Final report for SW23-944
Project Information
Ranchers and land managers in the United States spend billions of dollars annually to control invasive plants, such as cheatgrass – an invasive annual grass that pushes out native plant species and increases wildfire risk on western rangelands. Efforts to restore native perennial species through seeding projects are also costly, often with low and unpredictable germination rates. A growing body of evidence suggests that livestock grazing, particularly fall and spring targeted grazing, may reduce invasive annual grass abundance through biomass and litter removal of annual grass while perennial species are dormant. However, treatment of invasive species in isolation will be unsuccessful if there is no longer a sufficient seed source for native species to return without further assistance. With this project, we therefore examine relationships between soil seed bank composition and aboveground vegetation composition along a gradient of cheatgrass invasion. In doing so, we set the stage to examine the potential for targeted grazing treatments to remove unwanted species and facilitate recovery of perennial species via regeneration from the seed bank.
The soil seed bank is a critical and understudied component of rangeland plant communities. Seed banks contribute to plant species persistence and recovery after disturbance and provide a signature of past, present, and future characteristics of the aboveground plant community. Furthermore, the seed bank can be used as a resource for restoration efforts after undesirable species (such as cheatgrass) have been removed. In relatively understudied, higher elevation areas at the leading edge of cheatgrass invasion, sheep grazing could play an integral role in both the cheatgrass removal and restoration processes, thereby sustaining the natural resource base on which livestock production depends.
The primary near-term objective of our study is to examine relationships between aboveground vegetation composition and belowground seed bank composition across a range of cheatgrass invasion levels. Our results will help rangeland managers better assess and incorporate the often neglected yet important belowground seed bank in their cheatgrass management decisions.
In addition to producing standalone research and education products, this project will build local capacity for researchers and students to conduct rangeland seed bank studies. It will also expand the scope of an existing funded WSARE targeted grazing project (SW22-938) by collecting baseline data on the soil seed bank and a heatmap of sheep use within the project’s targeted grazing treatments. These data will be used in a planned follow-up seed bank assessment to determine the efficacy of sheep grazing to decrease cheatgrass invasion directly (by reducing the abundance of cheatgrass and thereby cheatgrass seeds in the seed bank) and indirectly (by removing biomass and litter to make space for native seed germination), as well as through other potential mechanisms ripe for future investigation and collaboration.
Expected outcomes:
- Improved understanding of the impacts of cheatgrass invasion on the soil seed bank.
- Broader knowledge about the value of the soil seed bank and the opportunity to utilize it and sheep grazing as a tool for rangeland restoration.
- Increased capacity to conduct rangeland seed bank studies to inform management decisions.
Research objectives
- Quantify the differences in plant community composition between the soil seed bank and aboveground vegetation.
- Assess the effects of cheatgrass invasion on the soil seed bank by quantifying differences in soil seed bank composition across a range of aboveground vegetation invasion levels.
- Quantify the amount of sheep activity within targeted grazing treatment plots for more precise evaluation of sheep-vegetation interactions.
Education objectives
- Increase knowledge of the soil seed bank and how it can be affected by annual grass invasion and sheep grazing.
- Increase understanding and capacity to conduct seed bank research at the host institution.
- Increase land managers’ knowledge of the use of the seed bank as a complement to grazing for restoration in rangelands.
Cooperators
- - Technical Advisor - Producer
- - Producer
- (Researcher)
Research
Objective 1
We utilized data already being collected in 2023 from an existing WSARE project (SW22-938) in 32 previously established plots. In addition, we collected aboveground vegetation data in 8 new “low invasion” plots established in 2023 for this project using the same protocol as the existing WSARE project. We collected line-point intercept data for plant species cover at 1-m intervals along the three 50-m transects in each plot, by dropping a pin and recording each plant species it intercepts (Herrick et al., 2017). We also recorded all other plant species observed along the transects and noted these as “also present.”
We collected soil samples (5 cm diameter, 5 cm depth) at 8-m intervals along each transect with a hole saw and cordless drill to minimize soil disturbance (Byrne, 2021). We composited the first three samples and the second three samples (n=6 total cores per transect, or 2 composited samples). If we were unable to take a core at the sampling location (due to rocks, woody plant stems, etc.), we collected a core at the closest adjacent area along the transect where sampling was possible. This produced a total of 6 composited samples per plot and 240 samples across the entire experiment.
On the Boise State University campus, we first air-dried the samples, then sieved them to remove stones and wood, and weighed them. We sprayed each composited sample with deionized water and cold stratified them for 4 weeks at 2 °C to encourage germination of dormant seeds (Baskin & Baskin, 2014). Once cold stratification was complete, we thinly spread each composited sample over potting soil in trays and conducted a greenhouse emergence study in the experimental greenhouse in the Biology Department at Boise State, following the protocol described by Nunes and Byrne (2022). We placed each tray and two control trays (with potting soil but no samples from the experiment) in the greenhouse and randomly rotated trays on a biweekly basis to account for any environmental differences within the greenhouse benches. As seedlings emerged, we identified them to species and removed them from the trays. If we were unable to identify the seedlings, we transferred unknowns to pots and raised them until they developed identifiable characteristics. We maintained the trays and pots at or near field capacity with automatic irrigation and a 16-hour photoperiod to promote germination of all viable seeds. After 3 months, and no new seedling emergence for more than 14 days, we scraped the top 0.5 cm of soil to promote germination of smaller buried seeds, then continued to monitor seedling emergence for another 3 months, or until there was no new germination in a tray for 21 consecutive days. The seed bank was fully exhausted after 10 months.
To assess differences in the types of species present in the aboveground community and emerging from the seed bank, we compared the proportions of species above- and belowground across functional groups (i.e., forb, graminoid, and shrub species), annuals vs. perennials, and native vs. introduced species. For this we performed chi-square goodness of fit tests in which we compared the proportion of species present in each group in the seed bank relative to their proportions aboveground. Findings of significant differences indicate that the proportion of species in each group in the seed bank do not match what would be expected based on the proportion of species in each group aboveground.
To characterize aboveground plant cover and belowground seed species composition in each plot, we used the ‘vegan’ package in R to calculate the Bray-Curtis dissimilarity index and perform nonmetric multidimensional scaling (NMDS) on relativized seed bank density data and aboveground proportional species cover data (Bray & Curtis, 1957, McCune & Grace, 2002, Oksanen et al., 2025). We then used the ‘adonis2’ function to perform a permutational multivariate analysis of variance (PERMANOVA) to test for differences in the composition of the above- and belowground communities.
Objective 2
Managers make decisions to treat cheatgrass invasion or to plant desired species based on visible, aboveground vegetation conditions. Given the expected dissimilarity in aboveground plant and belowground seed bank compositions (Research Objective 1), we used the data described above to identify possible relationships between aboveground cheatgrass cover and a range of response variables. To do so, we first visually assessed the data and calculated Pearson’s correlation coefficients as an initial means of detecting significant relationships (1) between each species’ aboveground cover and seedling emergence density, and (2) between aboveground cheatgrass cover and seed bank species richness, seed bank species evenness, and seed bank species Shannon diversity. To assess the relationship between aboveground cheatgrass cover and the seed bank community as a whole, we performed a distance-based Redundancy Analysis (dbRDA) using the ‘vegan’ package in R (Bakker 2024, Oksanen et al. 2025). We did not proceed as planned with generalized additive mixed models (GAMM) to identify “tipping points” in the relationship between aboveground cheatgrass cover and the response variables due to the lack of relationships found between the above- and belowground communities in our data.
Objective 3
Sheep may brush against mature seed heads and trample seeds into the soil in ways that could affect seed bank conditions in the longer term (Puntenney et al. 2025). Quantifying sheep activity levels in each plot will therefore enable us to draw more nuanced conclusions about sheep-seed bank interactions in the follow-up study we hope to launch after additional years of sheep grazing treatments. Starting in Fall 2023, we used low cost, commercial off-the-shelf GPS units to monitor sheep movements during the grazing treatments (Karl and Sprinkle 2019). We designed customized 3D-printed cases to house the GPS units and batteries, which we then affixed to collars. We collared twenty sheep per band each before each grazing treatment period. Because sheep spend most of their time near each other (McGranahan et al., 2018), 20 collars per 1000 ewes was sufficient to estimate the general location of the band, and is at least double the percentage of collared sheep (2%) used in other range sheep studies (0.1-1.0%) (Young et al., 2019; Baum et al. 2022). GPS positions logged every five minutes between 5:00 and 23:00 daily (Karl and Sprinkle 2019; Baum et al. 2022).
We fit separate occurrence distributions for each sheep each day using the ‘amt’ and ‘ctmm’ R packages, with the Ornstein-Uhlenbeck Foraging model selected as the best fit for the sheep’s movements (Fleming and Calabrese 2023). These occurrence distributions described the probability of a sheep being in location at a given time, given its previous and subsequent locations and the time elapsed between them (Fleming et al. 2016). We then used ctmm’s “mean” function to estimate an average occurrence distribution for each day across all collared ewes in an allotment. For each of the two allotments, we then calculated the mean probability mass across all days. Finally, we weighted the allotment occurrence distributions across all days by the number of days sheep were marked in that allotment to estimate the average study area-wide occurrence distribution. Each raster pixel value in the resulting heatmap represents the predicted relative amount of time sheep spent in that pixel across the entire collared period.
Objective 1
In the greenhouse, we found that 50 unique species emerged from the soil samples collected across all forty plots. There were 36 forb species, 13 graminoids, and 1 shrub (Artemisia tridentata ssp vaseyana). Of these, 25 species were native, 8 were introduced, and 17 were still unknown at the time of this report, as some are still growing in the greenhouse until they produce identifiable features. Of those that were identifiable, 19 were annuals and 14 were perennials, with many of the remaining “unknowns” also suspected to be perennial species.
In the aboveground plant community, we intercepted a total of 114 unique species across all plots. There were 82 forb species, 18 graminoids, and 14 shrubs. Of these, 13 were introduced species and 35 were annuals, comprising 11% and 31% of the species, respectively. In comparison, at least 16% of the species emerging from the seed bank were introduced and 38% were annuals. The proportion of introduced and native species emerging from the seed bank differed significantly from those in the aboveground community (𝛘2 = 4.6, p = 0.03), as did the proportions of annuals and perennials (𝛘2 = 9.8, p = 0.002), with a higher proportion of introduced and annual species emerging from the seed bank. The proportions of species in each functional group also differed between the aboveground and seed bank communities (𝛘2 = 13.7, p = 0.001), with equal percentages of forb species in each (72%) but more shrub species aboveground (12% of all species) than in the seed bank (2%) and a higher percentage of graminoid species emerging from the seed bank (26% of all species) than present aboveground (16%).
When considering not just the identity of species present but also accounting for their aboveground cover and density of seedlings in the seed bank, we also found that the above- and belowground communities differed significantly (PERMANOVA: F1,78 = 35.3, p < 0.001). Cheatgrass comprised a major proportion of both aboveground cover and seedling density, ranging from 2-58% of aboveground cover (mean = 30.8%) and 0-2,441 seedlings m-2 (mean = 623 seedlings m-2). Species that tended to be associated more strongly with the seed bank community than the aboveground community included Verbascum thapsis (a non-native perennial forb with highly persistent seeds) and Sisymbrium altissimum (a non-native annual/biennial forb with prolific seed production). Collinsia parviflora (a native annual forb) was also more strongly associated with the seed bank, but typically in plots less dominated by cheatgrass.
Objective 2
Cheatgrass was abundant in the seed bank and significantly, positively correlated with aboveground cheatgrass cover (R = 0.50, p < 0.001). The densities of two non-native mustard species in the seed bank were also marginally significantly correlated with aboveground cheatgrass cover (Sisymbrium altissimum (R = 0.32, p = 0.05; Camelina microcarpa (R = 0.81, p = 0.09)). However, with the exception of mountain big sagebrush (Artemisia tridentata ssp vaseyana, R = 0.41, p = 0.008) and mountain hollyhock (Iliamna rivularis, R = 0.38, p = 0.02), the density of native perennial species in the seed bank did not correlate strongly with their own aboveground cover, nor with aboveground cheatgrass cover (all correlations p > 0.10). Aboveground cheatgrass cover also showed no relationship to the diversity of seedlings emerging from the seed bank as measured by species richness, evenness, and Simspon’s diversity (p > 0.24). Notably, several perennial species commonly used in restoration (e.g., Sandberg bluegrass (Poa secunda) and mountain big sagebrush (A. tridentata ssp vaseyana)) emerged in over 20% of the plots (up to 63% of plots for P. secunda), across the full range of cheatgrass cover. The dbRDA results revealed that only 7.2% of the variation in the overall composition of the seed bank community was explained by aboveground cheatgrass cover in each plot, further underscoring the relatively weak relationship between the aboveground plant community and seedling emergence from the seed bank.
Objective 3
Our customized GPS collars functioned well for capturing sheep locations during treatment implementation periods. Sheep location data revealed that the estimated relative amount of time sheep spent in each plot varied within each treatment, with sheep spending more time or achieving better coverage in some plots than others. Plots in which sheep spent more time tended to be rated as having more disturbance, according to researchers’ direct observations in the field, but this pattern was also variable. Areas where sheep bedded at night or “nooned” (rested midday) emerged as hotspots with the most sheep presence on the landscape, but the location data confirmed that sheep typically spent more time in fall and spring targeted grazing plots than in summer control plots that were intended to receive lighter, “once-over” grazing. While we do not expect targeted grazing treatments to have affected the seed bank during this initial sampling year, these measurements will provide crucial data to eventually assess sheep effects on the seed bank and aboveground vegetation composition over the longer term.
Research Outcomes
Restoring cheatgrass-invaded rangelands to more resilient native plant communities would benefit rangeland-based agriculture in the western U.S. We found that aboveground cheatgrass cover is a strong indicator of the amount of cheatgrass seed simultaneously present in the soil seed bank. However, the composition of above- and belowground communities as a whole differed significantly, suggesting that it will be challenging to use aboveground species cover as the sole indicator of native species’ viability in the seed bank.
Many of the species that emerged from the soil seed bank were also present in the aboveground community, but the seed bank community was more dominated by annual and non-native species. This finding is expected, given many annual and non-native species’ tendency toward ruderal growth strategies that would promote their emergence in a greenhouse study. However, several perennial species commonly used in restoration (e.g., Sandberg bluegrass (Poa secunda) and mountain big sagebrush (A. tridentata ssp vaseyana)) emerged in over 20% of the plots, across the full range of cheatgrass cover, suggesting that the seed bank continues to act as a reservoir for ecologically important species after cheatgrass invasion.
Taken together, our results suggest that managers are unlikely to be able to identify a cheatgrass cover threshold that could inform decisions about whether active re-seeding efforts would be needed as part of a restoration strategy to accompany cheatgrass removal. However, the emergence of native perennial species across a range of invasion levels suggests that these desirable species could be capable of natural regeneration following cheatgrass reduction by sheep targeted grazing, and that the seed bank could therefore be an economically practical strategy for aiding restoration of some sagebrush communities.
Future seed bank emergence trials could use a more diverse suite of strategies to encourage seeds to break dormancy, thereby enabling more native and perennial species to emerge from the seed bank (Kildisheva et al. 2019). Alternatively, techniques such as environmental DNA (eDNA) analysis could also potentially detect a wider diversity of species present in the seed bank to form a more complete picture of seed bank composition (Carrasco-Puga et al. 2020). Ultimately, a follow-up study after more years of targeted grazing implementation could shed light on the extent of natural regeneration of native species in the longer term.
Although the GPS collars worked well for tracking the bands’ movements during the treatment periods, we found that estimates of the relative amount of time that sheep spend in each plot cannot substitute for on-the-ground observations of vegetation utilization and other soil impacts, which we hypothesize are also strongly affected by sheep behavior, topography, and other factors. Future studies could more thoroughly investigate the relationships between time spent in plots, other behavioral and environmental factors, and their cumulative effects on aboveground vegetation. In addition, follow-up studies to assess the seed bank after additional years of targeted grazing, as well as more mechanistically-oriented studies, could help untangle the effects of sheep herbivory and trampling on seed presence in the seed bank to provide a more complete understanding of sheep’s potential role in helping to restore cheatgrass-invaded ecosystems. The GPS collar design could also be used in studies with other objectives for which sheep location data over relatively short periods (~2-3 weeks of GPS battery life) would be useful.
Education and Outreach
Participation Summary:
Our education plan focused on creating materials for different audiences to increase knowledge and understanding about the soil seed bank and how it can complement sheep grazing as a tool for restoring annual grass invaded rangelands. These audiences included: (1) students and researchers at the host university, who we reached through a guest lecture, research opportunities for students, development of field and greenhouse protocols and an open-source GPS case design, and direct consultations; (2) land managers and other native plant enthusiasts, who we reached through presentations to the Idaho Native Plant Society and educational videos posted online; and (3) sheep producers and other industry stakeholders, who we reached through a presentation at the West Central States Wool Growers Convention. Presentations at the Society for Range Management annual meeting also reached all target audiences, as will a forthcoming scientific publication and accompanying fact sheet.
We assessed the effectiveness of our outreach and education activities using Western SARE’s evaluation survey after outreach activities when possible. The survey was administered to attendees at our presentation to the Idaho Native Plant Society meeting (Education Objectives 1, 3) and to undergraduates who attended the Co-PI’s guest lecture in a “Restoration Ecology” course (Education Objective 2). Regrettably, due to a misunderstanding with our evaluation forms, we were unable to evaluate sheep producers’ reception of our presentation at the West Central States Wool Growers Convention. For education and outreach videos, we collected digital metrics of their views as a metric of their reach. We will also track instances in which researchers have elected to use our seed bank protocols in their research (Education Objective 2).
Objectives 1 and 3
After the Co-PI’s presentation to a chapter of the Idaho Native Plant Society, with an audience composed of mostly land managers and plant enthusiasts, 13 attendees took the Western SARE evaluation survey. Of these, 12 reported that the presentation improved their awareness of the soil seed bank’s use in restoration, and 11 reported gaining knowledge. They estimated that they would share some aspect of the presentation with over 70 others. The chapter’s Secretary wrote afterward: “Our group enjoyed learning how naturally occurring seed banks in the soil can oftentimes harbor enough native materials to allow for restoration to occur after dominant weedy species (such as invasive annual grasses) have been removed from the landscape…Utilizing targeted sheep grazing to remove such grasses helped demonstrate the practicality of this approach. Personally, I am a fan of reducing our dependence on pesticides, external seed sources, rangeland drills, and other disturbance-causing equipment in restoration work. [This] work feels like an important step in that direction.” Furthermore, webmetrics indicate that over 270 individuals have viewed the recording of the presentation, which is posted on the Idaho Native Plant Society’s website.
After the Co-PI’s lecture in the “Restoration Ecology” course at Boise State University, 10 of 12 students who took the Western SARE evaluation survey reported that they gained new knowledge about the soil seed bank as a potential restoration tool, and 70% reported that they gained new skills. They estimated that they would share some aspect of the presentation with over 75 other people, with one student stating that they would share what they had learned with “probably everybody I work with.”
In addition to the reported results to date, we will create a fact sheet to accompany the forthcoming scientific publication, which we will share with relevant stakeholders (producers and land managers) and post on the PI and Co-PI’s websites.
Objective 2
Researchers at Boise State gained direct knowledge of how to conduct seed bank emergence studies. One graduate student, 2 undergraduates, and a postdoc directly assisted with our seed bank study, and an additional graduate student consulted with the Co-PI about seed bank emergence studies for use in their research. Another professor is now conducting a seed bank study in her lab as well.
Education and Outreach Outcomes
As part of our planned outreach activities, Co-PI Byrne gave a presentation to the Idaho Native Plant Society. This talk was recorded and posted on their YouTube channel (https://www.youtube.com/watch?v=bIx4Go-e1pE). To date, this presentation has received over 270 views and currently comes up on Google’s search engine as the second video result for “soil seed bank.” Two journalists found the video by searching for “soil seed bank” and contacted Byrne for interviews and later published articles about the soil seed bank using information informed by the interviews (Washington Post and High Country News). Based on this experience, we recommend that future researchers utilize YouTube or other searchable video platforms to expand the reach of their educational and outreach activities, ideally partnering with organizations with existing online followings, such as a native plant society.
We received low response rates to our paper evaluation surveys following presentations. We had higher engagement when we and our colleagues in our presentation sessions used more interactive, digital evaluation tools. We therefore recommend a mix of paper and digital evaluation tools, or even a full transition to digital tools accessed via QR code, for larger audiences who may generally be less inclined to participate in evaluation activities.
- Knowledge of the soil seed bank and how it can be affected by annual grass invasion
- How to conduct seed bank research at the host institution
- The use of the seed bank as a complement for restoration in rangelands
The use of the seed bank as a complement for restoration in rangelands