Abstract
Restoration efforts in North America targeting lands dominated by the non-native and persistent Agropyron cristatum (crested wheatgrass) have failed to sufficiently reduce the species in favor of native plant communities. Restoration efforts have largely focused on traditional vegetation control methods, but there is evidence crested wheatgrass alters soil properties and processes. This may contribute to its persistence and the poor success of native plant community reconstruction. We conducted a field trial in northeastern Montana, USA, to evaluate reconstruction methods that aim to condition the site for two years prior to native species seedings. We used combinations of herbicide, tillage, biological soil amendment, and cover crop, with the goals of reducing crested wheatgrass cover and improving soil nutrient content and microbial activity. We observed treatments including herbicide or tillage were effective at reducing crested wheatgrass cover from 20–35% in the control to less than 10% cover after two years of treatment. When tillage and herbicide were coupled with cover crops, we minimized soil exposure and reduced the risk for site degradation. Herbicide applications alone offered the cheapest method for effectively reducing crested wheatgrass, and cost increased as treatment complexity increased. Soil carbon and nitrogen pools and microbial abundance and composition were surprisingly uniform and stable across native, control, and treatment areas. At this location, which has been dominated by crested wheatgrass for over 80 years, we did not observe severe degradation of important soil health indicators, and we effectively reduced crested wheatgrass cover without degrading soil health.
Restoration Recap
Northeastern Montana, USA hosts large tracts of public land dominated by Agropyron cristatum (crested wheatgrass), which diminishes wildlife habitat and livestock grazing value.
Restoring native grassland species into crested wheatgrass stands has proven difficult across its North American range but has primarily focused on aboveground plant control strategies.
We conducted a field evaluation of restoration techniques combining vegetation control strategies with soil enhancement methods for two years to condition the site prior to native species seedings.
Soil properties were stable across treatments, indicating long-term dominance by crested wheatgrass has not irreversibly depleted the soil nutrient and microbial assets.
We recommend combining herbicide or tillage with a cover crop to maximize crested wheatgrass control and to minimize soil impacts. Herbicide combined with cover crops was more cost effective than tillage at achieving these goals.
Agropyron cristatum (crested wheatgrass) is a perennial, cool-season bunchgrass that is now one of the most widespread invasive grass species in Great Plains and western North American rangelands (DiAllesandro et al. 2013). It is native to the steppes of central Asia and has spread throughout the arid sagebrush steppe and mixed-grass prairie biomes in the USA (Zlatnik 1999). The species was intentionally planted on an estimated 6–10 million hectares (Ambrose and Wilson 2003, Wilson and Pärtel 2003), with hundreds of thousands of these hectares within the public lands of eastern Montana of the USA (Lesica and Cooper 2019). Crested wheatgrass was used to stabilize degraded lands and provide reliable livestock forage throughout western North America. It provided soil cover after the drought of the 1920s–30s, it is considered desirable due to its hardy attributes, and it easily establishes in cold and drought-prone environments (Lesica and Deluca 1996). However, its introduction and widespread dominance have led to a reduction in native plant species diversity and soil quality (Christian and Wilson 1999). Currently, many land managers seek to convert monotypic stands of crested wheatgrass to a native plant community for enhancing wildlife habitat, ecological diversity, and soil health (Misar et al. 2016, Miller et al. 2021).
Approaches to improving land use value and ecosystem services in crested wheatgrass-dominated grasslands have primarily focused on vegetation control strategies (e.g., herbicide, tillage, grazing, and fire) (Wengreen et al. 2016). Since crested wheatgrass dominates both aboveground cover and the seedbank for multiple decades, successful reduction of crested wheatgrass has required repeated herbicide or soil disturbance applications (Morris and Monaco 2019). These methods have been successful in suppressing crested wheatgrass seed production and removing the existing plants with repeated applications, but they have negative implications for soil health (Henderson and Naeth 2005, Trognitz et al. 2016), and do not directly encourage native species establishment. The most successful restoration strategies reported have combined crested wheatgrass suppression methods with strong plant competition (Wilson and Pärtel 2003, Cox and Anderson 2004).
Given the limited success of aboveground techniques for restoring native species to crested wheatgrass stands (Hulet et al. 2010, Fansler and Mangold 2011, Davies et al. 2013, Lesica and Cooper 2019), it is reasonable to turn our attention to understanding how belowground factors relate to restoration of lands dominated by crested wheatgrass. While mechanistic studies examining relationships between crested wheatgrass, soil properties, and invasiveness are lacking, there is evidence that soil properties beneath crested wheatgrass are fundamentally different from those characteristic of pre-colonization and native reference communities, and these differences may facilitate crested wheatgrass persistence (Palit et al. 2025).
The observed differences in soil properties associated with crested wheatgrass are summarized below, and they relate to the plant’s root morphology, tissue chemistry, and exudation habits, which alter soil carbon and nitrogen pools. Crested wheatgrass has a coarse root system with lower root biomass than that of native species (Christian and Wilson 1999), limiting the mass and turnover rates of root-derived organic inputs into soil. Wilson and Pärtel (2003) reported that cultivated land planted with crested wheatgrass in the late 1940s had lower soil carbon and nitrogen than similar age fields dominated by native grasses. Klein et al. (1988) compared root exudates from crested wheatgrass to those produced by two native species, Pascopyrum smithii (western wheatgrass) and Bouteloua gracilis (blue grama), in a simulated growing season and found significantly lower total carbon and nitrogen released from crested wheatgrass root exudates than from the native species.
Crested wheatgrass is also associated with weak soil aggregates and low microbial activity as indicated by soil respiration rates, compared to levels in native rangeland (Broersma et al. 2000, Norton et al. 2012). Soil reclaimed with crested wheatgrass after surface mining had substantially lower microbial abundance compared to undisturbed and reclaimed soils occupied by native species (Gasch et al. 2016). Crested wheatgrass has been shown to either lack or have fewer symbiotic arbuscular mycorrhizal fungal molecular root signatures and taxa compared to those of native grass and shrub species surveyed (Reinhart and Rinella 2021). Jordan et al. (2012) found soils conditioned with crested wheatgrass had significantly lower arbuscular mycorrhizal fungal richness when compared to soils supporting native species. Based on these studies demonstrating reduced soil carbon, nutrients, and beneficial microbes beneath crested wheatgrass compared to higher levels of these components in native plant-occupied soils, it is possible that establishment of a diverse native plant community may require enhancing these soils with organic matter, microbial inoculants, or nutrients.
Collectively, these differences in belowground properties may play a role in both promoting crested wheatgrass dominance and in preventing native species establishment by reducing soil nutrients and beneficial microbial partners. Crested wheatgrass dominance appears to favor its own growth, as evidenced by its enhanced competition in soils conditioned by crested wheatgrass (Hooker and Stark 2008, Nafus et al. 2020). Similarly, a controlled environment study evaluating crested wheatgrass invasiveness in soils from European and Canadian ranges indicated strong soil-mediation and enhanced growth in Canadian soils, assumed to relate to the microbial makeup of the respective soils (Villasor et al. 2024). Clearly, there is enough evidence to warrant evaluation of restoration techniques that address the entire above- and belowground habitat of crested wheatgrass and native plant species.
Restoration approaches combining above- and belowground methods may be the most effective strategies to restore grasslands dominated by crested wheatgrass. Soil-focused treatments can be coupled with traditional vegetation control measures to accomplish this. Wallace et al. (2009) found that using a biosolid treatment on crested wheatgrass-dominated rangelands successfully increased the soil’s water-stable aggregate, carbon, and nitrogen contents. Soil amendments aiming to boost soil carbon, nutrient cycling, and microbial activity may improve the soil condition for native species establishment (Kulmatiski 2011, Lal 2018, Gravuer et al. 2019). Cover crops offer another option for activating plant competition, boosting organic inputs (through labile biomass and exudate contributions) and stimulating microbial activity in soil in preparation for native species seedings. Plant root exudates have nutrient-rich compounds like amino acids, sugars, and proteins that recruit both beneficial and pathogenic microorganisms (Akiyama et al. 2005, Badri and Vivanco 2009). The composition of root exudates depends on the host plant identity, which indicates higher diversity of cover crop species leads to higher diversity of the microbes attracted to the exudates (Broeckling et al. 2008, Bardgett and van der Putten 2014). We hypothesize that restoration approaches incorporating soil improvements will be critical for repairing crested wheatgrass-dominated lands.
The goal of this study was to evaluate above- and belowground approaches to restoring a grassland dominated by crested wheatgrass to a native plant community. In terms of vegetation management, we aimed to reduce crested wheatgrass cover and to maintain a living plant community and surface cover. We suspect that to effectively reconstruct the native ecosystem’s structure and function, we need to combine targeted control of crested wheatgrass with soil improvement strategies. We also suspect an approach that facilitates plant community succession through a multi-year, two-phase approach will lead to long-term restoration success. This study encompasses the first phase of the restoration, where we implemented combinations of crested wheatgrass control (herbicide and tillage) and soil conditioning (cover crops and a biological soil amendment) treatments for two years in northeast Montana, USA. We report the costs of implementing each treatment method. We evaluated vegetation and soil responses across treatments to identify effective crested wheatgrass reduction methods and soil carbon, nitrogen, and microbial responses to treatments. This study aims to develop recommendations for site conditioning, which would precede a second reconstruction phase focused on native species establishment.
Methods
The Mooney Coulee study site (near 48° 18' 53.9994", -106° 39' 46.7994") is located about 16 km northwest of Glasgow, MT, USA. The study area is approximately 49 hectares (805 × 610 m). Soils at the site are characterized as fine, smectitic, frigid Torrertic Natrustalfs and Aridic Haplustalfs (Soil Survey Staff 2024). The vegetation was historically mixed-grass prairie, but the site was cultivated for crop production, abandoned, and revegetated with crested wheatgrass in the 1930s. Currently, the land is owned by the US Department of the Interior, Bureau of Land Management and is managed for wildlife habitat and as a livestock grazing allotment. The climate at the site is classified as cold semi-arid (Köppen climate type BSk), with temperate continental patterns of hot, dry summers and cold, snowy winters. The 30-year average total annual precipitation is 34 cm with nearly 85% of the precipitation arriving April through October, and the 30-year average frost-free period is 140 days (National Weather Service 2025).
Mooney Coulee treatments and their abbreviations. Treatments were installed in duplicated strips on 49 hectares of land dominated by crested wheatgrass. Control strips flanked all treatment strips, and the native mixed-grass prairie reference site was 4.3 km from the site.
We established nine experimental treatments (Table 1) across the Mooney Coulee site in 2022 and re-applied them in 2023. We duplicated each treatment strip (approximately 30.5 m × 0.8 km); untreated control strips flanked the 18-strip treatment area. In addition to the Mooney Coulee experimental site, we identified a nearby intact native mixed-grass prairie site located 4.3 km northeast of Mooney Coulee. The reference area (referred to as “Native”) is also managed for wildlife habitat and as a livestock grazing allotment, but it has never been cultivated. On each treatment strip, we established six sample point locations, for a total of 12 sample points per treatment. We evenly positioned the points along the length of the strip and centered to avoid edges. In the native reference site, we only established six sample points total.
We began treatment applications at Mooney Coulee in spring of 2022. On May 23, 2022, to treatment strips assigned a herbicide treatment, we applied a broad-spectrum contact herbicide formulation (glyphosate, Glystar 5 EPA Reg No 42750-61 [68 L], with Brimstone [activator, 10 L], Crosshair [drift control, 2 L], and Bronc Max [water treatment, 4 L], at a Glystar application rate of 12,730 L/ha). On May 25, 2022, we established the tillage treatment strips with a Degelman Pro-Till high-speed disk (Degelman Industries, Hillsboro, ND), which fractures the soil surface and partially inverts the soil. The disturbance depth in our treatments was approximately 15 cm.
Cover crop seed mix and seeding rates. The cover crop seed mix used in experimental treatments at Mooney Coulee. The total seeding rate was approximately 32 kg/ha.
On May 27, 2022, we applied the commercial liquid amendments and seeded the cover crops with a John Deere 1890 disk drill with stainless liquid tubes. This equipment allows individual control of the seed and liquid applications, so they can either be applied alone, or they can be applied together. We applied the Rejuvenate and SeaShield liquid amendments (Advancing Eco Agriculture, Middlefield, OH) at a rate of 131 L/ha (45 L water plus 4 L of each amendment) on strips assigned to receive amendments. The Rejuvenate product contains a proprietary mix of complex carbohydrates, humates, and magnesium intended to stimulate soil microbial activity and residue decomposition. The SeaShield product is a cold pressed crab, fish, and shrimp concentrate intended to deliver nutrients and stimulate soil biological activity. We seeded the cover crop mix at approximately 32 kg/ha at a 2.5 cm depth with 19 cm spacing on strips assigned to receive cover crop. The cover crop species and individual seeding rates are shown in Table 2; the mix was designed to include cool- and warmseason grasses, broadleaves, and legumes. We allowed the cover crops to grow until they senesced or were killed by frost. We did not remove any cover crop biomass from the treatments, so all plant materials, including seeds, remained within the treatment strips.
We re-applied the same treatment methods in the spring of 2023 on the same strips; we applied herbicide on May 18, tillage on June 6, and the cover crops and amendment on June 7. We calculated all costs for installing treatments (materials, custom rates, equipment, etc.) for each year and report these in the results.
In July of 2022 and 2023, we quantified ground and vegetation species cover at each of the sampling points following a modified Bureau of Land Management National Assessment, Inventory, and Monitoring (AIM) protocol (Herrick et al. 2021). Transects were 12.5 m long and oriented at a bearing of 8 degrees. From each transect we extracted and calculated percent cover of crested wheatgrass and all other vegetation (including cover crops). We include a full plant species list in the Supplementary Material (Supplementary Material Table S1), which indicates species presence in each treatment in each survey year. For brevity, we present the cover results for crested wheatgrass and the sum cover for all other species. In mid-September of each year, we clipped aboveground herbaceous biomass within a 0.5 m² frame at each of the sample points. We only collected vegetation at the Native sample points in 2023. We dried and weighed the clippings. Plant species were indistinguishable, so we were unable to separate clippings by species.
In mid-September of 2022 and 2023, we collected soil samples at each sample point, using a shovel (0–15 cm depth) and a volumetric core using a hammer-driven coring device (0–15 cm depth × 5 cm diameter). We kept all samples in coolers until they were returned to the lab. We stored subsamples for microbial community analysis in a freezer (-20 degrees C), lyophilized them, then shipped them to a commercial lab. We stored remaining soil samples in a refrigerator (4 degrees C) until they were processed in the lab (within one week of sampling). We analyzed each bulk soil sample for both descriptive soil properties and response variables. Descriptive soil properties included particle size (only measured in 2022, Gee and Or 2002), and pH and electrical conductivity in a 1:1 soil:water slurry (Rhoades 1996, Thomas 1996). We measured bulk density (Blake and Hartge 1986) and gravimetric water content (Gardner 1986) on the volumetric core. These properties were stable across treatments throughout the duration of the study, and they are listed in Table 3.
Our response variables focused on soil carbon and nitrogen pools, as well as microbial community structure. We measured total organic carbon and total nitrogen with dry combustion (Nelson and Sommers 1996); these pools are considered relatively stable over time. We measured permanganate oxidizable carbon (POXC; Weil et al. 2003), which represents a small fraction of organic carbon in soil (phenolic and polyphenolic structures) that is responsive to soil management practices (Christy et al. 2023). We measured nitrate- and ammonium-nitrogen (Mulvaney 1996), which sum to the highly dynamic inorganic nitrogen pool. Phospholipid fatty acid analysis (PLFA; Buyer and Sasser 2012) provides an estimate of microbial group abundance at major taxonomic levels, including bacteria, decomposer fungi, actinomycete (filamentous bacteria), arbuscular mycorrhizal fungi, and eukaryotic microbes, which sum to estimate total microbial abundance. We also calculated the fungal-to-bacterial ratio from these biomarkers. We only analyzed soil samples collected in 2023 for PLFAs (analysis performed by Ward Labs, Inc., Kearney, NE).
Descriptive soil properties for field sites. Values are averages of soil samples collected from the Mooney Coulee experimental treatments (n = 120 × 2 years = 240) and the native mixed-grass prairie reference location (n = 6 × 2 years = 12). Gravimetric water content is reported as a range, rather than average.
The experimental design of this trial has duplicated treatment strips, each with six observations. Due to the lack of true replication and the low sample size, we chose to apply non-parametric statistical tests for all means comparisons. For each response variable, we calculated summary statistics (mean, standard deviation, and coefficient of variation) for each treatment strip in each treatment and each year and plotted all observations by strip, treatment, and year (see Supplementary Material). Based on the summary statistics, figures, and Mann-Whitney U tests to compare the strip means within treatments by year, we pooled all 12 observations per year to proceed with means comparisons at the treatment-level. We used a non-parametric means comparison test (Kruskal-Wallis) and post hoc tests (Dunn’s test) to compare response variable means across treatments within each year (n = 12 per treatment, n = 6 in native area). For response variables collected in both years, we also compared the means within each treatment between years using the non-parametric Mann-Whitney U test. All summary statistics and statistical test parameters are included in the Supplementary Materials. We used R (R Core Team 2024) along with the ‘FSA’ (Ogle et al. 2023), ‘ggplot2’ (Wickham 2016), ‘ggpubr’ (Kassambara 2023), ‘pastecs’ (Grosjean and Ibanez 2024), and ‘rcompanion’ (Mangiafico 2024) packages for data organization, analysis, and visualization.
Results
Our objective with the field trial was to evaluate both vegetation and soil responses to treatments that reduce crested wheatgrass and enhance soil health. The treatments aimed to prepare the invaded grassland for the second phase of native plant community reconstruction (seeding a high diversity mix of native species). Here, we evaluated the treatment costs, the effectiveness of treatments in reducing crested wheatgrass, and the soil carbon, nitrogen, and microbial indicator responses to treatments.
The costs for each treatment and their combinations are listed in Table 4. The least expensive treatment method was a sole herbicide application, followed by cover crop, and then by amendment. The costs increased with the increasing complexity of the treatments, with the most expensive option (CATH) costing nearly eight times more than an herbicide application alone. To treat the entire Mooney Coulee area (49 hectares, 1 year), the average expected total cost ranged from $3,822 USD (herbicide only) to $29,792 USD (CATH).
In the first year of the trial, crested wheatgrass cover across treatments was similar to the average cover in the control strips (mean cover = 20%, standard deviation =12, Figure 1A, Supplementary Material Table S2.1, Supplementary Material Figure S2, Supplementary Material Table S2.2). The combination treatments including herbicide or tillage had the lowest mean crested wheatgrass cover with means and standard deviations (SDs) as follows: CH mean = 10%, CH SD = 10, CT mean = 9%, CT SD = 5, CAT mean = 7%, CAT SD = 6, CATH mean = 5%, CATH SD = 6. The reductions persisted in these treatments into 2023, resulting in lower average crested wheatgrass cover in all combination treatments that included herbicide or tillage (CH mean = 1%, CH SD = 2, CT mean = 8%, CT SD = 12, CAT mean = 9%, CAT SD = 4, CATH mean = 1%, CATH SD = 2). The highest average crested wheatgrass cover (greater than 30% crested wheatgrass cover) occurred in treatments that had only amendment, cover crop, or their combination.
Treatment expenses. Costs for each treatment per hectare (in US dollars) calculated from installation costs (materials, custom rates, and equipment) for each year, their average, and the total of both years.
Vegetation cover. Box-and-whisker plots for treatment-level mean percent cover of A) crested wheatgrass and B) all other species in 2022 (light bars) and 2023 (dark bars). Significant differences (p < 0.05) between treatment means within a year are indicated by different letters (lowercase for 2022, upper case for 2023). Significant differences (p < 0.05) between years and within a treatment are indicated by an asterisk (*). Treatment abbreviations are: A = amendment, C = cover crop, H = herbicide, and T = tillage.
Herbaceous biomass. Box-and-whisker plots for treatment-level mean herbaceous biomass in 2022 (light bars) and 2023 (dark bars). Significant differences (p < 0.05) between treatment means within a year are indicated by different letters (lowercase for 2022, upper case for 2023). Significant differences (p < 0.05) between years and within a treatment are indicated by an asterisk (*). Treatment abbreviations are: A = amendment, C = cover crop, H = herbicide, and T = tillage.
Cover of other species, including the cover crops, was also similar across treatments in both years (Supplementary Material Table S1). Aside from cover crops and crested wheatgrass, the species composition within this category included many native forbs, grasses, and subshrubs including Antennaria spp. (pussytoes), Artemisia spp. (sagebrush), Aristida purpurea (purple threeawn), B. gracilis, Gutierrezia sarothrae (broom snakeweed), and P. smithii. We did not observe any species that were considered noxious, invasive, or problematic in any treatment. Mean cover across treatments ranged from 11% (amendment in 2022) to 58% (CATH in 2023), while average cover in the control strips was 27% (SD = 11) in 2022 and 39% (SD = 17) in 2023. (Figure 1B, Supplementary Material Table S3.1, Supplementary Material Figure S3, Supplementary Material Table S3.2). Cover crop treatments did not necessarily have more cover than treatments without cover crops, but the mean cover values were higher in the most intensive CATH treatment (mean = 43%, SD = 10 in 2022, and mean = 58%, SD = 14 in 2023) compared to cover crops alone (mean = 16%, SD = 11 in 2022, and mean = 24%, SD = 14 in 2023). The high mean cover of the CATH treatment (58%) in 2023 was similar to the mean cover in the Native area in 2023 (mean = 51%, SD = 21).
Total clipped herbaceous biomass was uniform and similar to the control (mean = 342 kg/ha, SD = 252) across treatments in 2022 but significantly increased in 2023 in plots combining cover crops with herbicide and tillage (Figure 2, Supplementary Material Table S4.1, Supplementary Material Figure S4, Supplementary Material Table S4.2). The mean biomass in the CH treatment nearly tripled (mean = 160 kg/ha, SD = 77 in 2022 and mean = 479 kg/ha, SD = 274 in 2023), and nearly doubled for CT (147 kg/ha to 375 kg/ha), CAT (152 kg/ha to 277 kg/ha), and CATH (429 kg/ha to 1000 kg/ha) treatments from 2022 to 2023. Based on the reduction of crested wheatgrass cover in those treatments, we can assume the increase in biomass was due to increased growth of cover crops and other plant species, including native species. Biomass in the native area was low (mean = 71 kg/ha, SD = 39) compared to all experimental treatments and untreated control.
The POXC pool of carbon was similar across treatments in 2022 (control mean = 238 mg/kg, SD = 84), with one exception: mean POXC in the CT treatment (mean = 148 mg/kg, SD = 63) was significantly lower than the mean POXC values in H (mean = 281 mg/kg, SD = 93) and CATH (mean = 296 mg/kg, SD = 78) treatments (Figure 3A, Supplementary Material Table S5.1, Supplementary Material Figure S5, Supplementary Material Table S5.2). Mean POXC values were again uniform in 2023 (control mean = 398 mg/kg, SD = 67), but many of the treatments had significant increases in POXC from 2022 to 2023, including the control treatment. Total organic carbon pools were also stable across treatments and years (control mean in 2022 = 10,083 mg/kg, SD = 2,539, control mean in 2023 = 12,417 mg/kg, SD = 3,579, Figure 3B, Supplementary Material Table S6.1, Supplementary Material Figure S6, Supplementary Material Table S6.2). For both carbon pools and in both years, the variation across treatments and the native area was within the range of variability observed in the control strips.
Soil carbon pools. Box-and-whisker plots for treatment-level mean pool size of A) permanganateoxidizable carbon (POXC) and B) total organic carbon in 2022 (light bars) and 2023 (dark bars). Significant differences (p < 0.05) between treatment means within a year are indicated by different letters (lowercase for 2022, upper case for 2023). Significant differences (p < 0.05) between years and within a treatment are indicated by an asterisk (*). Treatment abbreviations are: A = amendment, C = cover crop, H = herbicide, and T = tillage.
The mean inorganic nitrogen values (including both nitrate-nitrogen and ammonium-nitrogen) were generally small (less than 15 mg/kg) and similar across treatments in both years (Figure 4A, Supplementary Material Table S7.1, Supplementary Material Figure S7, Supplementary Material Table S7.2). In 2022, some of the treatments had higher mean inorganic nitrogen concentration than the control strips (H mean = 7.05 mg/kg, AH mean = 12.98 mg/ kg, CT mean = 5.94 mg/kg, CAT mean = 5.81 mg/kg, and CATH mean = 6.90 mg/kg, compared to control mean = 2.93 mg/kg and SD = 0.73). These differences persisted in 2023 in H (17.11 mg/kg), AH (14.60 mg/kg), and CATH (13.78 mg/kg) (control mean = 5.21 mg/kg, SD = 2.1). As with POXC, many of the treatments had an increase in the inorganic nitrogen pool from 2022 to 2023, including the control strips. Total nitrogen did not differ across treatments or between years (means ranged from 1,075 mg/ kg in C in 2023 to 1,417 mg/kg in A in 2023, Figure 4B, Supplementary Material Table S8.1, Supplementary Material Figure S8, Supplementary Material Table S8.2).
Soil nitrogen pools. Box-and-whisker plots for treatment-level mean pool size of A) inorganic nitrogen (nitrate- plus ammonium-nitrogen) and B) total nitrogen in 2022 (light bars) and 2023 (dark bars). Significant differences (p < 0.05) between treatment means within a year are indicated by different letters (lowercase for 2022, upper case for 2023). Significant differences (p < 0.05) between years and within a treatment are indicated by an asterisk (*). Treatment abbreviations are: A = amendment, C = cover crop, H = herbicide, and T = tillage.
The microbial community structure, as indicated by the PLFA method, was fairly stable across treatments and most of the mean microbial group sizes were within the range of variation observed in the control strips (mean total abundance = 2,019 ng/g) and Native areas (mean total abundance = 1,691 ng/g, Figure 5, Table 5, Supplementary Material Table S9.1, Supplementary Material Figure S9). The A treatment had high mean bacterial (1,729 ng/g) and fungal (758 ng/g) abundances, but the relative distribution of microbial groups was similar across all treatments. Eukaryotic microbial values were extremely low in all samples (13 ng/g or less). The fungal-to-bacterial ratio indicated that the Native area had relatively more fungi (ratio = 0.59) than the control (0.37), H (0.38), AH (0.37), and CAT (0.37) treatments.
Soil microbial community distribution. Stacked bar plot of mean microbial group abundance for each treatment in 2023, as determined by phospholipid fatty acid (PLFA) analysis. Treatment abbreviations are: A = amendment, C = cover crop, H = herbicide, and T = tillage.
Discussion
In this study, we established combinations of restoration treatments aimed at reducing crested wheatgrass; treatments were designed to increase soil carbon, nitrogen, and microbial activity in a grassland setting where crested wheatgrass has dominated for decades. Our goal was to identify the most cost-effective approach for conditioning the site for native species seedings. In general, we observed reduced crested wheatgrass cover in treatments that included herbicide and tillage, and we observed that soil responses to treatments were subtle.
Our results align with other studies that show successful reduction of crested wheatgrass following treatments that include tillage and/or herbicide application. (Vaness and Wilson 2007, McAdoo et al. 2017, Morris and Monaco 2019). Crested wheatgrass cover was effectively reduced by herbicide alone as well as in combination with cover crops and tillage. We also observed that reductions were reinforced with two years of treatment, which also aligns with other studies demonstrating that repeated treatment is necessary to effectively suppress crested wheatgrass (Morris and Monaco 2019, Davies et al. 2020). Solely using the biological soil amendment appeared to provide a benefit to crested wheatgrass cover, and cover crops alone or in combination with the biological amendment maintained crested wheatgrass cover at control levels. Crested wheatgrass has competitive growth and nutrient acquisition (Gunnell et al. 2010), so we may have observed a fertilization response with the amendment on crested wheatgrass and the microbial community, which had elevated cover and abundance respectively in the amendment treatment. Without more detailed nutrient cycling data, we cannot confirm this potential mechanism. When we pooled the cover of all other species together (including cover crops), the mean cover response to treatments was more variable and within the range of values observed in both control strips and the Native area. However, it did appear the most intensive treatment (CATH) had higher mean cover of other species compared to less intensive treatments (A, C, H, AH, and CA). At least, we can conclude that an intense treatment that was effective at reducing crested wheatgrass was not detrimental to other plant species. These results are further confirmed with the herbaceous biomass measurements, wherein the highest biomass occurred in the second year in the most intensive treatment (CATH). All together, we can conclude that effectively reducing crested wheatgrass liberated space and resources for cover crops to establish and grow, and they maintained adequate cover to minimize the risks of soil exposure and erosive loss.
Soil microbial community summary. Microbial group means (ng PLFA / g soil), and standard deviations collected from soil in crested wheatgrass reconstruction treatments (n = 12 except Native where n = 6). Different superscript letters indicate significant differences (p < 0.05) in abundance of microbial groups across treatments. The sum of microbial groups represents the mean total microbial abundance in each treatment. Treatment abbreviations are: A = amendment, C = cover crop, H = herbicide, and T = tillage.
We expected the carbon pool represented by POXC to increase in treatments including soil improvement methods (amendment and cover crops), since it represents a fraction of soil carbon that is dynamic in response to soil management (Hurisso et al. 2016, Bongiorno et al. 2019). However, that was not what we observed. The POXC values were not consistently higher in treatments that included any single method, but we did see significant increases in POXC from 2022 to 2023 in four of the treatments and the control strips. The increase from 2022 to 2023 may reflect interacting effects of a slow soil response to treatments as well as a difference in environmental conditions between the two years (such as precipitation timing and quantity). The total organic carbon values were generally stable across treatments and years. The limited changes in total carbon and POXC could be attributed to the extensive time it takes soil carbon to change. Changes in soil organic carbon vary in soil type and land use history but have been shown to take 6–10 years to be detectable, especially if organic matter additions are minimal (Smith 2004). Regardless, observed soil carbon values were within the range of those measured in the Native area, indicating soil carbon pools are resilient to land use and vegetation changes. Within the two years of this study, we may not have observed strong soil carbon responses to treatments, but we have established the baseline for monitoring into the second phase of the restoration.
Soil nitrogen results were similar to carbon in that the labile, plant-available nitrogen pool (nitrate-nitrogen and ammonium-nitrogen) increased from 2022 to 2023 in a number of treatment strips and the control. Like POXC, inorganic nitrogen is dynamic in response to soil temperature and soil moisture (Guntiñas et al. 2012), and 2023 conditions may have been better for organic matter mineralization and inorganic nitrogen release. There were no consistent increases in inorganic nitrogen that aligned with a particular treatment method. The highest mean inorganic nitrogen content was in the herbicide only and AH treatments. We do not know if this response is a direct effect of herbicide application (Gaupp-Berghausen et al. 2015), or if these treatments initiated a change in nitrogen cycling or plant uptake dynamics. We do not have evidence from our vegetation response to suspect plant uptake was reduced enough to result in soil inorganic nitrogen accumulation. Total soil nitrogen was stable across treatments and years, so we can conclude treatment methods neither depleted nor increased soil nitrogen in these soils. As with carbon, we expect substantial changes in the total soil nitrogen content will require more time to respond to treatments, but nitrogen pools appear stable across the treatments in Mooney Coulee.
Soil microbial communities, as measured by the PLFA method, were uniform across treatments in terms of total abundance and microbial group composition. The exception was the A treatment had higher abundance of multiple groups, especially compared to the H treatment. It is possible that the amendment application alone stimulated microbial activity, but when combined with other treatment methods, this response to the amendment was not apparent. The amendments used in this study are sold as biological stimulants; perhaps the stimulating effect is compromised when it is applied alongside soil and vegetation disturbance. We did not observe major changes in POXC and inorganic nitrogen, which aligns with a lack of microbial response to treatments since both nutrients are often correlated with one another and with microbial activity (Culman et al. 2012, Morrow et al. 2016). We also cannot conclude from these results that cover crops boosted microbial communities compared to other treatments. The expected belowground benefits of cover crops relate to their exudate contributions, rhizosphere habitat, and soil structure support (Housman et al. 2021, Gentsch et al. 2024, Seitz et al. 2024), and these benefits are reported to exist in proportion to cover crop biomass production (Wagg et al. 2021). While we observed cover crop establishment and growth, the cover crops may not have been dense or large enough to substantially enhance the soil microbial habitat or affect carbon and nitrogen pools (Blanco-Canqui 2022, De Notaris et al. 2025). The only microbial metric that differed between the control strips and Native area was the fungal-to-bacterial ratio, which was lower in the control. Most microbial group averages across treatments fell within the range of these two references. Bacteria abundance made up between 57–65% of the total microbial community. Generally, undisturbed, diverse native grassland communities host relatively more fungi compared to disturbed soils with fewer plant species (Elliott et al. 1988, Maharning et al. 2009, Culman et al. 2010), which aligns with our observations. From a restoration standpoint, it is important to note the microbial community (at the level of these large taxonomic groupings) at Mooney Coulee does not appear to be depleted or imbalanced in areas dominated by crested wheatgrass for decades, or in response to any of the experimental treatments.
It is possible the environmental conditions at the Mooney Coulee site impose strong controls on soil processes, which override potential responses to our treatments. For example, soil water content is a regulating property for many biogeochemical transformations and microbial activity in grassland soils (Tiemann and Billings 2011). Our gravimetric soil water content values, only measured in September of each year, were quite low (2–22% water by mass); the capacity of these soils to store water is higher than the values that we observed, based on the measured bulk densities. Total precipitation from April through September in Glasgow, MT was 17.4 cm in 2022 and 22.8 cm in 2023; both were below normal precipitation for this period (26.3 cm; National Weather Service 2025). Under arid conditions, plant growth, soil organic matter mineralization, nitrogen conversions, nutrient and exudate diffusion, and microbial activity may be limited (Housman et al. 2021), which is an important consideration when setting restoration expectations.
Given the cost of applying the amendment, we would not recommend it as a treatment element for this setting due to its apparent enhancement of crested wheatgrass stands. The H, CH, and CT methods met the vegetation management goals and maintained soil properties within the range of values that we observed in the native reference area; however, tillage and seeding substantially increase the equipment needs, cost, and reduce the scalability of these methods. If a project is limited by budget, equipment, or site accessibility, which is the case in much of the study region, a multi-year herbicide application alone may suffice. However, if possible, we recommend coupling the herbicide application or tillage treatment with a cover crop. We also recognize that while herbicide application is the cheapest treatment, it is not appropriate for all locations. For example, if a site contains surface waters, or it has native grass diversity expressed within the crested wheatgrass stand, an herbicide application may not be the best or cheapest option. While this study rules out some treatment approaches, it does not evaluate all possible options. In the study region, cattle are already present on the landscape, and adjustments to cattle grazing plans, in concert with prescribed fire, may be the most effective and affordable approach. Future research will incorporate these aspects into crested wheatgrass control in the region, especially in stands that host native species diversity or are inaccessible to large equipment.
As we move into the second phase of this restoration, we will monitor the longevity of the treatments that reduced crested wheatgrass and will evaluate which treatments support the establishment of a native plant community. It is important to note that this trial was only conducted on one site, so the scope of inference is limited. Even within the study region, allotments vary in their plant composition and crested wheatgrass dominance. A few future research goals specific to the soil aspects of this study would be to characterize the microbial community with more genetic or functional specificity, and to understand if soil enhancement techniques aiming to increase fungal biomass would facilitate restoration of native plant communities. Alternatively, as the native plant community at Mooney Coulee undergoes reconstruction, the fungal-to-bacterial ratio may offer a belowground metric useful for monitoring soil restoration. It may also be interesting to evaluate if an irrigation regime or passive water catchment and storage techniques are practical restoration tools for these sites, and if they might elicit soil nutrient and microbial responses and accelerate restoration. Finally, it would be worthwhile to experiment with different cover crop mixes and seeding densities to better understand if cover crops can be an effective tool to boost mineral nutrients and microbial activity in preparation for native seedings. Ultimately, the second phase of this restoration will enlighten us to the true impacts of these treatments, and if soil conditioning paired with vegetation control offers a path forward in managing crested wheatgrass stands in semi-arid grasslands.
Crested Wheatgrass (Agropyron cristatum [L.] Gaertn.) in Colorado. A.C. Hull. Provided by USDA Forest Service. USDA-NRC Plants Database.
Acknowledgements
This research was supported by a Cooperative Agreement (#L21AS00472) from the US Department of Interior Bureau of Land Management, managed by the Glasgow Field Office, Montana. Findings and conclusions in this article are those of the authors and do not necessarily represent the views of the Bureau of Land Management. The authors also wish to thank field and lab crews for their assistance: Joel Bell, Michael McKenna, Nathan Derby, private contracts, and seasonal BLM staff and individuals who provided project input and manuscript reviews: Marisa Sather and Laura Aldrich-Wolfe.
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (https://creativecommons.org/licenses/by-nc-nd/4.0/) permitting copying and distributing the material in any medium or format in unadapted form only, for noncommercial purposes only, provided the original work is properly cited
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.