Characterization of root-associated fungi and reduced plant growth in soils from a New Mexico uranium mine

ABSTRACT Characterizing the diverse, root-associated fungi in mine wastes can accelerate the development of bioremediation strategies to stabilize heavy metals. Ascomycota fungi are well known for their mutualistic associations with plant roots and, separately, for roles in the accumulation of toxic compounds from the environment, such as heavy metals. We sampled soils and cultured root-associated fungi from blue grama grass (Bouteloua gracilis) collected from lands with a history of uranium (U) mining and contrasted against communities in nearby, off-mine sites. Plant root-associated fungal communities from mine sites were lower in taxonomic richness and diversity than root fungi from paired, off-mine sites. We assessed potential functional consequences of unique mine-associated soil microbial communities using plant bioassays, which revealed that plants grown in mine soils in the greenhouse had significantly lower germination, survival, and less total biomass than plants grown in off-mine soils but did not alter allocation patterns to roots versus shoots. We identified candidate culturable root-associated Ascomycota taxa for bioremediation and increased understanding of the biological impacts of heavy metals on microbial communities and plant growth.


INTRODUCTION
Fungi are well known for their mutualistic associations with plants (Clay 1988;Rodriguez et al. 2009) and, separately, for their roles in the accumulation of toxic heavy metals in the environment (Burges et al. 2017;Deng and Cao 2017;Harms et al. 2011;Hassan et al. 2019). Most studies on heavy metal toxicity study fungi in isolation (e.g., in laboratory cultures; Yang et al. 2012;Zhang et al. 2008), whereas studies on plant-fungal symbioses focus on fungal benefits to plants rather than the synergistic potential for plants to influence how fungi interact with their soil environment. For example, the best candidate fungi from laboratory studies on bioremediation of heavy metals (phylum Ascomycota) are also separately documented to grow commonly in plant roots (Loro et al. 2012;Mandyam and Jumpponen 2005), especially in arid regions (Herrera et al. 2010;Knapp et al. 2012). Some arbuscular mycorrhizal (AM) fungi alleviate stress for plants in mine soils by accumulating heavy metals such as cadmium (Cd) at 10-20 times higher rates than plant roots (Janousková et al. 2007). In stressful systems, AM fungi can be stress alleviators, increase increasing overall plant growth and enhancing heavy metal bioaccumulation and translocation into the plant shoot (Miransari 2010). However, recent studies show that Ascomycota fungi may be better suited for bioremediation than AM fungi, in part because they have been found to be the dominant phylum in areas of high contamination, such as wastewater sludge and mine sites (Evans and Seviour 2012;Jia et al. 2018;Rosenfeld et al., 2018). Thus, nonmycorrhizal fungal endophytes may feature as key microbes in interactions between heavy metals and plants through their influences on soil biochemistry.
Plant-fungal interactions may provide unrealized microbial potential to stabilize metal mixtures, but few studies have characterized fungal communities in roots of plants in metal-contaminated sites, particularly for nonmycorrhizal fungi. Taxa in the Ascomycota have not been well studied in plants under heavy metal exposure, even though they are equally or more abundant in plant roots than arbuscular mycorrhizal (AM) fungi, particularly in dryland grasses (Jumpponen et al. 2017;Márquez et al. 2012). Specifically, Ascomycota root endophytes can have primary roles in alleviating plant drought and nutrient stress (Colpaert et al. 2011;Govarthanan et al. 2019;Kivlin et al. 2013;Newsham 2011). Plant-associated Ascomycota with high melanin production may rank among the best candidates for stress resistance in plants. Melanin pigments can adsorb radionuclides, including uranium (U) (Keith et al. 2013;McLean et al. 1998;Yang et al. 2012), and melanic fungi commonly dominate contaminated sites as well as dryland ecosystems (Cordero et al. 2017;Dighton et al. 2008;Khidir et al. 2009). Thus, contaminated sites in drylands could provide rich biorepositories to prospect for fungal taxa that maximize bioremediation. Common Ascomycetes with potential to stabilize metal mixtures include Aspergillus, Chaetomium, Fusarium, Penicillium, Talaromyces, and Trichoderma (Akhtar et al. 2007;Iram et al. 2009;Lh et al. 2019;Mohd et al. 2017;Segura et al. 2018). These taxa also commonly occur in symbiosis with plant roots as reported in published literature and biogeographic surveys (Jumpponen et al. 2017;Lagueux et al. 2021;Porras-Alfaro and Bayman 2011).
Mining legacies can not only alter microbial communities but also directly impact plant productivity and human health. For example, heavy metals in untreated wastewater can increase metal concentrations in plant tissues (Hussain et al. 2019;Lu et al. 2015). Especially in the Southwest of the United States, abandoned uranium mines disproportionally impact Native American communities, including agricultural lands in which produce is primarily bought by and sold to local community members (Arnold 2014;Yazzie et al. 2020). Heavy metal toxicity has major human health impacts, including lung and bone cancers, kidney damage, high blood pressure, autoimmune disease, and loss of reproductive function (Keith et al. 2013;Zhang et al. 2020). Expanding our understanding of how heavy metals affect plants and plant-microbe interactions can help improve both agricultural and health outcomes for local communities.
We aimed to identify candidate fungi for bioremediation by sampling replicated fungal communities in plant roots across sites with versus without mining legacies. We focused on U mining-contaminated areas in the semiarid drylands of central New Mexico. Here, the Grants mineral belt contains the second largest known U ore reserve in the United States, where average U concentrations average 5 ppm higher than the national average (Brookins 1979;Keith et al. 2013). High levels of U as well as other toxic metals (V, As) surround abandoned mine sites (Blake et al. 2017;Hoover et al. 2017). We surveyed the composition, abundance, and diversity of root-associated endophytic fungi. In addition, we used greenhouse experiments with the focal plant species, blue grama grass, to determine the effects of contaminated soils on plant production. We addressed the following questions.
(i) Do plants growing in mine sites have different root fungal communities than plants outside of mine sites? We hypothesized that fungal communities associated with plants growing in mine soils will differ from fungi in plants growing in off-mine soils, possibly driven by greater relative abundance of stress-tolerant species. (ii) Do soils from mine sites reduce plant growth and fungal colonization relative to soils outside of mines? We predicted that plants grown in soils from mine sites would have lower germination and survival, produce less biomass, and allocate more biomass to roots (larger root:shoot ratio) than plants grown in off-mine soils, possibly due to the toxic impacts of heavy metal contamination and associated stress-tolerant microbial communities.
The information obtained from this study provides new insights about fungal community composition and plant productivity in drylands affected by mining legacies, with implications for the improvement of remediation strategies for toxic metals.

MATERIALS AND METHODS
Study sites.-Samples were collected from three mine sites and three off-mine sites in the Pueblo of Laguna on 25-26 Aug 2020. Mine sites were sampled from the interior pit of the Jackpile-Paguate uranium mine. This mine was the largest open-pit uranium mine of its time, consisting of two main pits affecting 1271 hectares of land in operation from 1953 to 1982; it was backfilled in 1985 (Zehner 1985). Radiation at mine sites ranged from 20 to 77 cmp (counts per minute) during the root collection period. Off-mine sites were sampled from the wild landscape of the nearby Pueblo of Laguna in the village of Paguate. Mine soils ranged in U concentration from 4.4 to 48.4 ppm and off-mine soils ranged in U concentration from 1.5 to 1.9 ppm, compared with the average soil U concentration in the United States of 3 ppm (Keith et al. 2013) (SUPPLEMENTARY TABLE 1). All sites had similar soil pH (8.3-8.9) and shared similar concentrations of Ba, Ca, Cr, Cu, Fe, Mn, Na, Ni, P, Sr, Zn, As, U, V, and Pb. All sites were within a 3-km radius of off-mine site 1; thus, we assumed that climate was consistent across sites (SUPPLEMENTARY FIG. 1). Off-mine sites, rather than fully noncontaminated reference sites, were chosen in order to maintain results relevant to the impacted Native American community and to provide controls with similar geology, climate, and community composition to the mine sites.
Sample collection.-Root samples were taken from the most widespread native grass species in North America, blue grama grass (Bouteloua gracilis). We selected three mine and three off-mine sites for a total of six sites (SUPPLEMENTARY FIG. 1). At each site, we randomly selected nine plants for root sampling at ~5 m spacing. We also collected ~1667 g of soil from the rooting zone of each plant (soil depth 0-15 cm), for use in bioassay experiments in the greenhouse. Blue grama roots are shallow and primarily occupy 0-20 cm soil depth in desert grasslands, although roots have been reported as deep as 1.8 m for this species, for which the geographic range extends north into Canada (Anderson 2003;Weaver 1958). Due to COVID-19 and limited access to supplies and laboratory facilities, sample processing was prolonged; roots were surface-sterilized and plated within 23 days after sampling, with n =12 root samples plated within 48 h after sampling (2 plants from each site) and n = 42 plated within 17 days after sampling (N = 54). Because sample processing was randomized over time, the delay in plating is unlikely to have influenced the detection of differences in fungi among sites but may have increased or reduced absolute species richness across the sample set.

Root sterilization
We stored root samples at 4 C before processing. Roots were washed with deionized (DI) water to remove excess soil then cut into ~10 3-cm segments, placed into tissue cassettes, and stored in 70% ethanol (EtOH) for future microscopy. The remaining ~10 5-cm segments were surface-sterilized as follows: washed 1 min in 95% EtOH, transferred to 1% sodium hypochlorite for 2 min, transferred to 70% EtOH for 2 min, and rinsed three times with DI H 2 O (Higgins et al. 2011;Massimo et al. 2015). We placed roots in small strainers to transfer them among solutions to avoid contamination.

Root culturing/subculturing
We surface-sterilized roots and cut them into ~3-cm sections, using sterile technique, and then placed root segments into 10-cm malt extract agar Petri plates amended with antibiotics (streptomycin and penicillin, 1 mg/L) ("MEA plates" hereafter) to prevent bacterial contamination We inserted 10 3-cm segments of root into each MEA plate using surface-sterilized forceps, burying both ends of the root segment into the medium. We organized culture plates such that one plant from each site was processed at the same time so that a full representation of sample types was cultured on any one day. Once fungi began to grow from the roots, individual isolates were subcultured and transferred, using sterile technique, onto a 5-cm MEA plate. Isolates received unique IDs tracking their plot and individual plant of origin. We separated isolates into morphotypes, which were then processed for DNA extraction and Sanger sequencing for taxonomic identification.

Root staining and microscopy
Roots were cleared and stained using the modified Vierheilig method (Herrera et al. 2010;Vierheilig et al. 1998). At 200× on a bright-field microscope, we estimated the proportion of 100 intercepts that intersected fungal structures (McGonigle et al. 1990), including melanized (dark) septate hyphae, hyaline septate hyphae, arbuscles and vesicles of arbuscular mycorrhizal fungi, and aseptate hyphae of Glomerales and basal fungal lineages.

Fungal identification
We extracted DNA from a subset of isolates (96 of 181 isolates) representing each fungal morphotype using the QIAGEN DNeasy 96 Plant Kit (Germantown, MD, USA). Isolates were amplified via polymerase chain reaction (PCR) using standard fungal primers ITS1-FL and TW13 (Gardes and Bruns 2008;Taylor et al., 2016) and Sanger sequenced (Genewiz, South Plainfield, New Jersey; Lyons et al. 2021). We acknowledge that the internal transcribed spacer (ITS) sequencing region cannot parse some Ascomycota genera, including Fusarium and Darksidea (Romero-Jiménez et al. 2022;Torres-Cruz et al. 2022). We lacked additional resources necessary to sequence these loci due to the large size of our study. Given our questions and the robustness of community composition metrics to differing cutoffs for clustering (because additionally splitting typically produces more rare species) (Matute and Sepúlveda 2019), the impact of additional resolution on species identifications on pairwise dissimilarity scores is minimal. The forward and reverse reads were assembled using Geneious, sorted by length using the sortbylength command, and clustered into operational taxon units (OTUs) at 97% similarity via the cluster_smallmem command in USEARCH (11.0.667, 32 bit) (Edgar 2010). Each isolate was matched to an OTU at 97% similarity using usearch_global from USEARCH. The taxonomic identity for each OTU was determined with SINTAX (Edgar 2016) set to a cutoff of 0.8 and using the UNITE database (8.2, general FASTA release for fungi) (Abarenkov et al. 2010). Sequence data were archived in GenBank, accession numbers ON815377-ON815468.
Soil chemistry.-At each site, we collected a composite soil sample by combining ~1 kg of soil in equal amounts by volume from the rooting zones (soil depth 0-15 cm) of 12 focal plants, including the nine plants used for culturing. Soils were stored at 4 C then subjected to the following chemical analyses. Soil samples were dried overnight in an oven at 60 C in a controlled-temperature oven (Devore et al. 2019). Samples were sieved with a US Standard no. 230 mesh (63 µm) then analyzed for total elements content (Ba, Ca, Cr, Cu, Fe, K, Mg, Mn, Na, Ni, P, S, Sr, Zn, As, U, V, and Pb) (Blake et al. 2015). Triplicate of homogenized samples (1 g each) were weighed in 50-mL polypropylene tubes for acid digestion. Soil samples were digested by adding 2 mL of concentrated nitric acid (HNO 3 ), 4 mL of concentrated hydrochloric acid (HCl), and 2 mL of concentrated hydrofluoric acid (HF) (Rahman et al. 2018;Velasco et al. 2019). All reagents used were trace metal grade and high purity. Following heating, acid digested solutions were filtered and diluted with 18 MΩ water to 25 mL to be analyzed using inductively coupled plasma optical emission spectrometry (ICP-OES; PerkinElmer Optima 5300 DV; Waltham, MA, USA) and inductively coupled plasma mass spectrometry (ICP-MS; PerkinElmer NexION 300D). Total organic and inorganic carbon content in soil samples was estimated by loss-on-ignition (LOI). The moisture was first removed from the samples by drying them overnight at 105 C. Then the samples were heated at 1100 C for 3 h in a muffle furnace to measure total organic and inorganic (carbonates) content percentage by weight (Dean 1974). The calculation of the total organic and inorganic content in mass was conducted by subtracting the mass of the burned sample from the initial mass of the dried sample (details on LOI calculations are provided in the supplementary information). LOI ranges between 2% and 6.8% across all sites.

Soil treatments
We grew plants in a total of 36 pots in the greenhouse. Half of the pots in the bioassay received soils from mine sites; the other half were inoculated with off-mine soils. We collected soil samples from the base of each blue grama plant where a corresponding root sample was obtained; thus, each pot received rooting-zone soil from a unique plant individual from the field sites ( SUPPLEMENTARY FIG. 1). We stored all soils at 4 C prior to setting up the experiment. During 16-17 Oct 2020, we sieved soils with 2 mm mesh sieves and transferred them to pots (6.8 × 17.7 cm) in the greenhouse. Pots were set up from bottom to top: autoclaved paper towel (4 cm) to hold soil in place, ⅓ volume of sterile soil (Pavestone Natural Play Sand; Tyrone, GA, USA; autoclaved twice for 180 min, at 121 ± 2 C on a gravity cycle), ⅔ volume of living field-collected soil (~840 g), then 1 cm of sterile soil. Pots were then watered to water holding capacity before seeds were planted.

Greenhouse setup
We placed plants in a greenhouse bay with daytime temperatures of 21-24 C and night temperatures of 15-18 C. We surface-sterilized blue grama seeds, sourced from Curtis & Curtis Seed (Clovis, New Mexico), in 70% EtOH for 3 min and washed them in DI H 2 O for 15 min. Eight seeds were carefully placed in each pot and monitored for germination. Plants were hand-watered to water holding capacity for 1 month to avoid overwatering and seed loss.

Bioassay: Response variables
Seedling emergence data were collected daily beginning on day 5, the first day of seedling emergence from the soil, until day 16. At day 16, plants were thinned to one healthy individual per pot to avoid intraspecific competition. A sterile seedling germinated on an autoclaved paper towel in a closed, clear plastic chamber was transplanted into any pot that had no seedling emergence at day 16 (n = 18 pots). On day 31, we installed a drip irrigation system, and plants were watered Monday to Friday for 2 min at 7 AM. Plants were censused for survival, plant height, number of leaves, and additional germination every 5 days, beginning on day 18, until harvest on day 85. During harvest, we collected and dried shoots, weighed the total amount of wet roots, cut off a portion of roots for staining, and dried the remaining roots. Root colonization was estimated using the modified Vierhling method as described above for fieldcollected roots. However, among the subset of roots saved for microscopy, only seven samples had detectable fungal hyphae. Thus, root fungal colonization results from the bioassay were inconclusive.

Fungal community structure
To compare fungal community composition between mine and off-mine sites, we used the matrix of taxonomically resolved isolates from Sanger sequencing data (OTU matrix), with the abundance of each taxon (number of isolates that grew in culture) from each plant as the unit of observation. From the OTU matrix, we determined the Bray-Curtis similarity using PRIMER 6.0 (Clarke and Gorley 2006). We applied permutational multivariate analysis of variance (PERMANOVA) with a nested mixed effects model using replicate site as a random effect nested within the fixed factor of site type (mine vs. off-mine) to account for the statistical non-independence of plants located at the same site. PERMANOVA models were run with 100 randomized restarts and 9999 randomizations, and the effect of site type was tested over variation among sites (n = 3) to obtain the pseudo-F ratio. We tested for multivariate homogeneity of variances and differences between site types in dispersion using Permutational analysis of multivariate dispersions (PERMDISP). Then, we used SIMPER (Clarke 1993) to identify which fungal taxa contributed the most to differences among sites.

Fungal diversity, richness, and abundance
All significance reporting is relative to a significance cutoff of P < 0.05. We used raw counts from the OTU matrix to calculate several metrics of fungal diversity, including OTU richness, the Shannon diversity index, and the inverse of Simpson's evenness, using the VEGAN package in R (Oksanen et al. 2019; R Core Team 2020). We used mixed effects general linear models on each diversity metric as well as fungal abundance, determined as the total number of isolates cultured and as percentage root colonization. Models used the lmer function in R package LME4 (Bates et al. 2015) to test for the effects of the fixed factor of site type (mine/offmine) on fungal diversity metrics and abundance, using site as a random effect nested within the site type factor (R Core Team 2020). We present analysis of deviance results using chi-square loglikelihood ratio tests from R package CAR (Fox and Weisberg 2019). All models met assumptions of homogeneity of variances and normality of residuals. Graphics were produced in GGPLOT2 (Wickham 2016).

Soil chemistry and fungal communities
We combined the matrix of data on soil chemistry with the root fungal composition matrix in exploratory analysis of relationships between soil chemical properties and fungal community composition. We used Mantel tests in PRIMER 6.0 (RELATE function) for matrix correlation (Spearman R) and distance-based linear models (DistLM function, PRIMER 6.0) to examine correlations between fungal community composition and individual soil properties. Using R, we also examined correlations between fungal diversity metrics and abundance against each soil chemical property, by replacing the factor of site type (mine/off-mine) in the general linear models described above with each soil chemical property as a continuous variable.

Bioassay experiment
We evaluated soil inoculum treatment effects on plant responses using general linear mixed effects models on the following response variables: proportion of seedlings emerged, total plant biomass, root: shoot ratio, and fungal colonization of roots. These models included the fixed effects of soil origin (mine/off-mine) and the random effect of site nested within the soil inoculum origin treatment. We used generalized linear mixed effects models of the same form, but with a binomial distribution of the response variable, to assess soil origin effects on plant mortality (live = 1, dead = 0).

RESULTS
Soil chemistry diverged between mine and off-mine sites. Mine soils contained more than 22 times the amount of U than off-mine sites on average; mine soils also contained 2.5 times more vanadium (V) (FIG. 1). Soil nutrients important for plant production, including P, K, Mg, Ca, Cu, and Z, were lower in mine sites compared with off-mine sites (SUPPLEMENTARY TABLE 1). Specifically, mine sites had 23% less P (marginally nonsignificant, P = 0.06), but other soil chemical differences between mine and off-mine soils were not statistically significantly different. Mine and off-mine sites did not differ in soil pH.
Root fungal richness and diversity diverged between mine and off-mine sites, but the total abundance of fungal isolates obtained did not significantly differ, indicating that there was no need to rarefy diversity metrics (Gotelli and Colwell 2001). Fungal OTU richness (P = 0.007) was 25% lower onmine than off-mine, and Shannon diversity was 20% less (P = 0.016) in roots from mine soils compared with off-mine (FIG. 2). Differences in evenness were nonsignificant. Fungal abundance cultured from roots and root colonization by fungi estimated via microscopy also showed no significant difference between mine and off-mine sites (FIG. 2).
The majority of fungi cultured from mine sites were classified to the phylum Ascomycota (99%, with the other 1% classified to phylum Basidiomycota). Of the Ascomycota, 40% of OTUs were classified into order  Pleosporales, 39% were order Hypocreales, 11% were order Xylariales, and 10% were order Sordariales (FIG. 4). Of these taxa, 13% were categorized as dark septate endophytic (DSE) fungi. Fusarium and Darksidea were the most common genera in mine sites. Fungal taxa that diverged most between mine and off-mine sites included Pleosporales sp. 1 (OTU12), which was 90% less abundant in mine sites, Fusarium sp. 2 (OTU8), which was 45% more abundant in mine sites, and Xylariales sp. 1 (OTU10), which was 83% less abundant in mine sites compared with off-mine sites. Other notable fungal taxa that were more abundant in mine than off-mine sites included Fusarium sp. 1 (OTU3; 65% more abundant), Darksidea sp. (OTU6; 74% more abundant), and uncultured fungus (OTU11; 130% more abundant). OTUs that contributed >5% to overall dissimilarity are listed according to SIMPER classification in TABLE 1.

Soil chemistry and fungal community composition
Comparison of biological OTU matrix data with soil chemistry data showed no significant matrix correlation between the fungal community and soil. However, potassium (K; P = 0.001), magnesium (Mg; P = 0.001), and sulfur (S; P = 0.007) were each significant predictors of fungal community composition through distance-based linear univariate models. Negative correlations of fungal abundance, richness, and diversity metrics against each soil chemical property indicated that community richness was correlated to phosphorus (P; P = 0.025, r = −0.986), manganese (Mg; P = 0.043, r = −0.959), and zinc (Zn; P = 0.054, r = −0.991). Diversity was negatively correlated with phosphorus concentration (P; P = 0.028, r = −0.986), but no other characterized element; abundance was not negatively correlated with soil chemistry composition.

Plant biomass
Mine soils reduced overall plant production (FIG. 5,  SUPPLEMENTARY TABLE 1). Both aboveground biomass and belowground biomass were smaller for plants grown in mine soils relative to off-mine. Aboveground plant biomass was 70% lower (P = 0.0028) and belowground root biomass was 60% lower (P = 0.0194) in mine soils compared with off-mine soils. Total plant biomass was 65% lower (P = 0.0077) in mine soils. Mine soils did not alter plant allocation between aboveground and belowground tissue types, as supported by the nonsignificant differences in root:shoot ratio between mine and off-mine soils (FIG. 5C). Additionally, mine soils had 52% lower seedling emergence after 16 days (P = 0.0435), and plant mortality doubled for plants grown in mine soils (P < 0.001).

DISCUSSION
Uranium mining legacy reduced culturable fungal diversity and richness in plant roots.-Roots of blue grama from mine sites had 20% lower fungal biodiversity than in soils outside of the mine, with no net difference in the abundance of isolates cultured. Previous studies have focused mainly on soil fungi and arbuscular mycorrhizal (AM) fungi in soils with mine legacies (Gagnon et al. 2020;Wang 2017). Our finding of low diversity of root-associated Ascomycota in uranium mine soils was consistent in magnitude with the low diversity of soil bacteria and soil fungi in soils with heavy metals in other studies (Rosenfeld et al., 2018). Similarly, root-associated fungi in mercurycontaminated soils had 8-10% lower diversity and 33-41% lower richness in mine soils compared with reference soils, where species were identified through culture-based sequencing (Pietro-Souza et al. 2017). Ascomycota were consistently the most prevalent phylum across metal-contaminated sites, constituting between 97% and 99% of all root-associated fungi across grass species as determined by culture-independent DNA analysis (Narendrula-Kotha and Nkongolo 2017). We found no overall significant divergence in overall composition of culturable root fungi between mine and off-mine sites, although some specific taxa did strongly differ in abundance (see next section). In our prior work, culture-based methods accounted for far fewer fungal taxa (~2.2%) as compared with culture-independent methods for grass root-associated microbes in similar environments, with 7608 OTUs from Illumina sequencing of IT2 compared with 1033 isolates of 166 OTUs from Sanger sequencing of pure cultures . However, because our intent was to characterize fungi for use in bioremediation, it was necessary to use culture-based methods to catalog and archive the isolates for future experiments; our past work importantly showed that we successfully cultured 56% of the hostplant indicator taxa that were identified by ITS2 environmental sequences from roots, and that general patterns in diversity, composition, and richness were robustly similar between culture-based and cultureindependent methods .
Microbial populations in arid ecosystems are not typically sensitive to small nutrient shifts, as demonstrated through phosphorous and nitrogen soil nutrient addition studies (Chen et al. 2019;Chung et al. 2017;Sinsabaugh et al. 2015). Similar levels of congruence in overall fungal community composition between contaminated sites and reference sites were reported by (Kerfahi et al. 2020), in which soil chemistry did not predict overall differences in root fungal community composition. Notable differences in root fungal community composition are often correlated to soil pH   Siciliano et al. 2014). Our sites did not differ in soil pH, which may explain the lack of dissimilarity between fungal communities. More work will be needed to understand the impact of soil chemistry on root-associated fungal abundance. Our research indicates no significant markers contributing to shifts in abundance, but our results are limited by sample size (n = 3 sites of each type for soil chemistry data). However, fungal community diversity and richness were correlated to soil phosphorus, which is known to impact nitrification and denitrification potential (Cui et al. 2020).  (Fogarty and Tobin 1996). Melanin is also known to have a high adsorption capacity for U (McLean et al. 1998). With the cultured isolates from this study, we are now poised to test whether translocation or bioprecipitation of heavy metals plays a role in endophyte plantfungal symbioses. Improvements to OTU taxonomy that utilize additional gene regions to more precisely classify fungal taxa will improve our ability to focus on specific species of Fusarium and identify other DSE taxa from our isolate collection. To our knowledge, there is no current data on threshold for U or V toxicity for either Fusarium or Darksidea. Further study of the stress capacity of these species and their functional roles in plants under stress will improve our ability to identify mechanisms involved in symbiotic plant relationships and their effects on heavy metals.

Targeting Fusarium and Darksidea for bioremediation
Mine soils reduced plant production.-Plants exposed to soils from mine sites had reduced seedling emergence, biomass, and survival relative to those in soils collected offmine that were grown in the same greenhouse environment. This result has important biogeographic consequences because our focal plant is the most widespread native grass in North America, blue grama (Bouteloua gracilis). A combination of depletion of nutrients and greater concentrations of toxic heavy metals is the most plausible cause of the observed shifts in the root fungal community as well as declines in overall plant fitness. However, limited sample size caused by compositing soils from each site for soil chemical analyses made it difficult to detect strong positive or negative correlations between soil chemistry, fungal communities, and plant production.
Reduced plant growth in soils from mine sites could be caused by the direct effects of heavy metals in soil inocula or the indirect effects of a mining legacy on soil microbial communities. Our bioassay experiment indicated that plant biomass was most strongly linked to the U concentration of the soil (FIGS. 1, 5). However, it remains unclear whether U concentration is directly linked to plant health or whether impacts of U on microbial communities caused the plant declines in our simple bioassay. Previously, exposure of roots to heavy metals decreased plant growth, and separately, certain microbial communities were shown to dramatically increase plant growth (Aydinalp and Marinova 2009;Nihorimbere et al. 2011;Rudgers et al. 2020). Fungi are well known to mediate plant responses to soil environment and respond to changes in soil environment (Johnson et al. 2005;Kivlin et al. 2013;Lynn et al. 2019;Rodriguez-Ramos et al. 2021;Rudgers et al. 2014). Thus, soil sterilization treatments in future experiments could help disentangle soil chemistry from soil biological/microbial mechanisms. We did not sterilize soils because our bioassay experiment was a first test to determine whether mine soils had negative effects on plants. Because sterilizations could dramatically alter soil chemistry, including the chemistry of heavy metals (Krauße et al. 2019), such additional manipulations should be applied cautiously.
Further testing of mine soils on specific plant-fungal symbiosis can help determine the degree to which plant responses are mediated by individual fungal species. Inoculating plants with one of the fungal isolates, now cataloged in our fungarium, and testing for responses in mine and off-mine soils will allow us to track differences in bioremediation functions among different fungal species. To explore plant-fungal symbioses further, plant function could be assessed along U concentration gradients to identify the toxicity threshold for soil U on specific plant-fungal partners.

CONCLUSION
In our study comparing roots and soils collected from mine and off-mine sites, U mining legacies had the strongest influence on culturable root fungal endophyte biodiversity and richness, with weaker declines in fungal abundance and no significant shift in overall fungal community composition. We identified Fusarium oxysporum, a second Fusarium sp., a Darksidea sp., and Setopoma terrestris as focal species to test as potential bioremediators because of their dominance in roots of Bouteloua gracilis in mine soils. Our results broaden general understanding of shifts in plant-associated fungal community structure under heavy metal contamination stress.

ACKNOWLEDGMENTS
We thank Paul Owen-Smith, Roxanne Marquez, Amanda Anderson, Purbendra Yogi, and Tonny Nyonga for laboratory assistance. Additional assistance was provided by C. DeVore and T. Busch (Department of Civil Engineering, University of New Mexico).

DISCLOSURE STATEMENT
No potential conflict of interest was reported by the authors.

FUNDING
This work was supported by National Science Foundation grants NSF-DEB 1911451 COLLABORATIVE RESEARCH: Quantifying the microbial contribution to community recovery from drought and NSF-DEB 1456955 COLLABORATIVE RESEARCH: Parsing the effects of host specificity and geography on plant-fungal symbioses under climate change.