Introduction

Brazil is the fifth largest producer of cotton in the world (^{FAO, 2012}); technology and high productivity have allowed Brazil to shift from the largest importer to the third largest exporter of cotton in 12 years. In a ten-year span, Brazilian cotton cultivation was transformed from manual to total mechanization. For 2013/2014, the area under cotton cultivation in Brazil was 1,076,900 ha with a production of 1 638 100 Mg (^{CONAB, 2013}).

According to ^{Christoffoleti (2002)}, the utilization of direct-sprayed selective herbicides during initial post-emergence, along with herbicides in pre-emergence and in late post-emergence, is an important weed management system for cotton in central Brazil. Glufosinate ammonium (GA) effectively controls weeds in cotton crops and, in Brazil, it is applied during post-emergence by direct spray on areas planted with cotton genetically modified to be resistant to this herbicide (^{Rodrigues and Almeida, 2011}). Glufosinate ammonium is a synthetic version of phosphinothricin, a degradation product of bialaphos produced by Streptomyces viridochromogenes and S. hygroscopicus (^{Duke et al., 2000}), whose antibiotic and herbicidal properties were discovered by ^{Droge-Laser et al. (1994)}. This led to the synthesis of GA as herbicide (^{Manderscheid and Wild, 1986}) for use in several crops, different production systems and in non-agricultural areas (^{Krausz et al., 1999}). It is a non-selective herbicide that controls a broad spectrum of weeds (^{Chompoo and Pornprom, 2008}).

Glufosinate ammonium inhibits the activity of glutamine synthetase (EC 6.3.1.2), which converts glutamate and ammonium into glutamine (^{Logusch et al., 1991}) and inorganic nitrogen into organic compounds. It is a key enzyme in the metabolism of nitrogen assimilating ammonia produced by nitrite reductase, and recycling ammonia produced by photorespiration and deamination reactions (^{Miflin and Habash, 2002}). The inhibition of GA activity leads to a rapid accumulation of ammonia, which causes cell destruction (^{Senseman, 2007}). Therefore, an increase in ammonium is used as an indicator of the performance of glufosinate (^{Pornprom et al., 2000}; ^{Petersen and Hurle, 2001}).

The mobility of GA is low in plants and it requires good coverage at the time of application for a broad spectrum of weed control (^{Corbett et al., 2004}; ^{Steckel et al., 1997}). Since GA is not promptly translocated within the plant, the symptoms occur principally in the leaves (^{Davis et al., 2013}), which present chlorosis and die two to five days after treatment (^{Shin et al., 2011}). Leaf death occurs due to a sharp elevation in ammonium levels and reduced glutamine levels in foliar tissue within a few hours (^{Chompoo and Pornprom, 2008}).

According to ^{Kumaratilake et al. (2002)}, absorption occurs in the first 24 h after GA treatment in Lolium rigidum and Avena sterilis. A 6 h period without rain after application is needed to obtain an efficient performance of this herbicide (^{Rodrigues and Almeida, 2011}).

Given the importance of GA to control weeds in cotton crops, and the lack of data on the absorption velocity and the intensity of the effects in the metabolism on cotton and important weeds (Brachiaria decumbens and Ipomoea grandifolia), our study aimed to evaluate GA absorption (velocity and quantity) of glufosinate ammonium and its effects on the metabolism of weeds and cotton.

Materials and methods

Three experiments were carried out in a greenhouse under temperatures between 15 and 28 °C, at the São Paulo State University, UNESP, campus at Botucatu/SP/Brazil. Pots (5 L capacity) were filled with a substrate composed of turf, expanded vermiculite and pine bark, enriched with macro and micronutrients, at pH 5.8 (±0.5).

The experimental design was completely randomized with four repetitions; four plants of the cotton cultivar FiberMax 910 were utilized, and ten plants of B. decumbens Stapf and I. grandifolia Dammer per repetitions. The treatments were GA application (0.4 kg a.i. ha^-1) and five periods before a simulated rain (times for absorption: 1, 3, 6, 24 and 48 h between application and simulated rain); plus a control without herbicide.

The herbicide was applied in a stationary sprayer equipped with a spray bar equipped with four Teejet XR 11002VS tips arranged 0.5 m apart and positioned at a height of 0.5 m in relation to the plants. The system was operated at a displacement velocity of 1 m s^-1, with a liquid consumption of 200 L ha^-1 and a constant pressure of 1.5 kgf cm^-2 using compressed air. The rainfall simulation of 40 mm in each treatment was performed using the same structure previously described with an automated system allowing for constant-pressure via an hydraulic pump using a spray boom with three TK20-SS high flow spray nozzles spaced 0.5 m apart at 1.45 m above the experimental unit surface. This structure enabled simulations of rainfall of various volumes. The traveling speed of the system was 3.16 m min^-1, with pressure of 79.4 kPa, enabling the formation of artificial raindrops with a median volumetric diameter of 1140 μm.

At the time of application, the cotton plants presented completely expanded second true leaves (and a third in expansion) at approximate heights of 30 cm, whereas height of B. decumbens was approximately 35 cm, and that of I. grandifolia was 18 cm.

The plants were harvested for laboratory analyses two days after the herbicide application, when the first visual symptoms injury began to appear. In these samples, the levels of ammonia, glutamine and glufosinate were analyzed.

Visual rating of cotton plant injury and weed control was rated at 3, 6, 8 and 10 d after application (DAA) of glufosinate ammonium. Ratings were based on a 0 to 100 % scale with 0 % equal to no plant response and 100 % equal to complete weed control or cotton death (^{Richardson et al., 2007}).

Results and discussion

In cotton plants, an increase of GA level (absorption) was observed up to 24 h without rain (Figure 1). Although the herbicide was absorbed slowly and the highest levels were found in plants after 24 h without rain, the ammonia accumulation occurred independently of herbicide absorption in greater quantities. The absorption until 2 to 5 h without rain, promoted a large accumulation of ammonia in the plants, which shows that the quantity of glufosinate absorbed in 5 h was enough to cause plant injury (Figure 2).

Figure 1 Glufosinate concentration (mg kg^-1 fresh weight) in plants of cotton, B. decumbens and I. grandifolia for different periods without rain at two days after application (DAA). Vertical bars represent ± confidence interval of the mean (n=4) by the t test (p≤0.05). 

Figure 2 Ammonia concentration (mg kg^-1 fresh weight) in plants of cotton, B. decumbens and I. grandifolia for different without rain periods at two days after application (DAA). Vertical bars represent ± confidence interval of the mean (n=4) by the t test (p≤0.05). 

The glufosinate quantity absorbed by I. grandifolia plants was superior to that by cotton plants and rose until 48 h after application without rain. For B. decumbens, the absorption was more rapid and intense in the first 6 h without rain, and there was an increase in the herbicide levels in the leaves until 48 h after application without rain simulation (Figure 1).

^{Everman et al. (2009a)} evaluated the quantity of ¹⁴C-Glufosinate absorbed by four species, (glufosinate-resistant corn, Eleusine indica, Digitaria sanguinalis and Senna obtusifolia), and observed that S. obtusifolia absorbed 80 % of the GA applied, whereas E. indica and D. sanguinalis absorbed less than 22 % and the resistant corn absorbed only 16 %. ^{Everman et al. (2009b)} assessed absorption, translocation, and metabolism of glufosinate in transgenic and nontransgenic cotton, Amaranthus palmeri, and Ipomoea lacunosa, and point out that these species could be divided into two groups based on GA absorption: high (A. palmeri) and low (both nontransgenic and glufosinate-resistant cotton and I. lacunose).

Similarly to the cotton plants, the inside ammonia concentrations of I. grandifolia and B. decumbens stabilized in approximately 2 to 5 h without rain, despite the increase in herbicide absorbed starting from these periods. The higher accumulation of ammonia shows that the quantities of herbicide absorbed in the first hours after application were able to increase levels of this compound in the plants, leading to higher injury levels. Cotton presented the lowest ammonia accumulation, followed by I. grandifolia, and B. decumbens had the highest levels (Figure 2). Ammonium buildup is used as an indicator for glufosinate performance (^{Petersen and Hurle, 2001}). ^{Sellers et al. (2004)} evaluated glutamine synthetase activity and ammonium accumulation as influenced by time of glufosinate application in velvetleaf, and after 160-320 g ha^-1 glufosinate application, glutamine synthetase activity was reduced by at least 50 % and ammonium concentration was as least 13 times higher in treated plants, compared to control plants. This amount of accumulation is comparable to ammonium concentrations that alter chloroplast ultrastructure and inhibit ATP formation (^{Krogmann et al., 1959}; ^{Puritch and Barker, 1967}; ^{Barker, 1968}).

According to ^{Pline et al. (1999)}, the foliar absorption of glufosinate in Solanum carolinense and Asclepias syriaca was fast in the first 12 h after treatment and increased significantly after 12 h. As glufosinate ammonium presents a structure analogous to glutamate, it inhibits glutamine synthetase, which is a key enzyme for controlling nitrogen utilization inside the cells (^{Coetzer and Al-Khatib, 2001}). This results in a rapid accumulation of ammonia content (^{Lea et al., 1984}; ^{Coetzer and Al-Khatib, 2001}) and, subsequently, a change in the amino-acid spectrum in plant cells (^{Lea et al., 1984}; ^{Sauer et al., 1987}). ^{Tsai et al. (2006)} reported drastic accumulation of ammonia, a biochemical marker of the inhibition of glutamine synthetase by glufosinate, in two rice lineages at 9 h after treatment with 0.26 mM glufosinate, and accumulation continued until 24 h after the treatment.

Plant cells avoid ammonium toxicity by rapidly converting the ammonium generated from nitrate assimilation or from photorespiration (^{Miflin and Habash, 2002}; ^{Taiz and Zeiger, 2013}). The inhibition of glutamine synthetase by glufosinate delays the photosynthesis rate in C3 and C4 plants (^{Lacuesta et al., 1993}; ^{Wild et al., 1987}), which is attributed to inhibition of photorespiration, causing the accumulation of glycolate (^{González-Moro et al., 1993}; ^{Wendler et al., 1990}). This accumulation inhibits ribulose-1,5-bisphosphate carboxylase and, consequently, diminishes CO₂ assimilation (^{González-Moro et al., 1997}). Photorespiration functions, along with the Calvin-Benson cycle, contribute to a wide array of chloroplast processes, from bioenergy to metabolism of carbon and assimilation of nitrogen; thus, it protects the photosynthetic machinery under stress conditions (^{Taiz and Zeiger, 2013}). This could explain the higher accumulation of ammonium by B. decumbens, as compared with cotton and I. grandifolia, because it has a C3 metabolism in the presence of photorespiration.

The exposure of B. decumbens plants to glufosinate for 1 h before rain was enough to cause the maximum reduction (approximately 80 %) in glutamine content (Figure 3). Cotton showed 60 % reduction in glutamine in 3 h without rain, whereas I. grandifolia showed the lowest glutamine levels at approximately 24 h without rain. This shows that plants must absorb a greater quantity of the herbicide to promote the most intense effects on glutamine content (Figure 3).

Figure 3 Glutamine concentration (mg kg^-1 fresh weight) in plants of cotton, B. decumbens and I. grandifolia for different without rain periods at two days after application (DAA). Vertical bars represent ± confidence interval of the mean (n=4) by the t test (p≤0.05). 

The inhibition of glutamine synthetase, by glufosinate, generates a glutamine decrease, amino-acid biosynthesis inhibition and ammonia accumulation in the plants (^{Hess, 2000}). After applying 1.02 mM glufosinate, the activity of glutamine synthetase in the rice cv TNG 67, FSK and R11-2 diminished drastically within 6 h after the treatment (^{Tsai et al., 2006}). In Amaranthus palmeri S.Wats, 80 % of the activity of glutamine synthetase was inhibited by glufosinate between 6 and 24 h after herbicide treatment (^{Coetzer and Al-Khatib, 2001}).

Visual injuries of the species analyzed increased throughout the evaluation period (Table 1). Cotton at 3 and 10 DAA, displayed 10 % and 95 % injuries, respectively, and this verified that the quantity of herbicide absorbed in 6 h after application was sufficient to promote the highest injury levels. Symptoms of injuries are evident 3 to 5 d after application of glufosinate ammonium, beginning with chlorosis and spots of liquid saturation followed by the rapid necrosis and death of the plant (^{Pline et al., 1999}).

Table 1 Percentage of visual injury in cotton plants (n=4) and percentage of control in B. decumbens (n=10) and I. grandifolia (n=10) plants for different without rain periods. 

Values are the average of replications ± confidence interval of the mean (n=4) by the t test (p≤0.05). The equations models were adjusted by t test (p≤0.05

At 3 DAA, B. decumbens plants presented 40 % control, which was higher to that observed in cotton in the same period, and at 10 DAA there was 97 % control in plants treated with 3 h of herbicide absorption without rains (Table 1). In I. grandifolia at 3 DAA, plants injuries increased during the longest periods without rain, which shows that greater quantities of glufosinate absorbed in the longest periods without rain promoted more intense symptoms (Table 1). Nevertheless, in the 6, 8 and 10 DAA there were no differences starting at 3 h without rains, reaching approximately 90 % control at 10 DAA.

^{Anderson et al. (1993)} evaluated the effects of simulated rains of 4, 9 and 22 mm on the efficacy of glufosinate ammonium on barley (Hordeum vulgare cv. Samson) and green foxtail (Setaria viridis). They observed that a period of 1 to 8 h without rain (after herbicide application) was required to achieve good control of barley, compared to less than 20 min for green foxtail. ^{Coetzer et al. (2001)} estimated glufosinate efficacy in amaranth as affected by relative humidity and temperature, and 4 d after treatment, glufosinate dose (205, 410 and 820 g ha^-1) controlled all species, on average higher than 80 % for plants grown at 90 % relative humidity, but 820 g ha^-1 glufosinate injured more than 80 % of plants grown at 35 % relative humidity.

Conclusions

Glufosinate ammonium absorption increased until 24 h in cotton and 48 h in B. decumbens and I. grandifolia. Despite the increment in glufosinate content in the longest periods without rain, levels of ammonia were increased and those of glutamine were reduced. Cotton leaves showed the lowest contents of glufosinate, ammonia and glutamine. The two weeds species had higher amounts of glufosinate and ammonia, which were lower for I. grandifolia as compared to B. decumbens. In the periods up to 24 h without rain, the reduction in glutamine was lower for I. grandifolia. Finally, injuries were already severe at three hours without rain for the three species.