Is Crested Wheatgrass Invasive in Sagebrush Steppe with Intact Understories in the Great Basin?

ABSTRACT Seeding crested wheatgrass (Agropyron desertorum [Fisch.] Schult.) in the sagebrush steppe is a controversial management action. There are concerns that crested wheatgrass may invade new areas and exclude native vegetation because many former crested wheatgrass seedings are near monocultures. However, crested wheatgrass is often seeded to prevent exotic annual grass invasion and stabilize soils after disturbances in areas where native vegetation is unlikely to establish. These areas can be difficult to identify and, consequentially, crested wheatgrass may be seeded in areas it is not needed, such as sagebrush steppe with understories dominated by native perennial grasses and forbs (intact understories). Thus, land managers need to know the potential effects of seeding crested wheatgrass, especially if it can invade and exclude native vegetation in sagebrush steppe with intact understories. We investigated the ability of crested wheatgrass to invade and exclude native vegetation by drill seeding crested wheatgrass into intact sagebrush steppe understories at five sites. To be invasive, crested wheatgrass would need to increase in abundance beyond initial establishment from drill seeding, causing decreases in native vegetation. We measured vegetation cover and density for 7 yr after seeding to evaluate the effects of seeding crested wheatgrass. Crested wheatgrass established with drill seeding but did not increase from its original density. Cover of crested wheatgrass increased with time but was < 2% seven yr after seeding. We found no evidence that seeding crested wheatgrass decreased the cover or abundance of native vegetation. These results suggest that intact sagebrush steppe understories are competitive with crested wheatgrass and, thereby, limit its recruitment and growth. Though crested wheatgrass did not show indications of being invasive in intact sagebrush steppe understories, longer-term evaluations with the inclusion of disturbances that might open safe sites in these communities would be prudent to fully understand its potential effects.


Introduction
Seeding of crested wheatgrass ( Agropyron desertorum [Fisch.] Schult.), a non-native bunchgrass, is a controversial management action ( Davies et al. 2021b ). However, crested wheatgrass is often seeded after wildfires in the sagebrush steppe because it is less expensive and establishes better than native species ( Pellant and Lysne 2005 ;Eiswerth et al. 2009 ;James et al. 2012 ;. Crested wheatgrass is also highly competitive and is fre-✩ Mention of a proprietary product does not constitute a guarantee or warranty of the product by USDA, Oregon State University, or the authors and does not imply its approval to the exclusion of other products. The USDA is an equal opportunity provider and employer. quently seeded to limit exotic annual grass invasion ( Arredondo et al. 1998 ;Davies et al. 2010 ). This competitiveness may be the reason that many crested wheatgrass seedings persist as near monocultures ( Marlette and Anderson 1986 ;D'Antonio and Vitousek 1992 ;Krzic et al. 20 0 0 ). In a crested wheatgrass-native bunchgrass coplanted study in Oregon, crested wheatgrass recruited substantially more individuals than native bunchgrasses Hamerlynck and Davies 2018 ), though these communities were purposefully understocked with bunchgrasses and other vegetation ( Ganskopp et al. 2007 ). Where crested wheatgrass seedings are near monocultures, it is unknown as to whether crested wheatgrass excluded native plants or these plant communities were denuded before seeding crested wheatgrass by prior disturbances and treatments applied to improve crested wheatgrass establishment ( Nafus et al. 2016 ;Williams et al. 2017 ).
Restoring plant communities with native species is generally preferred because these species coexist and provide important https habitat elements for native wildlife. However, seeding crested wheatgrass is often the only feasible option to limit exotic annual grass invasion and protect the soil resource after wildfires, especially in Wyoming big sagebrush ( Artemisia tridentata ssp. wyomingensis [Beetle and A. Young] S.L. Welsh) communities with a depleted understory before fire ( Davies et al. 2021b ). Crested wheatgrass is often planted preemptively to safeguard against fire-based transitions to postfire dominance of exotic annual grasses. However, this can also result in crested wheatgrass being planted in areas with abundant perennial bunchgrasses that are unlikely to transition to exotic annual grasslands. This occurs because it is difficult to immediately ascertain if enough perennial bunchgrasses remain after fire to occupy the site to limit annual grasses . Preemptive seeding of crested wheatgrass may be undesirable as revegetation-restoration resources are limited, but there may be ecological consequences ( Reynolds and Trost 1980 ;Heidinga and Wilson 2002 ;Henderson and Naeth 2005 ). Knowing the outcomes of preemptively seeding crested wheatgrass is needed to help managers weigh the risk-rewards of seeding crested wheatgrass into areas they are not sure will recover unassisted.
There is also concern that crested wheatgrass may spread from areas it has been seeded and, subsequently, displace native sagebrush plant communities (i.e., crested wheatgrass may be invasive in the sagebrush steppe). To spread into unseeded areas, crested wheatgrass would have to recruit new individuals. Thus, in this manuscript, we consider plants invasive if they increase in abundance (i.e., evidence they are recruiting in these communities, without assistance from humans) and these increases cause substantial declines in native plants ( > 20% decreases in cover and abundance). Crested wheatgrass's ability to recruit more rapidly and consistently than native bunchgrasses Hamerlynck and Davies 2018 ) suggests that it has potential to be invasive. In sagebrush rangelands of southern Idaho, Hull and Klomp (1967) reported that crested wheatgrass spread from where it was initially seeded but did not report the effects of it spreading, the abundance of crested wheatgrass in unseeded areas, or condition of the communities it was spreading into (other than some areas were plowed and burned before seeding). Thus, Hull and Klomp (1967) provide little insight into the risk of crested wheatgrass invading and dominating sagebrush steppe plant communities with intact understories.
In the Great Plains, crested wheatgrass is considered an invasive species because it invades native grasslands and decreases native cool season grasses and forbs ( Heidinga and Wilson 2002 ;Henderson and Naeth 2005 ). Extrapolating results from grasslands of the Great Plains to sagebrush steppe of the Great Basin is ill advised. Differences in the timing and amount of precipitation and, consequently, plant community composition between the Great Plains and Great Basin suggest the ecological ramifications of crested wheatgrass likely vary between these two regions. Determining if crested wheatgrass abundance substantially increases in sagebrush steppe communities, particularly those that have an intact understory, would be valuable for assessing if it is invasive in these communities. In addition to increasing in abundance, crested wheatgrass would also need to decrease native plants to be considered invasive. Evidence of crested wheatgrass increasing in abundance and causing a decrease in native species in sagebrush steppe plant communities with an intact understory is lacking and, thus, questions remain about its potential invasiveness in these communities.
The purpose of this study was to investigate the invasiveness of crested wheatgrass in sagebrush steppe communities with intact understories. We seeded crested wheatgrass in sagebrush steppe communities with intact understories 1) because we want to investigate the effects of crested wheatgrass as opposed to the ef-fects of prior degradation and 2) because there is concern that crested wheatgrass may spread into or inadvertently be seeded in sagebrush communities with intact understories. Seeding crested wheatgrass into sagebrush steppe communities with intact understories is not a recommended or commonly practice management action. To accomplish the purpose of our study, we drill seeded crested wheatgrass into Wyoming big sagebrush steppe communities with intact understories and compared it with areas not seeded. We expected that 1) drill seeding crested wheatgrass would establish crested wheatgrass into the communities, 2) over time crested wheatgrass would increase in cover and density, and 3), subsequently, lead to declines in native plant cover and density.

Study area
Study sites were located on the Northern Great Basin Experiment Range (NGBER) in southeast Oregon approximately 56 km west of Burns, Oregon, USA (lat 43 °29'N, long 119 °43'W). Wyoming big sagebrush was the dominant shrub across study sites. The dominant perennial bunchgrass was Thurber's needlegrass ( Achnatherum thurberianum (Piper) Barkworth) or bluebunch wheatgrass ( Pseudoroegneria spicata (Pursh) A. Löve) depending on the study site. Other bunchgrass species at the study sites included Sandberg bluegrass ( Poa secunda J. Presl), Idaho fescue ( Festuca idahoensis Elmer), prairie Junegrass ( Koeleria macrantha (Ledeb.) J.A. Schultes), and squirreltail ( Elymus elymoides (Raf.) Swezey). Understories were considered intact because they were dominated by native perennial bunchgrasses and forbs and cover values were within the range of intact Wyoming big sagebrush-bunchgrass communities in this region (e.g., Davies et al. 2006 ;Bates and Davies 2019 ). Climate at the NGBER is characteristic of the northwestern portion of the Great Basin with cold, wet winters and hot, dry summers. Long-term (1991-2020) average annual precipitation was 252 mm ( PRISM 2022 ). Annual precipitation was 96, 72, 64, 109, 109, 88, 103, and 60% of the long-term average in 2011, 2012, 2013, 2014, and 2018, respectively ( PRISM 2022. Though 2018 was exceptionally dry, critical spring precipitation for plant growth (March-May) was 85% of the long-term average. Elevation at the study sites ranges from 1 310 to 1 450 m above sea level. Topography at the study sites was relatively flat with slopes not exceeding 4%. Soils at the study sites are Aridisols and Andisols and are shallow to moderately deep.

Experimental design and measurements
We used a complete randomized block design with two treatments at five sites (blocks). Blocks varied in vegetation characteristics and were Thurber needlegrass ( n = 2), Thurber needlegrassbluebunch wheatgrass ( n = 1), bluebunch wheatgrass ( n = 1), and bluebunch wheatgrass-Idaho fescue ( n = 1) plant communities. Blocks were separated by up to 6 km. Each block consisted of two 5 × 10 m plots randomly assigned one of the treatments with a 2m buffer between treatments. Treatments were 1) unseeded control (unseeded) and 2) drill-seeded crested wheatgrass (seeded). Before treatment application, shrubs were removed by sawing stems off just above the ground surface in both unseeded and seeded treatment replicates. Shrubs were removed to simulate a disturbance that may initiate a management decision to seed and because drill seeding is more practical to apply without shrubs. After shrub removal, the plots assigned the seeded treatment were seeded with crested wheatgrass at a rate of 11.2 kg PLS ·ha −1 using a drill seeder (Versa-Drill, Kasco, Inc., Shelbyville, IN) in mid-September 2011. Drill rows were spaced 23 cm from center to cen-ter, and a roller was pulled behind the drill seeder to improve seed-soil contact.
Vegetation characteristics were measured annually in June 2012 through 2018. Herbaceous cover and density by species were measured in fifteen 40 × 50 cm (0.2 m ²) quadrats. Quadrats were located at 2-m intervals along three 10-m transects in each treatment replicate. Transects were deployed parallel to each other and perpendicular to the short edge of the treatment plot at 2-, 3-, and 4-m locations. Visual estimates of cover were aided by markings along the edges of the quadrats, dividing them into 5%, 10%, 25%, and 50% segments. Density was measured by counting perennial plants rooted inside the quadrats and by counting annual plants rooted inside a permanently marked 10% section of the quadrats. Bare ground, biological soil crusts, and litter cover were also estimated in the 0.2-m ² quadrats.

Statistical analyses
Repeated measures analysis of variance (ANOVA) using the mixed model procedure in SAS v. 9.4 (SAS Institute Inc., Cary, NC) with year as the repeated variable was used to investigate the effects of seeding crested wheatgrass into sagebrush steppe with intact understories. Treatment was considered a fixed effect in models. Block and block-by-treatment interactions were considered random variables in analyses. The treatment-by-year interaction was included in all models but only reported when significant. Covariance structure was selected for each model using the Akaike's Information Criterion ( Littell et al. 1996 ). Data that violated assumptions of ANOVAs were square root transformed. Nontransformed (original) data were presented in figures and text. Herbaceous vegetation was grouped into the following categories for analyses: 1) crested wheatgrass, 2) Sandberg bluegrass, 3) large perennial bunchgrasses (excluding crested wheatgrass and Sandberg bluegrass), 4) exotic annual grasses, 5) perennial forbs, and 6) annual forbs. Sandberg bluegrass was analyzed separately from the other bunchgrasses because it is smaller, initiates growth and senesces earlier, and often responds differently to disturbances and management ( McLean and Tisdale 1972 ;Yensen et al. 1992 ;Davies et al. 2021a ). The exotic annual grass group primarily consisted of cheatgrass ( Bromus tectorum L.). Herbaceous vegetation cover was also grouped into a total herbaceous cover category for analysis. Statistical significance was set at P ≤ 0.05.

Cover
The response of crested wheatgrass cover to treatment varied by year ( Fig. 1 A ; P = 0.035). The first yr after seeding, crested wheatgrass cover was similar between seeded and unseeded treatments (i.e., it was so low in the seeded area that it did not differ from zero). In subsequent years, crested wheatgrass cover was greater in seeded compared with unseeded treatments. Seven yr after seeding, crested wheatgrass cover in the unseeded and seeded treatments was 0% and a little < 2%, respectively. The response of Sandberg bluegrass cover to treatment varied by year (see Fig. 1 B; P < 0.001). Sandberg bluegrass cover was similar between treatments in 2012, 2013, and 2015; greater in the seeded compared with unseeded treatments in 2014, 2016, and 2017; and then greater in the unseeded compared with seeded treatments in 2018. Large perennial bunchgrass cover (excluding Sandberg bluegrass and crested wheatgrass) did not differ between treatments (see Fig. 1 C; P = 0.181) but varied among years ( P < 0.001). Exotic annual grass cover was low ( < 0.2%) in both treatments and did not vary between treatments or among years (see . Perennial forb and annual forb cover was similar between treatments (see Fig. 1 E and 1 F; P = 0.286 and 0.809, respectively) but varied among years ( P < 0.001). Total herbaceous cover did not differ between treatments ( Fig. 2 A ; P = 0.557) but varied among years ( P < 0.001). Bare ground and litter cover were similar between seeded and unseeded treatments (see Fig. 2 B and 2 C; P = 0.929 and 0.798, respectively) but varied among years ( P < 0.001 and 0.005, respectively). Biological soil crust was similar between treatments (see Fig. 2 D; P = 0.422) but varied among years ( P < 0.001).

Density
Crested wheatgrass density was greater in seeded compared with unseeded treatments ( Fig. 3 A; P = 0.012) but did not vary among years ( P = 0.370). Seven yr after seeding, crested wheatgrass density was approximately 2 and 0 individuals ·m −2 in seeded and unseeded treatments, respectively. The response of Sandberg bluegrass density to treatment varied by year (see Fig. 3 B; P = 0.014). Sandberg bluegrass density was generally greater in the seeded compared with the unseeded treatment, but in some years it was relatively similar between treatments. Large perennial bunchgrass density was similar between treatments and did not vary among years (see

Discussion
As expected, drill seeding established crested wheatgrass into sagebrush steppe communities with intact understories. Drill rows created by disks on the drill seeder likely created a seedbed conducive to crested wheatgrass establishment ( Hull and Klomp 1967 ). However, there was no evidence that crested wheatgrass increased from its original density over the 7 yr of the study. Crested wheatgrass cover increased over time in the seeded treatment as individual plants grew larger but amounted to < 2% after 7 yr. This rate of increase in cover is dissimilar to reports of rapid increases in crested wheatgrass cover in areas with less residual perennial vegetation. For example, seeding crested wheatgrass after the Cinder Butte Wildfire in Oregon increased large bunchgrass cover from 2.5% to > 7% in the first growing season post fire, and by the second growing season it had increased to over 12% ( Davies and Boyd 2021 ). Unseeded areas in the Cinder Butte Wildfire had large increases in exotic annual grasses, suggesting that many safe-sites were unoccupied ( Davies and Boyd 2021 ). In another study, crested wheatgrass seeded after invasive annual grass control with burning and herbicide application increased from 0% to 2.5% cover in the first yr and by the third yr it had increased to > 10% . These results, in contrast to our current findings, suggest that crested wheatgrass cover will increase rapidly if sites are not largely occupied by vegetation (i.e., there are excess resources), but increases will be minimal with competitive residual vegetation. The lack of substantial increase in crested wheatgrass cover in our current study is likely due, in part, to its inability to recruit new individuals into a community where safe-site availability is low. In support of this, there was a substantial increase in crested wheatgrass density and, consequently, cover between the second and third yr after it was seeded into exotic annual grasslands where competition had been greatly reduced following herbicide control treatments . Our results suggest that intact understories in sagebrush steppe communities are resistant to the recruitment and growth of crested wheatgrass. Native plant groups showed no evidence of decreasing in response to seeding crested wheatgrass in sagebrush communities with intact understories. This included large native perennial bunchgrasses that are expected to be most affected by crested wheatgrass because of similarity in function ( Nafus et al. 2020 ). Perennial forbs, a critical habitat element for wildlife ( Stephenson et al. 1985 ;Beck and Peek 2005 ;Pennington et al. 2016 ), were also unaffected by seeding crested wheatgrass. This implies that crested wheatgrass, particularly at the seedling stage, is not highly competitive with established native perennial grasses and forbs. These results suggest that near monocultures of crested wheatgrass may be caused by degradation of the sites before seeding crested wheatgrass. Others have also speculated that the degraded condition of rangelands at the time of seeding and treatments applied to improve crested wheatgrass establishment (e.g., burning, plowing, disking) likely contributed to near monocultures of crested wheat-grass ( Nafus et al. 2016 ;Williams et al. 2017 ). This may also explain why some crested wheatgrass seedings are not near monocultures ( Williams et al. 2017 ;Nafus et al. 2020 ), though environmental characteristics likely also play a role in variation in crested wheatgrass dominance ( Nafus et al. 2020 ).
Restoration of native perennial herbaceous species into areas seeded with crested wheatgrass has generally failed, and this has been attributed to crested wheatgrass competition ( Hulet et al. 2010 ;Fansler and Mangold 2011 ;Morris et al. 2019 ). However, establishment of native perennial herbaceous vegetation in these types of rangelands is rarely successful even without competition from crested wheatgrass ( Knutson et al. 2014 ;Svejcar et al. 2017 ), though exceptions exist, particularly in cooler and wetter sagebrush communities (e.g., Knutson et al. 2014 ;. This, combined with our current study, suggests that crested wheatgrass may not be, or at least not be the sole fac- Figure 2. Cover (mean + standard error) of cover groups in areas seeded with crested wheatgrass (seeded) and not seeded (unseeded) for 7 yr post seeding. tor, limiting the success of co-occurring native vegetation. In support of this, successful control of crested wheatgrass released exotic annual grasses, not native perennial vegetation ( Hulet et al. 2010 ). Increases in exotic annual grass limit native vegetation establishment ( Nasri and Doescher 1995 ;Rafferty and Young 2002 ). Crested wheatgrass, however, may be competitive with bunchgrass seedlings even if it does not recruit into established native perennial communities. A similar situation exists with exotic annual grasses in sagebrush communities, where established native perennial vegetation limits exotic annual grasses ( Davies and Johnson 2017 ), but these native perennials are not competitive with exotic annual grasses at the seedling stage ( Nasri and Doescher 1995 ;Clausnitzer et al. 1999 ;Young and Mangold 2008 ). Effort s investigating the effects of crested wheatgrass on the recruitment of native perennial seedlings in sagebrush steppe communities with intact understories are needed to determine if, over extended time periods, crested wheatgrass may affect the populations of native perennials.
The ability of crested wheatgrass to recruit into native rangelands and exclude native vegetation appears to be different between the sagebrush steppe and Great Plains. Dissimilar to our results, crested wheatgrass invades native grasslands in the Great Plains, leading to substantial reductions in native grasses and forbs ( Heidinga and Wilson 2002 ;Henderson and Naeth 2005 ) and, subsequently, greatly reduces species diversity ( Heidinga and Wilson 2002 ). The dominance of the sagebrush steppe with intact understory by native cool season (C 3 ) bunchgrasses may limit the ability of crested wheatgrass to substantially invade and dominant these communities. Species with more overlap in resource acquisition patterns would be expected to compete more with each other. In support of this, native cool season compared with warm season (C 4 ) grasses were more effective at excluding crested wheatgrass in the northern Great Plains ( Bakker and Wilson 2001 ). Thus, we speculate that the risk of ecological damage from crested wheatgrasses is very different between sagebrush steppe of the Great Basin and grasslands of the Great Plains.
Though our study was relatively long term compared with typical 2-to 3-yr studies, longer evaluations of crested wheatgrass's ability to recruit new individuals into intact communities are warranted. The effects of disturbances, such as fire, that open safe-sites on crested wheatgrass recruitment into these communities should also be investigated. However, these disturbance-induced safe-sites are now generally filled by exotic annual grasses ( Chambers et al. 2007 ). Whether or not these disturbances promote increases in crested wheatgrass may be irrelevant if crested wheatgrass recruitment is preventing exotic annual grass dominance, but it would be important to know if crested wheatgrass is limiting native perennial recruitment where it would otherwise occur. The effects of disturbances on crested wheatgrass recruitment and any subsequent effects on annual grasses should be investigated since exotic annual grass invasion often increases fire frequency to the detriment of native perennials ( Stewart and Hull 1949 ;D'Antonio and Vitousek 1992 ;Balch et al. 2013 ). Clearly, information on the longer-term plant community dynamics in areas where crested wheatgrass will be introduced is needed to better understand its potential effects.

Management Implications
The concern that seeding crested wheatgrass in sagebrush steppe will result in substantial degradation and loss of native vegetation in areas with intact understories and lacking substantial . Density (mean + standard error) of plant groups in areas seeded with crested wheatgrass (seeded) and not seeded (unseeded) for 7 yr post seeding. disturbance was not supported by this study. We found no evidence that crested wheatgrass was invasive in sagebrush steppe communities with intact, undisturbed understories. We also found no evidence that it would spread from where it was seeded, thus it seems unlikely crested wheatgrass will substantially spread from seed areas into sagebrush communities with intact understories. Seeding crested wheatgrass where the understory is not depleted is, however, a waste of limited restoration-revegetation resources and is not considered a recommended practice. Clearly, only seeding crested wheatgrass where the understory is depleted and there is a substantial risk of exotic annual grass dominance is ideal ( Davies et al. 2021b ), but our current research suggests that a state shift to a crested wheatgrass monoculture will not occur if it is inadvertently seeded in areas with an intact understory. Land managers will have to evaluate the risk of not seeding crested wheatgrass (e.g., exotic annual grass invasion and soil erosion) with the cost of seeding it (e.g., seed, fuel, labor, distur-bance from seeding method, opportunity cost of not seeding elsewhere), but we found no indication that they should be concerned that crested wheatgrass is invasive in intact, undisturbed sagebrush steppe understories. Crested wheatgrass is a valuable management tool for preventing exotic annual grass invasion in degraded sagebrush steppe, and our research alleviates some of the concerns that it will be invasive in intact communities without disturbance. Our results should not be misinterpreted to apply to other ecosystems as crested wheatgrass is invasive in grasslands of the Great Plains. Similarly, it would not be appropriate to extrapolate results from the Great Plains to the sagebrush steppe of the Great Basin.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.