Glyphosate Resistance Conﬁrmation and Field Management of Red Brome ( Bromus rubens L.) in Perennial Crops Grown in Southern Spain

: The excessive use of the herbicide glyphosate on annual and perennial crops grown in Southern Spain has caused an increase in resistant weed populations. Bromus rubens has begun to spread through olive and almond cultivars due to low glyphosate control over these species, whereas previously it had been well controlled with ﬁeld dose (1080 g ae ha − 1 ). Characterization using Simple Sequence Repeat (SSR) markers conﬁrmed the presence of B. rubens collected in Andalusia. A rapid shikimic acid accumulation screening showed 17 resistant (R) populations with values between 300 and 700 µ g shikimate g − 1 fresh weight and three susceptible (S) populations with values between 1200 and 1700 µ g shikimate g − 1 fresh weight. In dose–response experiments the GR 50 values agreed with previous results and the resistance factors (RFs: GR 50 R/GR 50 S (Br1)) were between 4.35 (Br9) and 7.61 (Br19). Foliar retention assays shown no differences in glyphosate retention in both R and S populations. The tests carried out in a resistant ﬁeld (Br10) demonstrated the control efﬁcacy of pre-emergence herbicides since ﬂazasulfuron in the tank mix with glyphosate had up to 80% control 15 to 120 days after application (DAA) and grass weed postemergence herbicides, such as propaquizafop + glyphosate and quizalofop + glyphosate, had up to 90% control 15 to 90 DAA. Results conﬁrm the ﬁrst scientiﬁc report of glyphosate-resistant B. rubens worldwide; however, the use of herbicides with another mode of action (MOA) is the best tool for integrated weed


Introduction
Weed control has been performed by the application of multiples herbicides with different modes of action (MOAs) since the 1940s [1]. Specifically, the herbicide glyphosate (N-(hosphonomethyl)glycine) has been commercialized worldwide since the 1970s and is used as a broad-spectrum and postemergence treatment for weed control due to its translocation ability in plants [2]. The MOA of glyphosate is by aromatic amino acids biosynthesis inhibition [3]. The broad-spectrum activity of this herbicide is due to the inhibition of 5-enolpyruvyl-3-shikimate phosphate synthase (EPSPS), which is present in all plants [4]. The EPSPS enzyme acts in the shikimic acid pathway in the biosynthesis of aromatic amino acids, such as phenylalanine, tyrosine, and tryptophan [5]. These amino acids are essential for plants and when they are inhibited by the action of glyphosate all susceptible plants die.
Nowadays, it is know that the weed resistance is the consequence of selection pressure by farmers coupled with the high evolution capacity of weed populations [6]. The resistant Seedlings of the 20 populations were transplanted into pots (one plant plot −1 ) with 240 g of substrate (soil:peat moss (1:1)) that was previously irrigated. All populations were transferred to a greenhouse and watered daily to field capacity before and during assays.

Molecular Characterization of Bromus spp.
Four populations previously identified as B. sterilis, B. tectorum, B. diandrus, and B. madritensis [24] plus twenty populations of B. rubens identified in situ were used for molecular characterization. A total of 24 populations were characterized using Simple Sequence Repeat (SSR) markers following the methodology described by Ramakrishnan et al. [25] with some modifications. For this step, ten individuals from each population were used. Samples of~100 mg of young leaf tissue from each plant at BBCH 13-14 stage [26] were taken to obtain DNA. Forward primers were tailed with the M13 sequence (5 -TGTAAAACGACGCCAGT-3 ) at the 5 ends for fluorescent labelling of PCR fragments [27]. Amplification of DNA was carried out in a 15 µL reaction mixture containing 20 ng of DNA, 5x Buffer (50 mM KCl, 10 mM Tris-HCl, 0.1% Triton X-100), 2.5 mM MgCl 2 , 250 µM of dNTPs, 0.1 µM of forward primer, 0.4 µM of reverse primer, 0.4 µM of 6-FAM, and 0.25 units of Taq DNA polymerase (BIOTOOLS). PCR reactions were performed in a Biometra ® thermocycler and conditions of the PCR amplification were as follows: one cycle of 15 min at 95 • C, then 40 cycles of 30 s at 95 • C, 30 s at 60 • C, and 30 s at 72 • C, followed by 8 cycles of 30 s at 90 • C, 45 s at 53 • C, and 45 s at 72 • C, and one final extension step of 10 min at 72 • C. Subsequently, the PCR products were separated using an automatic capillary sequencer (ABI 3130 Genetic Analyzer Applied Biosystems, Madrid/HITACHI, Madrid, Spain) from the University of Cordoba, Spain. The results were analyzed using Genotyper software 3.7 (Applied Biosystems). A DNA standard (400HD-ROX) was used to calculate the size of the amplified PCR fragments (alleles) for each SSR marker alleles. Genetic distances between all individuals were calculated using Jaccard's coefficient of similarity. Grouping of the genotypes was determined using the unweighted pair group method with arithmetic mean (UPGMA) and a dendrogram was generated with the NTSYS program [28].

Resistant Fast Screening by Shikimic Acid Accumulation Assay
The main objective of this assay was to differentiate resistant and susceptible populations, knowing that the increase in shikimic acid accumulation referred to the action of glyphosate and therefore, were considered susceptible (S). However, those populations that accumulated very little or nothing were labeled as resistant (R). Discs were cut from the youngest leaf of ten plants and then were pooled. In total, 50 mg from each mix per population was transferred into 2 mL Eppendorf tubes containing 999 µL of monoammonium phosphate (NH 4 H 2 PO 4 10 mM, pH 4.4) plus 1 µL of glyphosate (1000 µM). The shikimic acid accumulation was performed according to methodology described by Vázquez-García et al. [29] with some modifications. Four treated replications and four nontreated samples were used in a completely random design test. Finally, the results were indicated in µg of shikimic acid g −1 fresh tissue.

Glyphosate Dose-Response Curves Assay
Whole plants at  of each population were treated with glyphosate (Roundup Energy ® SL, 45% as isopropylamine salt, Monsanto) doses ranging from 0 to 3000 g ae ha −1 . Herbicide application was performed by chamber (SBS-060 De Vries Manufacturing, Hollandale, MN, United States) equipped with an 8002 flat fan nozzle delivering 200 L ha −1 at 250 KPa. After application, the plants were taken into the greenhouse and irrigated daily as necessary. Ten replications (one plant = one replication) per glyphosate dose were used in a completely random design test. At 21 days after application, the survival plants were evaluated to estimate the lethal dose to kill 50% of population (LD 50 ). In addition, plants were weighed after dried them at 60 • C for 48 h. Subsequently, the dose that inhibits the plant growth to 50% (GR 50 ) was estimated.

Glyphosate Foliar Retention Assay
The retention experiment was performed in six plants of each B. rubens population. According to González-Torralva et al. [30], a glyphosate dose of 360 g ae ha −1 plus 100 mg L −1 Na-fluorescein was applied to B. rubens plants. The treatment equipment was described in the previous section. Two hours after application, the plants were cut and transferred to test tubes which contained 50 mL of 5 mM NaOH. Then, test tubes were shaken for 30 s to remove the spray solution. Subsequently, the washed solution was transferred to glass vials to measure the fluorescein absorbance; for this step, a spectrofluorometer (Hitachi F-2500, Tokyo, Japan) with an excitation wavelength of 490 nm and absorbance at 510 nm was used. Finally, plants were weighed after 48 h at 60 • C drying. A completely randomized design was performed with two repetitions (one repetition = six plants per population). The results are expressed in µL spraying solution per gram dry matter.

Chemical Alternatives In Situ
This trial was carried out during two growing seasons at winter-spring time (2018-2019 and 2019-2020) in a field where some G-R populations originated. In "La Reina" (37.737492, −4.646091), almond groves infested with B. rubens (80% infestation) were treated with glyphosate and other herbicides ( Table 2) to test their performances. A randomized complete block design with four replications was used. The herbicide treatments were performed in plots of 10 m 2 at two different stages: (a) pre-emergence and (b) postemergence. A Pulverex backpack sprayer equipped with four flat fan nozzles Teejet 11002, at a spraying pressure of 200 kPa and calibrated to deliver a volume of 200 L ha −1 was used for applications. The control was evaluated 30, 60, 90, and 120 days after application (DAA) at pre-and 30, 60, and 90 DAA at postemergence stages for the percentage of effectiveness in B. rubens control. Control ratings were expressed on a 0 (no control) to 100 (plant dead) scale.

Statistical Analyses
Parameters GR 50 and LD 50 described in dose-response curve assays were estimated with a nonlinear regression using Equation (1).
where Y is the dry weight, or plant mortality expressed as a percentage of the value for the untreated control, c and d are the coefficients corresponding to the lower (fixed at 0) and upper asymptotes, respectively, b is the slope of the curve point (i.e., GR 50 , LD 50 ), x (independent variable) is the glyphosate doses, and g is the herbicide rate at the point of inflection curve. The nonlinear regression analyses were conducted using the R package "drc" [31]. In addition, GR 50 and LD 50 resistance factor (RF) was calculated with Equation (2). RF = (GR 50 or LD 50 R/GR 50 or LD 50 S) (2) where "R" is the resistant population and "S" is susceptible population. For the rest of experiments, the normal error distribution and the homogeneity of the variance were verified for each set. Finally, data were assessed via analysis of variance (ANOVA) using the Statistix software version 10.0 (Analytical software, Tallahassee, FL, USA). A Tukey (p < 0.05) test was conducted to compare the means.

Bromus spp. Molecular Characterization
Seven SSR markers (Bt03, Bt04, Bt05, Bt12, Bt26, Bt30 and Bt33) were enough to discriminate the Bromus species tested. The dendrogram shows five groups at a similarity coefficient of 0.5 ( Figure 1). Group I was formed by B. sterilis individuals. The second group corresponded to B. rubens and included the 20 populations from Andalusia. All B. rubens populations grouped together regardless of the province where they were collected, and R and S individuals could not be distinguished by the seven SSR markers used in this study. B. diandrus and B. madritensis were differentiated in groups III and IV, respectively. Finally, group V was comprised of the B. tectorum population. criminate the Bromus species tested. The dendrogram shows five groups at a similarity coefficient of 0.5 ( Figure 1). Group I was formed by B. sterilis individuals. The second group corresponded to B. rubens and included the 20 populations from Andalusia. All B. rubens populations grouped together regardless of the province where they were collected, and R and S individuals could not be distinguished by the seven SSR markers used in this study. B. diandrus and B. madritensis were differentiated in groups III and IV, respectively. Finally, group V was comprised of the B. tectorum population.

Resistant Fast Screening by Shikimic Acid Accumulation Assay
Overall, the B. rubens populations response was different, which resulted in multiple patrons of resistance to the herbicide glyphosate. This fast screening at 1000 µM of glyphosate, showed 17 resistant populations out of 20, which had accumulated the least shikimic acid ( Figure 2). Br1, Br3, and Br11 populations accumulated the highest amount of shikimic acid at a rate of 1200 to 1700 µg g −1 fresh weight, whereas the other 17 populations accumulated 300 to 700 µg g −1 fresh weight.

Resistant Fast Screening by Shikimic Acid Accumulation Assay
Overall, the B. rubens populations response was different, which resulted in multiple patrons of resistance to the herbicide glyphosate. This fast screening at 1000 µM of glyphosate, showed 17 resistant populations out of 20, which had accumulated the least shikimic acid (Figure 2). Br1, Br3, and Br11 populations accumulated the highest amount of shikimic acid at a rate of 1200 to 1700 µg g −1 fresh weight, whereas the other 17 populations accumulated 300 to 700 µg g −1 fresh weight.

Glyphosate Dose-Response Curves
The estimated dose-response curve parameters demonstrated different resistanc levels in B. rubens populations (Table 3). Seventeen populations were resistant becaus they required at least 1080.33 to 2100.40 g ae ha −1 to reduce their mortality to 50%, indicat ing that the glyphosate field dose should be doubled or even tripled for total control. Br and Br19 populations were the most resistant these required 2100.4 and 2024.47 g ae ha −1

Glyphosate Dose-Response Curves
The estimated dose-response curve parameters demonstrated different resistance levels in B. rubens populations (Table 3). Seventeen populations were resistant because they required at least 1080.33 to 2100.40 g ae ha −1 to reduce their mortality to 50%, indicating that the glyphosate field dose should be doubled or even tripled for total control. Br4 and Br19 populations were the most resistant these required 2100.4 and 2024.47 g ae ha −1 , respectively, to kill 50% of the population, whereas Br3 and Br1 populations needed only 274.22 and 229.87 g ae ha −1 , respectively. In contrast, GR 50 values showed a S population (Br1) needed only 140.64 g ae ha −1 , whereas the most resistant needed 1031 g ae ha −1 to reduce the dry weight at 50%. This meant that the RF referred to a dry weight reduction (GR 50 ) varying from 1.05 to 7.61 (Table 3, Figure 3). According to the RF results, we characterized 17 populations as G-R. (Figure 3).

Foliar Retention Assay
There were not differences between the 20 B. rubens populations. They had values from 371.6 to 416.75 µL glyphosate g −1 dry weight. Both R and S B. rubens populations had

Foliar Retention Assay
There were not differences between the 20 B. rubens populations. They had values from 371.6 to 416.75 µL glyphosate g −1 dry weight. Both R and S B. rubens populations had no significant differences in this assay; thus, foliar retention was not involved in the low susceptibility to glyphosate.

Chemical Alternatives In Situ
Field trials carried out in "La Reina" (Br10 (GR 50 factor: 6.79 and LD 50 factor: 7.41)) in almond trees, demonstrated the potential alternatives to glyphosate. Overall, the percentage B. rubens control was similar in the two seasons. Treatments applied pre-emergence were the least promising because only the glyphosate and flazasulfuron tank mix had the best results in both seasons (Figure 4). This treatment maintained an efficacy close to 80% (±) against B. rubens from 30 DAA to 120 DAA (Figure 4). Otherwise, the application with chlorotoluron and diflufenican had the worst result, with poor control (20%) from 30 to 120 DAA in both growing seasons. Diflufenican plus iodosulfuron and glyphosate had satisfactory control in nontarget species, such as Lolium sp., Vulpia sp., and Conyza sp. but not against B. rubens (Figures 4 and 5). Herbicides applied in postemergence were a more promising chemical alternative. Grass weed herbicides (ACCase inhibitors), such as propaquizafop and quizalofop in tank mix with glyphosate, controlled B. rubens up to 90% from 30 to 90 DAA ( Figure 5). Additionally, this tank mix controlled other important weeds, such as Lolium sp., Hordeum murinum, and Bromus sp. Glyphosate resistance was visualized with applications at 1080 g (control under 60%) and 1800 g ae ha −1 (control close to 80% but only at 30 DAA). Flazasulfuron (20 g ai ha −1 ) plus glyphosate (860 g ae ha −1 ) is a commercial product (Chikara Duo ® ) that showed to be a good treatment, but only for the first 30 DAA, after which time B. rubens control was poor (Figure 4).

Chemical Alternatives In Situ
Field trials carried out in "La Reina" (Br10 (GR50 factor: 6.79 and LD50 factor: 7.41)) in almond trees, demonstrated the potential alternatives to glyphosate. Overall, the percentage B. rubens control was similar in the two seasons. Treatments applied pre-emergence were the least promising because only the glyphosate and flazasulfuron tank mix had the best results in both seasons (Figure 4). This treatment maintained an efficacy close to 80% (±) against B. rubens from 30 DAA to 120 DAA (Figure 4). Otherwise, the application with chlorotoluron and diflufenican had the worst result, with poor control (20%) from 30 to 120 DAA in both growing seasons. Diflufenican plus iodosulfuron and glyphosate had satisfactory control in nontarget species, such as Lolium sp., Vulpia sp., and Conyza sp. but not against B. rubens (Figures 4 and 5). Herbicides applied in postemergence were a more promising chemical alternative. Grass weed herbicides (ACCase inhibitors), such as propaquizafop and quizalofop in tank mix with glyphosate, controlled B. rubens up to 90% from 30 to 90 DAA ( Figure 5). Additionally, this tank mix controlled other important weeds, such as Lolium sp., Hordeum murinum, and Bromus sp. Glyphosate resistance was visualized with applications at 1080 g (control under 60%) and 1800 g ae ha −1 (control close to 80% but only at 30 DAA). Flazasulfuron (20 g ai ha −1 ) plus glyphosate (860 g ae ha −1 ) is a commercial product (Chikara Duo ® ) that showed to be a good treatment, but only for the first 30 DAA, after which time B. rubens control was poor (Figure 4).

Discussion
Molecular markers have been successfully employed for genetic diversity and genetic characterization in a wide range of plant species. Particularly, SSR markers are very reliable and suitable for the study of genetic diversity between species of the same genus because of their transferability and power to detect closely related polymorphic individuals. It has been reported that, in general, cross-species transferability within genera is moderate to high (50-100% success) [32,33]. In this work, we used seven SSRs developed in B. tectorum [25] to discriminate five Bromus species. All SSR markers were polymorphic and transferable to the five species. Although we could not genetically distinguish R and S populations of B. rubens, the seven SSRs were useful tools for discriminating between the Bromus species. Thus, the dendrogram constructed in this study revealed that the five Bromus species are genetically distinct from each other and the 20 populations collected in Andalusia belong to B. rubens.
In contrast, the rapid screening using the leaf disc shikimic acid accumulation allowed us to separate glyphosate-R and -S populations. In our study, the Br1 population

Discussion
Molecular markers have been successfully employed for genetic diversity and genetic characterization in a wide range of plant species. Particularly, SSR markers are very reliable and suitable for the study of genetic diversity between species of the same genus because of their transferability and power to detect closely related polymorphic individuals. It has been reported that, in general, cross-species transferability within genera is moderate to high (50-100% success) [32,33]. In this work, we used seven SSRs developed in B. tectorum [25] to discriminate five Bromus species. All SSR markers were polymorphic and transferable to the five species. Although we could not genetically distinguish R and S populations of B. rubens, the seven SSRs were useful tools for discriminating between the Bromus species. Thus, the dendrogram constructed in this study revealed that the five Bromus species are genetically distinct from each other and the 20 populations collected in Andalusia belong to B. rubens.
Glyphosate field doses recommended for the control of weeds in Southern Spain under field conditions (1080 g ae ha −1 ) can control S (Br1, Br3, and Br11) B. rubens populations (LD 50 between 229.87 to 563.46 g ae ha −1 ). However, R populations required greater field doses than those used by farmers (Table 3). These results are supported by the definition of Herbicide Resistance Action Committee (HRAC), which defines tolerant and/or resistant plants as those that survive higher doses of glyphosate than those usually used by farmers [13]. However, this definition is very subjective since there are countries where glyphosate doses are lower than in others; therefore, a weed may be resistant in one country but not in another [15,29,34,35,37,39] (Figure 3). The high RF and low accumulation of shikimic acid observed in the different R B. rubens are in agreement with those plants that have acquired resistance to the addition of more than one resistance mechanism, which could be a non-target site (NTSR) and/or mechanism of resistance to the target site (TSR). This scenario has been demonstrated in other grass weed species [40][41][42]. Our results conclude that RF values also separated the 20 populations of B. rubens into one group -S and another much larger group (17 populations) of glyphosate-R ( Figure 3). Therefore, we can observe that all R populations meet the requirements of RF values greater than 4 to be considered resistant [29,35].
Overall, this study revealed different levels of G-R in B. rubens harvested from different agricultural areas in Southern Spain, where there are a variety of soils and climatic differences. The proposed response of herbicides between different places depends on local ecological factors, such as a variation in climate, soil type, tillage practices, types of crops, and fertilizers, among others. [34,43,44]. Additionally, the use of different glyphosate formulations and the dose rate, application time per year, application technique used by farmers, and environmental conditions could respond to the differences found [45,46]. In addition, an increase in relative humidity and temperature increases the absorption, translocation, and effectiveness of glyphosate in many species of grass weeds, which could help us understand the differences between populations of B. rubens [47,48].
Only in some cases did differences in plant architecture or total leaf surface area contribute to a plant's sensitivity to glyphosate, as a change in the fitness of R versus S plants can alter the growth of R plants to reduce glyphosate retention [49][50][51]. Our results determined that the R and S populations of B. rubens did not exhibit differences in fitness and herbicide retention. In addition, glyphosate retention was similar to that found in other grass weeds, such as Hordeum murinum [29], among others.
When a weed begins to predominate due to a lack of control, it is necessary to carry out a study of alternative herbicides that will help in short-and medium-term management in the field. The study must include pre-emergent or postemergence application alternatives [52]. We found that for pre-emergent applications, the best results were obtained when mixing glyphosate and flazasulfuron ( Figure 5C); it is obvious that the control of B. rubens offered by this mixture is attributed to flazasulfuron (an ALS herbicide). Similar results were obtained by Reeves and Hoyle [53], where the application of flazasulfuron resulted in acceptable controls against Poa anua up to 133 DAA. The application of postemergence herbicides used alone or in combination with pre-emergence herbicides is very frequent in plantations, as they help to carry out fewer applications per year. As for postemergence applications, in this work, the best results were obtained with the mixture of glyphosate plus propaquizafop or quizalofop ( Figure 5D). ACCase herbicides have multiple advantages of being applied in postemergence. However, they are specific to grass weeds, such as B. rubens, and have excellent selectivity in crops [54]. Other studies in B. tectorum and B. japonicus found that glyphosate is effective in reducing biomass [55][56][57].
There is little information related to the application of graminicides for the control of Bromus sp. [58,59]. Ball et al. [58] found better efficacy of quizalofop and fluazifop than sethoxydim. Our results are promising for the benefit of farmers, both in pre-and postemergence applications. The key to success in weed control is to alternate modes of action, use the recommended dose and apply it in a suitable phenological state [60]. These tools must be used correctly to preserve their efficacy. In addition, farmers must learn weed management lessons and use other nonchemical measures that can help decrease the seed bank and seedling density in the field.

Conclusions
The results confirmed the resistance of B. rubens populations to glyphosate collected in Southern Spain. This research is the first scientific report with established resistance parameters of G-R in B. rubens from Spain. Furthermore, field trials demonstrated that, at the moment, there are alternative herbicides to control these R populations. Flazasulfuron (pre-emergence herbicide) in the tank mix with glyphosate (postemergence herbicide), propaquizafop, or quizalofop (postemergence herbicides) plus glyphosate increase the control of B. rubens. Nowadays, we are aware of that fact and effective research is in progress to characterize resistance mechanisms NTSR or TSR involving these populations.