Spring-Applied Treatments Offer Another Window of Opportunity for Revegetation of Annual Grass–Invaded Rangelands

ABSTRACT Invasive annual grasses have degraded tens of millions of hectares of the sagebrush ecosystem of western North America. Restoration of perennial vegetation in annual grass–invaded rangelands is a management priority to decrease fire risk, increase livestock forage quality, and improve wildlife habitat. Annual grasses are traditionally controlled in the fall with preemergent herbicides, such as imazapic, and treated areas are often seeded with perennial bunchgrasses 1 yr later to avoid nontarget herbicide damage to revegetation species. However, there is a limited window of time in the fall to accomplish annual grass control treatments. Spring-applied control treatments may be another option compared with only fall control treatments, but they have received little attention. We imposed spring-applied annual grass control treatments followed by fall seeding of a perennial bunchgrasses and then measured vegetation response for the next 3 yr in cheatgrass (Bromus tectorum) and medusahead (Taeniatherum caput-medusae)–invaded communities. Spring treatments that included imazapic application (at a low rate), followed by fall seeding of perennial bunchgrasses, successfully controlled annual grasses and substantially increased perennial bunchgrass cover and density. Spring burning and glyphosate herbicide application, without imazapic, were not successful in promoting substantial increases in perennial bunchgrass cover. Spring burning before imazapic application was the most successful treatment for rehabilitation seeding. By the third yr after seeding, perennial bunchgrass cover was 17% in the spring burn–imazapic treatment, greater than what is generally found in intact Wyoming big sagebrush (Artemisia tridentata ssp. wyomingensis)-bunchgrass communities in this region. The results of this study provide strong evidence that spring-applied control treatments including imazapic can be part of successful revegetation efforts, thereby decreasing some of the logistical challenges associated with revegetation of annual grass–invaded sagebrush rangelands.


Introduction
Invasive annual grasses are degrading rangelands around the world ( Purdie and Slatyer 1976 ;D'Antonio and Vitousek 1992 ;Brooks et al. 2004 ).Cheatgrass ( Bromus tectorum L.) and medusahead ( Taeniatherum caput-medusae [L.] Nevski) are two common invasive annual grasses that are especially problematic in the big ✩ Mention of a proprietary product does not constitute a guarantee or warranty of the product by the US Dept of Agriculture, Oregon State University, or the authors and does not imply its approval to the exclusion of other products that also may be suitable.The USDA is an equal opportunity employer and provider.
sagebrush ( Artemisia tridentata Nutt.) ecosystem of western North America, where they have invaded tens of millions of hectares ( Meinke et al. 2009 ;Brady et al. 2018 ;Smith et al. 2022 ).Annual grass invasion of sagebrush communities reduces biodiversity, degrades wildlife habitat, and decreases livestock forage ( Torell et al. 1961 ;Crawford et al. 2004 ;Davies and Svejcar 2008 ;Davies 2011 ).These problems are a product of invasive annual grasses being highly competitive with native vegetation, particularly at the seedling stage, often leading to the exclusion of native plants ( Melgoza et al. 1990 ;Nasri and Doescher 1995 ;Rafferty and Young 2002 ;Humphrey and Schupp 2004 ;Young and Mangold 2008 ).
Annual grass invasion also increases the amount and continuity of highly flammable fine fuels ( Davies and Nafus 2013 ), frequently resulting in fire return intervals that are too short for the persistence of many native plants and the development of an annual grass−fire feedback cycle ( D'Antonio and Vitousek 1992 ;Brooks et al. 2004 ;Crist et al. 2023 ).
Restoration of perennial dominance of invasive annual grass−dominated sagebrush communities is often a land management objective to improve livestock forage, decrease fire probability, and enhance wildlife habitat.The most successful effort s usually involve controlling annual grasses with a fallapplied preemergent herbicide followed 1 yr later with seeding perennial vegetation ( Davies 2010 ;Kyser et al. 2013 ;Davies et al. 2015 ;Davies et al. 2024 ).Seeding is usually delayed 1 yr after preemergent herbicide application to limit nontarget damage to seeded species ( Davies 2010 ;Kyser et al. 2013 ;Davies et al. 2014 ).Imazapic is a commonly used preemergent herbicide in effort s to restore perennial vegetation to annual grass−invaded rangelands ( Davies 2010 ;Elseroad and Rudd 2011 ;Davies and Boyd 2018 ).Burning before imazapic application can substantially improve revegetation success by removing accumulated plant litter, which serves to increase herbicide-soil contact and create a more favorable seedbed ( Davies 2010 ;Davies et al. 2015 ).Though these methods can be successful, there are logistical challenges.There is only a small window of time in the fall that these treatments can be applied before annual grass emergence, at which point preemergent herbicide control effectiveness drastically decreases.It is also logistically challenging to prescribe burn these annual grass−invaded rangelands in the fall because there is often a short period of time when it is safe enough to prescribe burn prior to it being too wet and cool for fire to carry effectively.Hence, it would be beneficial if there were other times of the year when annual grass control treatments could be applied.Spring control treatments may be especially valuable if perennial vegetation could be seeded in the fall of the same year, leading to successful establishment in the next year.This would decrease the logistical challenge of having revegetation projects that require multiple years to implement and potentially restore rangeland productivity faster than traditional effort s.
Spring control of invasive annual grasses, followed by effort s to restore perennial vegetation, have received limited attention.Spring treatments with imazapic and contact herbicides such as glyphosate can reduce invasive annual grasses, but their effectiveness at promoting establishment of seeded vegetation remains inconclusive ( Kyser et al. 2013 ;Morris et al. 2016 ).It also remains unknown if herbicide toxicity from spring imazapic application abates enough for fall-seeded vegetation to be successful.One major advantage of spring treatments is that burning at this time of year is generally safer and less likely to have fire escape compared with fall burning ( Allen et al. 2008 ).Spring burning improved fallapplied imazapic control of medusahead and increased the cover and density of bunchgrasses seeded 1 yr after imazapic application ( Davies 2010 ).However, it remains unknown if spring burning before spring application of imazapic can improve annual grass control and, subsequently, establishment of bunchgrasses seeded in the fall of the same year.Nontarget herbicide damage to seeded bunchgrasses may limit success with this treatment; therefore, evaluations of spring treatments to control invasive annual grasses followed by fall seeding of perennial bunchgrasses are needed.
The objective of this study was to determine the efficacy of spring-applied invasive annual grass control treatments, followed by fall seeding a perennial bunchgrass for restoring dominance of perennial vegetation to annual grass−invaded rangelands.Specifically, we investigated the effects of prescribed burning, glyphosate application, imazapic application, combined glyphosate and imazapic application, and combined burning and imazapic application in the spring to control annual grasses and promote the establishment and growth of a fall-seeded perennial bunchgrass.Imazapic was applied at a low rate to decrease the likelihood of nontarget herbicide damage to bunchgrasses seeded in the fall of the same year.We expected that treatments including imazapic would have less annual grass and greater perennial bunchgrass cover and density.We also expected that combination treatments would improve annual grass control and, subsequently, result in greater bunchgrass cover and density.

Study area
The study was conducted in annual grass−invaded sagebrush steppe communities in southeastern Oregon approximately 30 km east of Crane, Oregon.Six annual grass−invaded sites were included in the study.The six sites are located on R010XC021OR-SR Clayey 9-12 PZ ecological sites.Potential natural plant communities at the study sites would have been Wyoming big sagebrush ( Artemisia tridentata ssp.wyomingensis [Beetle & A. Young] S.L. Welsh)−bunchgrass steppe.Dominant bunchgrasses would have been bluebunch wheatgrass ( Pseudoroegneria spicata [Pursh] A. Löve) and Thurber's needlegrass ( Achnatherum thurberianum [Piper] Barkworth).Before treatments, study sites were dominated by invasive annual grasses (medusahead and cheatgrass) with no shrub overstory and almost no residual perennial bunchgrasses or perennial forbs.Elevations at study sites were ∼1 0 0 0 m above sea level, and slopes were relatively flat (0−3%).Soils were moderately deep, and textures were clay loams.Long-term average annual precipitation was 267 mm ( PRISM 2017 ), with the majority occurring as snow or rain from October to May.Crop year (Oct-Sept) precipitation was 75% and 92% of the long-term average during the pretreatment and treatment years (2012 and 2013), respectively ( PRISM 2017 ).Crop year precipitation was 69%, 99%, and 98% of the long-term average for the 3 yr (2014−2016) following treatment application, respectively ( PRISM 2017 ).Livestock were excluded from the study sites with barbed wire fences for the duration of the study.Wild herbivores were not excluded from the study sites.

Experimental design and measurements
We used a randomized complete block design with six sites (blocks) to evaluate treatment effects on vegetation.Treatments were randomly assigned to one 6 × 6 m plot in each block and were spring burning (SB), glyphosate application (G), imazapic application (I), spring burning and imazapic application (SB-I), imazapic and glyphosate application (I-G), untreated control (C-S), and untreated control without seeding (C).All treatment plots, except for the unseeded control (C), were drill-seeded with desert wheatgrass ( Agropyron desertorum [Fisch.Ex Link] Schult) at a rate of 46 kg pure live seed •ha −1 in October 2013.Treatment plots had a 3m wide buffer between them and other adjacent treatment plots.Spring burns were conducted on 7 May, 2013 using drip torches to ignite head-fires.During the burns, wind speeds varied from 2 to 6 km •h −1 , relative humidity ranged from 28% to 32%, and air temperature was 18 °C.Burns were complete across burn treatment plots consuming nearly 100% of the fine fuel.Herbicides were applied with a utility terrain vehicle (UTV)-mounted boom spray with a nozzle height of 60 cm and a tank pressure of 207 kPa.Imazapic was applied at a rate of 52 g ai •ha −1 on 8 May, 2013.This rate of imazapic is 25% of the highest rate recommended by the manufacture.Glyphosate was applied at a rate of 1 065 g ai •ha −1 on 30 May, 2013.Seeding was accomplished by using a Versa-Drill (Kasco, Inc, Shelbyville, IN) pulled behind a UTV with drill rows spaced 23 cm apart.Treatment plots burned in the mid-July 2014 Buzzard Wildfire Complex, but burn severity was low because there was not any woody vegetation on the study sites and herbaceous vegetation was dormant at the time.
Vegetation cover and density were measured in June in the 3 yr post seeding (2014−2016).Foliar cover by species was visually estimated in 40 × 50 cm (0.2 m −2 ) quadrats located at 1-m intervals along four, 5-m transects, resulting in 16 quadrats per treatment replicate.The 5-m transects were located parallel to each other and separated by 1 m.The 0.2 m −2 quadrats had markings along the edges dividing them into 5%, 10%, 25%, and 50% segments to aid in visual estimates of cover.Density of perennial herbaceous species was measured by counting all individuals rooted in the sixteen 0.2 m −2 quadrats in each treatment replicate.Density of annual herbaceous species was measured by counting all individuals rooted inside a 10% permanently marked section of the 0.2 m 2 quadrats.Shrubs were not measured because no shrubs occurred in any of the treatment replicates.

Statistics
We used repeated measures analyses of variance (ANOVAs) using the mixed model method (PROC MIXED SAS v. 9.4) to evaluate treatments.Year was the repeated variable and block and block-by-treatment interactions were treated as random variables in analyses.Data were square root transformed when assumptions of ANOVAs were not met.Text and figures report nontransformed data.We selected covariance structures for analyses using the Akaike's information criterion ( Littell et al. 1996 ).Vegetation cover and density were categorized into five groups for analyses: perennial bunchgrasses (excluding Sandberg bluegrass [ Poa secunda J. Presl]), Sandberg bluegrass, invasive annual grasses, perennial forbs, and annual forbs.Sandberg bluegrass was analyzed individually because it is smaller in stature, develops earlier, and responds differently to management and disturbances than other perennial bunchgrasses in this ecosystem ( McLean and Tisdale 1972 ;Yensen et al. 1992 ;Davies et al. 2021a ).The invasive annual grass group was composed of medusahead and cheatgrass.The annual forb group was predominantly non-native species ( > 90%).Shrubs were not analyzed because they did not occur in any treatment replicate.Treatment means were separated with the LS function in SAS (SAS v. 9.4).Means were considered different at P ≤ 0.05 and reported with standard errors.

Results
The response of perennial bunchgrass and invasive annual grass cover to treatments varied among years, with the magnitude of differences among treatments increasing with time ( Fig. 1 A and B; P < 0.001 and < 0.001).Treatments that included imazapic resulted in greater perennial bunchgrass cover than other treatments.Spring burning followed by spring application of imazapic resulted in the greatest cover of perennial bunchgrasses.By the end of the study, perennial bunchgrass cover was 1.4-and 21.4-fold greater in the spring burn followed by imazapic application compared with the imazapic-only treatment and the untreated control, respectively.We did not find evidence that treatments without imazapic increased perennial bunchgrass cover compared with the untreated control.Invasive annual grass cover appeared to only be substantially reduced in treatments that included imazapic.Spring burning before imazapic application resulted in the greatest reduction in invasive annual grass cover.We did not detect a difference in invasive annual grass cover among treatments that did not include imazapic and the untreated control.The response of annual forb cover to treatment varied among years ( Fig. 1 C; P < 0.001).Spring burning resulted in an initial increase in annual forb cover, but this decreased substantially in subsequent years in spring burn treatments that included imazapic.Treatments that included imazapic resulted in substantial declines in annual forb cover compared with other treatments.Other treatments did not result in declines in annual forb cover compared with the untreated control.Sandberg bluegrass and perennial forb cover were not influenced by treatment ( Fig. 2 A and B; P = 0.925 and 0.437, respectively).
The response of perennial bunchgrass and invasive annual grass densities to treatments varied among years ( Fig. 3 A and B; P < 0.001 and P = 0.006, respectively).Perennial bunchgrass density decreased substantially from the first to second yr in seeded treatments, but not in the untreated control.Perennial bunchgrass density was greater in all seeded treatments compared with the untreated control.Densities of perennial bunchgrass were greater in treatments that included imazapic compared with other treatments.Combination treatments (SB-I and I-G) resulted in the greatest perennial bunchgrass densities at the end of the study.Treatments that include imazapic, glyphosate, or spring burning decreased invasive annual grass densities compared with the untreated control.Spring burning followed by imazapic application resulted in the greatest reduction in annual grass density.At the end of the study, annual grass density was > 50% less in the spring burn followed by imazapic application treatment compared with the untreated control.The response of annual forb density to treatments varied among years, with some of the treatment differences becoming greater with time ( Fig. 3 C; P < 0.001).Treatments that included imazapic, burning, or glyphosate resulted in lower annual forb densities compared with the untreated control.Spring burning followed by imazapic application resulted in the greatest reduction in annual forb densities.Sandberg bluegrass and perennial forb densities were not influenced by treatments ( Fig. 4 A and B; P = 0.940 and 0.650, respectively).Densities of both plant groups were very low and remained low throughout the duration of the study.

Discussion
Restoration of perennial dominance in annual grass−invaded rangelands is a challenging endeavor ( Svejcar et al. 2017 ;Davies et al. 2021b ;Davies et al. 2024 ).The limited window of opportunity to control annual grasses in the fall, the risk of fire escape with fall burning, and having to plan revegetation efforts over several years has made this challenging problem even more difficult.Our results suggest that spring-applied annual grass control treatments are a promising alternative to traditional fall control treatments for revegetating annual grass−dominated sagebrush communities.Having the option of applying control treatments in the spring provides more opportunities to restore perennial vegetation to annual grass−invaded rangelands.
In support of our hypothesis, revegetation treatments that included imazapic were generally more successful.Application of a preemergent herbicide, imazapic in this study, appears critical to controlling invasive annual grasses and, thereby, improving seeded bunchgrass establishment.Spring burning without imazapic was ineffective for revegetation efforts in our studied annual grass−invaded rangelands.These results mirror a prior attempt to use prescribed burning to revegetate medusahead-invaded rangelands ( Davies 2010 ), though in that study, seeding of perennial bunchgrasses was delayed until the fall of the next year.Burning, in general, is not very successful at controlling invasive annual grasses ( Young et al. 1972 ;Davies and Sheley 2011 ).Glyphosate treatments were also not effective at promoting perennial bunchgrasses, though they can control annual grasses when they are actively growing ( Kyser et al. 2013 ).A likely shortcoming with both prescribed burning and glyphosate application is that they only control the current population of actively growing annual grasses.Thus, a single point in time control does not adequately reduce annual grasses as subsequent cohorts may maintain annual grass dominance and limit revegetation success, whereas imazapic and other preemergent herbicides can provide continued control of an-  nual grasses (including subsequent cohorts) because of their soil activity.
Prescribed spring burning before imazapic application substantially improved revegetation success.Three yr after seeding, perennial bunchgrass cover was 17% in this treatment combination.This amount of perennial bunchgrass cover is greater than what is generally observed in intact Wyoming big sagebrush communities in this region ( Davies et al. 2006 ;Bates and Davies 2019 ).Prescribed burning before imazapic application has improved annual grass control and the establishment of seeded perennial vegetation in other annual grass revegetation efforts ( Davies 2010 ).Burning, if applied prior to seed maturity, may offer some control of annual grasses ( Murphy and Lusk 1961 ;DiTomaso et al. 2006 ;Davies et al. 2013 ).However, we speculate the biggest contribution to revegetation success is the removal of the annual grass litter, leading to 1) greater preemergent herbicide control of annual grasses and 2) an improved seedbed for desert wheatgrass establishment.Burning improves imazapic control of annual grasses by removing litter that would otherwise intercept soil active herbicides ( Monaco et al. 2005 ;Davies 2010 ;Davies and Sheley 2011 ;Clark et al. 2019 ).Removing the litter is also beneficial because soil surface litter creates a microenvironment that is favorable for annual grass emergence and growth ( Facelli and Pickett 1991 ;Newingham et al. 20 07 ;Adair et al. 20 08 ) and can be a barrier to perennial plant establishment ( Evan and Young 1970 ).Prescribed burning before imazapic application appears critical for improving revegetation success of annual grass−invaded rangelands.
Abundance of nonseeded perennial plant groups (Sandberg bluegrass, perennial forbs, and shrubs) remained low in or, in the case of shrubs, absent from all treatment plots regardless of treatment.The lack of a response from unseeded species to annual grass control has been noted in other effort s when the pretreatment plant community is nearly an annual grass monoculture ( Davies 2010 ;Davies and Boyd 2018 ;Davies and Hamerlynck 2019 ).Though not conclusive, this suggests that these species may be seed limited in many annual grass−invaded rangelands.Whether or not augmenting the seedbank by seeding these species will lead to increases in their abundance remains to be seen as  perennial vegetation faces many challenges to establishment in these ecosystems ( Svejcar et al. 2017 ;Davies et al. 2021b ).However, if these perennial plant groups establish and persist after annual grass control, they may increase the resistance of these plant communities to reinvasion.Established perennial vegetation, regardless of plant group, is critical to preventing invasive annual grasses from redominating plant communities after annual grass control ( Davies and Johnson 2017 ).More diverse assemblages of perennial plant communities likely decrease safe sites for invasive plants ( Sheley et al. 2006 ).Additional research is warranted investigating whether seeding these plant groups after annual grass control can successfully increase their abundance and limit reinvasion.

Implications
Spring control treatments before fall seeding of desert wheatgrass can be used to successfully revegetate annual grass−invaded sagebrush steppe.Additional evaluations of spring control treat-ments followed with seeding of other species in different years is warranted to determine the utility of these treatments.Our results suggest that land managers are not solely reliant on applying annual grass control treatments in the fall.Including a preemergent herbicide, imazapic in this study, appears essential for improving control of annual grasses and establishment of seeded perennial bunchgrasses.It should be noted that we used a low rate of imazapic, which may be vital to limiting nontarget herbicide damage when seeding in the fall after spring application.Treatments without imazapic did not result in substantial increases in perennial bunchgrass abundance, providing evidence that they were ineffective unless combined with a preemergent herbicide.Prescribed burning before preemergent herbicide application can greatly improve revegetation success.Land managers are often hesitant to apply fall burning because of the risk of fire escape.The ability to use spring burning may increase the likelihood that managers will use prescribed burning in revegetation efforts because the risk of fire escape is much less.Fuels, other than annual grass litter, are generally actively growing and too wet to burn in the spring, thus only areas with annual grass litter accumulations are likely to burn.The duration of revegetation efforts can also be shortened with spring treatments compared with traditional fall control treatments followed 1 yr later with seeding.This may make revegetating annual grass−invaded rangelands less logistically challenging since all revegetation actions, including seeding, can be implemented in a single year.This study provides strong evidence that spring-applied control treatments as part of revegetation efforts can be effective.These results suggest there are more windows of opportunity to implement revegetation in annual grass−invaded sagebrush communities than traditional fall treatments alone.

Declaration of Competing Interest
We do not have any conflict of interest to declare.

Figure 1 .
Figure 1.Perennial bunchgrass, invasive annual grass, and annual forb cover in the treatment plots from 2014 to 2016.SB-I indicates spring burn and imazapic application; C, untreated control; CS, untreated control with seeding bunchgrasses; G, spring glyphosate application; I-G, spring imazapic and glyphosate application; I, spring imazapic application; and SB, spring burn.All treatments, except for the untreated control (C), were seeded with desert wheatgrass ( Agropyron desertorum [Fisch.Ex Link] Schult) in the fall of 2013.

Figure 2 .
Figure 2. Sandberg bluegrass and perennial forb cover in the treatment plots from 2014 to 2016.SB-I indicates spring burn and imazapic application; C, untreated control; CS, untreated control with seeding bunchgrasses; G, spring glyphosate application; I-G, spring imazapic and glyphosate application; I, spring imazapic application; SB, spring burn.All treatments, except for the untreated control (C), were seeded with desert wheatgrass ( Agropyron desertorum [Fisch.Ex Link] Schult) in the fall of 2013.

Figure 3 .
Figure 3. Perennial bunchgrass, invasive annual grass, and annual forb density in the treatment plots from 2014 to 2016.SB-I indicates spring burn and imazapic application; C, untreated control; CS, untreated control with seeding bunchgrasses; G, spring glyphosate application; I-G, spring imazapic and glyphosate application; I, spring imazapic application; SB, spring burn.All treatments, except for the untreated control (C), were seeded with desert wheatgrass ( Agropyron desertorum [Fisch.Ex Link] Schult) in the fall of 2013.

Figure 4 .
Figure 4. Sandberg bluegrass and perennial forb density in the treatment plots from 2014 to 2016.SB-I indicates spring burn and imazapic application; C, untreated control; CS, untreated control with seeding bunchgrasses; G, spring glyphosate application; I-G, spring imazapic and glyphosate application; I, spring imazapic application; SB, spring burn.All treatments, except for the untreated control (C), were seeded with desert wheatgrass ( Agropyron desertorum [Fisch.Ex Link] Schult) in the fall of 2013.