Invasive winter annual grasses, such as cheatgrass (Bromus tectorum L.) are considered serious threats to regional biodiversity. Pollinator populations that depend on the native flora are likely to be negatively impacted as these native species may be displaced by the invasive grass species. Colonization by cheatgrass is also predicted to increase risk of wildfires, as dead plant parts provide fuel in the already dry and arid regions of the western United States. Biocontrol, grazing, prescribed burning, or use of broad-spectrum nonselective herbicides have been suggested as possible means to control B. tectorum. Efficient control may facilitate regrowth of native flora that could in turn support other ecosystem functions. Reporting our findings as a case study, we describe here the results of the application of a preemergent herbicide, indaziflam, that limits germination of B. tectorum seeds. Herbicide was applied to the study locations during the months of December 2016, January 2017, and February 2017. The data reported here on the diversity of flowering plants were collected between May through September 2018. Herbicide-treated plots showed an increase in diversity and abundance of flowering plants compared to the untreated control within two seasons after cheatgrass control was implemented, suggesting that effective reduction of the population of the invasive annual cheatgrass may help facilitate the growth of native forbs. Further studies are necessary to understand mechanisms that facilitate reestablishment of native flowering species, the long-term consequences of reducing invasive annual grasses and to document any residual effects of the herbicide on ground-nesting pollinators.
Introduction
Invasive annual grasses have been shown to have devastating consequences on native biodiversity, environmental quality, and ecosystem services (Bartz and Kowarik 2019; Jones and McDermott 2018; Kumar Rai and Singh 2020; Pejchar and Mooney 2009). These invasive annual grasses effectively displace native vegetation (Pyšek et al. 2012), thus altering decomposition cycles and soil food webs (Lenz et al. 2003), disrupting ecosystem networks such as the plant–pollinator reproductive mutualisms (Schweiger et al. 2010; Traveset and Richardson 2006), altering historic fire regimes (D'Antonio and Vitousek 1992), and displacing ecosystem diversity and stability (Musil et al. 2005). Although controlling invasive species has received global and regional attention, the success of control measures and the positive impacts on the ecosystems following control may not always be uniform or generalizable (Adams et al. 2020; Skurski et al. 2013) and may vary across different levels of ecological complexity (Vilà et al. 2011). It has been suggested that effective control measures for invasive species, should also emphasize ecological processes that prevent reinvasion, possibly combining control with simultaneous restoration to retain broader ecosystem functions (D'Antonio et al. 2004; Flory and Clay 2009; Monaco et al. 2017).
Invasive species contribute to biodiversity losses by compounding effects of habitat destruction, agricultural intensification, and climate change, as recently discussed by Wagner (2020) in a report on global decline in insect biodiversity. Although the extent of impact may vary across different ecosystems, declining populations of insects, specifically pollinators, could compromise reproductive success of native flora (Gilgert and Vaughan 2011) and affect ecosystem functioning (Blüthgen and Klein 2011). While the impact of invasive species on native vegetation is relatively well described, and studies demonstrate targeted ecosystem trade-offs resulting from controlling invasive species (Adams et al. 2020; Pyšek et al. 2012; Skurski et al. 2013), few studies explore the relation between control of invasive plants and the subsequent impact on pollinator-friendly forbs. In the rangelands of Colorado, this relationship is especially critical, as the well-documented bee diversity of this region (Goldstein and Scott 2015) is important for the reproductive success of the native forbs. While studies indicate that the bee populations in these rangelands may not be currently experiencing concerning declines (Kearns and Oliveras 2009b) and have been conserved over several decades (Kearns and Oliveras 2009a), the spread of invasive annual grasses could compromise the habitat quality of these rangelands, negatively impacting bee populations in the long run.
It has also been suggested that winter annual invasives such as cheatgrass or downy brome (Bromus tectorum L.) affect regional ecosystem functions (Boyte et al. 2016; Knapp 1996). A systematic review spanning 64 yr (Monaco et al. 2017) suggests that of the different methods of control currently available, only one method, herbicide application, decreased B. tectorum and increased perennial grass abundance over the long term, lending support to herbicide-based control methods. Recent research reports from this region describe the efficient control of B. tectorum following winter application of a preemergent herbicide, indaziflam, a chemical whose cellulose biosynthesis-inhibiting action inhibits root development following seed germination, a mechanism different from previously used herbicides (Clark et al. 2019; Sebastian et al. 2016, 2017a). These studies also report that the residual effects of indaziflam application may last up to nearly three years, allowing for further reduction of B. tectorum seedbank in the soil (Sebastian et al. 2017b), improving the potential for native forbs to reestablish after continued germination suppression of the invasive grass seeds. Taken together, the inhibited seed germination and longer residual effect suggest that controlling B. tectorum during the winter months could improve reestablishment of the spring-emerging native flora in these rangelands (Sebastian et al. 2016). With this in mind, we test the hypothesis that herbicide-mediated control of B. tectorum has a positive impact on the native flora in the rangelands of Colorado by identifying the diversity and abundance of pollinator-friendly flora in the herbicide-treated and control plots. We report the findings as a case study.
Materials and Methods
Study Locations and Treatments
Three geographic locations within Boulder County, CO, shown in Figure 1 were identified for the study such that each location had paired herbicide-treated and untreated plots. Plot sizes depended on the terrain, but all plots had at least one side measuring 100 m in length. During the winter months, between December 2016 through February 2017, the area where the treated plots were demarcated received application of the preemergent herbicide indaziflam (Esplanade™, Bayer Crop Science, St. Louis, MO 63167, USA) at the rate of 102 g ai ha–1. The exact dates of application varied based on accessibility over the terrain and weather conditions. The paired treated and untreated plots were in similar habitat types with vegetation cover dominated by B. tectorum and field brome (Bromus arvensis L.; syn.: Bromus japonicus Thunb.) and 0% to 10% canopy cover of scattered co-occurring species (for a list of co-occuring species, see Sebastian et al. 2017a). The coordinates of the three locations, Rabbit Mountain Open Space West (RM 1, herbicide applied to treated plots in January 2017), Rabbit Mountain Open Space East (RM 2, herbicide applied to treated plots in February 2017), and Colp (herbicide applied to treated plots in December 2016) are shown in Figure 1. Every effort was made to ensure that the treated and control plots were in the same vicinity, but in the case of Colp, this was not feasible due to the lack of suitable locations of the required size close to the treated location. Therefore, as shown in Figure 1, the treated and control plots at Colp are farther apart than at the other two study locations, and the control plot at the Colp measured 80 by 100 m. Based on our observations during the 2018 study season, we are confident that this did not significantly affect the results being presented here.
Transect Sampling
Eight permanent 100-m belt transects were established at each survey plot, spaced evenly across the vertical and horizontal axes of the plots. A meter tape was stretched between the ends of the transects to demarcate the transect line. A 1-m2 frame was placed 1 m away from the tape at 10-m intervals, on alternating sides of the transect line. All flowering plants within the frame were identified and the number of plants of each species were counted before moving to the next frame-stop that was 10 m away. Plants that were not flowering during the sampling weeks were not recorded. When frames landed in areas with no flowering individuals, researchers moved to the next 10-m stop. For data analysis, the number of plant species in bloom was pooled across all quadrats for each transect. Sampling was conducted for a period of 8 wk (9 wk in Colp) beginning in May through September. Through the season, there were a total of 48 belt transects completed across all locations and plots. Each of the three study locations had a total of 16 belt transects, with 8 each in the treated and control plots.
Random Walk Sampling
Flowering plants that did not fall within the sampling frames of the belt transects could not be recorded during the entire study period. We conducted focal-flowering plant sampling in the entire plot using random walk sampling to obtain a census or inventory of all flowering plants that were not counted in the transect sampling. We walked the plot in an organized fashion beginning at one end spanning the entire plot, specifically targeting all flowering plants that did not fall into the frames, recording all blooming species in the plot on any given sampling day. The data collected by random walk sampling were used to create an inventory of all flowering plants recorded in our study.
Data Analyses
Standard ecological indices for plant species diversity, richness, and abundance were calculated for the treated and untreated plots for each sample event in each geographic location. Species richness is simply the total number of unique species during each sampling event. Shannon diversity index 
and Simp son's diversity index 
were calculated as the diversity measures for pollinator-friendly flora in the three geographic locations (Magurran 2013; Ortiz-Burgos 2016; Pielou 1966; Simpson 1949; Whittaker 1972). The Shannon diversity index combines evenness and richness into a single measure and assumes that all species are represented in a sample, while Simpson's diversity index gives more weight to common species and assumes that the few rare ones with only a few representatives will not affect the diversity values.
Data from the belt transect sampling were analyzed using a general linear model for multiple dependent variables. Treatments and geographic locations were fixed effects; sampling week was a covariate; and species richness, Shannon diversity index, and Simpson's diversity index values were dependent variables. Treatment by location interaction was also determined to analyze any location-specific response. As needed, logarithmic transformations were performed for nonnormal species richness data before analysis.
Figure 1.
Study locations in Boulder County Parks and Open Space area in Colorado. State map of Colorado on the left and the inset study area on the right. The locations and their coordinates: RM 1, Rabbit Mountain 1 (Untreated: 40.2547°N, 105.2139°W; Treated: 40.2495°N, 105.2143°W); RM 2, Rabbit Mountain 2 (Untreated: 40.2468°N, 105.1984°W; Treated: 40.2463°N, 105.2015°W); Colp (Untreated: 40.1861°N, 105.2526°W; Treated: 40.1396°N, 105.2819°W).
Results and Discussion
Here we present a case study showing the richness and diversity of pollinator-friendly flora in three locations where a preemergent herbicide, indaziflam, was applied to control the invasive annual grass B. tectorum. There was a significant treatment effect on the different diversity measures. Herbicide-treated plots had higher richness and alpha-diversity measures across all three locations (Table 1; Figure 2; species richness: F(1, 41) = 23.25, P = < 0.0001; Shannon diversity index: F(1, 41) = 20.29, P = 0.001; Simpson's diversity index: F(1, 41) = 15.87, P = 0.001), suggesting that the control of B. tectorum could result in reduced competition allowing for the reestablishment of native flowering plants. There was no significant effect of location on these measures (species richness: F(2, 41) = 0.61; Shannon diversity index: F(2, 41) = 0.08; Simpson's diversity index: F(2, 41) = 1.09) and no significant interaction effect between treatment and location (species richness: F(2, 41) = 3.58; Shannon diversity index: F(2, 41) = 2.96; Simpson's diversity index: F(2, 41) = 1.81).
To visualize these diversity measures across seasons, the data were grouped into early (May to early June), mid (June to July), and late (August to September) seasons, as presented in Figures 3 and 4. Table 2 provides the list of pollinator-friendly plant species that were blooming during the study period in the three locations. The impact of herbicide application was consistent in the three locations, suggesting the possibility that previously demonstrated herbicide-mediated control of the invasive grass, B. tectorum (Clark et al. 2019, 2020; Sebastian et al. 2016, 2017a) could be responsible for the growth of pollinator-friendly flora. A noteworthy caveat is that our study did not measure the abundance of B. tectorum in the study plots. Therefore, reduced competitive pressure as a possible means for reestablishment of flowering plant species is a proposed mechanism.
Invasive annual grasses have been shown to impact community composition in ecosystems where they are invasive, leading to potential reductions in abundance and diversity of native species in these ecosystems. However, the intensity of displacement likely depends on the ecological context, specifically the ability of one species to preempt another (Fridley et al. 2021; Lenz et al. 2003; MacArthur and Levins 1967; Pyšek et al. 2012). A decrease in the richness of native species and reduced ecosystem functionality in the presence of invasive species is evident even at smaller spatial scales (Bernard-Verdier and Hulme 2019). Decrease in species richness has also been previously described in rangelands experiencing B. tectorum invasion (Clark 2020; Clark et al. 2019, 2020). Our results support this premise that controlling the invasive annual grass B. tectorum can have beneficial impacts on the rangelands by improving the richness and abundance of native flora in the region. It is to be noted that the results of the case study we present is from one flowering season immediately following the winter application of the herbicide.
Ongoing studies on biological invasions and their control suggest that the long-term impact of invasive species removal on native species richness needs further investigation. The benefits of increased species richness and diversity observed soon after control may be modest and may not be long lasting (Adams et al. 2020; Kettenring and Adams 2011). In regions experiencing long-term establishment of invasive plant species, it is likely that the diversity of native species in the ecosystem has been compromised (Duncan et al. 2004), though communities with native annual forbs can be impacted (Meyer-Morey et al. 2021). Our study shows that flowering species reappearing in the year following indaziflam application include annuals, biennials, and perennials (Table 2), many of which are native to the region, agreeing with the earlier report that indaziflam application for B. tectorum control does not appear to negatively impact native species richness in the natural areas and rangelands of Colorado (Clark et al. 2019). It has been suggested that implementing control measures when there is still remaining native vegetation may yield better success in restoration of native species (Davies and Sheley 2011). Our case study shows reduced flowering plant species diversity in control plots (Figure 2), reiterating the possibility that controlling B. tectorum populations in these rangelands could improve native flowering plant populations. In addition, planning restorative actions needed for assisted reestablishment of native forbs in combination with the application of herbicide for invasive grass control may further promote flowering plant reestablishment.
Table 1.
Multivariate general linear model showing the effect of treatment on diversity measures calculated from belt transect data.
As mentioned earlier, one limitation of our case study is that we focused on diversity of flowering plants and did not determine the abundance of B. tectorum in the control and treated plots. While there are few studies that explore the direct impacts of invasion by nonnative plant species on pollinators, it is evident that abundance of native flowering species is reduced when ecosystems are dominated by invasive species (Bernard-Verdier and Hulme 2019). Thus, there is a high likelihood that plants that support the nutritional and nesting needs of pollinators (Blüthgen and Klein 2011; Giannini et al. 2015; Soliveres et al. 2016; Tscharntke et al. 2012) are reduced in such invaded areas. Although this report is a single case study from three locations in the rangelands of Colorado, the immediate benefits of controlling the invasive annual grass B. tectorum are compelling. An earlier study conducted in the same geographic region suggests very little if any residual effects of the herbicide indaziflam (Clark et al. 2019). However, the nesting biology of the pollinators previously reported in this rangeland ecosystem (Goldstein and Scott 2015; Kearns and Oliveras 2009a; Scott et al. 2011) indicates that many of the bee species are ground nesting, wherein the female bees tunnel into the soil, lay eggs, and provision the larvae with pollen that is consumed over the larval developmental period (Buchmann and Nabhan 1996; Michener 1974). It would be critical to determine the extent of herbicide residue in the soil and its potential to impact the development of ground-nesting bee larvae (Buckles and Harmon-Threatt 2019; Harmon-Threatt 2020). Continued monitoring of these locations will help strengthen data on the diversity of native plant species as invasive grasses continue to be controlled. This would also provide critical information on the long-term effectiveness of herbicide use and invasive species control on ecosystem functions.
Figure 2.
Average (±SD) of floral diversity measures from belt transects. Statistical comparison is across treatments within each location. Different letters indicate significant differences at P < 0.001 using a post hoc Bonferroni comparison (Table 1).
Figure 3.
Shannon diversity index values from belt transects for treated and untreated control plots across the season in the different geographic locations.
Acknowledgments.
The study design was based on input from Harry Quicke and Steve Sauer. JH and AS would like to thank Nicholas DiMascio and Jim Sebastian for help in the field. The authors received funding from Boulder County Parks and Open Spaces Small Grants and a Bayer vegetation management grant. The authors express their gratitude to two anonymous reviewers and the subject editor whose suggestions greatly improved the quality of the article. The statements made in the article represent authors' views and should not be interpreted as endorsement from their respective employers or the funding agencies. Mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture (USDA). USDA is an equal opportunity provider and employer.
