Introduction

There is increasing recognition of the roles that belowground factors play in the dynamics of non-native plant invasions (Reinhart and Callaway 2006; Pringle et al 2009; Dawson and Schrama 2016). Many invasive species have traits that make them strong belowground competitors, allow for rapid utilization of soil resources and modify soil ecosystem processes (Liao et al. 2008; Aerts et al. 2017; Ni et al. 2018; Caplan et al. 2019). For example, among eastern North American understory woody plants, greater rates of leaf litter production, root productivity and increased soil moisture, in conjunction with high aboveground productivity, promote faster soil nitrogen cycling and subsequent uptake in invasive compared to native species (Jo et al. 2015, 2017a). The belowground aspects to plant invasions are tightly linked to microbial communities, with invasives driving changes in diversity or shifts in taxonomic and functional community composition (Lekberg et al. 2013; Anthony et al. 2017; Farrer et al. 2021). These changes can impart positive effects on invaders or negative effects on co-occurring natives, which may offer invaders a competitive advantage (Callaway et al. 2004; Reinhart and Callaway 2006; Dawson and Schrama 2016). Given that microorganisms such as fungi are among the most diverse lineages of life on Earth, understanding how invaders influence soil microbial communities is not only important for uncovering mechanisms of invasion but also for understanding how invaders impact native biodiversity.

The mycorrhizal symbiosis can influence plant invasions and shape soil fungal communities. For example, northern hemisphere ectomycorrhizal (EM) fungi and Pinus spp. co-invade southern hemisphere habitats, driving large compositional change in both plant and soil fungal communities (Nunez et al. 2009; Dickie et al. 2010). However, the majority of plant species form symbioses with arbuscular mycorrhizal (AM) fungi, whose interactions with invasive species may be more cryptic and context specific. Relative to co-occurring natives, invasive species may have reduced AM dependency (Vogelsang and Bever 2009; Pyšek et al. 2019; Moyano et al. 2020; Ebert et al. 2023), relaxed specificity for AM fungal species (Moora et al. 2011; Rodríguez-Caballero 2018), differential preference for AM fungal species (Zhang et al. 2010) or increased AM dependency (Marler et al. 1999), any of which may promote plant invasion (Pringle et al. 2009). Furthermore, AM plants can sometimes have detrimental effects on co-occurring EM plants (Fernández et al. 2022), although the potential negative effects of invasion by AM plants on native EM plants or fungi has rarely been addressed (Grove et al. 2017a). In all cases, the impact of non-native invaders shifts fungal communities in invaded soils (Batten et al. 2006; Hawkes et al. 2006; Mummey and Rillig 2006; Zhang et al. 2010; Busby et al. 2013; Day et al. 2015), which may cause long-term alterations to ecosystems.

Pathogens also influence the dynamics of plant invasions. Dispersal into new geographic ranges may lead to release from enemies present in a plant’s native rage (Keane and Crawley 2002; Dawson and Schrama 2016). For example, Prunus serotina (black cherry) is a successful invader of European habitats due in part to its release from highly virulent pathogenic Oomycetes present in soils of its native North American range (Reinhart et al. 2010). Further, compared to invasives, native species can sometimes generate more negative plant-soil feedbacks due to pathogen buildup in the soil (Klironomos 2002). In other cases, invasives appear to facilitate the accumulation of soil pathogens to the detriment of more susceptible native vegetation (Flory and Clay 2013). In the broader context, negative plant-soil feedbacks tend to be stronger for AM plants than EM plants (Bennett et al. 2017), potentially due to the pathogen protection provided by EM fungal mantles around fine root tips. These lines of evidence suggest that invasive AM and native EM plant species should both exhibit decreased pathogen loads relative to native AM plant species.

We used DNA metabarcoding to explore the influence of plant nativity and mycorrhizal type on fine-root-associated fungal communities of 11 co-occurring eastern North American woody understory species. Our first hypothesis was, H1) fungal communities differ between invasive AM and native AM species. This may be driven by several non-exclusive causes such as invasives having different root traits relevant to their interactions with AM fungi (Ebert et al. 2023), escape from pathogens leading to a reduction of pathogenic fungi in roots of invasives, and differences in how natives and invasives influence soil chemistry. These differences may result in fungal communities on roots of invasive plants having distinct compositions, decreased diversity, and/or a decrease in pathogens. We further hypothesized that H2) fungal composition on roots of co-occurring native EM plants should be the most distinctive from all AM plant species (native and invasive). This is expected because they host EM fungal species and have different effects on soil processes than AM plants, but may also be driven by a reduction in pathogens on native EM plants. The inclusion of EM plant species provides a comparison to understand the ecological relevance of influences that nativity status may have on root-associated communities of AM plants.

Materials and methods

Study site and plant species

Our study took place in a temperate broadleaf secondary forest near Pompey, New York, USA (42.912698°, − 76.036587°, 436 m a.s.l). Long-term mean annual temperature and precipitation at the site are 7.75 °C and 124.4 cm (Brun et al. 2022). Canopy species include Acer saccharum (AM; 65% of stems > 10 cm diameter), Ostrya virginiana (EM; 15% of stems), and a mix of less abundant AM (e.g., Fraxinus americana, Prunus serotina) and EM species (Carya cordiformis, Tilia Americana; W.T. Starmer unpublished data). The majority of the site was clear-cut at least once in the nineteenth century, with some selective logging during the last century. However, one block near the edge of the sampling area (see below) included scattered old domestic apple trees (Malus pumila) within the forest canopy, suggesting it was part of an orchard in the past. The understory is composed of saplings of native AM canopy trees, localized patches of invading AM woody species, isolated saplings of native EM canopy trees, and scattered, native and invasive herbaceous AM or non-mycorrhizal species. Within an ~ 8 hectare area we delineated six non-overlapping sampling blocks ranging from 1000 to 2000 m2 in area, within which a single individual of each of the 10 most common understory woody plant species were targeted for sampling (Table 1), including three non-native invaders (Webster et al. 2006). Sampling locations under targeted individuals were at minimum 0.5 m from any other plant and were typically three or more meters apart from other sampled individuals. Carya cordiformis, which did not occur as a sapling in all blocks, was collected as an 11th species when present (n = 3, Table 1) in order to increase the representation of EM plant species. Roots collected from one of the Rosa multiflora individuals were later found to be unhealthy/dead, so were excluded from the sample set (Table 1). Heights of the sampled plants were restricted between 0.5 and 1.5 m tall (mean ± 1SD = 1.48 ± 0.60; Online Resource 1). The mycorrhizal status (AM or EM) of the lineages containing all sampled plant taxa are well established in the literature (Soudzilovskaia et al. 2020), and was corroborated with preliminary microscopic examination of cleared and stained roots of each species (LJ Lamit, unpublished data).

Table 1 The plant taxa used in this study
Full size table

Fungal community characterization

In late June 2017, roots were sampled from a single location within 0.5 m of the main stem(s) of each plant, to a depth of 15 cm. Sampling was done by hand, aided with a small shovel, and fine roots were traced to larger roots emanating from the target individual. Three or more distinct segments of fine roots (~ 1st to 3rd order) were obtained from each plant, placed in sterile sample bags using a clean gloved hand, kept on ice during field collection, and placed in a -20 °C freezer for long-term storage.

DNA was extracted from 20 to 50 mg of freeze-dried fine roots using the DNeasy PowerSoil Pro Kit (Qiagen). Prior to extraction, dry root material was pulverized to a powder by adding two 3.2 mm diameter chrome-steel beads to the initial extraction tube (also containing the zirconium beads added by the manufacture) and ground for two rounds of 30 s on a Beadmill 24 (Fisher Scientific) prior to adding the initial extraction buffer. After the extraction buffer was added, samples were ground again on the bead mill for two 30 s intervals, followed by incubation at 65 °C for 30 min. The rest of the extraction followed the manufacture’s protocol, with a 100 ul final elution. Resulting DNA was quantified with a Qubit Fluorometer (Invitrogen, Life Technologies). For general fungal community characterization, PCR (2 min at 96 °C; 30 cycles of 30 s at 94 °C, 40 s at 58 °C; 2 min at 72 °C; ten min at 72 °C) was conducted with the ITS2 region primers ITS4-fun and 5.8S-fun (Taylor et al. 2016), which also contained Illumina Nextera adaptors (with the forward adaptor attached to ITS4-fun) separated from each primer with 1–4 random bases as spacers. These primers capture a broad suite of fungal lineages, including good coverage of AM fungal lineages (Gao et al. 2019) not often capture by ITS primers. ITS2 amplicons were submitted to the Genomics Core Facility at Cornell University Institute of Biotechnology where 8 bp indexes plus Illumina sequencing adaptors were ligated to each end of the amplicons and Ampure bead clean-up (Beckman Coulter) was conducted after pooling samples. The Illumina MiSeq platform was used to obtain 2 × 250 bp paired end reads.

Bioinformatic processing of sequence data proceeded as follows. First, residual Illumina adapters, Nextera adaptors and PhiX 174 were filtered with BBDuk (sourceforge.net/ projects/bbmap/), while 3′ and 5′ PCR primer artifacts were trimmed with cutadapt 3.2 (Martin 2011). Conserved sequences flanking the ITS2 region were removed with a hard trim of 131 bp from the 5’ end of the forward reads to remove the 5.8 s region and 29 bp from read 2 to remove bases of the 28S region. Sequences were then imported into Qiime2 (Bolyen et al. 2019) where DADA2 (Callahan et al. 2016) was used for read joining, ASV (Amplicon Sequence Variant) formation, quality filtration (expected error rate = 2), and removal of chimeras. Because the ITS2 region has natural length variation among fungal taxa, the 3’ ends of reads 1 and 2 were trimmed to a quality score of 5 but were not subject to trimming to a uniform length prior to read joining and ASV formation. ASVs were then de novo clustered into operational taxonomic units (OTUs) at 97% similarity using the Qiime2 VSEARCH plug-in (Rognes et al. 2016). Taxonomic assignments were conducted using the Qiime2 naïve Bayes feature-classifier (Pedregosa et al. 2011; Bokulich et al. 2018) with the UNITE 9.0 all eukaryote dataset (Nilsson et al. 2019; Abarenkov et al. 2022), and OTUs not classified as fungi were removed from the dataset. Functional groups were assigned using FungalTraits (Põlme et al. 2020), followed by manual curation. Scaling with ranked subsampling (Beule and Karlovsky 2020) was applied to the dataset (4500 sequences per sample) prior to all further analyses.

Additional plant and soil data

A subset of fine roots (1st–3rd order) were scanned at 400 dpi (higher dpi values were tested on a subset of samples and did not provide increased accuracy), and used for root trait measurements (Online Resource 1). Total root length and average diameter were obtained from images with WinRHIZO (Regent Instruments Inc.). Scanned roots were dried and weighed to calculate specific root length (SRL; m g−1) and root tissue density (RTD) assuming a cylinder (g cm−3).

A set of soil variables were collected from the same plants in coordination with root sampling (Online Resource 1). We utilized soil data originally reported in Hull et al. (2020), included average leaf litter depth, total soil carbon and nitrogen, carbon to nitrogen ratio, soil ammonium and nitrate concentrations, lab-based soil nitrogen mineralization potential (ug N [g soil]−1 d−1), and lab-based carbon mineralization potential (mg C [g soil]−1d−1). To obtain soil for these measurements, five cores (3.5 cm diameter, 5 cm depth) from beneath the canopy of the target plant were pooled. pH was measured with a handheld meter in a 1:2 (mass:volume) soil to water slurry made from archived dry soil collected by Hull et al. (2020).

Statistical analyses

Unless otherwise noted, analyses were run in R 4.0.3 (R Core Team 2020). Univariate fungal, plant and soil variables were tested with linear mixed models (lmerTest package; Kuznetsova et al. 2017) that included the fixed effect of plant type (native AM, invasive AM, native EM). Models also included plant species as a random intercept effect nested within plant type, and survey block as a random effect. F-ratios were obtained using the Kenward-Rogers approximation, and marginal means were estimated with the emmeans package (Russell 2020).

Multivariate OTU composition was tested using permutational multivariate analysis of variance (PERMANOVA; run in Primer 6.1.15 with permanova + 1.0.5, Anderson et al. 2008) and visualized using non-metric multidimensional scaling (NMDS; using the R vegan package, Oksanen et al. 2020), both with Bray–Curtis dissimilarity. PERMANOVA followed the same model structure described for univariate analyses, above. When relevant, post-hoc pairwise PERMANOVA were used to compare OTU composition among plant types. The aforementioned multivariate analyses were each conducted using the full OTU matrix, the OTU matrix after removing mycorrhizal taxa (AM and EM), and the OTU matrix with only AM fungi with AM plant species. PERMANOVA was initially run on the full OTU matrix using sequence counts relativized as a proportion of the total sequences in a sample and then again after Wisconsin double relativization to down-weight the influence of the most abundant OTUs; in contrast, all NMDS ordinations, and all PERMANOVA on subsets of the community, were only run after Wisconsin double relativization. Vectors of individual plant and soil variables were fit to NDMS ordinations using the ‘envfit’ function in vegan. Soil and plant variables exhibiting the strongest significant relationships in ordination vector analyses were then added as covariates to additional PERMANOVA tests following the model structure outlined above.

Indicator species analysis (R indicspecies package; De Cáceres and Legendre 2009), was used to examine the affinity of the most abundant OTUs to each plant type (native AM, invasive AM, native EM) plus native AM + invasive AM plants combined, and to each individual plant species. Only OTUs present in at least two samples and comprising the top 20% of the most abundant OTUs (when ranked by total sequence count in the dataset) were utilized (277 OTUs).

Results

The fungal community

A total of 1385 fungal operational taxonomic units (OTUs, 97% similarity) were recovered from 279000 total sequences (4500 seq per plant × 62 plants). The OTUs represented 12 fungal phyla, comprised of at least 46 fungal classes, in addition to a number of less taxonomically resolved lineages (Online Resource 2). The majority of OTUs were Ascomycota (51.13% of sequences, 40.07% of OTUs, 14 classes) followed by Basidiomycota (34.69% of sequences, 30.47% of OTUs, 11 classes) and Glomeromycota (11.32% of sequences, 14.37% of OTUs, 3 classes; Online Resource 2). Saprotrophs dominated the dataset (45.21% of sequences, 33.50% of OTUs), followed by EM fungi (14.19% of sequences, 6.50% of OTUs) and AM fungi (i.e., Glomeromycota) (Online Resource 2; Fig. 1). Although AM and EM fungal sequences were found in association with most root systems (Online Resource 2; Fig. 1), AM fungal sequences had lower relative abundances in samples from EM plants and EM fungal sequences were far less common in samples from AM plants (Online Resource 2; Fig. 1) suggesting they were only incidentally associated with non-host species.

Fig. 1
figure 1

Relative abundances of fungal functional groups by plant species. Species codes: LOBE = Lonicera x Bella, ROMU = Rosa multiflora, RHCA = Rhamnus cathartica, FRAM = Fraxinus americana, PRSE = Prunus serotina, PRVI = Prunus virginiana, RICY = Ribe cynosbati, ASCI = Acer saccharum, CACO = Carya cordiformis, OSVI = Ostrya virginiana, TIAM = Tilia Americana. AM = Arbuscular mycorrhizal plant, EM = ectomycorrhizal plant

Full size image

Fungal diversity

Consistent with H2 but not H1, fungal OTU diversity was higher in AM plants than EM plants, but not different between native and invasive AM plant species. Shannon’s diversity differed significantly by plant type (Table 2) with diversity decreasing by nearly 25% from the invasive AM plants to native EM plants, and native AM plants being somewhat intermediate but more similar to invasive AM plants (Fig. 2a). OTU richness and evenness, the component parts of Shannon’s diversity, showed similar overall patterns among the three plant types as Shannon’s diversity, however these were not statistically significant (Table 2; Fig. 2b, c). OTU richness and Shannon’s diversity (to a lesser extent) also varied within species (Table 2; Fig. 2). Carya cordiformis was distinct by having particularly low OTU diversity.

Table 2 PerMANOVA and linear mixed model results for fungal responses to fixed effects (bold = P < 0.10, italics = P ≤ 0.05)
Full size table
Fig. 2
figure 2

Species means (± 1 SE), and marginal means (± 1 SE) of three types of plants, for Shannon’s OTU diversity index (a), OTU richness (b), and OTU evenness (c). Species codes: LOBE = Lonicera x Bella, ROMU = Rosa multiflora, RHCA = Rhamnus cathartica, FRAM = Fraxinus americana, PRSE = Prunus serotina, PRVI = Prunus virginiana, RICY = Ribe cynosbati, ASCI = Acer saccharum, CACO = Carya cordiformis, OSVI = Ostrya virginiana, TIAM = Tilia Americana, Inv_AM = Invasive arbuscular mycorrhizal, Nat_AM = native arbuscular mycorrhizal, Nat_EM = native ectomycorrhizal

Full size image

Fungal pathogens

Inconsistent with H1 and H2, there was no evidence that pathogens varied with host nativity, mycorrhizal type or plant species. The results were corroborated when examining overall relative abundances of pathogens, observed pathogen OTU richness, and pathogen OTU richness expressed relative to the total number of pathogen sequences in an observation (Table 2; Fig. 1). We note that many of the fungal taxa annotated as pathogens or pathogens/saprotrophs were not taxa that target plant roots (Online Resource 2).

Fungal OTU composition

OTU composition of the full fungal community was similar between native and invasive plants, but distinct between AM vs EM plants, consistent with H2 but not H1. PERMANOVA results indicated that plant type influenced OTU composition when sequences were relativized as proportion of the total within observations (Table 2). However, plant type was reduced to marginal significance after Wisconsin relativization, indicating that the effect of plant type was driven by dominant OTUs (Table 2). Ordination suggested a division in fungal composition between invasive and native AM plant species (Fig. 3a). However, post hoc pair-wise PERMANOVA using relative abundances indicated that the effect of plant type was driven primarily by a distinction in composition between AM plants and EM plants (relative abundance data: Invasive AM versus Native AM t6.21 = 1.06, P = 0.272; Invasive AM versus Native EM t4.27 = 1.28, P = 0.056; Native AM versus Native EM t7.33 = 1.40, P = 0.036), and down-weighting dominant OTUs emphasized that the distinction was primarily between native AM and native EM plant species (Wisconsin relativized data: Invasive AM versus Native AM t6.26 = 1.04, P = 0.253; Invasive AM versus Native EM t4.33 = 1.06, P = 0.154; Native AM versus Native EM t7.71 = 1.12, P = 0.060). Individual plant species nested within plant types also had a distinctive influence on fungal OTU composition, which was statistically significant before and after down-weighting dominant OTUs (Table 2; Fig. 3a).

Fig. 3
figure 3

NMDS ordinations of the full fungal community (a, b), all fungi excluding mycorrhizal fungi (c, d) and only arbuscular mycorrhizal fungi (e, f), showing means (± 1 SE) of ordination axes for each plant species labeled with species codes (a, c, e), and arrows represent significant envfit vectors (b, d, f). Small symbols in the background represent fungal communities associated with individual plants. Arrow lengths are scaled to the magnitude of their correlations through ordination space; dashed arrows = P < 0.10, solid arrows = P ≤ 0.05 (see Online Resource 4). Species codes: LOBE = Lonicera x Bella, ROMU = Rosa multiflora, RHCA = Rhamnus cathartica, FRAM = Fraxinus americana, PRSE = Prunus serotina, PRVI = Prunus virginiana, RICY = Ribes cynosbati, ASCI = Acer saccharum, CACO = Carya cordiformis, OSVI = Ostrya virginiana, TIAM = Tilia americana, Inv_AM = Invasive arbuscular mycorrhizal, Nat_AM = native arbuscular mycorrhizal, Nat_EM = native ectomycorrhizal. Vector arrow variables: Nmin = soil nitrogen mineralization potential, LitDepth = average litter depth, pH = soil pH, Cmin = soil carbon mineralization potential, SRL = specific root length

Full size image

Additional analyses with subcomponents of the community pointed towards the importance of mycorrhizal fungi (AM and EM) as drivers of the patterns observed with analysis of the full community. First, after removing the mycorrhizal fraction of the fungal community (i.e., excluding AM and EM fungi), the effect of plant type became non-significant (Table 2; Fig. 3c). Second, considering only AM fungi in association with AM plants revealed a compositional distinction between native and invasive plants (Table 2; Fig. 3e). This result shows that the AM component of the fungal communities was sorting by host nativity.

Indicator species analysis highlighted plant type and plant species effects on the most common OTUs (Online Resource 3). When pooling all plant species by plant type, invasive AM plants had similar numbers of indicators (17 OTUs) as the pooled invasive AM + native AM plants category (18 OTUs), while native AM plants on their own had far fewer indicators (4 OTUs; Online Resource 3). In all cases, each of the AM plant types (or combined AM plant types) had AM fungal OTUs as indicators, while the invasive AM plants and the pooled invasive AM + native AM group both also had a number of saprotrophs as indicators. Indicator analyses at the plant species level revealed that Rhamnus cathartica had 18 indicator OTUs, primarily AM and saprotrophic fungi, while all other AM plant species only had two to six indicators OTUs. Cross examination of the results for R. cathartica with those from the analysis with plants pooled by plant type reveals that many of the OTUs indicative of R. cathartica were also indicators of the invasive AM plant type or the pooled invasive AM + native AM plants groups (Online Resource 3). When pooled together, indicators of EM plant hosts (14 OTUs) were primarily EM fungi, while analysis at the plant species-level suggested a degree of specificity of some of the EM fungi for individual host plant species (Online Resource 3).

Soil traits and fungal OTU composition

Fungal composition was associated with soil variables in ways that were independent of host plant type or species. Soil variables only varied significantly among sampling blocks, highlighting their independence from plant type or species (Table 3). Vector analysis (Online Resource 4) with the full fungal community revealed a gradient associated with increasing pH and decreasing carbon mineralization potential, and an orthogonal gradient associated with increasing nitrogen mineralization potential and litter depth (Fig. 3b). After excluding mycorrhizal OTUs from the dataset, the first of these gradients remained but the second did not (Fig. 3d). Ordination with only AM fungi in association with AM plants also indicated a shift in composition with increasing pH and decreasing soil respiration potential, with nitrogen mineralization potential also significant (Fig. 3f). After inclusion of the strongest soil vector, pH, as a covariate in PERMANOVA with the full OTU community, the effect of plant type became significant where it had been only marginally so without the covariate (Table 2). This effect was driven by a distinction between native AM and native EM plants and a marginal difference between invasive AM and native AM communities (Wisconsin relativized data: Native AM versus Native EM t7.71 = 1.12, P = 0.031; Invasive AM versus Native AM t6.26 = 1.05, P = 0.079; Invasive AM versus Native EM t4.33 = 1.07, P = 0.143). In contrast, the addition of pH as a covariate did not alter the results for plant type and plant species when only considering non-mycorrhizal OTU composition in the full dataset or only AM fungal OTUs in association with AM plants (Table 2).

Plant traits and fungal OTU composition

Plant traits were variable among plant types and species but not sampling blocks (Table 3). Invasive AM plants had higher SRL than native AM species, while native EM plants were intermediate (marginal mean ± 1SE: invasive AM = 43.67 ± 0.10 m g−1, native AM = 29.23 ± 0.07 m g−1, native EM = 34.41 ± 0.10 m g−1; Table 3). In contrast, fine root diameters of native AM plants were greater than those of native EM plants, with invasive AM plants intermediate (marginal mean ± 1SE: native AM = 0.34 ± 0.01 mm, native EM = 0.27 ± 0.02 mm, invasive AM = 0.30 ± 0.02 mm; Table 3). RTD was similar for native and invasive AM plants, whereas native EM plants exhibited lower RTD than the AM plants (marginal mean ± 1SE: native AM = 0.39 ± 0.03 g cm−3, invasive AM = 0.35 ± 0.04 g cm−3, native EM = 0.53 ± 0.04 g cm−3; Table 3).

Of the examined plant traits, only SRL was related to fungal community composition. In NMDS ordinations, SRL was a significant vector (Online Resource 4) when considering AM fungi in association with AM plants (Fig. 3f), but not when considering the full fungal community or only non-mycorrhizal fungi (Fig. 3b, d). When SRL was added as a covariate to PERMANOVA with AM fungi in association with AM plant species, the effects of plant type (i.e., native vs non-native AM plants) and SRL were both significant (Table 2).

Discussion

We show that non-native, invasive understory woody plants can have distinct root-associated fungal communities relative to co-occurring native plants. These effects were more evident in some aspects of the fungal community than others. For example, nativity influenced dominant fungal species and AM fungal composition, but not pathogens and overall fungal diversity. These patterns may be due to the tendency of invaders to alter litter and soil chemistry (Batten et al. 2006; Mummey and Rillig 2006; Jo et al. 2015, 2017a), but two lines of evidence suggest the nativity effect in our study was due to direct interactions between roots and fungi. First, the clearest distinction between fungal communities of native AM and invasive AM plants was exhibited by the AM fungal portion of the community, which colonized the root cortex of all the non-EM plants in our study. Second, soil properties associated with nutrient cycling, pH and litter input were similar in soil beneath native and invasive species (Hull et al. 2020), making it unlikely that fungal communities were structured indirectly through plant nativity-specific alterations to soil conditions. At our site, the distribution of invasive species was represented by lone individuals or localized patches dispersed among native vegetation, which indicates the early stages of invasion. This suggests that the initial influence of invasive plants on fungal communities can occur through direct interactions in the rhizosphere, independent from changes in soil chemistry that may accrue over a longer time period.

Of the root traits examined, SRL showed a clear link to AM fungal community composition. Invaders, including a wide range of Eastern North American understory woody plants (Jo et al. 2015), often have high SRL and other fine root traits indicative of rapid resource utilization. Root economics theory posits that AM plants with high SRL should also be less dependent on mycorrhizal fungi for resource acquisition (Bergmann et al. 2020). This contention is supported by a recent trial where invasive species from multiple Eastern North American understory woody plant families grew faster, had greater SRL and received less growth benefit from AM fungal inoculation, than congeneric native species (Ebert et al. 2023). It is possible that AMF OTUs preferentially associated with invasives in our study system may be less beneficial mutualists because they are associating with plants that do not have a strong dependence on them, although inoculation trials with individual AM fungal isolates are needed to verify this idea.

Table 3 Means (± 1standard deviation) and mixed-model results for plant trait and soil variables (bold = P < 0.10, italics = P ≤ 0.05)
Full size table

We found that individual plant species were associated with distinct fungal communities. For example, C. cordiformis had consistently low OTU diversity relative to all other taxa, likely because it was strongly dominated by a single fungal genus, the ectomycorrhizal Hymenogaster (~ 50 of sequences on average; supplemental document Online Resource 2). Of the invasives, R. cathartica stood out as having a unique fungal community, including a high number of AM fungal and saprotrophic indicator OTUs. R. cathartica may impact soil processes through its elevated production of fast decomposing litter that is high in nitrogen, and it also produces chemicals that may have allelopathic effects in soils (Knight et al. 2007; Warren et al. 2017). While both of these factors could impact root-associated fungi, the lack of a plant species effect on soil variables in our system indicate soil nutrient and carbon cycling process were not the dominant driver of fungal community change under R. cathartica. We also suggest that negative effects of allelopathic chemicals on fungi is unlikely, given that R. cathartica had some of the highest fungal diversity, and Lonicera x bella lacked the same distinctive effects on fungal communities despite some Lonicera species being putatively allelopathic (Skulman et al. 2004; Dorning and Cipollini 2006). Our results point to factors related to direct interactions with fungi at the root-soil interface as drivers of the unique communities associated with R. cathartica roots, potentially related to its high SRL and low RTD. Interestingly, R. cathartica is now the most abundant woody species in our study region (supplanting the dominant native tree A. saccharum within the last decade; City of Syracuse 2016). Although the nativity status of its fungal associates is unknown, their unique community composition is alarming, given the potential for ‘invasional meltdown’ (sensu Simberloff and Von Holle 1999) as R. cathartica may accelerate the shift of both plant and soil fungal communities away from native species.

The distinction between AM and EM plants necessitates another consideration when assessing the impacts of invasive species on soil fungi. Forests dominated by AM trees tend to be more prone to invasion by understory plants (Jo et al. 2017b). Moreover, competition with non-EM woody plants, including invasive species, can negatively impact EM plant performance and shift EM fungal communities (Meinhardt and Gehring 2012; Grove et al. 2017b; Fernández et al 2022). Although we found that components of the fungal communities were unique between invasive AM and native AM plants, invasive AM plants could in theory still host some of the AM fungi that associate with native AM plants. In contrast, AM invaders have no capacity to host EM fungi associated with native EM plants. Increases in the density of aggressive AM invaders to the point of competitive exclusion of native EM plants, would lead to the total loss of EM fungal taxa. This may be a plausible outcome if invasives become dominant in our system over time, as they have in other areas of central New York state, given that EM saplings are a small subset of the forest understory.

Belowground enemy release may play a role in some invasions (Reinhart et al. 2010), but our study did not provide support for this mechanism. We did not observe a greater relative abundance or OTU richness of pathogens on native AM plants compared to invasive AM plants. Furthermore, the majority of putative pathogens annotated in our dataset are not root specialists, indicating that the majority of this small fraction of the fungal community was not functioning as active pathogens. A caveat to this conclusion is that relative abundances based on DNA sequence data do not necessarily reflect absolute abundances. It is therefore theoretically possible for pathogenic species to be more abundant on native taxa if there are more fungal individuals overall, even if there is no difference in their relative abundances. An additional point to consider is that many root pathogens, including nematodes and Oomycetes, were not targeted with our fungal primer set. Consequently, we cannot entirely rule out release from belowground enemies as a mechanism promoting invasion in our system.

We also found that soil fungal communities were associated with some soil properties, independent of plant composition. One of the strongest gradients was associated with pH and soil C mineralization potential, indicating the presence of more labile C and/or greater saprotrophic fungal activity at lower pH in our sites. Although the lack of connection between plant identity and soil properties can in part be explained by the relatively diffuse distribution of invaders at our site, it is surprising that EM plants also did not shift soil processes, given that EM symbioses are associated with distinct soil nutrient cycling dynamic compared to those of AM symbioses (Read and Perez-Moreno 2003; Phillips et al. 2013). However, as with the invasive AM plants, small individual native EM saplings were likely not at great enough density to alter the likely overwhelming signature of the overstory trees.

The results presented here highlight the impact that invasive AM woody plants can have early in the invasion process. This has distinct ramifications for native AM and EM plants in forest understories, which present unique consequences for fungal diversity even in the initial absence of other modifications by invasives to the soil ecosystem (i.e., nutrient and carbon cycling). Our results with R. cathartica also point to the non-uniformity of how invaders interact with soil fungi; each invader may have unique impacts on soil fugal communities beyond those common to invaders as a group. Surprisingly, we did not detect signs that fungal pathogens were influenced by plant nativity, although further work is needed to fully understand the potential role of enemy release on the invasion process of our site. Some of the most impacted fungi were mycorrhizal taxa, likely because of their more intimate relationship with plant hosts than the saprotrophs that dominate the remainder of the fungal community; this suggests that we can look to plant-mycorrhiza interactions for early signs that invaders are influencing a system. Taken together, our results point to the sensitivity of fungal communities to host plant nativity, which may have long-term consequences for soil biodiversity and feedback to shape the further invasion of non-native woody plants in forest understories.