability and subsequent persistence in soils previously occupied by S. gigantea plants.

An alternative type of legacy effect of invasions to changes in soil is allelopathy. Extracts and soils of the congener S. canadensis have been shown to have allelopathic effects on germination of several native species in Europe (Abhilasha et al. 2008) and in China (Yuan et al. 2013). Allelopathic compounds released by S. gigantea plant parts could in theory inhibit growth of native plants in invaded relative to uninvaded soils. Del Fabbro et al. (2014) showed that, through application of activated carbon to remove allelopathic compounds, there was no difference in native species germination between field soils invaded and not invaded by S. gigantea. However, our study design was not suitable for discerning the effects of allelopathy from S. gigantea on native plant growth. A second caveat was that we took a ‘black-box’ approach to assessing soil-biota effects in invaded versus uninvaded soils on plant growth. We can clearly deduce that solutions from unsterilized soils have a net antagonistic effect on native plant growth, relative to sterilized soil solutions, and that the antagonistic soil organisms are likely to be >20 µm in size. We cannot say for certain what these antagonistic soil biota are, only that their negative effects on native plant growth appear to outweigh mutualistic effects from biota such as mycorrhizal fungi. In addition, regardless of the identity of the antagonistic soil biota, their effects do not strongly depend on whether or not the soil has been invaded by S. gigantea. One final limitation of our study is that we only considered the short-term effects of soil biota on early first-season growth of native plants. While we found that biota from invaded soils affected native plant growth similarly to biota from uninvaded soils, we do not know if long-term survival and competition with S. gigantea itself could be affected. Thus, future studies should try to separate potential allelopathic effects from soil-biota effects, assess the most important soil biota, and test for long-term legacy effects.

In summary, we find little evidence of a legacy effect of invasion by S. gigantea on native plant species via soil biota, at least in the context of first-season growth of native plant species. It is likely that the existence and strength of legacy effects in invaded soils and the soil micro-organisms involved will be site-specific, with initial soil properties, and plant and soil community compositions playing a key role. The results of our study suggest that, at least for early re-establishment of native species, efforts to restore communities previously invaded by S. gigantea are unlikely to be severely hampered by shifts in soil microbial community composition.

SUPPLEMENTARY MATERIAL

Supplementary material is available at Journal of Plant Ecology online.

FUNDING

W.D. acknowledges funding from the Deutsche Forschungsgemeinschaft (AZ DA 1502/1-1).

ACKNOWLEDGEMENTS

Thanks go to Otmar Ficht and Claudia Martin for assistance during the experiment.

Conflict of interest statement. None declared.

REFERENCES