In grassland agroecosystems, perennial weed invasions may be strongly affected by plant–soil biota interactions (Hallett 2006; Raizada et al. 2008; Reinhart and Callaway 2006; Wolfe and Klironomos 2005). In particular, site occupancy by invasive plant species may have strong soil-mediated “legacy” effects that influence subsequent plant invasion dynamics. Recent comparative studies have revealed that invasive species appear to experience, on average, less-negative effects from root-zone soil microbe communities than do natives (Kulmatiski et al. 2008), which may experience intense negative feedback from these microbes (Petermann et al. 2008). This evident difference in plant–soil interactions between invasive and native species implies that invasives may be capable of particularly strong soil-legacy effects, as the result of prolonged site occupancy by invasives resulting in changes to multiple attributes of soils.

Documented modifications of the biotic composition of soil on invaded sites include effects on soil food webs (Duda et al. 2003), total soil microbial communities (Kourtev et al. 2002), and mutualistic fungi (Allen et al. 2003; Hawkes et al 2006). Effects on physical or chemical attributes of soil include modification of inputs and cycling of nitrogen (N) and other elements (Ehrenfeld 2003; Haubensak et al. 2004; Hawkes et al. 2005; Sperry et al. 2006), soil organic matter and aggregation (Saggar et al. 1999), and pH (Kourtev et al. 2002), as well as release of allelopathic substances (Thorpe et al. 2009).

These effects of invasive species on soils will be significant to plant invasion if they confer a specific advantage to invasive relative to native species. Such advantages have been observed (Corbin and D'Antonio 2004; Ehrenfeld 2003; Jordan et al. 2008; Ortega and Pearson 2005; Richardson et al. 2000). For example, the invasive forb leafy spurge (Euphorbia esula L.) transforms the biotic composition of soil to its own advantage (Klironomos 2002), causing growth reduction in several prairie forbs at the seedling stage (Jordan et al. 2008). Similar effects have been observed in other invasive species (Allen et al. 2003; Bray et al. 2003; Callaway et al. 2003; Hawkes et al 2006; Stinson et al. 2006). Invasive species may also have soil-mediated facilitative effects on other invasive species (Jordan et al. 2008).

Additionally, invasive plant species may be less dependent than native species on mutualistic interactions with soil microbiota. For example, invasives may become less dependent on AMF mutualism than conspecifics in the native range (Seifert et al. 2009). AMF are “keystone” root symbionts with a wide range of effects on plant physiology and ecology (Smith and Read 1997) and many North American tallgrass prairie species are highly dependent on AMF (Wilson and Hartnett 1998) for growth and reproduction.

Based on emerging evidence of qualitative functional differences between invasives and natives in plant–soil biota interactions, we propose that there are qualitative differences between soil-legacy effects of invasive and native grassland species, such that soil-mediated effects of invasives on natives differ from the reciprocal effects. Specifically, we hypothesize that (1) soils conditioned by invasives have neutral or facilitative effects on conspecifics and other invasives while exerting negative effects on natives and (2) soils conditioned by natives have negative effects on conspecifics and positive or neutral effects on nonconspecific natives and invasives.

To address these hypotheses, we compared soil-legacy effects of previous soil occupancy by multiple invasive and native species in a single experimental study. Most previous assessments of legacy effects of soil occupancy by invasive species have focused on effects of a single invasive species and have examined responses to soil modifications by only a few native species, exotic species, or both. We conducted an extensive experimental comparison of interspecific effects on growth, mediated by soil occupancy, among a group of co-occurring native and invasive species from the mixed-grass prairie grassland communities of North America. We focused on legacy effects mediated by AMF, given the importance of these fungi in grassland plant communities (Hartnett and Wilson 2002). All native and invasive species examined are mycorrhizal (Wilson and Hartnett 1998, Carey et al. 2004). We estimated effects of soil occupancy by three exotic invasive species: smooth brome (Bromus inermis Leyss.; hereafter all species will be referred to by genus), crested wheatgrass [Agropyron cristatum (L.) Gaertn.], and leafy spurge (E. esula). All are “strong invaders” (Ortega and Pearson 2005), able to become community dominants and to form nearly monospecific stands in invaded grasslands. We derived species-specific profiles of growth response to soil occupancy by other species, which allowed us to evaluate our hypothesis that exotic species differed categorically from natives in soil-mediated legacy effects on other species (Jordan et al. 2008).

Materials and Methods

Glasshouse Experiments

Experiments were conducted in a glasshouse on the University of Minnesota campus, St. Paul, MN, from July 2003 to November 2004. At the inception of the experiment, two soil-biotic treatments were imposed, pasteurized (−AMF) and not pasteurized (+AMF). Half of the field soil from each reserve was steam-pasteurized to kill all field soil biota by steaming twice (24 h apart) for 40 min at 80 C, followed by a 20-min aeration to mitigate nitrogen accumulation (modified from Burrows and Pfleger 2002). Soils were mixed 1 ∶ 1 soil ∶ pasteurized sand and placed in 2.5-L pots, keeping soils from each reserve separate. To create the −AMF treatment, we added a “microbial wash” to each pot (using methods of Koide and Li 1989); this was an aqueous suspension created by shaking. The inoculum wash was prepared by shaking 20 g (0.7 oz) of each soil “inoculum” from each reserve in 1 L distilled water for 10 min. The suspension was passed through an 11-µm filter to exclude spores and other AMF propagules. Pots were inoculated with 20 ml of the filtered suspension (Burrows and Pfleger 2002). This suspension serves to inoculate pasteurized soils with soil microbes smaller than AMF, including fungal pathogens. It reliably reduces differences in non-AMF soil microbial communities between pasteurized and non-pasteurized treatments, so that these treatments can be interpreted as −AMF and +AMF treatments, respectively.

Effects of soil occupancy by plant species and mixtures were estimated by creating soil “conditioning” treatments in three separate soils, collected from nature reserves as above. Each soil was conditioned by natives alone and in mixture, as well as by the invasive plant species that was dominant at the respective soil collection site (e.g., Bromus was used in soils from Lostwood NWR). In each soil, five soil conditioning treatments were established in 16 replicates: (1) monoculture invasive (species specific to soil collection site), (2) monoculture Stipa viridula Trin, (3) monoculture Bouteloua gracilis (Willd. ex Kunth) Lag. ex Griffiths, (4) monoculture Linum perenne L. var. lewisii (Pursh) Eat. & Wright, and (5) mix of six native species (Stipa, Bouteloua, Linum, Aster ericoides L., Koeleria macrantha (Ledeb.) Schult., and Ratibida columnifera (Nutt.) Woot. & Standl. (nomenclature according to Great Plains Flora Association 1986). Seeds of Bromus, Agropyron, and Euphorbia were collected at sites of soil collection; seeds of native species were purchased¹; all native seed was sourced from within the ecoregion of the three soil collection sites and occurred at sites where we sampled.

Soil conditioning treatments were repeated through three 3-mo growth cycles (I: July 28, 2003, to November 10, 2003; II: February 25, 2004, to May 17, 2004; III: June 21, 2004, to August 20, 2004) with cold treatments (∼ 4 C) in between. All conditioning treatments began with 15 to 20 established seedlings per pot and produced abundant aboveground biomass (with the sole exception of Linum in the −AMF treatment, as noted below). Aboveground biomass produced in each growth cycle was harvested by pot at the end of each cycle, as was any litter. Pots that lacked regrowth of the conditioning species were reseeded after each of the first two cycles. The conditioning treatments were imposed in a glasshouse with 400-watt high-pressure sodium lamps used for supplemental lighting (14- to 16-h day length) from September to May. Average summer glasshouse temperatures were 27/23 C (day/night); spring and fall temperatures were 21/19 C. Pots were watered as needed. Osmocote 14–14–14², a slow-release fertilizer, was added at the start of the second growth cycle (1.25 g pot⁻¹). Pots containing soils from a given collection site were held on adjacent glasshouse benches with no intermingling during the first two growth cycles. In the third cycle, all pots were combined in a randomized complete-block design. A final growth cycle (September 15, 2004, to November 19, 2004) was used to assess effects of treatments on seedling growth. Prior to this cycle, all pots were frozen at ∼ 4 C to eliminate regrowth; crowns that did regrow were pinched back.

In the final cycle, half of each pot was planted with six native and half with six invasive species in a fixed pattern (Figure 1); consequently, the experimental unit for growth responses is a sowing position within a pot conditioned by a given species. Seedlings were thinned to one plant per species per pot. Invasives were Bromus, Agropyron, Euphorbia, Cirsium arvense (L.) Scop., Centaurea maculosa Lam., and Poa pratensis L.; natives were Stipa, Bouteloua, Linum, Aster, Koeleria, and Ratibida. Centaurea seed was obtained from R. Callaway, University of Montana; Cirsium seeds were collected in Ramsey County, MN; and Poa was obtained from the University of Minnesota. Pots were monitored for regrowth from perennating structures remaining in the soil. Little regrowth occurred; it was clearly distinguishable from emerged seedlings and was immediately removed by pinching upon observation. After 12 wk, shoot biomass was harvested; at this time, seedlings of each species were small and well separated from other species in each pot; seedlings were assumed to not interact with other seedlings. For four replicates, soil samples were taken for N and pH, and for these four and one additional replicate, total plant biomass (root and shoot) was harvested for all seedlings. Plants were dried at 60 C for 3 to 7 d and weighed. N (nitrate and ammonium) was determined by a 2-M KCl extraction of 5 g of soil (Robertson et al. 1999). Extracts were analyzed on an OI Corporation SF3000 autoanalyzer.³ Soil pH was measured electrometrically. Equal amounts of soil and neutralized nanopure water were mixed 1 ∶ 1 and allowed to equilibrate for 30 min. The supernatant was decanted and allowed to resettle for 30 min, the pH of the supernatant was then measured once with an electrode (Robertson et al. 1999).

Figure 1

Protocol for estimating effects of soil conditioning by individual invasive species (Agropyron [I1], Bromus [I2], and Euphorbia [I3]), individual native species (Stipa [N1], Linum [N2], and Bouteloua [N3]), and native mixture (Stipa, Bouteloua, Linum, Aster ericoides [N4], Koeleria macrantha [N5], and Ratibida columnifera [N6]). Seedling growth was assessed on six native species (N1 to N6) and six invasive species (I1 to I3 and Cirsium arvense [I4], Centaurea maculosa [I4], and Poa pratensis [I6]; NB, Aster (N4) and Koeleria (N5) were removed from final analysis. Soil conditioning treatments were applied to each of three soils collected from separate prairie nature reserves as noted in text; each soil was conditioned by natives alone and in mixture, and by the single invasive species that was most dominant at the respective soil collection site. Diagram depicts conditioning treatments applied to one of the three soils.

By comparing seedling biomass production in experimental units originally containing pasteurized or unpasteurized soil and non-AMF soil microbes introduced by a filtrate, we sought to assess species-specific effects of soil conditioning in the presence of AMF indigenous to reserves where each species had invaded, and in the absence of AMF in the control soils. During the three glasshouse growth/vernalization cycles, various soil organisms undoubtedly colonized our experimental soils. However, there was likely little AMF colonization of control soils during the experiment, as we observed no AMF colonization in plants grown in a separate validation experiment (below, and Table 1).

Table 1

Arbuscular mycorrhizal colonization from greenhouse experiments for soils with arbuscular mycorrhizal fungi (+AMF ) from three prairie nature reserves. Estimated colonization rates are mean percentages of observed root samples that contained AMF ± 1 SE. No colonization was observed in pots not receiving AMF inoculum (−AMF). Colonization of invasive species (Bromus, Agropyron, and Euphorbia) was assessed only in soil from nature reserve where that species was particularly abundant.

Results and Discussion

Consistent qualitative differences were evident between exotic and native species in response to repeated rounds of soil conditioning, but these differences depended on whether AMF were present (conditioning treatment by species type by AMF interaction, F  =  4.95; df  =  1, 60; P  =  0.030). Thus, the effects of AMF were contingent on the identity of the responding species and the main effect of AMF was marginally nonsignificant (F  =  3.46; df  =  1, 45; P  =  0.069). In soil with AMF, the biomass of all invasive species was greatest in soils conditioned by the three individual native species or by the native mixture, whereas biomass production by invasives was lower in soils conditioned by invasives. Seedling growth of exotic species was generally facilitated by soils conditioned by natives: five of the six exotic species had greater biomass (P < 0.05, preplanned comparisons in single-species ANOVA) in soils conditioned by at least two of the three individual native species used for soil conditioning (Figure 2). Native species frequently facilitated invasive species growth by more than 100%, relative to mean performance in soils conditioned by exotics (Figure 2). In contrast, for four of the six exotic species, the native species mixture did not have a facilitative effect on seedling growth. In pasteurized (−AMF) soils, facilitative effects of native species on exotic species were converted to neutral effects relative to invasive effects on these invasive species, and thus the general level of native facilitation on invasive biomass production was much weakened (Figure 2). In −AMF experimental units, maximum biomass production by all exotic species occurred in response to conditioning by Linum, which itself produced very little biomass in the absence of AMF. We note that N levels in these units were not anomalously high relative to other treatments (Table 2), suggesting that the observed exotic growth responses do not simply reflect high N levels in these units.

Figure 2

Mean aboveground biomass production by invasive species in soils conditioned by invasive and native species and by a native species mixture, in soils with AMF (+AMF; field soil + microbial wash) or without AMF (−AMF soil; pasteurized soil + microbial wash). All means are based on growth in three soils collected from separate prairie nature reserves as noted in text. Asterisks denote significant differences (P < 0.05, preplanned comparisons in single-species ANOVA) between a native-species conditioning treatment and mean biomass production in soil conditioned by three invasive species.

Table 2

Nitrate, ammonium, and pH (mean ± SE, n  =  4) soil measurements for each collection site/conditioning species and mycorrhizal treatment combination used in this study. Soils conditioned by invasive species are labeled by collection site, i.e., Lostwood soils were conditioned by Bromus, Medicine Lake soils by Agropyron, and TRNP by Euphorbia.^a

Although native species generally facilitated the growth of exotics in the presence of AMF, the reciprocal effect—facilitation of aboveground biomass production of native species by invasive species, relative to mean performance in soils conditioned by natives—was not observed (with a single exception, Figure 3). Typically, conditioning by exotic species did not significantly reduce native biomass production, but such antagonistic effects (P <0.05, preplanned comparisons in single-species ANOVA) were evident in Bouteloua, Linum, and Stipa (Figure 3). As was true for the invasive species, the effect of invasive conditioning on natives was affected in some cases by the microbial composition of conditioned soils. In particular, the invasive forb Euphorbia had an antagonistic effect on all native species in +AMF soils; in −AMF soils these effects were converted to neutral or facilitative effects for five of the six native species. We note that our experiment does not enable the various direct and indirect effects of AMF colonization (e.g., on physical structure and nutrient availability of soils) to be resolved, although our use of soil chemistry covariates indicates that the observed effects of AMF are not mediated by effects of soil nitrate, ammonium, and pH.

Figure 3

Mean aboveground biomass production by native species in soils conditioned by invasive and native species and by a native species mixture, in soils with AMF (+AMF; field soil + microbial wash) or without AMF (−AMF soil; pasteurized soil + microbial wash). Note difference in scale on y-axis for Figures 2 and 3. Means in soils conditioned by invasive species are based on growth in a single soil (collected in different nature reserves as noted in text); means in soils conditioned by native species are based on growth in all three of the collected soils. Asterisks denote significant differences (P < 0.05, preplanned comparisons in single-species ANOVA) between an invasive-species conditioning treatment and mean biomass production in soil conditioned by three native species.

Our results underscore the importance of AMF in the functional ecology of grassland perennials (Wilson and Hartnett 1998) and raise the possibility that restoration of AMF in soils may be important to efficient establishment of the diversified biofuel grasslands that are now attracting much interest (Tilman et al. 2006a, b). The effects of soil conditioning were strongly affected by the presence of AMF, suggesting that observed legacy effects were mediated by changes in the composition of AMF communities in experimental soils. The facilitative effects of natives on invasives observed in +AMF treatments might reflect some growth-inhibiting artifact of pasteurization, but the multiple rounds of conditioning used in our protocol reduce the likelihood of such an effect. Moreover, our group has applied recently developed methods for identifying AMF taxa in plant roots via molecular methods (Aldrich-Wolfe 2007) to exotic and native species grown in this experiment. Preliminary results (N. R. Jordan, unpublished data) indicate that invasive perennials of grassland agroecosystems do not host AMF taxa that occur in desirable native perennials, suggesting that repeated rounds of conditioning may create substantially differentiated AMF communities in experimental soils. These same exotic species have been shown to create soil microbial legacies that sharply limit growth of a number of these same native perennials (Jordan et al. 2008).

There is growing recognition of the significance of aboveground–belowground interactions in plant community dynamics (van der Putten et al. 2009); meta-analyses (Kulmatiski et al. 2008) suggest that plant interactions with soil biota can be comparable in magnitude to factors such as herbivory and competition. A variety of evidence on soil-mediated interspecific effects of exotic perennial invasive species (Mangla et al. 2008; Vogelsang and Bever 2009) suggests a conceptual model of an exotic weed invasion process in grassland agroecosystems. In this model, certain exotic species, upon invasion, occupy soils long enough to affect some biotic or abiotic modification that reduces fitness of natives relative to exotics, facilitating further invasion and impeding restoration of native perennial communities. Such differential soil-mediated effects on native vs. invasive species could support self-reinforcing invasion processes, i.e., an “invasional meltdown,” driven by cofacilitation by multiple invasive species (Best and Arcese 2009; Jordan et al. 2008) or by reducing the ability of native species to establish and persist in modified soils (Standish et al. 2008) even when management actions are taken to aid establishment (Lombardo et al. 2007). Such processes may establish a “stable degraded state” in a grassland agroecosystem (Kulmatiski 2006; Suding et al. 2004;), enforced by a legacy of extensive invasive modification of soil (Peltzer et al. 2009; Rout and Callaway 2009) that facilitates reinvasion even after effective control or removal of invasives.

This conceptual model hinges on a key functional difference between soil-mediated effects of invasive and native species on plant community dynamics. Specifically, the premise of the model is that exotics have neutral or facilitative effects on conspecifics and other exotics (Kulmatiski et al. 2008), and exert negative effects on natives (Batten et al. 2008; Jordan et al. 2008; Vogelsang and Bever 2009). Natives are presumed to have negative effects on conspecifics (Kulmatiski et al. 2008) and nonnegative effects on nonconspecific natives and on exotics. Thus, soil-mediated effects of invasives on natives are hypothesized to differ from such effects of natives on invasives. Evidence regarding such qualitative differences in interspecific effects of invasives and natives is limited, despite the potential importance of these effects to community dynamics. In particular, soil-mediated interspecific effects of native species on exotics are poorly known and theoretical expectations are not clear (Reynolds and Haubensak 2009).

Most previous comparisons of interspecific effects of soil occupancy by invasive and native exotic species have focused on effects of a single invasive species and have examined responses to invasive soil occupancy in only a few native or invasive species, or both. Damaging effects of invasive species on natives were observed in some cases (Batten et al. 2008; Mangla et al. 2008; Niu et al. 2007; Reinhart and Callaway 2006), but not others (Batten et al. 2008; Rowe and Brown 2008; van Grunsven et al. 2007). Most previous work has compared the effects of invasive soil occupancy to that of a mixture of natives or used field-collected soil from a native plant community (Allen et al. 2003; Niu et al. 2007; Stinson et al. 2006; Vogelsang and Bever 2009; Yu et al. 2005) as the basis of comparison for interspecific soil-mediated effects of exotic species. Only a few studies (Batten et al. 2008; Rowe and Brown 2008; van Grunsven et al 2007) have taken the important step of comparing individual interspecific legacy effects of invasive and native species in a single experimental study, and none has examined multiple invasive and native species in a single study. Our results provide a more comprehensive assessment of the effects and responses to soil occupancy of individual co-occurring exotic and native species.

In accord with the key premise of the conceptual model outlined above, we observed a consistent qualitative difference in soil-mediated interspecific effects of the co-occurring invasive and native species used in this experiment: natives facilitated invasives, but invasives did not facilitate natives. Among the range of observed soil occupancy effects of three invasives on natives, a few were strongly antagonistic, relative to the corresponding effect of natives. However, many of these invasive effects were not significantly more antagonistic than the corresponding effect of natives. Our results thus provide the first indication of the variability of soil-mediated interspecific effects of invasives on natives, among a group of co-occurring exotic and invasive plant species. We note that no native species was affected negatively by all three invasive species, and the variability in responses among native species suggests that soil occupancy by invasives will have complex effects on native community dynamics.

We observed a marked functional difference between soil-mediated interspecific effects of natives and invasives that was not anticipated from previous research: the extensive facilitation of exotic species growth by native species, when evaluated relative to the corresponding effect of invasives on natives. As was the case for native responses to invasives, variability in observed responses among these invasives suggests that soil occupancy by natives will create a complex pattern of facilitation among invasive species that enter a plant community. Four of the six invasive species examined in this experiment experienced significant facilitation by all three native species; the other two invasives were significantly facilitated by at least one native species. Estimated facilitative effects of individual natives differed in magnitude for some invasives (e.g., Centaurea), whereas other invasives had more uniform responses to native soil occupancy (Poa). Soil conditioning by the native mixture was less facilitative to invasives than individual native species effects; Vogelsang and Bever (2009) observed a similar negative effect of a mixture. The mechanistic basis of a difference between soil-conditioning effects of mixtures vs. individual species is not clear, but additional experimental comparisons of such effects are warranted to better understand soil-mediated invasion in plant communities.

If invasive perennials create legacy effects that damage native species of interest in grassland restoration, then a cost-effective method for remediating these soil effects will be needed to reduce the risk of further weed invasion and other problems (e.g., lack of diversity, reduced stands and yields) that would result from poor establishment of these species in grassland restoration efforts. Certain native species show potential for cost-effective remediation, via the microorganisms with which they associate (Azcon-Aguilar et al. 2003; Barni and Siniscalco 2000; Smith et al. 2003), potentially countering effects of invasive species on soil microbial communities and soil attributes. The term “nurse species” has been applied to native plant species that first reoccupy a site after disturbance, and that have a facilitative effect on other native species (Lockwood and Samuels 2004). Many kinds of facilitative effects by such species have been demonstrated, including protection from physical stresses or herbivory, increasing nutrient supply, and enabling effective pollination and dispersal (Bertness and Callaway 1994; Callaway 1997; Larson and Siemann 1998; Lortie et al. 2004; Padilla and Pugnaire 2006), but soil-mediated facilitative effects have not been evaluated experimentally. For example, Canada wildrye (Elymus canadensis L.)—a native species widely regarded as a nurse species by grassland managers—is able to establish and produce substantial aboveground biomass in situations where large seedbanks of annual weeds are present (N. R. Jordan, unpublished data); such aboveground growth presumably creates potential for substantial effects on soil microbial communities. Canada wildrye has been shown (Noyd et al. 1995) to develop high levels of AMF colonization in soils where other native perennial grasses were poorly colonized. Therefore, it is plausible that such species can ameliorate damaging soil-mediated legacy effects of invasive species, and may restore populations of mutualistic soil biota (e.g., AMF) that may not be supported during soil occupancy by invasive species. Such facilitative plant species may serve, in effect, to couple plant and microbe dispersal processes on highly disturbed lands, such that other plant species and microbial symbionts co-occur and can form effective mutualisms (Enkhtuya et al. 2005; Oba et al. 2004; Smith et al. 2003). Hence, nurse plants may be crucial to the establishment of grassland agroecosystems that benefit from an “environmental filter” (Belyea 2004) that appears to result from successful restoration, in which plant community development increasingly limits invasion (Bezemer et al. 2004, Blumenthal et al. 2005).