Legacy of pine invasion disrupts plant–soil feedback of native and non‐native grasses

Summary

• Plant–soil feedback (PSF) is a standing concept helping to explain plant species coexistence. Accumulating evidence suggests that PSF depends on the environmental context. One particularly impactful disruption of the environmental context is the invasion of alien species into a community. While PSF between native and non‐native species has been linked to invasion success, the effect of an invader on PSF between resident plants has received little attention.

• We explored how Pinus contorta (lodgepole pine) invasion impacts PSF between four resident grasses (two native and two non‐native species). We grew the plants in live soil to condition species‐specific soil communities, followed by a response phase where the plants were grown in conspecific and heterospecific soil. Both phases were performed with soil from uninvaded, invaded, or controlled invasion areas.

• Pairwise PSF between the species in uninvaded soil was negative and became neutral to positive in invaded soil. This effect persisted even after P. contorta removal. Our results suggest a destabilizing effect of P. contorta invasion on the coexistence of the resident grasses, benefitting non‐natives over natives.

• This work exemplifies an overlooked mechanism by which an invader can alter interactions between resident species and adds a novel dimension to the context dependence of PSF.

Introduction

Plants form intricate relations with soil biota influencing their associated soil communities in species‐specific ways (Fitzpatrick et al., ²⁰¹⁸), and in turn, these communities impact plant performance (Wardle et al., ²⁰⁰⁴; van der Heijden et al., ²⁰⁰⁸). This constitutes the basis for the reciprocal interaction known as plant–soil feedback (PSF). Since the inception of the concept in the early 1990s (Van der Putten et al., ¹⁹⁹³; Bever, ¹⁹⁹⁴), PSF has developed into an experimental and analytical framework helping to explain species coexistence and the maintenance of species diversity in plant communities (Bever et al., ¹⁹⁹⁷; Bever, ²⁰⁰³; Revilla et al., ²⁰¹³; Teste et al., ²⁰¹⁷). PSF is considered a two‐phase process: In the conditioning phase, soil communities are structured by a plant species, and in the response phase, the performance of plants subsequently grown is affected by the soil communities that were conditioned by either conspecifics or heterospecifics. The associated metric, the interaction coefficient I _s, compares the relative performance of the plants in conspecific against heterospecific conditioned soil for a given species pair (Bever et al., ¹⁹⁹⁷). Negative PSF (I _s < 0) occurs if plants exhibit a lower performance in conspecific compared with heterospecific conditioned soil, whereas positive PSF (I _s > 0) occurs if plants perform relatively better in conspecific conditioned soil. Acting as an intraspecific density‐ or frequency‐dependent mechanism (Chesson, 2000; Eppinga et al., ²⁰¹⁸), negative and positive PSF can stabilize (Kulmatiski et al., ²⁰⁰⁸; Petermann et al., ²⁰⁰⁸; Mangan et al., ²⁰¹⁰) or destabilize (Smith & Reynolds, 2012; Crawford & Knight, ²⁰¹⁷; Dudenhöffer et al., ²⁰²²) plant diversity in plant communities, respectively.

Importantly, PSF does not act in isolation but is shaped by the environmental context (Smith‐Ramesh & Reynolds, 2017; Beals et al., ²⁰²⁰; Gundale & Kardol, ²⁰²¹; De Long et al., ²⁰²³). The context dependence of PSF has recently gained great attention and motivated research on moderators of PSF, such as the consequences of climate change (Pugnaire et al., ²⁰¹⁹; Dudenhöffer et al., ²⁰²²; Hassan et al., ²⁰²²; de Vries et al., ²⁰²³), resource availability (McCarthy‐Neumann & Kobe, 2019; in't Zandt et al., ²⁰¹⁹), and the role of biotic factors, such as herbivory (Bezemer et al., ²⁰¹³; Bennett et al., ²⁰²⁰; Heinze et al., ²⁰²⁰). Increasing evidence suggests that the intensity and direction of PSF between particular plant species is not a static characteristic but rather a process that can be shaped by external moderators, which impairs the predictive power of PSF for plant community composition (Reinhart et al., ²⁰²¹). Thus, a better understanding of how external moderators can impact the direction and intensity of PSF, and consequently, its manifestation, as a diversity stabilizing or destabilizing mechanism, is critical to our ability to utilize PSF theory to explain and predict plant community dynamics in a changing environment (van der Putten et al., ²⁰¹⁶; De Long et al., ²⁰¹⁹).

One particularly impactful disruption of the environmental status quo is the invasion of alien species into a resident plant community. While PSF between invading and native species has been linked to invasion processes (Klironomos, 2002; Callaway et al., ²⁰⁰⁴; Levine et al., ²⁰⁰⁶; Smith & Reynolds, ²⁰¹²; Suding et al., ²⁰¹³; Aldorfová et al., ²⁰²⁰), the indirect effects of an invader on PSF between other resident plants have yet received less attention. However, non‐native plant species get involved in PSF dynamics in the receiving community potentially changing their outcome. If such changes benefit non‐native over native species, this might even contribute to a so‐called ‘invasional meltdown’ (Simberloff & Von Holle, 1999) where one invader facilitates invasion by other non‐native species (Zhang et al., ²⁰²⁰; Chen & van Kleunen, ²⁰²²). In addition to being involved in PSF, invading plants can also modify soil biotic and abiotic properties of the receiving habitat (Ehrenfeld, 2003; Hawkes et al., ²⁰⁰⁵; Vilà et al., ²⁰¹¹; Meisner et al., ²⁰¹⁴), which, aside from direct consequences for the resident plant community, alters the environmental context for PSF dynamics.

The spread of alien conifers, predominantly from the pine family (Pinaceae), into open areas in the southern hemisphere (Richardson et al., ¹⁹⁹⁴; Simberloff et al., ²⁰¹⁰; Nuñez et al., ²⁰¹⁷) is an invasion process with the potential to severely alter the environmental conditions of the receiving habitat. Originally introduced for timber production, erosion control, and more recently carbon sequestration (Richardson et al., ¹⁹⁹⁴; Essl et al., ²⁰¹⁰), many species have become invaders in the introduced areas threatening native biodiversity and ecosystem functioning (Rouget et al., ²⁰⁰²; Sapsford et al., ²⁰²⁰). Invading pines are often accompanied by ectomycorrhizal symbionts (Dickie et al., ²⁰¹⁰; Policelli et al., ²⁰²³), promoting their successful establishment and spread (Nuñez et al., ²⁰⁰⁹). The co‐invasion of pines with their fungal symbionts can impact soil nutrients, soil biota community composition, and functioning (Dickie et al., ²⁰¹⁴; Peralta et al., ²⁰¹⁹; Sapsford et al., ²⁰²²; Mujic et al., ²⁰²³; Sapsford & Dickie, ²⁰²³), thereby altering the conditions for PSF between native and non‐native species in the plant community. The induced changes to soil conditions can persist even after the removal of the invader (Dickie et al., ²⁰¹⁴; Peralta et al., ²⁰²⁰; Sapsford & Dickie, ²⁰²³), altering the baseline soil conditions for future PSF dynamics, and consequently, the trajectories of plant communities after invasive pines have been controlled.

Here, we used Pinus contorta (lodgepole pine) invasion into New Zealand grassland to evaluate the potential indirect effect of an invading species on PSF between native and non‐native resident grasses. Being the most aggressively spreading conifer species in New Zealand (Ledgard, 2001), it is considered a threat to biodiversity, farming productivity, and landscape values (Peltzer, 2018; Howell, ²⁰¹⁹; Wyse & Hulme, ²⁰²¹). As PSF can be masked (Crawford & Knight, 2017) or accentuated (Lekberg et al., ²⁰¹⁸) by plant competition, we evaluated PSF between native and non‐native grass species when grown individually and together in competition with each other in soils representing different stages of P. contorta invasion. With this study, we specifically aim to address the following questions:

1. Does P. contorta change the PSF dynamics between native and non‐native resident grasses?

2. Do changes in PSF persist after invasion control?

3. Does competition alter or mask individually observed PSF?

4. What are the implications of altered PSF for the coexistence of the resident species?

By answering these questions, we explore an overlooked dimension of context dependence by assessing the indirect effects of plant invasion on the receiving community via the alteration of PSF dynamics. The assessment of invader‐mediated alterations in PSF between resident species is a novel yet potentially important aspect for evaluating the risk of co‐invasion and invasional meltdown following invasion.

Materials and Methods

Experimental design

We conducted a two‐phase PSF experiment comprising a conditioning phase where the test species were grown in live soil to condition species‐specific soil communities and a response phase where the same species were subjected to the soil communities conditioned by conspecifics or heterospecifics in all pairwise combinations. To test for the indirect effect of Pinus contorta Douglas ex Loudon invasion on PSF dynamics between the four species, we conducted both the conditioning phase and the response phase using soil from three different sources: uninvaded grassland areas, areas invaded by P. contorta, or areas where P. contorta invasion had been previously controlled (Fig. 1).

Fig. 1.

Experimental setup to test the effect of Pinus contorta invasion on plant–soil feedback (PSF) dynamics between four resident grass species. In the conditioning phase, plants were grown in live soil sourced from three different invasion stages (uninvaded, invaded, and controlled; brown background). In the response phase, plants were grown either in single pots (individual) or together in one pot (competition) in pasteurized background soil (grey background) inoculated with live soil from the conditioning phase (coloured strips). Each treatment was replicated eight times, preserving the correspondence between inoculum soil and background soil regarding the source locations in the field.

Soil collection and preparation

We collected soil from the south‐eastern Lake Pukaki region (44.15°S, 170.22°E) in the Mackenzie District, New Zealand, in October 2023 (austral spring). The area comprises a mosaic of open grassland and stands of the invasive P. contorta (Fig. 2a). While large stands of P. contorta had been treated and largely killed with herbicide (TDPA; Supporting Information Table S1) via aerial foliar spray application (AFSA) in December 2020 (Fig. 2b), invasion is still ongoing, and young P. contorta trees are growing in several areas (Fig. 2c). The dominant grassland vegetation of the area is composed of three of our test species, the introduced grasses Anthoxanthum odoratum L. and Agrostis capillaris L., and the native tussock Festuca nova‐zelandieae (Hack.) Cockayne (Fig. S1). While not dominant, our fourth test species, the native Pentapogon avenoides (Hook.f.) P.M. Peterson, Romasch. & Soreng, also occurs in the area.

Fig. 2.

Sampling area in the south‐eastern Lake Pukaki region, Mackenzie District, New Zealand. (a) Overview of the landscape showing areas of grassland vegetation interspersed with invading and controlled Pinus contorta. (b) A stand of P. contorta that had been controlled in December 2020. (c) Ongoing P. contorta invasion into grassland.

We took soil samples from uninvaded grassland areas at least 70 m away from the nearest P. contorta individual. Soil samples from invaded grassland areas were taken in stands of young P. contorta (3–5 yr old) within a 1 m radius of an individual tree. Soil samples from controlled P. contorta invasions were collected from stands that had been treated and largely killed in late 2020, c. 3 yr before our sampling took place. We collected individual soil samples from eight locations for each invasion stage that were at least 200 m apart from each other, resulting in a total of 24 soil samples. We collected c. 20 l of soil from the upper 10 cm for each sample. We paid attention not to sample directly from underneath one of our target grasses to avoid samples with strong soil legacies of a particular species that might carry over into the subsequent conditioning and response phases (Wubs & Bezemer, 2018). The tools used for soil collection were cleaned and wiped with 70% ethanol before collecting a new sample. Subsamples (c. 200 g) of fresh soil from three locations of each invasion stage were kept in a cooler and were analysed for nutrients within 72 h of soil collection (Table S2). Nutrients (including P, K, Ca, Mg, S, and Na) were extracted using Mehlich III extractant and determined by inductively coupled plasma optical emission spectroscopy (Mehlich, 1984). Nitrate–nitrogen (NO₃–N) was extracted from soils using a 2 M KCl solution. Nitrate was determined by reduction in nitrate (NO₃–N) to nitrite (NO₂–N) using a cadmium column followed by spectrophotometric measurement (Keeney, 1982). The individual soil samples were sieved through a 5‐mm mesh. One part of the sieved samples was used as live soil for the conditioning phase within 1 wk after field collection (Kushwaha et al., ²⁰²⁴). The remaining part was pasteurized (holding a core temperature of 85°C for 1 h) and kept refrigerated at 4°C in sealed bags for later use as background soil for the response phase.

Seed collection and preparation

We collected seeds of all target grass species in February and March 2023 (late austral summer) from an uninvaded grassland area in the same general region (south‐eastern Lake Pukaki). The area of seed collection itself was not used for any soil collection. For each species, we collected a pooled sample of seeds from a total of 15–20 individual plants. The seeds were stored dry at 4°C until further processing. Before planting, the seeds were surface‐sterilized in a 10% household bleach (50 g l⁻¹ sodium hypochlorite) water solution for 6 min and rinsed with autoclaved water three times. The sterilized seeds were then spread out in moist autoclaved sand and germinated at 20°C before transferring the seedlings into pots.

Conditioning phase

For the conditioning phase, we grew the plants in 0.7‐l pots filled with 500 g dry weight equivalent of live field soil. Each of the four grass species was grown in each of the 24 collected field soil samples (96 pots in total). A single 2‐wk‐old seedling was transplanted into each pot. The tools used for transplanting were wiped with 70% ethanol before setting up the next pot. All pots of the conditioning phase were set up on the same day. The pots were suspended in racks and were watered with fixed amounts of 30 ml twice a week over the growth period. The plants were grown for a total of 140 d in the glasshouse until harvest. We rinsed the roots from adhering soil and determined below‐ and aboveground biomass. The plant material was dried at 70°C for 72 h before weighing. The conditioned soil of each pot was collected and individually stored at 4°C for a maximum of 2 wk before being used as inocula for the response phase.

Response phase

The response phase was set up in two variants: one to evaluate PSF on individual plants and one to evaluate PSF on the plants when all four were grown in competition with each other. For the first, we grew the test species individually in single 0.4‐l pots filled with 285 g dry weight equivalent of pasteurized background soil inoculated with 15 g dry weight equivalent of live soil from the conditioning phase. Each of the four grass species was grown with each of the 96 soils from the conditioning phase (384 pots in total). The pots were set up within three consecutive days. On the first day, all pots were prepared and filled with background soil. On the second day, we added and mixed the inoculum soils into the background soil. On the third day, a single 2‐wk‐old seedling was transplanted into each pot. The pots were watered with fixed amounts of 25 ml twice a week.

For the competition PSF response phase experiment, four 2‐wk‐old seedlings, one from each test species, were transplanted together into 1‐l pots filled with 760 g pasteurized background soil inoculated with 40 g of live soil from the conditioning phase (96 pots in total). The pots were set up the same way as the pots for the individual response phase and watered with fixed amounts of 50 ml twice a week. For both response phase experiments, we mixed the inoculum soil into the pasteurized background soil while preserving the correspondence regarding their source locations in the field. The tools used for transplanting and mixing the soil were cleaned and wiped with 70% ethanol before setting up another pot. The plants were grown for 114 d in the individual response phase and 118 d in the competition response phase until harvest. For the individual PSF response phase, we harvested and determined below‐ and aboveground biomass as described. For the competition PSF response phase, we determined the aboveground biomass of the single plants, while belowground biomass could only be determined per pot, as the roots of the plants were intertwined and could not be unambiguously separated.

Statistical analyses

We adopted a model‐fitting approach to estimate average pairwise PSF and its uncertainty. We used generalized linear mixed‐effect models (R package glmmtmb v.1.1.10; Brooks et al., ²⁰¹⁷) and fitted total plant biomass (individual plants) or aboveground biomass (plants in competition) to a Gaussian distribution using a log‐link function. We fitted separate models for each invasion stage and included only the conditioning species by response species interaction term and no intercept as a fixed effect. We included the source location of the soil samples within each invasion stage as a random intercept in the model. For plants grown in competition, we additionally included pot identity as a random intercept. This way, the estimated fixed effect coefficients represent the average biomass production for each combination of conditioning species by response species at the log scale (total biomass for the individually grown plants and aboveground biomass for plants grown in competition). From there, pairwise PSF (I _s) was calculated according to Bever et al. (1997) as:

$$I_{\mathrm{s}}={\log }_{\mathrm{e}}\left({\mathrm{\alpha }}_{\mathrm{A}}\right)-{\log }_{\mathrm{e}}\left({\mathrm{\alpha }}_{\mathrm{B}}\right)-{\log }_{\mathrm{e}}\left({\mathrm{\beta }}_{\mathrm{A}}\right)+{\log }_{\mathrm{e}}\left({\mathrm{\beta }}_{\mathrm{B}}\right)$$

where log_e(α_A) is the biomass of Species A in conspecific conditioned soil, log_e(α_B) is the biomass of Species A in soil conditioned by Species B, log_e(β_A) is the biomass of Species B in the soil conditioned by Species A, and log_e(β_B) is the biomass of Species B in conspecific conditioned soil, all estimated at the log scale as fitted by the model. We applied a nonparametric bootstrap and refitted the models to estimate 95% confidence intervals of I _s (BCa method) based on 5000 bootstrap replicates (R package boot v.1.3‐28; Davison & Hinkley, ¹⁹⁹⁷; Canty & Ripley, ²⁰²¹). We tested for the effect of the invasion status of the habitat from which soil was sampled on average pairwise PSF using linear mixed models accounting for the specific species pair as random intercept (R package lme4 v.1.1‐32; Bates et al., ²⁰¹⁵) and assessed its significance with likelihood‐ratio chi‐squared tests. To test whether P. contorta invasion‐induced changes in PSF persist after invasion control, we assessed the correlation of the I _s values between the invaded and controlled grassland sites.

To assess the implications of PSF for the coexistence of the species, we adopted a graphical analysis approach of mutual invasibility (Bever, 2003; Revilla et al., ²⁰¹³), assuming equal carrying capacities for the species following (Ke & Miki, 2015). Therefore, we partitioned total PSF into the differences between the host plants' performance in conspecific conditioned soil compared with heterospecific conditioned soil (log_e(α_A) − log_e(α_B) for Species A, and log_e(β_B) − log_e(β_A) for Species B). Depicting these parameters in two dimensions creates four regions of potential competition outcomes between the species (dominance of Species A, dominance of Species B, dominance of either Species A or B depending on initial abundance, and coexistence of Species A and B). All analyses were performed in R v.4.2.2 (R Development Core Team, ²⁰²²).

Results

The nutrient composition was relatively similar between uninvaded and invaded soil, whereas the soil from controlled areas had higher concentrations of mineral nitrogen (NO₃–N; Table S2). Pairwise PSF estimates from the total dry biomass of plants grown individually in the response phase were predominantly negative across all species pairs when grown in soils from uninvaded areas. The pattern became more idiosyncratic in soil from areas invaded by P. contorta and where invasion had been controlled (Fig. 3a). Overall, we observed a shift to neutral or even positive PSF for the invaded and controlled invasion stage (Fig. 3b) and only A. odoratum and F. nova‐zelandieae exhibited a constantly negative pairwise PSF (Fig. 3a). The general pattern of PSF, as observed in invaded soil, persisted after pines were controlled (Fig. 3c).

Fig. 3.

Pairwise plant–soil feedback (PSF) between the four resident grass species. The plants are grown in individual pots (a–c) and in competition with each other (d–f) with soils representing different Pinus contorta invasion stages. Pairwise PSF has been estimated based on the plants' total biomass for individually grown plants and on aboveground biomass for plants grown in competition. (a, d) Pairwise PSF of each species pair; error bars represent 95% bootstrap confidence intervals. (b, e) Boxplot illustrating the general shift of PSF along the invasion stages; box and whiskers indicate the median, 25^th/75^th percentile, and 1.5 × IQR. (c, f) Scatterplot illustrating the correlation between pairwise PSF in invaded soil and after invasion control. Species codes: AC, Agrostis capillaris; AO, Anthoxanthum odoratum; FN, Festuca nova‐zelandieae; PA, Pentapogon avenoides. Native species are indicated in blue and non‐native species in red.

We also observed mainly negative PSF in uninvaded soil when the plants were grown in competition with each other (Fig. 3d) as well as a general shift towards more neutral to positive PSF for the invaded and pine control sites (Fig. 3e). Pairwise PSF in invaded and pine control soil was correlated in the competition assay, too (Fig. 3f). The intensity of PSF was generally lower in the competition experiment, where it was calculated based on aboveground biomass compared with PSF calculated for the individually grown plants. This was also the case when PSF of individually grown plants was calculated based on aboveground biomass only (Fig. S3a).

Graphical analysis of mutual invasibility indicated a shift from potentially coexistence stabilizing effects of PSF in uninvaded soil towards more destabilizing effects in invaded and controlled invasion soil, especially when the plants were grown in competition. For the individually grown plants, two of the six species pairs fell into the region where PSF can stabilize coexistence in uninvaded soil (Figs 4a, S2a). In invaded soil, PSF contributed to coexistence for only one species pair. PSF consistently benefitted the non‐native over the native species in controlled invasion soil. When the plants were grown in competition, four species pairs fell into the region of coexistence stabilizing PSF (Figs 4b, S2b). This shifted to only one or even no species pair in invaded and controlled invasion soil, respectively. In controlled invasion soil, four species pairs were in the region where PSF benefits the initially more abundant species.

Fig. 4.

Implications of plant–soil feedback (PSF) for the coexistence of the species pairs. The plants were grown in individual pots (a) and in competition with each other (b) with soils representing different Pinus contorta invasion stages. The x‐axis reflects the effect of PSF on the growth of Species A given as the log‐response ratio of biomass in conspecific soil (αA) vs soil conditioned by Species B (αB), while the y‐axis reflects the same for Species B. The quadrants represent the possible outcomes of competition for each species pair A|B based on PSF only. Regions A and B: The respective species outcompetes its competitor; Region A/B: The species with the initially higher abundance outcompetes the other; Region C: Species A and B can coexist. Species codes: AC, Agrostis capillaris; AO, Anthoxanthum odoratum; FN, Festuca nova‐zelandieae; PA, Pentapogon avenoides. Native species are indicated in blue and non‐native species in red.

Discussion

We have shown that the invasion of P. contorta into an existing grassland community can change PSF dynamics between the already resident plant species from negative to neutral and positive. The observed negative PSF between the four species in uninvaded soil is consistent with findings of predominantly negative effects of conspecific conditioned soil for graminoids (Kulmatiski et al., ²⁰⁰⁸; Meisner et al., ²⁰¹⁴; Cortois et al., ²⁰¹⁶). The shift towards more neutral or positive PSF suggests an alleviated pressure of host‐specific pathogens, an increased benefit from mutualists, or a combination thereof (van der Putten et al., ²⁰¹⁶). The invasion of P. contorta into grassland systems can cause changes to abiotic and biotic soil conditions that might affect these aspects. For example, Dickie et al. (2014) have shown that P. contorta invasion increased soil nutrient availability and shifted the system towards bacterial dominance. A reduced pressure of fungal pathogens and bolstered plant defences due to a better nutrient supply might contribute to less negative PSF in invaded and controlled invasion soil, albeit increased nutrient concentration (NO₃–N) was only evident for the latter in our experiment (Table S2). While the soil nutrient status is clearly of importance for PSF (McCarthy‐Neumann & Kobe, 2019; in't Zandt et al., ²⁰¹⁹; Dostál, ²⁰²¹), pine invasion is also accompanied by significant changes to the composition and functioning of the soil biota communities (Peralta et al., ²⁰²⁰; Sapsford et al., ²⁰²²; Sapsford & Dickie, ²⁰²³) that mediate PSF between the resident grasses. For example, Sapsford et al. (2022) have shown that invasion by the ectomycorrhizal Pinus nigra reduced the abundance of pathogenic fungi and saprotrophs, of which the latter can also be facultative pathogens (Smith et al., ²⁰¹⁷). The change in PSF from uninvaded to invaded soil, which exhibits a relatively similar nutrient status, suggests that altered soil biota communities are a significant driver for the observed shift in our experiment.

The changes in PSF induced by P. contorta invasion persisted even c. 3 yr after invasion control. Pairwise PSF in invaded soil was largely mirrored by PSF in controlled invasion soil for both the individually grown plants and plants grown in competition. Invasion‐induced changes to the composition and functioning of the soil microbiome were shown to last for several years (Sapsford & Dickie, 2023), and our results suggest that this can have consequences for soil‐mediated plant–plant interactions after invader removal. In addition to the direct legacy effects of plant invasions on biotic and abiotic conditions (e.g. Grman & Suding, ²⁰¹⁰; Elgersma et al., ²⁰¹¹; Grove et al., ²⁰¹²; Reynolds et al., ²⁰¹⁷), the resulting changes in interactions between the resident species in – this case mediated via PSF – complicate the prediction of postinvasion plant community trajectories. This might impede restoration efforts to the preinvasion state and should be accounted for in postinvasion management strategies (Dickie et al., ²⁰²²).

The general shift from predominantly negative PSF in uninvaded soil towards neutral to positive PSF in invaded and controlled invasion soil was consistent, irrespective of whether the plants were grown individually or in competition with each other, while the overall intensity of PSF was weaker for competing than individual plants. This might partially be because the calculation of PSF is based on total biomass for the individually grown plants in contrast to aboveground biomass for plants grown in competition. However, the intensity of PSF for individually grown plants was still higher when PSF calculation was based on aboveground biomass only (Fig. S3a). The pattern in PSF between the specific species pairs for the individually grown plants was largely consistent irrespective of whether PSF estimation was based on total or aboveground biomass (Fig. S3b). Furthermore, pairwise PSF of competing and individually grown plants based on aboveground biomass was only weakly correlated (Fig. S3c). Thus, we assume that direct competition between the plants influenced and partly weakened PSF as observed for the individually grown plants. This is in line with the findings of Crawford & Knight (2017) that direct plant competition can overwhelm PSF observed for individual plants. In our case, the non‐native A. odoratum seemed to be the superior competitor when the plants were grown together. This is indicated by its comparatively higher biomass in the competition assay, while the biomass of individually grown plants was more evenly distributed between the four species (Fig. S4). This observation is in line with A. odoratum exhibiting an exploitative strategy within the root economics spectrum characterized by rather unselective root placement and high specific root length (Mommer et al., ²⁰¹¹).

The change in PSF between the resident grasses from negative to more neutral or even positive following P. contorta invasion corresponds to a shift from potential coexistence stabilization to destabilization (Bever, 2003) and benefited predominantly the non‐native species. The release from negative density dependence would likely further foster non‐native over native grasses given their higher growth rates and stronger response to increased soil fertility (Scott, 1970) as observed in controlled invasion soil. For example, A. odoratum was not only the dominant species in the competition assay but also showed the highest relative biomass increase in controlled invasion soil in the response phase (Fig. S4) and the conditioning phase (Fig. S5). Thus, a release from negative conspecific feedback might further compound its competitiveness in postinvasion scenarios. This fits the observation of A. odoratum (together with A. capilaris) being a dominant species in areas where P. contorta invasion had been controlled (Dickie et al., ²⁰¹⁴).

A further implication for plant community dynamics is that a shift to more positive PSF benefits abundant over rare species. Indeed, four of the six species pairs in controlled invasion soil fell into the region where PSF would benefit the more abundant species when grown in competition with each other (Fig. 4b). This would increase the impact of priority effects (Chase, 2003; Fukami, ²⁰¹⁵), potentially rendering plant community trajectories less predictable (e.g. Dudenhöffer et al., ²⁰²²).

The disruption of coexistence stabilizing negative PSF dynamics between resident native and non‐native plant species highlights a yet underexplored mechanism by which an invading plant can alter the interactions between species in the receiving habitat. By benefiting predominantly the non‐native species, this can potentially contribute to an invasional meltdown promoting the secondary invasion by non‐native grasses. In addition, our results exemplify a novel dimension to the mounting evidence that PSF can be dependent on the environmental context, specifically representing a form of biotic context dependency.

Competing interests

None declared.

Author contributions

J‐HD developed the study conception and design. J‐HD conducted the experiment, analysed the data and wrote the original draft with input from PEH. PEH secured the project funding. J‐HD and PEH contributed to revisions and the final manuscript.

Disclaimer

The New Phytologist Foundation remains neutral with regard to jurisdictional claims in maps and in any institutional affiliations.

Supporting information

Fig. S1 Example of grassland vegetation in the study area.

Fig. S2 Implications of plant–soil feedback for the coexistence of the species pairs.

Fig. S3 Pairwise plant–soil feedback for individually grown plants estimated from aboveground biomass.

Fig. S4 Biomass of the four grasses at the end of the response phase.

Fig. S5 Biomass of the four grasses at the end of the conditioning phase.

Table S1 Herbicide formulation for aerial control of Pinus contorta infestation.

Table S2 Soil nutrients of three random samples of each invasion stage.

Please note: Wiley is not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

Data availability

The data and R‐scripts that support the findings of this study are available in the Open Science Framework (OSF) repository at https://osf.io/j6ysc/.