Indaziflam controls nonnative Alyssum spp. but negatively affects native forbs in sagebrush steppe

Abstract Nonnative plant invasions can have devastating effects on native plant communities; conversely, management efforts can have nontarget and deleterious impacts on desirable plants. In the arid sagebrush steppe rangelands of the western United States, nonnative winter annual species affect forage production and biodiversity. One method proposed to control these species is to suppress the soil seedbank using the preemergent herbicide indaziflam. Our goal was to evaluate the efficacy of indaziflam to control nonnative annual mustards (Alyssum spp.) and to understand potential nontarget effects of management on the diverse mountain sagebrush steppe plant communities within Yellowstone National Park. Six sites were established along an elevation gradient (1,615 to 2,437 m), each with high and low Alyssum spp. infestations. We applied 63g ai ha−1 of indaziflam in late summer of 2018 and evaluated plant community cover in situ for 2 yr after treatment and emergence of forb species from the soil seedbank ex situ. Indaziflam was highly effective at controlling emergence of Alyssum spp. for 2 yr. Richness and Shannon’s diversity of the nontarget plant community were significantly lower in sprayed plots than in the control, and both decreased along the elevation gradient. These reductions were due to a decrease in perennial forbs and native annual forbs in the sprayed plots; perennial graminoids were not affected. Overall, the aboveground and seedbank community composition was negatively impacted by indaziflam, and these effects were strongest for the native annual forbs that rely on annual regeneration from the seedbank. The effects of this herbicide to the nontarget community should be evaluated beyond the length of our study time; however, we conclude that indaziflam should likely be reserved for use in areas that are severely invaded and have seedbanks that are composed of nondesirable species rather than diverse, native mountain sagebrush communities.


Introduction
Nonnative plant invasions can have devastating impacts on native plant communities (Elton 1958;Mack and D'Antonio 1998;Tilman 1999;Vilà et al. 2011). However, chemical efforts to control nonnative plants can also damage co-occurring native species and communities (Crone et al. 2009;Kettenring and Adams 2011;Rinella et al. 2009;Wagner and Nelson 2014), sometimes causing more harm than the invader itself (Ortega and Pearson 2011;Skurski et al. 2013). Understanding the efficacy of herbicides on target species and potential adverse impacts on nontarget desirable species is key to developing effective management strategies.
Controlling invasive plants in semiarid grasslands is critical to maintaining the essential functions and services this ecosystem provides to humans around the world (Lund 2007;O'Mara 2012). The sagebrush steppe in the western United States is a diverse ecological community that provides forage for livestock operations and habitat for wildlife species (Beck et al. 2012). This semiarid ecosystem is threatened by land use change, climate change, and nonnative plant invasions (DiTomaso 2000;Knapp 1996;Vasquez et al. 2010), particularly winter annual grass species that can alter fire regimes and disrupt ecosystem functions (Balch et al. 2013;Billings 1994;Young and Fay 1997), creating novel plant communities far less diverse than the native communities they replaced (Allen and Knight 1984).
The preemergent herbicide indaziflam is promoted as an option to control invasive annual grasses in sagebrush steppe Sebastian et al. 2016). Indaziflam is a nonselective herbicide that inhibits cellulose biosynthesis (Brabham et al. 2014) and provides residual control in the top few centimeters of soil (0 to 2.5 cm). Originally developed for turf and orchard use, indaziflam was recently approved for use in natural areas and grazed rangeland (Bayer 2020) and may be used to provide long-term control of invasive annual grasses by depleting the soil seedbank (Sebastian et al. 2017a(Sebastian et al. , 2017b. Indaziflam shows great promise to reduce target species, as this herbicide remains active in the soil for up to 3 yr (Sebastian et al. 2016) and would potentially require fewer applications than other less persistent herbicides (Sebastian et al. 2017a).
The efficacy of indaziflam in sagebrush rangeland has been evaluated primarily in highly disturbed areas dominated by invasive annual grasses with little remaining native vegetation (Sebastian et al. 2017a). The control of invasive annual grasses results in an increase in growth of existing perennial plants due to reduced competition (Sebastian et al. 2017a. Established perennial vegetation is largely unaffected by indaziflam (Clark et al. 2019), likely because roots of perennial plants often extend below the zone of herbicide activity. However, new recruitment of a nonnative perennial species was inhibited by indaziflam (Sebastian et al. 2017c), and it is likely that any germinating seeds in the zone of herbicide activity will not emerge, as indaziflam inhibits cellulose biosynthesis in the radicle and is nonselective (Brabham et al. 2014). Species with annual life cycles and short-lived seedbanks will be more impacted by indaziflam than either perennial species that do not rely on annual germination or species with long-term seedbanks. Only a couple of indaziflam studies have examined impacts to annual forbs (Clark et al. 2019;Sebastian et al. 2020), and one of those found a slight increase in native annual cover after indaziflam treatment .
The forb desert alyssum or desert madwort (Alyssum desertorum Stapf) is a nonnative winter annual species common to the sagebrush steppe (Noack 2020) and is purported to displace native vegetation (Mosley 2014), though quantitative evidence of its impacts is lacking. Two other nonnative annual mustards [pale madwort (Alyssum alyssoides (L.) L.) and alyssum (Alyssum simplex Rudolphi)], can co-occur but receive less attention. In Yellowstone National Park (YNP), A. desertorum has recently been found above its documented elevational range, primarily in areas disturbed by wildlife, tourists walking off designated routes, or recent construction (H Anderson, personal communication). This spread has prompted YNP land managers to consider controlling localized infestations of A. desertorum to maintain the native diversity of these areas. Managers are controlling infestations of A. desertorum with indaziflam in one area of YNP that has experienced over a century of intense human land use. Using this herbicide to control Alyssum spp. in less disturbed and more species-diverse areas of YNP needs to be evaluated. Previous indaziflam studies in other semiarid grasslands have primarily focused on the response of perennial forbs and perennial grasses, as they are desirable components of rangeland plant communities; however, annual ephemeral forbs are also key components of rangelands (Pokorny et al. 2004). They provide critical spring forage for many wildlife species (Drut et al. 1994;Luna et al. 2018) and occupy an early successional niche. We sought to assess the impacts of indaziflam on the native plant community with a primary focus on annual forbs.
Our study examined the efficacy of preemergent indaziflam to control nonnative Alyssum spp. and evaluated impacts to the diversity of the nontarget plant community. The objectives were: (1) to evaluate the efficacy of indaziflam to control the target species; (2) to assess the effect of indaziflam on the richness and diversity of the whole plant community, distinguishing between perennial and annual species; and (3) to evaluate the effect of indaziflam on perennial and annual forb germination from the soil seedbank.

Study Area and Experimental Design
Six field sites were established along an elevational gradient (1,615 m to 2,347 m) in the northern range of YNP, USA (Table 1). Sites were randomly selected from a roadside survey of A. desertorum populations conducted with aid of the park botanist (H Anderson). However, further evaluation found a mix of A. desertorum with A. alyssoides and A. simplex at these sites (Alyssum spp. hereafter). The plant communities at all sites were characterized as mountain big sagebrush steppe (dominated by Artemisia tridentata Nutt. ssp. vaseyana).
We defined high and low invasion levels of Alyssum spp. at each site, with high levels equal to or above 10 individuals m −2 , and adjacent low levels with fewer than 10 individuals m −2 . Before spray application we assessed Alyssum spp. in each plot, using density classes (1: 0; 2: <10; 3: 10 to 100; and 4: >100 individuals m −2 ). High-invasion plots were mainly in the high (4) category (75%, 78 plots) with the rest in the medium (3) level (25%, 26 plots [104 plots total]). Most of the low-invasion plots had no Alyssum spp. (94%, 96 of 102 plots [2 plots lost to the study]); only 6 had between 1 and 10 Alyssum spp. At five of the sites, eighteen 1-m 2 plots were established in each of the high-and low-invasion areas. In each of the invasion treatments, 12 plots were sprayed with indaziflam (Esplanade®, Bayer CropScience, Cary, NC, USA, 27513) and six plots were left unsprayed to serve as controls. At the sixth and highest site, fourteen 1-m 2 plots were established in each high-and low-invasion area, with seven

Management Implications
The sagebrush steppe rangelands of the American West are threatened by infestations of annual nonnative plant species, and an effective approach to controlling these species is through seedbank depletion. Indaziflam is a preemergent herbicide that can effectively deplete the soil seedbank of invasive annual grasses. This herbicide shows great promise for control of these annual invaders due to its residual soil activity of up to 3 yr. There have been no documented negative effects to established perennial vegetation; however, nontarget impacts to the native seedbank are expected. A field study was conducted to examine the impacts of indaziflam on nontarget species aboveground, and a greenhouse study addressed impacts to the soil seedbank. The mountain big sagebrush (Artemisia tridentata ssp. vaseyana) sites had localized infestations of nonnative annual mustards (Alyssum spp.) and high levels of species richness and diversity. Overall indaziflam controlled Alyssum spp. for 2 yr but also reduced richness and diversity of the nontarget forb community. Annual forbs were the most impacted, both abovegroundwith reductions of up to 50% cover-and in the soil. Perennial forbs were also prevented from regenerating from the soil seedbank. Native annual forb species are of particular concern, as they rely on annual regeneration from the seedbank, whereas established perennial species can repopulate after the herbicide has degraded. The soil seedbank is often relied upon for passive restoration efforts, and broadscale use of indaziflam may not be appropriate in areas with minimal infestations. Our findings highlight the importance of considering nontarget impacts when using indaziflam in areas with diverse native plant communities and low infestations of annual nonnative species. of the plots in each invasion treatment sprayed with indaziflam and seven plots left unsprayed as controls. We had a high level of replication and chose our plot size to ensure we had a high detection rate for small ephemeral annual species, allowing adequate representation of the plant community. Plots were randomly assigned to the spray or control treatment. All spray plots were treated in August 2018 with 63 g ai ha −1 indaziflam using a backpack sprayer fitted with a XR11002 flat spray nozzle (TeeJet® Spraying Systems, P.O. Box 7900, Wheaton, IL 60187) at 195 L ha −1 at 138 kPa. Ocular estimates of foliar cover (to the nearest 1%) for each species and ground cover were conducted 1 and 2 yr after treatment during peak vegetative season, for the central 0.75 m 2 of each plot to account for potential edge effects of the herbicide. Species with "trace" (<0.5%) cover were recorded and analyzed as 0.1% cover. Estimates were allowed to exceed 100% to account for understory canopy structure. The sites were sampled from low to high elevation with the aim of sampling each site at comparable growth phases. Half of the spray plots were sampled 1 mo after the other plots at each site; this enabled us to us to determine whether we had missed any species and provided a better understanding of the plant community (Pokorny et al. 2004).

Evaluating the Effect of Herbicide on the Soil Seedbank
To assess the effect of herbicide on the seedbank, soil samples were collected at the two lowest sites in April 2019. At each site, soil cores were taken in each of six spray plots and six control plots in both areas of invasion treatments (n = 48). Six soil cores (10cm diameter by 6-cm deep) were taken per plot in the area outside the vegetation quadrat but within the 1-m 2 plot and combined into one soil sample. Soil samples (2,832 cm 3 ) were stored in a cold storeroom at 5 C for 2 wk, then spread out in trays (28 by 13 cm) on top of 2.5 cm of sterilized soil (1:1:1 ratio of mineral soil, sphagnum peat moss, and washed concrete sand) in the Montana State University Plant Growth Center (Bozeman, MT, 59717). Trays were placed in a greenhouse (22 to 18 C, 16-h photoperiod) and watered twice each day for 5 min using a drip irrigation system. Seedling emergence above the soil surface was recorded at least once per week. Seedlings that emerged were identified, counted, and then removed from the tray. Seedlings that were unidentifiable at the 1-to 2-leaf stage were repotted and grown until they were identifiable. Emerging graminoids were removed and not counted.
There was no germination for the first 45 d; therefore, trays were moved to a cold, wet, and dark stratification chamber (4 C) for 30 d to break dormancy, after which the trays were returned to the greenhouse. After germination diminished again (6 mo) the trays were returned to the stratification chamber for 56 d, and then returned to the greenhouse. Any additional germinants were identified and removed. The experiment was terminated after 17 mo.
To evaluate the effects of indaziflam on the abundance of Alyssum spp. (objective 1), the high invasion data were subset, and a linear mixed model following a Gaussian distribution was fit. The saturated model included herbicide treatment (spray, control), elevation, and year (2019, 2020) as main effects. To account for repeated measures of plots between years, the unique tag number for each plot was included as a random effect. For objective 2, indaziflam effects on the plant community, the full data set was used, including both levels of Alyssum spp. invasion. Species richness and Shannon's diversity index were assessed for the whole community, perennial graminoids, perennial forbs, and annual native forbs. Percent cover estimates were used to calculate Shannon's diversity index for each plot. Shannon's diversity index was calculated as where p i is the proportion of species i, S is the number of species, and logb is the logarithm base 10. Species richness was evaluated with generalized linear mixed-effects models following a Poisson distribution, and Shannon's diversity was evaluated following a Gaussian distribution. A quasi-Poisson distribution was used when overdispersion was present. Saturated models with all possible interactions were fit with main and random effects as noted earlier and including invasion (high, low) as a main effect. Any significant interactions were maintained in the final models. Normality and heteroscedasticity assumptions were assessed, and no transformations were necessary. The model intercept was the lowest elevation site, control, and low invasion. A type II ANOVA was used to evaluate treatment effects. For objective 3, to evaluate impacts to perennial and annual forbs in the soil seedbank, the abundance, richness, and Shannon's diversity were analyzed with linear models similar to those previously described. Abundance and Shannon's diversity responses were fit following a Gaussian distribution, and the richness responses were fit with a quasi-Poisson distribution to account for overdispersion; treatment effects were evaluated with a type II ANOVA.

Results and Discussion
Indaziflam Efficacy on Alyssum spp.
Indaziflam provided excellent control of Alyssum spp. 1 and 2 yr after treatment (P < 0.01). There was no difference by elevation (P = 0.47); however, there was a difference by year (P < 0.01) and a significant interaction between year and indaziflam (P < 0.01) ( Table 2). This is likely explained by natural variation of Alyssum spp. abundance in the control plots (Figure 1). The mean (±SE) cover of Alyssum spp. in spray plots was 0.1% (±0.04%) in 2019 and 0.2% (±0.1%) in 2020; in control plots, mean cover was 3.7% (±0.43%) in 2019 and 2.5% (±0.32%) in 2020. This reduction in sprayed plots equates to mean Alyssum spp. cover of 97% less in 2019 and 91% less in 2020 compared with control plots.

Impacts of Indaziflam on the Whole Plant Community
There were 160 species across all sites: 100 perennial forbs, 32 perennial graminoids, 1 annual graminoid, 16 annual forbs, and 11 shrubs (Supplementary Table S1). There were some co-occurring nonnative species: 4 perennial graminoids, 1 annual graminoid, 7 perennial forbs, and 2 annual forbs (Supplementary Table S1). Mean (±SE) cover of nonnative species across all elevations was 3.9% (±0.86), which included the targeted Alyssum spp. populations. Total species richness at each site ranged from 38 to 76 species (Table 1). The mean species richness of the whole plant community was lower in sprayed plots (P < 0.01) and high Alyssum spp. invasion plots (P = 0.05), decreased as elevation increased (P < 0.01) but did not differ by year (P = 0.26) (Table 3; Figure 2). Sprayed plots also had lower Shannon's diversity than control plots (P < 0.01), and diversity decreased as elevation increased (P < 0.01) but did not differ between levels of invasion (P = 0.39) or by year (P = 0.18) ( Table 4). Species richness has shown a decrease or hump-shaped response along elevation gradients in mountain habitats globally (Haider et al. 2018;Pauchard et al. 2009).
The significant difference in total species richness between high-and low-invasion treatments was generally fewer than two species and mainly related to the differences in richness between invasion treatments at the two highest elevations (Figure 2). To evaluate the impacts of an invader on richness and diversity, the values for the invader should be removed to avoid artificially inflating the response (Thomsen et al. 2016). This additional analysis was performed and found similar patterns (data not shown). However, when annual forb richness and diversity were evaluated, neither differed between the invasion treatments (Meyer-Morey 2021). These results suggest that Alyssum spp. is a weak invader in the mountain sagebrush plant community. However, further evaluation of Alyssum spp. impact is necessary to determine whether and which populations to prioritize for management (Rew et al. 2007).
Our findings that richness and diversity decreased with herbicide application is in contrast with other rangeland studies that found increased richness and abundance 1 yr after treatment  and no effects on species richness after 2 yr (Clark et al. 2019) for the whole community. These studies were conducted in areas with dense infestations of downy brome (Bromus tectorum L.) (between 30% and 70%) and another four to five nonnative species, and the increased abundance and richness was attributed to release from competition. Our study sites were in relatively undisturbed mountain big sagebrush plant communities that are more resilient to disturbance and resistant to annual species invasion than lower-elevation sagebrush communities (Chambers et al. 2014).
Richness was also higher at our sites than in the study sites of Clark et al. (2019; 33 to 35 native species), and we only observed B.  tectorum at one site, where its cover was very low (0.5%). To better understand the differences we observed, richness and diversity of perennial graminoids, perennial forbs, and annual forbs were analyzed separately.

Perennial Species Response
Established perennial graminoid richness and Shannon's diversity were largely unaffected by indaziflam (P = 0.96 and P = 0.50, respectively) (Tables 2 and 3, Supplementary Figure S1), as expected from previous studies (Clark et al. 2019. Perennial forb mean species richness was lower in sprayed plots (P < 0.01) and under high invasion (P = 0.02), decreased with elevation (P < 0.01), and differed between years (P = 0.01; Table 3, Supplementary Figure S1). Shannon's diversity of perennial forbs was lower in sprayed (P = 0.04) and high-invasion plots (P = 0.05), decreased with elevation (P < 0.01), and was lower in 2020 than 2019 (P = 0.01). Other studies have shown no injury to or reduction in abundance of existing perennial vegetation (Clark et al. 2019;Sebastian et al. 2017a); therefore, we theorize that reduced recruitment from the seedbank may have been the cause of perennial forb reduction, which our seedbank study results corroborate. Sebastian et al. (2017c) also concluded that indaziflam reduced recruitment of a perennial nonnative forb at their field study site.

Annual Forb Response
Annual forbs were greatly impacted by indaziflam. Mean species richness of annual native forbs was lower in sprayed plots (P < 0.01) by at least 50% (Figure 2), increased with elevation (P < 0.01), and was greater in high-invasion areas (P < 0.01) due to the presence of Alyssum spp., but did not differ between years (P = 0.14) (Table 3; Figure 2). Shannon's diversity of annual forbs was reduced by indaziflam (P < 0.01), increased significantly with elevation (P < 0.01), and did not differ by year (P = 0.08) or level of invasion (P = 0.17; Table 4). The increase with elevation was explained by blackfoot groundsmoke (Gayophytum racemosum The reduced richness and diversity of annual forbs after herbicide application is in contrast with another study that found no effect of indaziflam on annual forbs ) and suggests that individual species of annual forbs may have differing sensitivities to indaziflam. There were 12 species of annual forbs plus Alyssum spp. at our sites, six were observed in both control and spray plots: Alyssum spp., tiny trumpet (Collomia linearis Nutt.), woodland draba (Draba nemorosa L.), L. septentrionalis, slender phlox [Microsteris gracilis (Hook.) Greene], and Douglas' knotweed (Polygonum douglasii Greene). The other six were only observed in the control plots: pygmyflower rockjasmine (Androsace septentrionalis L.), maiden blue-eyed Mary (Collinsia parviflora Lindl.), tall annual willowherb (Epilobium brachycarpum C. Presl.), pinyon groundsmoke (Gayophytum ramosissimum Torr. & A. Gray), G. racemosum, dwarf purple monkeyflower (Mimulus nanus Hook. & Arn.), and Suksdorf's monkeyflower (Mimulus suksdorfii A. Gray). Because annual species rely on annual regeneration from the seedbank, long-lived residual preemergent herbicides like indaziflam may have long-term impacts on all these species, depending on the species' seed decay rates, which are poorly quantified. We did not evaluate the relative fecundity of the species that emerged in control and sprayed plots, but they did produce seeds in both. Our results suggest the species only observed in the control plots are likely to be most affected, because for these species to remain in a community after herbicide application will require recolonization via seed movement from adjacent unsprayed areas.

Impacts of Indaziflam on Soil Seedbank Recruitment
Indaziflam greatly suppressed germination of both perennial and annual forbs from the soil seedbank (Figure 3). There were 29 forb species that emerged: 18 perennials and 11 annuals (Supplementary Table S1). Mean abundance of perennial forbs in the control soils was 23 (±15) individuals and there were no perennial forbs that emerged in the sprayed soils (Figure 3). This supports the results of our field study, which showed reduced richness and diversity  of perennial forbs in sprayed plots, and the results of Sebastian et al. (2017c), which documented reduced recruitment of seven perennial nonnative forbs in their seedbank study.
The few species that emerged in the sprayed soils were all annual forbs: herb sophia [Descurainia sophia (L.) Webb ex Prantl], C. linearis, D. nemorosa, and M. gracilis. These were very low in abundance compared with the control soils; mean annual forb abundance was 1 (±0.83) individual in sprayed soils compared with 33 (±11.3) individuals in the controls ( Figure 4A). The latter three species (C. linearis, D. nemorosa, and M. gracilis) were also found in the sprayed plots in the field experiment but also occurred at greater abundance in the control plots. Annual forb richness was lower in the sprayed soils (P < 0.01) and differed between the two sites (P < 0.01) and levels of invasion (P = 0.03). Mean annual richness in the controls was 1.5 (±0.3) species at the lower site (M) and 3.6 (±0.4) species at the higher site (B) and, in sprayed soils, was reduced to 0 (±0) and 0.5 (±0.34), respectively ( Figure 4B). Shannon's diversity was also lower in sprayed soils (P < 0.01) and differed between sites (P < 0.01), though it did not differ between levels of invasion ( Table 5). The soil seedbank at our sites had a high diversity of species (29 species) despite Alyssum spp.
infestation, and these nontarget species, particularly the native annual forbs, were shown to be negatively impacted by indaziflam.
Previous indaziflam studies have primarily focused on the response of perennial forbs and perennial grasses, as they are desirable components of rangeland plant communities; however annual forbs are also key components of rangelands (Pokorny et al. 2004). They occupy an early successional niche and provide critical spring forage for many wildlife species (Drut et al. 1994;Luna et al. 2018). Additionally, native annual forbs are phenologically similar to nonnative winter annual grasses and likely use similar resource pools (Forbis 2010). The annual forbs in our study area are early-season and early seral species; early seral annual forbs can compete with nonnative winter annual grasses, specifically B. tectorum (Uselman et al. 2015). Therefore, knowing which annual species are tolerant to indaziflam is useful for creating restoration seed mixes for use after indaziflam application.
In the western United States, approximately 23 million ha of rangelands are infested with nonnative annual species (DiTomaso et al. 2017), presenting a considerable challenge to restore and preserve these working lands. Areas with extensive monocultures of nonnative plants require novel and innovative   approaches, such as seedbank depletion, to achieve adequate longterm control; however, these same methods may not be appropriate at all levels of infestation or with all nonnative annual species. In areas where the seedbank is composed mostly of nonnative species that have formed persistent soil seedbanks, indaziflam is a valuable tool to deplete the seedbank of target species before active revegetation efforts . However, in areas with minimal infestations, an existing diverse plant community, and therefore likely a diverse seedbank, the use of indaziflam and its impacts on the seedbank of nontarget species should be carefully considered before broadscale use. Additionally, to determine whether active management is warranted, the actual impacts of an invader should be considered (Rew et al. 2007). Our results suggest Alyssum spp. is a weak invader, and using indaziflam to control it may do more harm than the target species itself in high-elevation sagebrush steppe. Indaziflam significantly reduced species richness and diversity, particularly of nontarget annual forb species, resulting in a community more dominated by perennial species. Native annual forb species are often overlooked, though they occupy an important early successional niche and contribute to the biodiversity of rangelands. Maintaining biodiversity should be a priority for land managers, as communities that are higher in diversity are typically more resistant to ecosystem alteration and more resilient after disturbances (Chambers et al. 2014;Hobbs and Huenneke 1992;Standish et al. 2014). Understanding and minimizing the nontarget effects of invasive plant control on existing intact vegetation is critical when developing management strategies (Rew et al. 2007), especially in arid rangelands. Future studies should address the long-term impacts of indaziflam to plant community composition and explore the tolerances and seedbank longevities of native species for restoration seed mixes.