Conyza bonariensis (L.) Cronq. Walker is a weed annual broad-leaved species in the Asteraceae family native to South America [1] , present in fallow in autumn and spring in Argentina. Previous to the widespread adoption of no-till systems, this species was confined to pastures and undisturbed areas [2] . No-till favours germination and emergence of small-seeded weed species including Conyza spp. [3] -[5] which typically emerges during the whole fallow period [6] -[8] . In the last years, applications of glyphosate in no-till fallow have increased significantly achieving initially very good control of Conyza spp. in the region of the study [9] . At present, control of some populations is difficult in many sites of the Argentine soybean region, although no glyphosate resistant Conyza spp. biotypes have been reported. The selection of these tolerant biotypes may be due to suboptimal growth stage at application [10] , overreliance on glyphosate and long-term use of sublethal doses [11] . Resistance to glyphosate of C. bonariensis was observed for the first time in 2003 in South Africa [12] . At present, resistance to five modes of action with Conyza canadensis (L.) Cronq. and four modes of action C. bonariensis have been reported [12] . Resistance biotypes were identified in Europe [13] -[15] , Brazil [16] and USA [17] . In Argentina, a better understanding of the differences in control between biotypes is essential for developing effective long-term management strategies.

The small size of weed plants at herbicide application is relevant for achieving an acceptable control as at this stage, weeds are more susceptible [18] -[21] . Glyphosate efficacy is particularly influenced by weed size [22] [23] . In some biotypes of Conyza spp., glyphosate control fails due to an inappropriate dose when there is a delay in application that allows plants to reach sizes that are difficult to kill consistently [11] .

In no-tillage fallow, control of annual weeds increases by applying residual herbicides combined with glyp- hosate [5] [24] [25] . Many of the residual herbicides for Conyza spp. control are acetolactate synthase (ALS)- inhibiting herbicides, particularly chlorimuron-ethyl, metsulfuron-methyl and diclosulam [26] [27] . Mixtures of such herbicides with glyphosate may control biotypes resistant to glyphosate [28] .

Other herbicides included in this group consist of mixtures containing sulfometuron or chlorsulfuron and are less studied, so the dose for adequate control is not clear.

The aim of this study was to compare control of three stages of C. bonariensis biotypes susceptible and tolerant to glyphosate with glyphosate and mixtures of acetolactate synthase (ALS)-inhibiting herbicides.

Experiments were conducted at the University of Rosario Experimental Farm at Zavalla (Lat. 33˚01'S Long. 60˚53'W), Argentina.

A correlation of r = 0.99; P ≤ 0.01 was obtaineb between C. bonariensis control at 30 DAT using visual evaluations and plant weight determinations. Therefore, only visual evaluations are presented. No significant interactions were detected in the analysis between years for any of the variables so data were pooled over the experiments for each biotype. A significant interaction occurred between herbicides, dose and growth stage for both biotypes. All herbicides achieved 100% of control with the 8X dose or lower, for rosette and bolting stages. At the reproductive stage, control was lower than 100% even with the 8X dose and thus this stage was not considered to build the dose-response curves.

For glyphosate with the recommended dose (1X), control at the rosette stage was 100% for biotype S and 80% for biotype T (Figure 1). At bolting, control was 100% for biotype S and 77% for biotype T (Figure 1). Injury symptoms were characterized by leaf chlorosis followed by necrosis of the tissue.

For the S biotype, the I₅₀ relative to the recommended rate was 0.08X at the rosette stage and 0.17X for the bolting stage, whilst for the T biotype, I₅₀ corresponded to 0.28X at the rosette stage and 0.80X for the bolting stage.

Resistance Index (RI) is defined as the ratio of I₅₀ values relative to that of a susceptible population or recommendation dosage. Resistant index at the rosette stage was approximately 4 for both weed stages.

I₉₀, relative to the recommended dose, was 0.18X for the S biotype at the rosette stage and 1.33X for the bolting stage, whilst for the T biotype, was 1.39X for the rosette stage and 11.8X for the bolting stage. This last value was fitted by the model, even though plants were 100% controlled with 8X (Table 1).

Dose-response curves for acetolactate synthase (ALS)-inhibiting herbicides are shown in Figures 2-4.

No differences were observed in the parameters between the T and S biotypes except for metsulfuron-

methyl + chlorsulfuron which showed higher injury of the biotype T relative to biotype S at low doses for the rosette stage (Table 2).

At the rosette stage, for all the herbicides of this group, I₅₀ ranged between 0.02 and 0.14X.

For each weed stage and for each herbicide, no significant differences for I₅₀ between S and T biotypes were observed. At rosette stage, values of I₅₀ were in average for both biotypes 1.7 (0.13X) for diclosulam, 0.5 (0.01X) for metsulfuron-methyl + chlorsulfuron and 2.1 (0.19X) for sulfometuron-methyl + clorimuron-ethyl. At bolting, 7.3 (0.58X) for diclosulam, (0.15X) for metsulfuron-methyl + chlorsulfuron and (0.52X) for sulfometuron-

methyl + clorimuron-ethyl. Injury sympthoms were characterized by stunting, purpling, general yellowing of new growth, terminal bud death.

I₉₀ relative to the recommended dose for the rosette stage for diclosulam was 0.25X for the S and T biotype, for metsulfuron-methyl + chlorsulfuron was 0.22X for S biotype and 0.15 for the T biotype, for sulfometuron- methyl + clorimuron-ethyl was 0.21X and 0.18X. At bolting for diclosulam was 60X for the S and 2X for the T biotype, for metsulfuron-methyl + chlorsulfuron was 0.41X for S biotype and 6.70 for the T biotype, for sulfometuron-methyl + clorimuron-ethyl was 0.26 for S biotype and 0.12X for T biotype. The value of 60X for diclosulam was fitted by the model, even though plants were 100% controlled with 8X.

At the reproductive stage plants were well developed and lignified at application time and control was poor for all the herbicides and doses. All these plants survived, flowered and produced abundant seeds (data not shown).

Results of the experiments provided evidence that C. bonariensis at the rosette stage the biotype S is easily controlled by both, glyphosate and the acetolactate synthase (ALS)-inhibiting herbicides studies whilst the biotype T, showed low response to glyphosate although it is sensitive to the ALS inhibiting herbicides.

The dose-response curves confirmed that injury of the biotype T relative to biotype S was significantly lower for both growth stages which concurs with other studies in C. canadensis [7] .

Plant stage at application time is a key factor which greatly affects glyphosate efficacy and weed control is higher at the first stages of weed development [31] . In C. bonariensis, in the present study, the glyphosate dose required for control increased as a function of plant age as was also observed with other studies in this species [13] and in C. canadensis [23] .

Resistance factors for glyphosate were similar or lower (approximately 4 for both weed stages) relative to other studies in C. bonariensis which showed resistance factors about 7 and 9 [13] or ranged between 3 - 9 [17] .

In those studies, resistance has arisen in orchards or vineyards after multiple applications of glyphosate as the only herbicide. In the soybean region in Argentina, glyphosate is mostly applied in fallows and in soybeans one or two times per year, which may account for the lower resistance factor of C. bonariensis. At bolting, acceptable (>90%) control, for the S biotype and the T biotype at both stages requires doses much higher than the recommended use dose, thus, application of glyphosate at these levels is not economical.

Acetolactate synthase (ALS)-inhibiting herbicides, at the rosette stage, provided control of the biotype S of C. bonariensis at doses significantly lower than the recommended rate. In other studies, acetolactate synthase (ALS)-inhibiting herbicides were also used in combination with glyphosate in fallows previous to soybean planting achieving adequate control of small size plants of Conyza spp. [4] . Herbicides of this group provide residual control and may help to reduce new seedling flushes not controlled by glyphosate alone [5] .

At the reproductive stage, control with glyphosate and acetolactate synthase (ALS)-inhibiting herbicides was very low, thus, a new application of a postemergence herbicide with alternate mode of action is needed (“double knock” technique) [32] [33] .

From these results it is clear that plants from C. bonariensis populations tolerant to control with glyphosate may be controlled using acetolactate synthase (ALS)-inhibiting herbicides. Among the herbicides within this group, diclosulam at the rosette stage showed the lowest ED₅₀ relative to the recommended rate and showed complete control at 0.5X. In another study, a biotype of C. bonariensis resistant to glyphosate was also sensitive to metsulfuron-methyl and clorimuron-ethyl, both acetolactate synthase (ALS)-inhibiting herbicides [16] . However, [28] and [31] found that C. canadensis exhibited multiple resistance to glyphosate and acetolactate synthase (ALS)-inhibiting herbicides such as cloransulam indicating that, it is necessary to adopt an adequate management of chemical control to avoid resistance to this mode of action.