Introduction
Canada thistle [Cirsium arvense (L.) Scop., Asteraceae] is a pervasive perennial weed found throughout temperate regions globally. In the western United States, C. arvense ranks as one of the highest among the most commonly occurring noxious weeds, posing significant threats to both managed and natural landscapes (Bodo Slotta et al. Reference Bodo Slotta, Foley, Chao, Hufbauer and Horvath2010; Moore et al. Reference Moore1975; Nuzzo Reference Nuzzo1997). In agricultural systems, C. arvense competes for light, nutrients, and water, leading to reduced crop yield and quality, generating significant economic losses for producers (Jacobs et al. Reference Jacobs2006; Moyer et al. Reference Moyer, Schaalje and Bergen1991; O’Sullivan et al., Reference O’Sullivan, Kossatz, Weiss and Dew1982). In natural systems, it similarly competes with and displaces native plant species (Jacobs et al. Reference Jacobs2006). Cirsium arvense is commonly found in disturbed areas, including roadsides, stream banks, ditches, clear-cuts, forest openings, and wet or wet-mesic grasslands and rangelands, as part of the initial postdisturbance community (Morishita Reference Morishita, Sheley and Petroff1999; Nuzzo Reference Nuzzo1997). It is prevalent in nearly every upland herbaceous community within its range, particularly prairie communities and riparian habitats (Nuzzo Reference Nuzzo1997).
Cirsium arvense survives and spreads through two reproductive strategies: sexual reproduction via seeds and clonal vegetative growth. Seeds are small and light, with highly variable germination success (Hodgson, Reference Hodgson1964; Moore Reference Moore1975; Nuzzo Reference Nuzzo1997); however, seeds are important for range expansion, as shown by the genetic diversity of North American populations (Bodo Slotta et al. Reference Bodo Slotta, Foley, Chao, Hufbauer and Horvath2010). Once established, the plant develops a creeping root system, up to 2 to 3 m belowground, with adventitious root buds resulting in clonal vegetative growth that enables rapid propagation and spread (Donald, Reference Donald1994; Lalonde and Roitberg Reference Lalonde and Roitberg1994). The adventitious buds develop into new rosettes and lateral roots that continue to grow throughout spring, summer, and into fall. As temperatures decrease in fall, the aboveground vegetation dies off, and the roots overwinter; in the spring, root growth resumes and new shoots emerge (Lalonde and Roitberg Reference Lalonde and Roitberg1994; Tiley Reference Tiley2010). Fragments of roots as small as 1 cm are capable of regenerating and forming new colonies (Nadeau and Vanden Born Reference Nadeau and Vanden Born1989; Thomsen et al. Reference Thomsen, Mangerud, Riley and Brandsæter2015). The complex and extensive root system of C. arvense allows for propagation, spread, and recovery, making it particularly problematic and challenging for management (Nadeau and Vanden Born Reference Nadeau and Vanden Born1989; Tiley Reference Tiley2010).
Several management tactics are regularly utilized in efforts to control C. arvense. Chemical control is common and effective, generally providing rapid results. Mowing can also be an effective control method by reducing photosynthetic capacity aboveground and depleting root reserves used for regrowth, leading to a reduction of new shoots the following season (Bourdôt et al. Reference Bourdôt, Hurrell, Skipp, Monk and Saville2011; Graglia et al. Reference Graglia, Melander and Jensen2006). Similarly, cultivation and tillage fragment the root system and force the plant to use root reserves for recovery; however, this may also promote new shoot development and further spread of C. arvense (Graglia et al. Reference Graglia, Melander and Jensen2006; Thomsen et al. Reference Thomsen, Mangerud, Riley and Brandsæter2015). Both tillage and mowing are most effective when integrated into a weed management program to control established populations. While most current management of C. arvense relies on herbicides, mowing, or cultivation, these tactics can be costly, labor-intensive, and often not ideal for environmentally sensitive areas (e.g., riparian zones) (Bourdôt et al. Reference Bourdôt, Hurrell, Skipp, Monk and Saville2011; Graglia et al. Reference Graglia, Melander and Jensen2006; Peterson et al. Reference Peterson, Merrett, Fowler, Barrett and Paynter2020; Thomsen et al. Reference Thomsen, Mangerud, Riley and Brandsæter2015). In contrast, biological control tactics are often more suitable for managing weeds in natural areas, as they can be self-perpetuating, and more economical in landscape-wide suppression of target species (Cripps et al., Reference Cripps, Gassmann, Fowler, Bourdôt, McClay and Edwards2011; Guske et al. Reference Guske, Schulz and Boyle2004; Peterson et al. Reference Peterson, Merrett, Fowler, Barrett and Paynter2020; Sciegienka et al. Reference Sciegienka, Keren and Menalled2011).
Effective biological control has long been sought for C. arvense. Puccinia suaveolens (syn.: P. punctiformis), an obligate biotrophic rust fungus, was first proposed as a control method for C. arvense in 1893 in North America (Wilson Reference Wilson1969). Puccinia suaveolens can be naturally found on C. arvense plants throughout its growing region, commonly co-introduced in the invasive range and causing periodic outbreaks of disease (Berner et al. Reference Berner, Smallwood, Cavin, Lagopodi, Kashefi, Kolomiets, Pankratova, Mukhina, Cripps and Bourdôt2013; French and Lightfield Reference French and Lightfield1990). Highly host specific to C. arvense, P. suaveolens has only been reported on two other thistles native to Eurasia: Silybum marianum (L.) Gaertner (Milk thistle) in 2002 under greenhouse conditions and Cirsium setosum (Willd.) M. Bieb. (field thistle) in 2023 under field conditions in China (Berner et al. Reference Berner, Paxson, Bruckart, Luster, McMahon and Michael2002; Liang et al. Reference Liang, Wang, Wang, Zhao, Li, Li, Zhang, Meng and Yan2024). The potential for P. suaveolens to be utilized more broadly as a biocontrol for C. arvense will increase with continued research (Bean et al. Reference Bean, Gladem, Rosen, Blake, Clark, Henderson, Kaltenbach, Price, Smallwood, Berner, Young and Schaeffer2024; Berner et al. Reference Berner, Smallwood, Cavin, McMahon, Thomas, Luster, Lagopodi, Kashefi, Mukhina, Kolomiets and Pankratova2015a, Reference Berner, Smallwood, Vanrenterghem, Cavin, Michael, Shelley, Kolomiets, Pankratova, Bruckart and Mukhina2015b; Cripps et al. Reference Cripps, Bourdôt, Saville and Berner2014; Thomas et al. Reference Thomas, Tworkoski, French and Leather1994).
The life cycle of P. suaveolens can be divided in two stages: the vegetative mycelium within the root system and the spore-producing aboveground systemic and local infections. It is thought that P. suaveolens remains latent within the root system until abiotic or biotic conditions are adequate to produce spore-bearing systemically infected stems (Mendgen and Hahn Reference Mendgen and Hahn2002). Puccinia suaveolens infection reduces C. arvense belowground biomass as root resources are parasitized by mycelia and allocated to plant defense compounds instead of growth (Chichinsky et al. Reference Chichinsky, Larson, Eberly, Menalled and Seipel2023; Clark et al. Reference Clark, Jahn and Norton2020). This reduces the aboveground shoots and competitive ability of C. arvense (Chichinsky et al. Reference Chichinsky, Larson, Eberly, Menalled and Seipel2023). Systemically infected stems typically do not flower, can die off early in the season, and may help to further reduce root resources of infected C. arvense colonies (Chichinsky et al. Reference Chichinsky, Larson, Eberly, Menalled and Seipel2023; Van Den Ende et al. Reference Van Den Ende, Frantzen and Timmers1987). The rate and distance of spread of P. suaveolens caused by underground mycelia or aboveground spores remains unknown.
Puccinia suaveolens produces five spore types that develop consecutively, beginning with emergence of systemically infected shoots in early spring that are deformed, chlorotic, strongly floral scented, and covered in yellow-orange pycnia pustules (Buller Reference Buller1950; Menzies Reference Menzies1953; Petersen Reference Petersen1974). After outcrossing, pycniospores give rise to chain-like formations of the dark orange-brown aeciospores, giving the systemically infected stems a characteristic rusty appearance (Berner et al. Reference Berner, Smallwood, Cavin, Lagopodi, Kashefi, Kolomiets, Pankratova, Mukhina, Cripps and Bourdôt2013; Connick and French Reference Connick and French1991). Production of urediniospores is indicated by the darkening of the spores (Buller Reference Buller1950; Petersen Reference Petersen1974). Urediniospores and aeciospores are morphologically indistinguishable as single-celled spores, but only urediniospores produce localized infections on neighboring plants throughout summer (Kirk et al., Reference Kirk, Cannon, David and Stalpers2001; Menzies Reference Menzies1953; Petersen, Reference Petersen1974). Localized infections produce pustules on the leaves, but the shoots do not have the same growth abnormalities associated with systemic infection, and will still appear relatively normal (Baka and Lösel, Reference Baka and Lösel1992; Thomas et al. Reference Thomas, Tworkoski, French and Leather1994). Localized infections can produce two-celled teliospores on leaves that senesce heading into fall, then the spores will either blow off and overwinter in the soil or germinate on new rosettes to initiate new vegetative infection in the roots or systemic infection aboveground (Alexopoulos et al. Reference Alexopoulos, Mims and Blackwell1996; Berner et al. Reference Berner, Smallwood, Vanrenterghem, Cavin, Michael, Shelley, Kolomiets, Pankratova, Bruckart and Mukhina2015b; Menzies Reference Menzies1953). Optimal teliospore germination occurs when temperatures are between 16 and 20 C (Berner et al. Reference Berner, Smallwood, Cavin, Lagopodi, Kashefi, Kolomiets, Pankratova, Mukhina, Cripps and Bourdôt2013; French and Lightfield Reference French and Lightfield1990) with optimal dew periods of 2 to 3 h (Morin et al. Reference Morin, Brown and Auld1992a, Reference Morin, Brown and Auld1992b).
Integrated weed management (IWM) is a holistic approach implementing one or more biological, physical, cultural, or chemical control tactics (Harker and O’Donovan Reference Harker and O’Donovan2013). IWM aims to reduce weed adaptation and resistance to any single control tactic by using several possible tactics that take into account threshold populations, critical periods, and environmental outcomes. Utilizing IWM may lead to reduced environmental impacts of any given control method, decreased control costs by reducing pests to economically and ecologically insignificant levels, increased sustainability, and reduced herbicide resistance (Harker and O’Donovon Reference Harker and O’Donovan2013). In a meta-analysis, Davis et al. (Reference Davis, Mangold, Menalled, Orloff, Miller and Lehnhoff2018) found that combined treatments had better long-term outcomes for control of C. arvense than reliance on herbicide treatment alone. Mowing and tillage have been shown to affect C. arvense populations, but results have varied between significant reductions in population size to virtually no impact at all (Beck and Sebastian Reference Beck and Sebastian2000; Bourdôt et al. Reference Bourdôt, Hurrell, Skipp, Monk and Saville2011). A stem-mining weevil, Hadroplontus litura, and a bacterial plant pathogen, Pseudomonas syringae pv. tagetis, have been shown to have an additive effect in suppressing C. arvense treated with herbicide (Sciegienka et al., Reference Sciegienka, Keren and Menalled2011). Puccinia suaveolens has also previously been used in conjunction with other control methods. Mowing combined with P. suaveolens strongly reduced the proportion of fertile flower heads of C. arvense compared with infection alone (Kluth et al. Reference Kluth, Kruess and Tscharntke2003). Demers et al. (Reference Demers, Berner and Backman2006) found that systemically infected P. suaveolens shoots increased, while healthy shoots decreased when combined with mowing. In a greenhouse experiment with a crop sequence of spring wheat (Triticum aestivum L.), forage pea (Pisum sativum L.), and safflower (Carthamus tinctorius L.), crop competition reduced C. arvense aboveground biomass, but the effect was increased by approximately 10% when C. arvense was inoculated with P. suaveolens (Chichinsky et al. Reference Chichinsky, Larson, Eberly, Menalled and Seipel2023). Variable levels of control have also been observed across different environments, likely based on a combination of genetic and local conditions. A recent study by Bean et al. (Reference Bean, Gladem, Rosen, Blake, Clark, Henderson, Kaltenbach, Price, Smallwood, Berner, Young and Schaeffer2024) saw stem densities decline at 77% of treated sites. The study also found that the pathogen’s effect was greater with increased inoculum, frequency of treatment, and broadcast application of spores (Bean et al. Reference Bean, Gladem, Rosen, Blake, Clark, Henderson, Kaltenbach, Price, Smallwood, Berner, Young and Schaeffer2024). The potential of using P. suaveolens in an IWM approach for C. arvense is supported, but more research is needed to develop and refine best management practices.
Cirsium arvense is a problematic weed that is difficult to control, and current methods have varying degrees of success in both managed and natural ecosystems. As a biological control for C. arvense, P. suaveolens has significant potential, as it can self-perpetuate, spread to surrounding areas, and contribute to population suppression at large scales when applied alone (Bean et al. Reference Bean, Gladem, Rosen, Blake, Clark, Henderson, Kaltenbach, Price, Smallwood, Berner, Young and Schaeffer2024). Our objectives were to determine the efficacy and compatibility of different control methods (mowing, tillage, herbicide, and P. suaveolens) when applied alone and in combination to suppress C. arvense. We highlight the benefits and limitations of using P. suaveolens in an IWM program, along with considerations for improved application efficacy.
Materials and Methods
Study Sites
Two experimental sites were established in 2020: one in the Tamarack Ranch State Wildlife Area of Colorado (CO; 40.8320°N, 102.80437°W) and the other in Park City, Utah (UT; 40.674330°N, 111.491324°W). The CO site is within the High Plains ecoregion, while the UT site is within the Wasatch and Uinta Range ecoregion (Omernik and Griffith Reference Omernick and Griffith2014). The CO site is a 12 by 600 m plot of land, characterized as a shortgrass prairie with seasonal water inundation. Historically, the site had been maintained as a food crop plot for wildlife. At the end of each growing season, glyphosate was applied, and the plot was tilled for several years, resulting in the formation of a near-monoculture of C. arvense (L Kokes, personal communication, May 2020). The CO site typically has precipitation occurring throughout the year (Supplementary Table S1). The UT site is a small preserve nestled within a suburban development. Historically, herbicides were applied, particularly for musk thistle (Carduus nutans L.), and goat/sheep grazing was occasionally employed at the UT site but had not occurred for many years; the area is largely left untouched (SJ Dickens, personal communication, 2020). The UT site generally has most of its precipitation occurring in the winter months (Supplementary Table S1) (PRISM Climate Group Reference Climate Group2023).
Puccinia suaveolens Inoculum
Dried inoculum was prepared following Berner et al. (Reference Berner, Smallwood, Cavin, Lagopodi, Kashefi, Kolomiets, Pankratova, Mukhina, Cripps and Bourdôt2013) and Bean et al. (Reference Bean, Gladem, Rosen, Blake, Clark, Henderson, Kaltenbach, Price, Smallwood, Berner, Young and Schaeffer2024). Briefly, C. arvense leaves bearing telia (small pustules on yellowing leaves) were collected in late summer from a site near Colorado Springs, CO. Leaves were harvested and stored in paper bags to allow foliage to dry at room temperature. Dried leaves were ground to a coarse powder in a kitchen blender and used as inoculum within the season or stored at −80 C for future use. Samples of ground leaf preparations were viewed under a microscope to ensure most of the spores were two-celled teliospores, which are necessary for initiating systemic infection (Berner et al. Reference Berner, Smallwood, Cavin, Lagopodi, Kashefi, Kolomiets, Pankratova, Mukhina, Cripps and Bourdôt2013; French and Lightfield Reference French and Lightfield1990; Van Den Ende et al. Reference Van Den Ende, Frantzen and Timmers1987).
Experimental Design
In both UT and CO, an experimental site was established using a randomized complete block design, consisting of 10 treatment combinations applied across replicates. Treatments included an untreated control; P. suaveolens inoculation alone; tillage; tillage plus P. suaveolens inoculation; mowing; mowing plus P. suaveolens inoculation; herbicide (aminopyralid and chlorsulfuron tank mix); herbicide plus P. suaveolens inoculation; herbicide, mowing, and tillage (HMT); and HMT plus P. suaveolens inoculation (Table 1). Each treatment was applied once in 2020 and 2021 to field plots (CO: 2 by 6 m; UT: 2 by 5 m). Plots were spaced (CO: 4 m; UT: 2 m) apart to avoid edge effects, with 8 replicates in CO and 4 in UT. Differences in experimental setup between sites were due to the size and accessibility of C. arvense populations.
Table 1. Overview of weed management tactics employed for treatment of Cirsium arvense at experimental sites in Colorado and Utah.
Herbicide and mowing treatments were applied in the fall during the first week of September. A herbicide tank mix was applied (aminopyralid at 122.5 g a.e./ha−1; chlorsulfuron at 52.5 g a.i./ha−1) using a backpack sprayer calibrated in the field. Aminopyralid was chosen specifically, as it is more effective at lower rates compared with other herbicides (e.g., picloram and clopyralid) and may also be used in areas where other chemicals are not appropriate or recommended (Enloe et al., Reference Enloe, Lym, Wilson, Westra, Nissen, Beck, Moechnig, Peterson, Masters and Halstvedt2007). In herbicide and mowing combination treatments, mowing was applied first to provide an opportunity for more even herbicide application and uptake given the physical damage to C. arvense (Carpinelli Reference Carpinelli2004). Fourteen days after initial treatments with mowing and herbicide, tillage (30-cm depth) and P. suaveolens inoculum (CO: 40 g; UT: 33.3 g) were applied to select plots. For P. suaveolens inoculum application, the entire plot was first sprayed with water using a backpack sprayer to create a mist on C. arvense leaves, then P. suaveolens was broadcast by hand no higher than 1 m above the ground to avoid excessive dispersal by wind. The 14-d period allowed the herbicide to translocate through the roots and other tissues before tillage, following the manufacturer’s (Corteva Agriscience, Indianapolis, Indiana, USA, 2020) guidelines. Puccinia suaveolens inoculum was applied last either alone or in combination (Table 1). The later timing for inoculum application in IWM treatments allowed for new growth and rosettes of C. arvense in response to mowing and possibly tillage, which may improve the chance for infection (Demers et al. Reference Demers, Berner and Backman2006). Applications of P. suaveolens inoculum before mowing, tillage, or herbicide spray would have been detrimental to germinating teliospores, which might have begun developing mycelia in the live tissue and subsequently been destroyed (Berner et al. Reference Berner, Smallwood, Cavin, Lagopodi, Kashefi, Kolomiets, Pankratova, Mukhina, Cripps and Bourdôt2013; Petersen Reference Petersen1974).
The initial monitoring of plots at both sites occurred in fall before first treatments. Monitoring occurred 2 wk before the optimal timing for P. suaveolens teliospore inoculum application at each respective site. At both sites, a 1-m2 quadrat was placed in the center of each plot, and the number of C. arvense stems was counted and percent ground cover of C. arvense, grass, forbs, litter, and bare ground was estimated visually. Across the entire plot, a 2-min timed count of C. arvense stems systemically infected with P. suaveolens was also performed.
Statistical Analyses
All analyses were performed with R (R Core Team Reference Core Team2023), using packages tidyverse, ggplot2, glmmTMB, DHARMa, emmeans, and car (Brooks et al. Reference Brooks, Kristensen, Benthem, van Magnusson, Berg, Nielsen, Skaug, Maechler and Bolker2017; Fox and Weisberg Reference Fox and Weisberg2019; Hartig Reference Hartig2022; Lenth Reference Lenth2024; Wickham et al. Reference Wickham, Averick, Bryan, Chang, McGowan, François, Grolemund, Hayes, Henry, Hester, Kuhn, Pedersen, Miller, Bache and Müller2019). Data from the two sites were analyzed separately because of the imbalance in design resulting from the difference in site accessibility and size of C. arvense populations. Stem count density change as a function of P. suaveolens inoculum application (Yes or No); management approach [Control, Herbicide (H), Mowing (M), Tillage (T), and HMT]; or year (Fall 2020, Fall 2021, and Fall 2022); and the interaction between combinations of these parameters were analyzed with a generalized linear model (GLM). Stem density was also modeled (with negative binomial) using GLM. Significance was tested using ANOVA type II Wald chi-square tests, followed by post hoc pairwise Tukey test. Finally, ground cover data were analyzed using a GLM (with beta distribution) testing significance with ANOVA type II Wald chi-square tests, followed by post hoc pairwise Tukey test. For all CO analyses, block 8 data were removed, as a Tukey’s fence method test determined that the herbicide and HMT plots were significant outliers. Block 8 lies in a section of the field that experiences significant seasonal water inundation, and herbicide effects were likely diluted in their effects.
Results and Discussion
In this study, conducted in two regions of the western United States, we evaluated the efficacy of P. suaveolens and its compatibility with other control methods for managing C. arvense by measuring stem density and vegetative cover. While stem density reflects the direct effects on the target weed, vegetative cover can represent biodiversity, forage availability, resiliency of the landscape, and nutrient cycling and may be used to predict production costs for livestock producers or fire risk. Consideration of both stem density and resulting vegetative cover will help land managers to make informed decisions about which treatments work in their IWM plan and how P. suaveolens can be incorporated.
Site Conditions and Stem Density
At the outset of the study, initial average stem density in CO was nearly four times that observed in UT (Figure 1). These differences may be attributed to prior management practices employed and variation in climate conditions experienced at each site (PRISM Climate Group Reference Climate Group2023; Supplementary Table S1). The UT site has higher average annual precipitation compared with the CO site, although they were about equal during the first year of treatments (Supplementary Table S1). At the UT site, most precipitation occurs as winter snowfall, resulting in extended dry periods that can stress C. arvense, reducing root and stem growth, and subsequently, stem density and coverage (Tiley Reference Tiley2010). In contrast, Tamarack Ranch, CO, receives more evenly distributed precipitation throughout the year. At the UT site, temperature ranged from −11.1 C during the coldest months to 29 C in the hottest months of the experiment period (2020 to 2022). At the CO site, the temperature ranged from −13.5 C during the coldest months to 33 C (2020 to 2022) during the hottest months.
Figure 1. Cirsium arvense stem count (m-²) in Fall 2020–2022 in (A) Colorado and (B) Utah following treatment with individual and combined weed management approaches.
Herbicide and HMT
Herbicide treatments, whether alone or in combination, were most effective in decreasing C. arvense stem density at both sites, (UT: P < 0.001; CO: P < 0.001). There was an immediate decline in stem density that continued even after the second application, with sparse regrowth observed (Figure 1). In CO herbicide-only treatments, stem density decreased 95% in year 1 and another 62% in year 2. When P. suaveolens was applied along with herbicide, stem density declined 91% in year 1 and 100% the following year. The UT site experienced similar decreases in stem density caused by herbicide: year 1 by 97% and year 2 by 100%. Herbicide plus P. suaveolens reduced stem density in year 1 by 98% and year 2 by 100% (Table 2).
Table 2. Annual average Cirsium arvense stem count change (%) and average stem count (m-²) with ± SE in Colorado and Utah from 2020 to 2022 following combined and individual treatments.
a MHT, mowing+herbicide+tillage.
In CO, the combined treatment (HMT) without P. suaveolens reduced stem density by 99% in year 1 and 85% in year 2. When P. suaveolens was applied along with the HMT treatment, stem density decreased 85% in year 1 and 95% in year 2 (Table 2). At the UT site, stem density in the HMT plots without P. suaveolens decreased 69% in year 1 and another 95% in year 2. When P. suaveolens was applied along with the HMT treatment, stem density decreased 84% in year 1 and 100% in year 2 (Table 2). Aminopyralid, which is selective against broadleaf weeds in rangelands and pastures, provided near 100% control of C. arvense in herbicide-treated plots with additive effects from P. suaveolens inoculum application. Limited C. arvense regeneration was observed, likely emanating from neighboring plants in buffer zones between plots, seeds, or remaining root fragments.
Puccinia suaveolens
Puccinia suaveolens was present at both sites in plots after treatments; however, there were only a few symptomatic stems. In CO, no symptomatic plants were found during the fall monitoring. In UT, the symptomatic shoots were found in the tillage plus P. suaveolens treatment, first appearing in year 1 (1 shoot) and also in year 2 (4 shoots). There was an overall lack of statistical significance (Table 3) of P. suaveolens impact at both sites, but a general declining trend in stem density indicating that the P. suaveolens had a slight suppressing effect on C. arvense (Figure 1; Table 2).
Table 3. Statistical results on the impact of the rust pathogen (Puccinia suaveolens), management practice, and their combination across seasons on Cirsium arvense stem count in Colorado and Utah.
In UT, P. suaveolens treatments alone appeared to slow the increase of stem density by 48% after 2 yr when compared with the untreated control, which had a stem density increase of 98% (Table 2). In CO, P. suaveolens treatment, reduced stem density 22% more than the untreated control (Table 2). Bean et al. (Reference Bean, Gladem, Rosen, Blake, Clark, Henderson, Kaltenbach, Price, Smallwood, Berner, Young and Schaeffer2024), also observed C. arvense stem density decrease after P. suaveolens application at 77% of treated sites in Colorado over 3 to 8 yr, stem density went from 87.9 ± 6.5 stems to 44.7 ± 4.2 stems on average. Sites with more frequent and higher quantities of P. suaveolens inoculum applied had a lower stem density over time. We suspect that C. arvense stem densities within P. suaveolens treated plots will continue to decrease with or without additional inoculations.
While stem decline was observed, the lack of symptomatic C. arvense stems could potentially be attributed to genotypic differences and associated resistance within C. arvense or the compatibility of the host–pathogen interaction. Alternatively, the abiotic or biotic factors that induce production of systemically infected stems may not have been met, though vegetative mycelium within the root system could still be present. Puccinia suaveolens may continue to have an impact on C. arvense or additional inoculation treatments might be required. This could be the case at both sites, but particularly at the CO site, where stem density decline was more obvious in Fall 2022 after a second inoculation (Figure 1).
Mowing
There was no significance between mowing plots in CO; however, a trend with mowing plus P. suaveolens showed a greater decrease in stem density (66%) compared with mowing alone (56%). In UT, mowing and mowing plus P. suaveolens resulted in small decreases in stem density of 1% and 13% respectively with no significance.
The reduction in stem density between mowing with P. suaveolens inoculation and mowing alone was not statistically significant in UT or in CO. In CO, mowing with P. suaveolens inoculation initially had a smaller impact compared with mowing alone but both still decreased stem density (24% and 41%). However, in year 2, mowing with inoculations showed a greater decrease in stem density compared with mowing alone (55% and 26%) (Table 2; Figure 1). In UT, C. arvense stem density following mowing (averaged over P. suaveolens) was significantly lower compared with control (averaged over P. suaveolens), (UT: P = 0.003; CO: P = 0.004). In CO, mowing plus P. suaveolens was significantly lower than P. suaveolens alone (P = 0.0178). In UT, mowing had significantly lower stem density (P = 0.006) compared with tillage, with no significance in CO. Mowing has been used to enhance the occurrence of systematically infected stems (Bourdôt et al. Reference Bourdôt, Hurrell, Skipp, Monk and Saville2011) and increases localized infection by spreading spores (Demers et al. Reference Demers, Berner and Backman2006). Very few systemically infected stems were found during the 2-yr study, which may have resulted in fewer additive effects from mowing with P. suaveolens compared with mowing alone. However, mowing should still be utilized with P. suaveolens in an IWM program, as the two treatments may be compatible and mutually beneficial based on reports of other studies.
Tillage
In CO, there was a significant difference (P < 0.001; Table 3) between management practice and an interaction between management practice and season. Further analysis showed that tillage treatments had significantly greater decline in stem density compared with control in 2022 (P = 0.009). There was no significant difference between tillage alone and tillage with P. suaveolens (UT: P > 0.05, CO: P > 0.05). The percent change in stem density is similar for both treatments: tillage in UT had an increase of 139% and in CO a decrease of 63%. In UT, tillage with P. suaveolens had a stem density increase of 142%, and in CO, a decrease of 70% (Table 2). In CO, tillage combined with P. suaveolens resulted in a slightly greater decrease in stem density in the first year (Figure 1) compared with tillage alone. The higher annual precipitation in the first year (Supplementary Table S1) may have contributed to P. suaveolens and tillage having a greater effect than in the second year. In our study, application of P. suaveolens inoculum did not cause a significant interaction with tillage but could be implemented in an IWM approach. Tillage has been used to manage C. arvense by reducing stem density through the depletion of root reserves and reduction in shoot biomass (Thomsen et al. Reference Thomsen, Brandsæter and Fykse2011; Weigel et al. Reference Weigel, Andert and Alt2025). Proper timing of tillage can be crucial, as early tillage can allow C. arvense to recover and rebuild root reserves for overwintering (Donald Reference Donald2000; Thomsen et al. Reference Thomsen, Mangerud, Riley and Brandsæter2015). Applying P. suaveolens inoculum 2 wk after tillage may enhance pathogen invasion of the smaller root fragments and increase systemic infection in the spring (Alexopoulos et al. Reference Alexopoulos, Mims and Blackwell1996; Berner et al. Reference Berner, Smallwood, Cavin, McMahon, Thomas, Luster, Lagopodi, Kashefi, Mukhina, Kolomiets and Pankratova2015a).
Ground Cover
Cirsium arvense
The trends observed in C. arvense cover align closely with those in stem density. When cover measurements are divided by stem density, an estimate of the biomass of individual shoots can be made, which may indicate the health of the population. Treatments with herbicide had the lowest amount of C. arvense cover, ranging from 0% to 25% in UT and 0% to 35% in CO (Supplementary Table S2). Of note, in 2022, plots treated with herbicide and P. suaveolens had zero C. arvense cover, while plots treated with herbicide alone still had a low density of C. arvense stems. Puccinia suaveolens may have an additive effect in herbicide-treated areas or may help prevent regrowth, suggesting that these two treatments are compatible. Mowing also significantly reduced percent C. arvense cover compared with control, although no significant difference occurred between mowing alone and mowing with P. suaveolens. The greatest C. arvense cover occurred in the control, which was more easily seen in UT than in CO (Figure 2). Cirsium arvense cover was greater in tilled plots as a result of possible (not documented) fragmentation and spread of C. arvense roots creating many small populations (Donald Reference Donald2000; Thomsen et al. Reference Thomsen, Mangerud, Riley and Brandsæter2015).
Figure 2. Average percent of the five ground cover types measured. (A) Colorado and (B) Utah experimental sites, 2020–2022, following treatment with individual and combined weed management approaches.
Vegetation
In UT, the herbicide treatment, which reduced broadleaf plants, including C. arvense, allowed more opportunity for grasses to grow (≤80% cover in year 2). Grass cover in UT was significantly higher in herbicide treatments compared with control and tillage (UT: P < 0.005), with greater effects observed in the second year. In a 3-yr study of management tactics for a non-native forb, a native grass cover increased as a result of herbicide treatments; however, a steady and significant increase in non-native grass cover was also recorded (Skurski et al. Reference Skurski, Maxwell and Rew2013). Grasses were not identified to the species level and could include undesirable invasive species of concern (e.g., cheatgrass) for management of natural areas. In contrast, grass coverage in CO showed minimal change across treatments (P > 0.05), although a significant interaction occurred between management and P. suaveolens (P = 0.003). Further analysis showed that HMT and P. suaveolens treatments had significantly lower grass cover compared with control, P. suaveolens, or mowing plus P. suaveolens (Table 4; Figure 2). Tillage to 20 cm distributes seeds throughout the entire tillage profile, reducing accumulation of seeds near the surface, and therefore may have caused reduced germination and grass cover, as seen in our results (Feledyn-Szewczyk et al. Reference Feledyn-Szewczyk, Smagacz, Kwiatkowski, Harasim and Woźniak2020). Differing climatic conditions between the two sites could also affect grass growth and contribute to the difference between treatments. Therefore, site characteristics need to be considered in conjunction with management strategy for potential revegetation or secondary invasion by non-native species.
Table 4. ANOVA table of the five ground cover types measured in Colorado and Utah sites as a function of rust inoculum application, management strategy, season, and the combined effects of these three parameters.
Before initial treatments, forb cover was low (0% to 10%) and remained below 30% (Supplementary Table S2) across most plots with no significant difference between treatments (UT: P > 0.05; CO: P > 0.05) (Table 4; Figure 2). Use of broadleaf herbicides against invasive forbs can be expected to also suppress both native and other exotic forbs within the treatment areas (Skurski et al. Reference Skurski, Maxwell and Rew2013). As expected, only treatments with broadleaf-selective herbicides showed a slightly greater decline in forb cover. In other studies, short-term changes in native forb cover remained insignificant after herbicide application, except for reductions in flowering and seed set for at least 4 yr posttreatment (Crone et al. Reference Crone, Marler and Pearson2009). There may be long-term implications in native forb recovery after herbicide is used to control non-native forbs like C. arvense.
Non-vegetation
Bare ground significantly increased as a result of HMT treatments with and without P. suaveolens in UT (P < 0.001). However, herbicide treatments with and without P. suaveolens did not result in a significant difference compared with other treatments (P > 0.05). Combined treatments had more of an impact on bare ground cover in UT than herbicide alone, perhaps due to significantly more grass cover in the herbicide alone treated plots in year 2 (P < 0.0001). Combined control tactics have been shown in both cropping and non-cropping systems to have better long-term control of C. arvense than herbicide alone (Davis et al. Reference Davis, Mangold, Menalled, Orloff, Miller and Lehnhoff2018). There may have been an additional effect from changes in seedbank availability due to tillage (Feledyn-Szewczyk et al. Reference Feledyn-Szewczyk, Smagacz, Kwiatkowski, Harasim and Woźniak2020) or from mowing, as mowing alone resulted in more bare ground compared with control (P = 0.03). In CO, bare ground cover was significantly greater in both herbicide- and HMT-treated plots compared with all other treatments (P < 0.001) (Figure 2). There was no significant difference found between bare ground cover in HMT and herbicide alone plots. Bare ground is an important aspect of land management, as it may necessitate reseeding to prevent soil erosion and the creation of niches for other noxious weeds. Revegetation should be included to promote native and desirable plants (Molvar et al. Reference Molvar, Rosentreter, Mansfield and Anderson2024; Rodriguez et al. Reference Rodriguez, McDonald, Bean and Larios2024; Weidlich et al. Reference Weidlich, Flórido, Sorrini and Brancalion2020).
At the UT site, herbicide treatments resulted in significantly more litter compared with tillage treatments (UT: P = 0.002). In CO, HMT treatments had significantly lower litter compared with all treatments except for control (Table 4; Figure 2). Litter cover can help to retain soil moisture and increase nutrient cycling (Perera et al. Reference Perera, Gruss and Szymura2024; Redmann Reference Redmann1978)
Effective strategies for controlling C. arvense vary with site conditions, management goals, and operation budgets, and the costs associated with the collection, processing, and distribution of P. suaveolens still need to be evaluated. Chemical treatments are often the cheapest and most effective approach for reducing C. arvense populations but frequently increase areas of bare ground for invasion from other weeds, including re-invasion by C. arvense, as observed in our study. The use of herbicides is also often restricted in natural landscapes or on organically certified farms, leaving stakeholders in search of alternative and complementary control tactics. Applications of P. suaveolens inoculum in addition to other tactics such as mowing show some promise for improved control of C. arvense. Continued monitoring will help determine whether additional treatment applications are needed, especially where P. suaveolens effects are enhanced on C. arvense. Although P. suaveolens acts slowly to suppress C. arvense, one of the long-term benefits is that recovery of native plant communities is more likely. Puccinia suaveolens is unique among tactics for managing C. arvense, as there are no direct effects on non-target plants. Applying P. suaveolens alone or with other tactics may provide a safe and sustainable means to enhance management of C. arvense in some natural areas.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/wsc.2025.10046
Acknowledgments
The authors wish to thank Sara Jo Dickens of Ecology Bridge LLC, Logan Jones of Park City Municipal Corporation, Levi Kokes of Colorado Parks and Wildlife, Tamarack Ranch State Wildlife Area, Colorado State University Extension Office, and the Utah Weed Supervisors Association for site access, Park City site information signage, and other logistical and outreach support.
Funding statement
This research was supported in part by USDA-NIFA (2019-70006-30452; RNS, SLY, and DWB) and APHIS Cooperative grants (AP20PPQFO000C386, AP21PPQFO000C237, and AP22PPQFO000C142). The multistate project was initiated and supported by USFS-BCIP agreement no. 17-CA-1142004-252 (DWB) in cooperation with Carol Randall of the USFS.
Competing interests
The authors declare no conflicts of interest.
Open access
