Introduction
Hairy fleabane and horseweed are annual broadleaf weed species present in diverse cropping systems including row crops, perennial crops, and noncropped areas. Their prolific seed production, long-distance seed dispersal, plasticity in seed germination requirements, and tolerance to harsh environmental conditions (Bajwa et al. Reference Bajwa, Sadia, Ali, Jabran, Peerzada and Chauhan2016) contribute to the invasive nature of these weeds. Conyza spp. are pervasive in production fields, and their management becomes more difficult with the evolution of herbicide-resistant biotypes. Reported cases of herbicide-resistant Conyza spp. are increasing, and they are present in multiple countries (Bajwa et al. Reference Bajwa, Sadia, Ali, Jabran, Peerzada and Chauhan2016). As of 2020, there are 20 unique cases of herbicide-resistant hairy fleabane to four different herbicide sites of action (SOAs), and 65 unique cases of herbicide-resistant horseweed to five SOAs (Heap Reference Heap2020). According to the International Herbicide-Resistant Weed Database, the herbicides with greatest reported resistance cases in these two Conyza spp. are glyphosate, an inhibitor of 5-enolpyurvylshikimate-3-phosphate (EPSP) synthase, and paraquat, a photosystem I (PSI) electron diverter. There are instances of multiple resistance to glyphosate and paraquat in both species (Heap Reference Heap2020).
The current understanding of the mechanism of the glyphosate resistance and glyphosate/paraquat resistance in Conyza spp. in the western United States is limited. Reduced glyphosate translocation is often reported in GR Conyza spp., and although the molecular aspects of the non–target site resistance mechanism remain elusive, it is believed to be related to herbicide sequestration within the vacuole (Gaines et al. Reference Gaines, Patterson and Neve2019). Similarly, paraquat resistance is often associated with reduced translocation and vacuole sequestration (Hawkes Reference Hawkes2014). A previous study indicated reduced translocation in both glyphosate and paraquat resistance in both species in orchard production systems of California (Moretti and Hanson Reference Hanson2017). Additional mechanisms of resistance in these biotypes are possible. Non–target site resistance mechanisms may confer cross-resistance to herbicides with distinct SOAs (Han et al. Reference Han, Yu, Beffa, González, Maiwald, Wang and Powles2020; Iwakami et al. Reference Iwakami, Kamidate, Yamaguchi, Ishizaka, Endo, Suda, Nagai, Sunohara, Toki and Uchino2019), and in some cases, to reactive oxygen species (ROS) generators (Ye and Gressel Reference Ye and Gressel2000).
In California, herbicide resistance, mainly glyphosate resistance, affects production practices in multiple cropping systems (Hanson et al. Reference Hanson, Wright, Sosnoskie, Fischer, Jasieniuk, Roncoroni, Hembree, Orloff, Shrestha and Al-Khatib2014). Glyphosate is the most used pesticide in California, whereas paraquat is the third most used herbicide in the state (CDFA 2017). GR Conyza spp. are widespread across the state and present in nearly all tested populations from tree nut crops; GPR cases are not known to be widespread but appear to be increasing (Moretti et al. Reference Moretti, Sosnoskie, Shrestha, Wright, Hembree, Jasieniuk and Hanson2016). Worldwide, herbicide mixtures and rotations are often the first strategies to manage resistant biotypes (Peterson et al. Reference Peterson, Collavo, Ovejero, Shivrain and Walsh2018). The success of herbicide tank mixtures and rotations are reported by several studies evaluating GR Conyza spp. (Eubank et al. Reference Eubank, Poston, Nandula, Koger, Shaw and Reynolds2008; Tahmasebi et al. Reference Tahmasebi, Alebrahim, Roldán-Gómez, da Silveira, de Carvalho, Alcántara-de la Cruz and De Prado2018; Urbano et al. Reference Urbano, Borrego, Torres, Leon, Jimenez, Dinelli and Barnes2007; Werth et al. Reference Werth, Walker, Boucher and Robinson2010), a strategy also successful for GPR Conyza spp. biotypes (Eubank et al. Reference Eubank, Nandula, Poston and Shaw2012; Moretti et al. Reference Moretti, Shrestha, Hembree and Hanson2015). The pillars of this strategy are based on the diversity of SOAs with efficacy on the targeted resistant biotypes. This premise can be jeopardized by herbicide cross or multiple resistance. POST control strategies for GR Conyza spp. in California orchards crops often depend on glufosinate, various inhibitors of protoporphyrinogen oxidase (PPO), and to a lesser extent 2,4-D (Hanson Reference Hanson2020); however, previous field research has suggested poor or inconsistent control of multiple-resistant hairy fleabane with 2,4-D alone or in combination with glyphosate (Moretti et al. Reference Moretti, Shrestha, Hembree and Hanson2015). There have been no reports on the response of GR and GPR Conyza spp. to different SOAs or characterization of cross-resistance patterns. This study has examined the presence of resistance to other herbicide SOAs in GR, GPR, and susceptible biotypes of hairy fleabane and horseweed.
Materials and Methods
Plant Material
The experiments were conducted at the University of California–Davis greenhouse facilities in Davis, CA (38.54 N, 121.76 W). The hairy fleabane and horseweed biotypes used in this study were the descendants of single seeds and were self-pollinated for five or more generations to ensure a uniform phenotypic response. The biotypes were characterized as glyphosate/paraquat–susceptible (GPS), glyphosate-resistant (GR), and glyphosate/paraquat–resistant (GPR) (Moretti et al. Reference Moretti, Sosnoskie, Shrestha, Wright, Hembree, Jasieniuk and Hanson2016).
Plant Growth and Herbicide Application
Seeds of each biotype were sown in flats filled with a commercial potting medium (Sungro Horticulture, Agawam, MA, USA) and grown under natural-light conditions. Seedlings at the first-leaf stage were transplanted as one seedling per 7.5- by 7.5- by 10-cm pot filled with the same potting medium. Plants were maintained in the greenhouses at 30/15 C day/night temperatures under natural-light conditions. Irrigation and fertilization were provided as needed to promote vigorous growth. The experiments were initiated when plants were at the 5- to 6-leaf stage for hairy fleabane and the 8- to the 10-leaf stage for horseweed. All biotypes of both species were tested simultaneously during each herbicide study.
Whole-plant dose–response assays were conducted from April through October 2015. Nine herbicides representing seven distinct SOAs were selected for the study (Table 1). The active ingredient and respective WSSA SOA groups were rimsulfuron in Group 2 (acetolactate synthase inhibitors); 2,4-D and dicamba (Group 4–synthetic auxins); hexazinone (Group 5–an inhibitor of photosystem at PSII site A); glufosinate (Group 10–an inhibitor of glutamine synthetase); flumioxazin and saflufenacil (Group 14–inhibitors of PPO); diquat (Group 22–PSI electron diverter); and mesotrione (Group 27–an inhibitor of 4-hydroxyphenylpyruvate dioxygenase; 4-HPPD). Each herbicide was tested at eight rates plus a nontreated control. Adjuvants were selected based on manufacturer label recommendations (Table 1). Treatments were applied to plant foliage using a spray chamber (Technical Machinery Inc., Sacramento, CA) calibrated to deliver 187 L ha–1 at 207 kPa. The chamber was equipped with an even flat-fan nozzle 8002E (TeeJet Technologies, Wheaton, IL) placed 45 cm above the canopy. Plants were evaluated 28 d after application as alive or dead, and aboveground biomass was collected, dried, and recorded.
Table 1. POST herbicides used in the experiments. Site of action, active ingredient, trade name, rates, reference rate, and manufacturers.
a Weed Science Society of America site of action (SOA) group number (https://wssa.net/wssa/weed/herbicides): 2, Acetolactate synthase inhibitor; 4, synthetic auxin; 5, inhibitor of photosynthesis at photosystem II; 10, glutamine synthase inhibitor; 14, protophorphyrinogen oxidase inhibitor; 22, photosystem I electron diversion; 27, inhibition of the hydroxyphenylpyruvate dioxygenase.
b All treatments included ammonium sulfate at equivalent of 2% wt/vol. Nonionic surfactant (Rainier; Wilbur Ellis, Aurora, CO) 0.25% vol/vol was included in rimsulfuron, hexazinone, 2,4-D, dicamba, diquat, and flumioxazin treatments. Methylated seed oil (Mor-act Crop Oil; Wilbur Ellis, Aurora, CO) was included at 1% vol/vol in saflufenacil and mesotrione treatments.
Statistical Analysis
The experiments were organized in a two-factor randomized complete block design. The three biotypes (GPS, GR, and GPR) were the first factor, and the nine herbicide rates were the second factor. There were four replicates per biotype by treatment level, and the experiment was conducted twice. Each species and herbicide was analyzed independently. Data were analyzed in R version 4.0.3 (R Core Team 2020). Plant mortality was analyzed using logistic regression with a mixed-effect model using the glm function from package stats version 3.6.2. Experimental blocks and experimental runs were treated as random factors. Herbicide rate, biotype, and interaction were tested using a Wald test (P < 0.05) (Table 2). Mortality data were analyzed using nonlinear logistic regression (drc package version 3.01) with two parameters (Ritz et al. Reference Ritz, Baty, Streibig and Gerhard2015). The parameters were the herbicide dose required to kill 50% of the plants (LD50) and the slope of the regression near the LD50 parameter. The ratio of LD50 for each biotype was used to compare biotypes by a z-test (P < 0.05) using the compParm procedure in the drc package. The resistance index (RI) was calculated as the ratio of LD50 values relative to the GPS biotype of each species. When the regression models were fitted for each biotype (full model), the model was compared to a simplified model using a common regression for all biotypes. Models were compared using a chi-square test with the ANOVA command in the drc package. The reduced model was used when no statistical difference in LD50 was reported or when the chi-test result was not significant. Plant biomass data were analyzed using a mixed model with the lmer function in the lme4 package version 1.123. The herbicide rate, biotype, and their interactions were tested as fixed factors (Table 2). Data were submitted to a nonlinear regression using the drc package. Multiple models were tested and compared using a Akaike information criterion to decide the best fit. Diagnostic plots were used to check normality and heteroskedasticity assumptions. The three-parameter log-logistic model was appropriate in all cases (Equation 1).
$${Y = {{d} \over \frac {{1 + {{\left( x \over {\frac{{{\rm{E}}{{\rm{D}}_{50}}}}} \right)}^{\rm{b}}}}}}}$$
Table 2. Fixed factors for logistic and ANOVA analysis for mortality and biomass of Conyza spp.
Abbreviations: R, rate; B, biomass; significance levels based on Wald Test for mortality and ANOVA for biomass: NS, not significant (>0.05); *, P < 0.05.
Where Y was the response measured, d references the upper limit, x refers to the herbicide rate, ED50 denotes the herbicide dose causing 50% reduction in the response measured, and b is the relative slope of the curve around the ED50. The three-parameter log-logistic model was with slope of the regression, upper as the maximum value of the regression, and GR50 as the herbicide rate required to reduce biomass accumulation by 50%. Biotype effects were also tested using compParm function and fitting a regression to each biotype (full model) compared to a common regression (reduced model). The reduced model was used when the models were not different.
Results and Discussion
All tested herbicides affected Conyza spp. mortality and biomass, but biotype response depended on the herbicide tested, and it differed between species (Table 2). The hairy fleabane mortality response to hexazinone and diquat depended on biotype tested, whereas the interaction of rate and biotype was significant for 2,4-D and flumioxazin. Biomass analysis also indicated a biotype and rate-by-biotype effect for 2,4-D and diquat, but not for the other herbicides. The mortality of horseweed differed among biotypes for flumioxazin, saflufenacil, diquat, and mesotrione. When considering biomass, the biotype effect was only significant for saflufenacil and diquat. A common regression model for the three biotypes was fitted for rimsulfuron and dicamba for both species. For the herbicides in which biotype or biotype-by-rate interactions were significant, a regression model was fitted for each biotype (full model). The comparison of LD50 or GR50 ratio indicated no differences among biotypes for hexazinone, glufosinate, flumioxazin, saflufenacil, and mesotrione. The full model did not differ from the reduced model (P < 0.05), so a single regression model for all three biotypes was fitted.
All the Conyza spp. biotypes tested were sensitive to the acetolactate synthase inhibitor (ALS) herbicide rimsulfuron. The rimsulfuron rates that caused 50% mortality (LD50) or reduction of biomass (GR50) of hairy fleabane were 78 and 1.1 g ai ha–1, respectively (Table 3). Similar values were observed for horseweed, and in all cases, the values were close to the reference 1× rate of 70 g ai ha–1. These results are comparable to previously reported results regarding the response of susceptible horseweed to chlorimuron, a sulfonylurea herbicide, with a GR50 of 0.1 g ai ha–1 (Zheng et al. Reference Zheng, Kruger, Singh, Davis, Tranel, Weller and Johnson2011). The response to the auxinic herbicide dicamba was similar among tested biotypes of both Conyza spp. The hairy fleabane LD50 was 879.9 g ae ha–1, and the GR50 was 72.6 g ae ha–1 of dicamba. Horseweed was more sensitive to dicamba than hairy fleabane, with an LD50 2.5-fold lower and a GR50 7.8 times lower (Table 3). The dicamba GR50 of both Conyza species is comparable to the findings of Zheng et al. (Reference Zheng, Kruger, Singh, Davis, Tranel, Weller and Johnson2011). Horseweed populations from Indiana were reported to have a GR50 ranging from 31 to 127 g ae ha–1 depending on the developmental stage of the plants (Kruger et al. Reference Kruger, Davis, Weller and Johnson2010; McCauley and Young Reference McCauley and Young2019).
Table 3. Regression parameter estimates and standard errors for plant mortality and biomass of hairy fleabane and Conyza spp. 28 d after foliar treatment with different herbicides. A common regression logistic regression was fitted across biotypes within each species and herbicide tested.a
a Abbreviations: LD50, effective dose killing 50% of the population; slope, relative slope around the LD50 or GR50; upper, upper limit of the response; GR50, effective doses reducing shoot dry biomass by 50%.
b Means are pooled across two experiments and three biotypes (n = 24) for hairy fleabane and horseweed. Standard errors are in parentheses.
The paraquat resistance mechanism of hairy fleabane was previously attributed to enhanced enzymatic detoxification of ROS conferring cross-resistance to oxidant stress (Shaaltiel et al. Reference Shaaltiel, Glazer, Bocion and Gressel1988; Shaaltiel and Gressel Reference Shaaltiel and Gressel1986). Several herbicides work by generating ROS upon light exposure, including PSII inhibitors, glutamine synthetase inhibitors, PPO inhibitors, PSI electron diverters, and inhibitors of 4-hydroxyphenylpyruvate dioxygenase (HRAC 2020). Hexazinone, a PSII inhibitor, is not used in tree nuts but is commonly used in small fruit crops like blueberry. Other chemicals with the same SOA most widely used in tree nuts are diuron and simazine. Hexazinone was effective in all biotypes of Conyza spp. The LD50 ranged from 27 to 62 g ai ha–1, whereas the reference rate is 1,320 g ai ha–1. There was no difference among tested biotypes for glufosinate, a glutamine synthetase inhibitor and ROS producer (Takano et al. Reference Takano, Beffa, Preston, Westra and Dayan2019). The LD50 for glufosinate was 574 g ai ha–1 or less, and the GR50 was 135 g ai ha–1; both parameters were below the reference rate of 980 g ai ha–1. Glufosinate is commonly used in tree nut crops, and resistance has been reported in Italian ryegrass [Lolium perenne L ssp. multiflorum (Lam.) Husnot] (Brunharo et al. Reference Brunharo, Takano, Mallory-Smith, Dayan and Hanson2019). Conyza biotypes responded similarly to flumioxazin and saflufenacil, the PPO herbicides tested. Flumioxazin LD50 was 1,473 and 1,326 g ai ha–1 for hairy fleabane and horseweed, respectively, values over three times greater than the reference rates for tree nut crops. Flumioxazin is used primarily for its PRE activity in Conyza spp. Flumioxazin in POST reduces Conyza biomass, as indicated by the GR50 lower than 153 g ai ha–1, but it often does not control the plants. Similar to these findings, flumioxazin applied POST did not control Conyza spp. populations from Europe (Tahmasebi et al. Reference Tahmasebi, Alebrahim, Roldán-Gómez, da Silveira, de Carvalho, Alcántara-de la Cruz and De Prado2018). In contrast, saflufenacil has an excellent POST activity on Conyza spp. with an LD50 of 2.2 g ai ha–1, and the GR50 was lower than 0.7 g ai ha–1. Saflufenacil is often used to manage GR Conyza spp. in tree nuts, and it has been shown to control GR horseweed in other cropping systems (Budd et al. Reference Budd, Soltani, Robinson, Hooker, Miller and Sikkema2016; Mellendorf et al. Reference Mellendorf, Young, Matthews and Young2013). Response to mesotrione did not differ among the tested biotypes of Conyza spp. (Table 3). The mesotrione doses to cause 50% mortality were 117.6 and 388.5 g ai ha–1 for hairy fleabane and horseweed, respectively. In general, more Conyza spp. survived mesotrione treatment relative to the other products evaluated in this research. This herbicide is often recommended as a tank-mixture partner to provide adequate control of horseweed (Armel et al. Reference Armel, Richardson, Wilson and Hines2009).
The Conyza spp. response to 2,4-D was species- and biotype-dependent (Table 2). The 2,4-D doses required to kill 50% of the hairy fleabane biotypes GR and GPR were 3.9 and 6.8 times greater than the GPS biotype, with an LD50 of 1,338.1 g ai ha–1 (Table 4 and Supplementary Figure S1). The biomass data also indicate that hairy fleabane GR and GPR are resistant to 2,4-D, with an RI of 2.7 and 2.9 (Table 4). The LD50 for horseweed GR and GPR biotypes showed an RI of 3.0 and 2.9, but biomass data indicate that the biotypes responded similarly, with a RI of 1.3 and 1.8, respectively (Table 4). The estimated 2,4-D GR50 of horseweed from California is comparable to levels reported elsewhere, such as in Indiana, where reported GR50 values ranged from 131.6 to 314.1 g ai ha–1 of 2,4-D (Kruger et al. Reference Kruger, Davis, Weller and Johnson2010; McCauley and Young Reference McCauley and Young2019). These data confirm the first case of 2,4-D–resistant hairy fleabane in California. Although no attempt was made to study the mechanism of resistance to 2,4-D, none of the plants exhibited the rapid-necrosis phenotype reported in 2,4-D–resistant Sumatran fleabane [C. sumatrensis (Retz.) E. Walter] from Brazil (de Queiroz et al. Reference de Queiroz, Delatorre, Lucio, Rossi, Zobiole and Merotto2020). It is important to note that the GR50 level of the GPS population was 766.7 g ai ha–1, which approaches the reference rate. For comparisons, hairy fleabane populations from Australia did not survive 2,4-D treatment of 875 g ae ha–1(Aves et al. Reference Aves, Broster, Weston, Gill and Preston2020). Previous studies have documented the low efficacy of 2,4-D in hairy fleabane from California (Moretti et al. Reference Moretti, Shrestha, Hembree and Hanson2015). Hairy fleabane biotypes resistant to 2,4-D curtail the benefits of 2,4-D use in the tree nut crops, including the potential use of low-volatility formulations like the 2,4-D choline salt (Peterson et al. Reference Peterson, McMaster, Riechers, Skelton and Stahlman2016). However, the absence of cross-resistance to dicamba suggests that other auxinic herbicides, such as halauxifen-methyl, could be effective against this biotype. Halauxifen-methyl has been shown to be effective against GR horseweed (McCauley et al. Reference McCauley, Johnson and Young2018).
Table 4. Estimated nonlinear regression parameters for hairy fleabane and horseweed biotypes from California in response to 2,4-D and diquat. Mortality and dry-biomass whole-plant dose–response assay were estimated for glyphosate/paraquat-susceptible (GPS), glyphosate-resistant (GR), and glyphosate/paraquat-resistant (GPR) biotypes of each species.a
a Abbreviations: LD50, effective dose killing 50% of the population; slope,relative slope around the LD50 or GR50; upper, upper limit of the response; GR50, effective doses reducing shoot dry biomass by 50%; RI, resistance index relative to susceptible biotype.
b Means are pooled across two experiments and three biotypes (n = 24) for hairy fleabane and horseweed. Standard errors are in parentheses. * Significance level based on z-test at P < 0.05.
The GPR biotypes of both Conyza spp. showed cross-resistance to diquat, whereas the GR biotypes were as sensitive to diquat as the GPS. Compared to the GPS biotypes within species, the hairy fleabane GPR LD50 was 4,961.8 g ai ha–1, with a RI of 90.2, and GR50 was 863.7 g ai ha–1, with a RI of 44 (Table 4 and Supplementary Figure S2). In horseweed, the LD50 of the GPR biotype was 1,791.2 g ai ha–1 (RI 11.3) and the GR50 563.1 g ai ha–1 (RI 15.6). Cross-resistance between PSI herbicides is commonly observed in paraquat-resistant weeds (Hawkes Reference Hawkes2014). A previous study reported that paraquat-resistant hairy fleabane exhibited a cross-resistance to diquat 10-fold lower than to paraquat (Vaughn et al. Reference Vaughn, Vaughan and Camilleri1989). In this study, both GPR biotypes were more tolerant to paraquat than diquat. The paraquat GR50 (RI) for these GPR biotypes were previously reported as 1,161 (RI 278) and 1,390 g ai ha–1 (RI 322) for the hairy fleabane and horseweed, respectively (Moretti et al. Reference Moretti, Sosnoskie, Shrestha, Wright, Hembree, Jasieniuk and Hanson2016). These RI values are 6- and 20-fold higher than the observed RI based on the diquat GR50 in this study. An absence of cross-resistance to various ROS-generating herbicides in GPR biotypes was observed in this study, suggesting that enhanced enzymatic detoxification of ROS is not associated with paraquat resistance in these biotypes. These findings agree with the previous report of a paraquat-resistant hairy fleabane biotype that showed no cross-resistance to ROS (Vaughn et al. Reference Vaughn, Vaughan and Camilleri1989).
The confirmation of multiple resistance to glyphosate, paraquat, and 2,4-D in hairy fleabane indicates that herbicide-based strategies to manage herbicide resistance are only a short-term approach. Increasing cases of cross- and multiple-herbicide resistance will curtail the SOA availability, the pillar for its success, jeopardizing this resistance management strategy. This finding underscores the critical need for nonchemical weed control alternatives in perennial crops.
Acknowledgments
This work was supported by the Almond Board of California and the California Walnut Board. No conflicts of interest have been declared.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/wet.2021.11
