Alternative Herbicides for Controlling Herbicide-Resistant Annual Bluegrass ( Poa annua L.) in Turf

: Poa annua is a cosmopolitan, cool-season grass species regarded as one of the most signiﬁcant weeds of turfgrass. It is mainly controlled by herbicides; however, repeated use of herbicides in golf turf has resulted in the evolution of multiple-herbicide resistant P. annua . Four ﬁeld experiments were performed in autumn and spring in golf turf to identify effective herbicide options to control multiple herbicide-resistant P. annua . In herbicide resistance screening, the trial site population (SA1) was found to be susceptible to amicarbazone and terbuthylazine, but resistant to simazine and metribuzin at the ﬁeld rate of each herbicide. Consistent with the results of the pot study, the PSII-inhibiting herbicides amicarbazone and terbuthylazine provided the best control (80–100%) of P. annua in both autumn and spring trials with minimal damage to the turf. In contrast, the other two PSII-inhibiting herbicides, metribuzin and simazine, were relatively ineffective in controlling P. annua in the ﬁeld. Indaziﬂam also performed well in both autumn trials and reduced P. annua occurrence by >75%. Pyroxasulfone and s-metolachlor only provided moderate weed control in both the autumn and spring trials, reducing P. annua occurrence by 50%. Among the nine different herbicides, amicarbazone and terbuthylazine were found to be most effective for spring and autumn application in turf. As resistance to some PSII-inhibiting herbicides has already evolved in this ﬁeld population, the use of amicarbazone and terbuthylazine needs to be integrated with other herbicide modes of action and non-chemical tactics to delay the onset of resistance to them.


Introduction
Annual bluegrass (Poa annua L.) is one of the most problematic weeds in sports turf, particularly in temperate climates [1,2]. It reduces the turf quality for sport and creates an uneven surface that affects ball roll [3,4]. This weed also competes for water and nutrients with the desired turfgrass species and reduces turf growth. It often produces panicles below the turf cutting height, which reduces the effectiveness of mowing for weed control [3]. P. annua is a genetically diverse weed species that typically germinates in autumn, grows in winter, and produces seed in spring; however, some germination can occur in spring as well [5,6]. During summer P. annua senesces, resulting in dead, bare, and unsightly patches that reduce the aesthetic value of turf [7]. It is considered the most problematic weed of golf courses in Australia and other countries, such as the USA. One feature that makes it difficult to control P. annua is its large seed bank of up to 200,000 seed m −2 [6] and potential for year-round germination [8].
Several alternative management strategies including manual, cultural, biological and chemical control are available for controlling P. annua [9,10]. However, chemical control tends to be most commonly used, because of the ease of application and reliability of weed control. Both PRE and POST herbicides are used to control P. annua in turf [11]. However, repeated use of herbicides has resulted in the evolution of herbicide resistant P. annua populations [12][13][14]. The loss of herbicides to resistance requires new weed management practices including additional herbicides to control P. annua.
Greenkeepers of the golf course used for this study had reported difficulty in controlling P. annua with several herbicides. Seeds were collected from the golf course trial site in October 2017 to screen for herbicide resistance. The population was confirmed resistant to four different herbicide modes of action HRAC (Herbicide Resistance Action Committee) group, 1, 2, 5 and 31 in initial research, with >9-fold resistance to HRAC group 1, 2 and 31 and >2-fold resistance to HRAC group 5. Therefore, alternate herbicides with different mode of action could be an option for controlling this population. Hence, four field trials were conducted at a golf course in spring and autumn seasons between 2018 and 2020. The treatments were selected based on some commonly used herbicides currently registered for controlling P. annua in golf turf in Australia and additional herbicides that are not currently registered for turf, but may be suitable for controlling P. annua. The objective of the study was to identify suitable herbicide options to control multiple herbicide-resistant P. annua in a bermuda grass (Cynodon dactylon) turf during autumn and spring.

Experimental Site and Field Trial Design
Field trials were conducted on a Bermuda grass (Cynodon dactylon) turf at a golf course (34.896 • S 138.51 • E) in spring and autumn between 2018 and 2020. The chosen site was a practice green, which received the same weed control measures as the remainder of the golf greens. The turf was well managed with regular mowing (twice a week) at a height of 9 mm and watered every third night depending on the weather conditions. The soil type of the trial site is sandy in texture with 1.48% organic matter (Table 1). Liquid fertilizer MP Brilliance (20-0-0 + 6Fe, 1Mg) was applied at 37 L ha −1 every 3-4 weeks to maintain the fertility of the golf course. Meteorological and soil properties of the experimental sites are presented in Tables 1 and 2. Nine different herbicide treatments were applied in the spring trial established in October 2018 and repeated in September 2019 (Table 3). Similarly, a field trial with nine herbicide treatments was undertaken in the autumn at the same location in March 2019 and repeated in March 2020 ( Table 3). Dates of herbicide application and weather conditions are presented in Table 1. The timing of herbicide application in these trials is consistent with the weed management programs used by local greenkeepers. A range of both PRE and POST herbicides were used in both spring and autumn trials as emergence of P. annua occurs over a long period. Therefore, some herbicides e.g., s-metolachlor, indaziflam, which are mainly recommended for PRE application, were applied in both autumn and spring trials to test for both PRE and POST herbicide activity. Additionally, herbicides not currently registered for P. annua control were tested in these field trials. The trial was established in a randomised complete block design with four replications. The plot size was 5 m × 2 m with a 0.25 m gap between plots. A non-treated check treatment was included in each trial. Herbicides were applied using a CO 2 -pressurized boom sprayer with medium droplet size flat-fan nozzles (TT01 and LD110-015) delivering 100 L ha −1 volume at a pressure of 200 kPa. The trials were assessed 28 DAT (days after treatment) and again 42 DAT.

Pot Trial Methodology
As most of the PRE herbicides applied in the trial were ineffective, a follow-up pot study was undertaken with the PRE herbicides (Table 4) to determine the resistance status of the P. annua population from the trial site (hereafter referred to as SA1). The P. annua population (SA1) used in this study was collected from the trial site during October 2018. One susceptible population collected from a non-golf course area was used as the susceptible control. Seed bulking was performed in 2019 and the pot trial conducted during July 2019 and repeated in June 2020 at the University of Adelaide following the method used by Barua, et al. [16]. Approximately 100 seeds were measured by volume (0.2 mL), placed onto the surface of standard potting mix [17] and herbicide applied directly onto the seed. Herbicides (Table 4) were applied using a laboratory moving boom sprayer equipped with a twin nozzle (Tee-jet 110 • flat fan Spraying Systems, Wheaton, IL, USA) delivering an output of 118 L ha −1 at a pressure of 250 kPa and speed of 1 m s −1 . Immediately after herbicide application, the seeds were covered with 5 mm of potting mix. The pots were placed outside and watered as required. The experiment was assessed for seedling emergence 28 DAT (days after herbicide treatment). Plants that emerged and grew to the two-leaf stage were considered resistant to the herbicide treatment. The experiment was repeated.
In order to further investigate resistance in P. annua to PSII inhibiting herbicides, an additional pot trial was undertaken. In this dose-response pot trial herbicide rates used were the following: amicarbazone (0, 26.3, 52.5, 105, 210, 420, 840 and 1680 g a.i ha −1 ), terbuthylazine (0, 32.8, 65.6, 131.3, 262.5, 525, 1050 and 2100 g a.i ha −1 ), simazine (0, 26.3, 52.5, 105, 210, 420, 840 and 1680 g a.i ha −1 ) and metribuzin (0, 13.1, 26.3, 52.5, 105, 210 and 420 g a.i ha −1 ). The methodology used was the same as Barua et al. [16]. The susceptible population mentioned above was used as the susceptible control. Seeds were sown in trays (330 by 200 by 50 mm) located outdoors. One to two leaf seedlings were transplanted into punnet pots (95 by 85 by 95 mm) containing the already described standard potting mix, with 5 plants per pot and replicated three times. At the 2-3 leaf stage, plants were treated with the herbicides using the laboratory moving boom sprayer mentioned above. Plants actively growing with new leaves after 28 days were classified as survivors, while plants with severe stunting or dead were considered susceptible [19]. The experiment was repeated.

Data Collection and Analysis
Prior to herbicide treatment, a low density of P. annua was present on the site in the autumn trials, but a much larger number of P. annua plants were uniformly distributed at the trial site in spring trials. The occurrence of P. annua in the trials after treatment was determined with the use of a 1 by 1 m grid divided into four hundred 50 by 50 mm squares. The grid was placed in the centre of each plot at three random locations and the number of squares containing P. annua plants counted to determine the % occurrence. The treatments were visually assessed for turf phytotoxicity at 7 DAT, 28 DAT and 42 DAT using a scale of 0−100 where 0 = no visual injury and 100 = no green tissue [20].
The data of percent occurrence of P. annua were subjected to two way analysis of variance (ANOVA) with GenStat version 19 (VSN International Ltd. Hemel Hempstead, UK) with herbicide and experiment run treated as variables. As there was a significant variation between experimental runs (p < 0.05), the data of each run is presented separately. The data were square-root transformed before statistical analysis to normalize the distribution of the residuals. Where treatment differences were significant, the means of the transformed data were compared using Fisher's Protected LSD at p = 0.05. As there were no significant differences in turf quality between the two runs, the data were pooled for statistical analysis.
Dose response trials were set up as a completely randomized design and repeated. Data was subjected to three way ANOVA for each herbicide with population, rate and run as variables. For every herbicide there was a significant effect of rate (p < 0.0001) and population (p < 0.0001), but not for run. Therefore, data for each herbicide were pooled across experiments. Survival at each rate was converted to mortality and the data analyzed using PriProbit (1.63) [21] with the LD 50 with 95% confidence intervals (CI) determined and resistance ratios calculated as LD 50 Resistant/LD 50 Susceptible. Population responses to the herbicides were considered different if confidence intervals of the LD 50 did not overlap.

Spring Trial Assessment
The occurrence of P. annua in the non-treated control for the two spring field trials varied between 86 and 100% of grids (Table 5), which indicates a relatively uniform spatial distribution at the site. In both spring trials, amicarbazone (2% occurrence) and terbuthy-lazine (15-20% occurrence) provided the greatest control. The remaining treatments were less effective, particularly endothall, which was ineffective in both spring trials due to herbicide resistance confirmed previously at this site Barua, et al. [16]. These less effective treatments showed inconsistency in weed control over the two years. In 2018, amicarbazone provided the highest level of control by reducing P. annua occurrence to 2% followed by terbuthylazine that reduced P. annua occurrence to 14% (Table 5). However, the other two PSII-inhibiting herbicides (simazine and metribuzin) provided moderate control and reduced the occurrence of P. annua between 60 to 66% ( Table 5). Other mode of action herbicides, such as indaziflam and propyzamide, also reduced occurrence to 60 to 62% (Table 5).
Amicarbazone and terbuthylazine were also the most effective weed control treatments in the 2019 spring trial, followed by pyroxasulfone, indaziflam and s-metolachlor. Amicarbazone reduced the occurrence of P. annua to 2% followed by terbuthylazine to 20% (Table 5). Simazine and metribuzin provided moderate control of P. annua similar to the results obtained in 2018. The next most effective treatments were pyroxasulfone and indaziflam that reduced P. annua occurrence to 39% followed by s-metolachlor to 41% (Table 5).

Autumn Trial Assessment
The occurrence of P. annua was high (>96%) in untreated plots for both 2019 and 2020 autumn trials (Table 6). Amicarbazone was once again the most effective herbicide in autumn 2019, followed by terbuthylazine, indaziflam, pyroxasulfone, and s-metolachlor. In contrast, pendimethalin, propyzamide and simazine were less effective treatments in both autumn trials. The occurrence of P. annua was reduced to 0% by amicarbazone, 6% by terbuthylazine, 22% by indaziflam, 43% by pyroxasulfone, and 45% by s-metolachlor (Table 6). In 2020, amicarbazone was again the best treatment and reduced P. annua occurrence to below 2% (Table 6). Terbuthylazine was the second best treatment and reduced P. annua occurrence to 8%, followed by indaziflam to 26%, pyroxasulfone to 40%, and s-metolachlor to 55% (Table 6). Consistent with the spring trials, simazine and metribuzin were less effective than amicarbazone or terbuthylazine.
A difference in the performance of two herbicides (propyzamide and simazine) was observed in the two autumn trials ( Table 6). As the soil type at the study site is sandy in texture (Table 2), higher rainfall in 2019 (Table 1) may have moved these herbicides through the soil profile reducing their performance in that year.
In all four trials, amicarbazone was the most effective treatment; however, it produced mild turf grass injury (data were not shown). This persisted for four weeks after herbicide application, followed by complete recovery of the turf at six weeks. No other treatments were phytotoxic to the turf. Perry, et al. [22] reported that amicarbazone provided superior control of P. annua through foliar and root uptake at 371 g ha −1 as compared to atrazine at 2025 g ha −1 . Another study suggested that amicarbazone is more active than other PSII inhibitors such as atrazine Dayan, et al. [23]. Amicarbazone has been recently registered for turf in Australia at 210 g ha −1 [24], a much lower rate than 700 g ha −1 used in our field trials that were initiated prior to the registration of amicarbazone. It is possible the lower rate on the label will provide less effective control of P. annua than observed in this trial and in other studies [22,23]. The efficacy of amicarbazone at lower rates on P. annua needs to be validated. Of all the herbicide treatments investigated, terbuthylazine consistently resulted in good turf quality followed by indaziflam, propyzamide and amicarbazone at six weeks after herbicide application. In both the autumn and the spring trials there was rapid recovery of turf grass and no difference in turfgrass density (data not shown here) at 6 weeks after herbicide application. However, terbuthylazine, which was the second most effective herbicide in these trials, is not registered for use in turf in Australia. Indaziflam was the next most effective herbicide for PRE application in autumn and is registered for use in Australia [25]. Two other herbicides, pyroxasulfone and s-metolachlor, provided similar levels of control (40−55%) in all the trials. S-metolachlor is registered in turf for P. annua control during the autumn season. Pyroxasulfone should be considered as an option for use in turf and would increase the range of herbicide options for P. annua. The use of herbicide rotations can help to slow the evolution of resistance in weeds.

Pot Trial Screening
To investigate why some PRE herbicides were not effective in the autumn field trial, a pot study was conducted with the trial site population (SA1) using these herbicides. Moderate levels of resistance (30 to 34% survival) were detected to metribuzin and simazine (Table 3), which could explain the poor efficacy of these two herbicides in the field. However, SA1 was susceptible to the other herbicides used in the field trials (Table 3). Therefore, other factors, such as weed growth stage, weed competition and environmental conditions appear to be responsible for the poor control of P. annua by these herbicides.

Dose-Response Trial
To further investigate the extent of resistance to the PSII inhibiting herbicides used in the trial, dose response experiments were conducted on SA1. Compared to the S population, SA1 had 2.6-fold resistance to simazine and 2.2-fold to terbuthylazine. The LD 50 of SA1 population to metribuzin was 1.5-fold higher than the S population and for amicarbazone was 1.2-fold higher (Table 7). Low-level resistance does not always result in a failure of the herbicide in the field. Terbuthylazine controlled the S population well below the field rate, so this herbicide was effective despite the low-level resistance present. In contrast, metribuzin at the field rate only controlled the S population, so the increase in tolerance of SA1 is the likely cause of poor performance in the field. The dose response study confirmed the SA1 population exhibited greater resistance to simazine and lower resistance to terbuthylazine, but little resistance to amicarbazone.

Discussion
Resistance to 10 herbicide modes of action has been reported in P. annua worldwide [12], which restricts the number of herbicide options available for its control in turf. Most of the herbicides used in the autumn trial exhibited poor control of P. annua in the field (Table 6), despite resistance not being confirmed in pot studies (Table 3). This indicates that factors other than herbicide resistance were responsible for the poor performance, such as high seed bank, herbicide application timing relative to weed emergence, soil type, thatch layer, and environmental conditions [10,26].
The timing of a herbicide application is important to maximise PRE herbicide effectiveness [11]. In the current study, herbicide timings were similar to those used by local greenkeepers. In our field study, the performance of indaziflam was greater in the autumn trial than the spring trial. Brosnan, et al. [27] reported that soil application of indaziflam was more effective than foliar application to control D. ischaemum and P. annua. They concluded that indaziflam needs to be absorbed by the roots to maximise POST control. Therefore, time of application is important for the success of indaziflam. It is likely that when this herbicide was applied in autumn some P. annua had already germinated, thus reducing the efficacy of the indaziflam. In previous research, indaziflam was shown to be more effective when used PRE and EPOST (early POST) [28], which could explain why it was less effective in our spring trials where weeds were established. Greenkeepers often reapply the same herbicide during the season to maximise activity on subsequent germinations, which was not done in these field trials. Multiple applications may be more effective than a single application. Thus, a follow-up application timing study is needed to determine the difference between a single and multiple applications of this herbicide.
Pronamide has been shown to provide good control P. annua with PRE and POST treatments [29]. However, in our study performance of pronamide only provided moderate efficacy (Tables 5 and 6). In a previous study, poor efficacy of pronamide used POST on P. annua in Georgia, USA was associated with reduced absorption and translocation [30]. Others have also claimed that unsuitable application timing (when plants are too large) can lead to the poor efficacy of pronamide [31].
Factors which influence the efficacy of PRE herbicides include solubility in soil solution, binding to organic matter and the half-life in soil [32]. Uptake of herbicides by the root occurs more readily when the herbicide is in soil solution. Out of the herbicides tested in the autumn trials, amicarbazone had the highest solubility in water [33] and was found to have the greatest efficacy. In contrast, terbuthylazine and indaziflam have lower water solubility that could account for their lower activity.
Soil organic matter content can also play an important role in PRE herbicide activity [34]. In our study, some the PRE herbicides were unsuccessful for controlling P. annua in the autumn trial, even though no herbicide resistance was detected. The soil at the trial site was sandy with 1.48% soil organic matter content (Table 2). Soil low in organic matter (low cation exchange capacity) has a lower tendency to bind herbicides and allows greater herbicide availability for uptake by weeds. However, heavy rainfall or frequent irrigation can move the herbicides through the soil profile before the compound has a chance to bind to the soil colloids and organic matter [32]. Since most P. annua seeds germinate in the top layer of soil [10,35], leaching of herbicides to lower layers of soil could result in reduced weed control.
Another factor that may have reduced the performance of PRE herbicides is the thatch layer, which is primarily organic matter (stems, stolons, roots), in turf that develops between the turf and soil surface [36]. This limits the movement of air, water and nutrition into the soil [37]. The thatch layer can also bind PRE herbicides, thereby reducing their effectiveness [38,39].
In conclusion, amicarbazone was the most effective herbicide for the reduction of P. annua occurrence (98-100%) followed by terbuthlazine (>80%), indazaflam (>63%) and pyroxasulfone (>57%) in both autumn and spring trials. Indaziflam performance was greater in the autumn trial than the spring trial indicating that indazaflam could be a good option for autumn application. The availability of multiple herbicides with different modes of action allows rotating herbicides in an herbicide management program, which is important to reduce the risk of resistance. Pyroxasulfone, although not currently registered in turf, could be a viable option for P. annua control. Amicarbazone and terbuthylazine controlled the trial site population (SA1) in pots indicating it was susceptible to both herbicides (Table 3). In contrast, simazine and metribuzin were less effective (Table 3). Terbuthylazine could be a potential candidate for P. annua control in turf. Given the extensive presence of herbicide resistance, it would be invaluable for the greenkeepers to test their populations for resistance status, so they can make informed choices of herbicides for P. annua control. The currently effective herbicides need to be used with care otherwise, they could also lose their effectiveness due to resistance.