Agronomic performance of rice to the use of urease inhibitor in two cropping systems 1

The use of urea coated with urease inhibitor may become a useful tool for increasing the efficiency of nitrogen top-dressing in rice crop, thereby reducing nutrient losses through volatilization of NH3 (ammonia). Thus, the aim of this study was to evaluate the volatilization of NH3 and the response of rice to the use of urea coated with urease inhibitor in two cropping systems, no-tillage and conventional. For this purpose, field experiments were developed in the agricultural years 2007/2008 and 2008/2009, in UFSM in Santa Maria-RS. The design was randomized blocks in bifactorial scheme (2x5) with two sources, urea and urea + NBPT and five intervals of water intake (0; 3; 6; 9; 12 days) after application of nitrogen sources. The results of two seasons show that the urease inhibitor present in urea slows and decrease the conversion of N to NH3, reducing the losses by volatilization, compared to urea without inhibitor. Among the systems, the losses are magnified in the no-tillage cropping system. The behavior of the response variable in relation to productivity is variable in two cropping systems used in this study and the stress caused to the rice plant by the late start of the irrigation is more damaging than the losses caused by the volatilization of NH3.


INTRODUCTION
In irrigated rice growing, the product most used in the supply of nitrogen (N) is urea due to its high level of N and lower cost per unit of element (XU et al., 2005). On the other hand, is a source that presents the greatest losses of N by processes such as volatilization of ammonia (NH 3 ), reaching up to 80% of the fertilizer applied (LARA CABEZAS;SOUZA, 2008;MARTHA JÚNIOR et al., 2004).
The directions of the research for the first top-dressing application of N in rice say that this shall be done preferably at tillering, preceding the intake of water at a maximum interval of three days between application and initiation of irrigation (SOCIEDADE SUL-BRASILEIRA DE ARROZ INTEGRADO, 2007). This recommendation is based on the fact that urea can be converted into ammonia (NH 3 ) and lost or, following with aerobic conditions, transformed into nitrate (NO 3 -), which can then be lost as N 2 or N 2 O when the soil is flooded. Thus, with flooding being performed immediately after application of urea, the probability of loss by evaporation, and, subsequently by denitrification, decreases, enhancing the efficiency of the fertilizer applied.
However, due to farming operational factors (farm size, irrigation capacity), this interval of time often cannot be met, reflecting on the efficiency of the fertilizer applied and consequently, on the productivity. One possibility to control or reduce the losses of N is to inhibit the rate of urea hydrolysis in soil by using an enzyme inhibitor, such as N-(n-butyl) thiophosphoric triamide (NBPT), currently considered the most promising urease inhibitor and is already being marketed in the U.S.A since 1996 (WATSON, 2000), reducing in 83% the losses by volatilization in rice (SCIVITTARO et al., 2010).
This study aimed to evaluate the N losses by volatilization and the response of irrigated rice using urea coated with urease inhibitor (NBPT) compared to urea, at different intervals of water intake after applying fertilizer in conventional and no-tillage cropping systems in two different seasons.

MATERIAL AND METHODS
Experiments were carried out in the seasons of 2007/2008 and 2008/2009 at the Universidade Federal de Santa Maria, in soil classified as typical albaqualf, belonging to the mapping unit of Vacacaí (EMPRESA BRASILEIRA DE PESQUISA AGROPECUÁRIA, 1999), with the following characteristics: pH water (1:1) = 5.0; P (Mehlich-1) = 11.8 mg dm -3 ; K (Mehlich-1) = 76 mg dm -3 ; M.O. = 2.4 m/v; Ca = 4.8 cmol c dm -3 ; Mg = 1.6 cmol c dm -3 ; Al = 0.5 cmol c dm -3 and clay = 25%. Each year two experiments were conducted simultaneously, allocated side by side, differing only in the cropping system (conventional system (CS) or no-tillage (NT)). To obtain the straw in no-tillage system, ryegrass (Lolium multiflorum Lam.) was sown at a density of 40 kg ha -1 of seeds. The climatic conditions of the years 2007/2008 and 2008/2009 were, respectively: 30 mm and 40 mm of precipitation, minimum temperature of 28 ºC and 27 ºC, in NT and CS, respectively, and maximum temperature of 30 ºC and 40 ºC, in NT and CS, for two years.
The experimental design was randomized blocks in bifactorial scheme (2x5) with four replications. The factor A (qualitative) was represented to sources of nitrogen (N), common urea and urea coated with NBPT (urea + NBPT) (Super N ® ), and the factor B (quantitative) five intervals of water intake, represented by: 0; 3; 6; 9 and 12 days after the application of the nitrogen source. Sowings occurred on 10/26/2007 and 11/10/2008, using cultivar IRGA 417 on the density of 90 kg ha -1 of seeds in both seasons. Fertilization, at the time of sowing, was 60 kg ha -1 of P 2 O 5 and 90 kg ha -1 of K 2 O of formula 05-20-30, using the total amount of 120 kg ha -1 of N, divided into 15 kg ha -1 at sowing in the form of urea, 75 kg ha -1 N before the final irrigation in the form of different sources and 30 kg ha -1 of N at panicle initiation, for both urea and urea + NBPT. The experimental units were 4.0 x 2.87 m (11.48 m 2 ) and floor area to estimate the yield was 3.0 x 1.19 m (3.57 m 2 ). The remaining practices were performed as recommended by the search for rice production (SOCIEDADE SUL-BRASILEIRA DE ARROZ INTEGRADO, 2007).
The variables analyzed were N losses for 288 hours, N accumulation in shoots of rice plants, SPAD readings in shoots, mineral N in the soil and yield of rice. The variables were subjected to analysis of variance by the F test, and the means of qualitative factors were compared by the Tukey test (P 0.05), whereas the quantitative factors were represented by regression analysis and equations adjusted for the water intake intervals on the variables analyzed. Regression analysis was aimed at identifying the effect of delaying the entry of water onto the rice agronomic performance. A confidence interval (upper and lower limit) at 95% probability was used to express the difference between the sources, and those were significantly different when, in comparison, their intervals did not overlap. The dates were analyzed in software Sisvar, version 5.3.
Evaluations of N losses through volatilization of NH 3 were performed with collectors and procedures according to Araújo et al. (2009). Evaluations were performed at 10; 24; 34; 48; 72; 96; 144; 216 and 288 hours after application of nitrogen source in the soil, determining NH 3 later, according to Tedesco (1995).
The evaluation of N accumulated in shoots of rice plants was determined by collecting whole plants at 0.5 m 2 of each plot at the last application of N, at panicle initiation, 43 days after emergence (DAE), using the methodology described by Tedesco (1995). On the other hand, for the second crop, plants were also collected at the time of full flowering, 93 DAE. The samples were dried at 70 °C in an oven of forced air and they were grounded right after that. To assess the mineral N in the soil, at the time of collecting the plants, a soil auger was used, taking a sample of 0;10 cm deep in each plot, subsequently freezing it at -5 °C until determining the content of mineral forms of nitrogen (ammonium and nitrate + nitrite), following the methodology described by Tedesco (1995).
In order to estimate the nutritional level of N in the culture, at the time of soil and plant sampling, SPAD readings (chorophyll meter SPAD 502) were also performed on the last fully expanded leaf, in three different positions of the leaf, in three plants per plot.
Grain yield was determined by manually harvesting the panicles of the useful area of the plots when the grains reached the average moisture of 20%. These were threshed and later, the grain weight was determined and the moisture was corrected to 13%.

RESULTS AND DISCUSSION N losses by ammonia volatilization
The NH 3 volatilization flows varied with the source of nitrogen (N) and cropping system in both years ( Figure 1). For the season of 2007/2008, no-tillage system (NT), the amount of NH 3 volatilized during the period of 288 hours, was approximately 28% of the applied amount (the total amount applied corresponded to 75 kg N ha -1 ), while the loss of urea + NBPT was 18%. The amount of NH 3 volatilization was lower in the conventional system (CS) compared to the NT in which the losses were 11% and 7% of the nitrogen applied to the common urea and urea + NBPT, respectively. In the NT system, NBPT delayed the conversion of nitrogen amidic (urea) into NH 3 , delaying the maximum emission of NH 3 from 34 to 72 hours (2-3 days), occurred with common urea to 144 hours (six days) in the treatment urea + NBPT, agreement with the results of Scivittaro et al. (2010) in similar work.
For the 2008/2009 crop, the results of NH 3 volatilization were similar to the previous season, showing different losses between sources and cropping systems. In the NT system, the initiation of a significant flow of NH 3 losses began after 72 hours, regardless of the source, and urea showed the greatest losses, reaching a maximum volatilization after 96 hours, a total cumulative loss of 47% of N applied during 288 hours. Urea coated with NBPT had lower losses of N, with maximums between 96 and 216 hours after application on the soil and accumulated losses of 20% of total N applied in top-dressing. In the NT system, NBPT delayed most part of the conversion of urea into NH 3 for 216 hours, whereas for the urea without inhibitor, most fertilizer was hydrolyzed between 72 to 96 hours. Similar to the first harvest, the urease inhibitor has not completely inhibited the hydrolysis and the loss of NH 3 , and its efficiency decreased gradually with time.
In this cropping, for the CS, the losses by volatilization have not exceeded 1.2%, and the maximum of losses occurred after 216 hours. In the treatment with urease inhibitor, total NH 3 loss was 0.014% of the total N applied. The urease inhibitor was effective in delaying and reducing the volatilization of NH 3 in the two systems because the product is an indirect inhibitor that, in aerobic conditions, is converted into a direct inhibitor, its oxygen analogue compound, N-(n-butyl) phosphoric triamide (NBPTO), within minutes or hours (DOMINGUEZ et al., 2008). However, under anaerobic conditions, this transformation can take several days (WATSON, 2000).
Despite the potential to inhibit the volatilization, the product had reduced its efficiency in the NT system, probably due to the presence of plant residue on the soil surface, since straw reduces the contact of fertilizer with the soil and may present significant activity of urease. This layer of straw covering the soil acts as a barrier between the nitrogen of the fertilizer and the soil, causing the NH 3 , hydrolysis product, to remain on the surface of the debris, reducing its adsorption to organic and inorganic colloids, thus facilitating the volatilization (CANTARELLA et al., 2008;LARA CABEZAS;SOUZA, 2008;TRIVELIN;VITTI, 2006). Moreover, the presence of plant residues on the soil surface concentrates great amounts of urease enzyme, which has its activity related to the presence of organic matter on the soil, promoting microbial activity and greater enzyme production, which shall accelerate the nitrogen fertilizer hydrolysis, resulting in the formation of NH 3 .
The pattern of N released from NBPT depends on the thickness and quality of the treatment. Urea passes from the interior of the granule into the soil by micro-pores, cracks or imperfections of granules or after the microbial degradation of the polymer coating covering the urea (CANTARELLA, 2007). With this, in the NT system, there is the possibility of a combination of higher concentration of urease have occurred coupled to the permanence of nitrogen on the surface, which may have jeopardized its efficiency, compared to its behavior in the CS. In CS, the volatilization losses were substantially lower, since the absence of plant residue provided greater contact between fertilizer and soil. Losses present in this system occurred due to a likely increase in temperature and decrease in moisture in the soil with no straw, which enhanced the volatilization E. Marchesan et al. process due to the increase in the rate of reactions linked to the urease activity, as well as facilitated the NH 3 upward diffusion to the atmosphere together with water evaporation from the soil (DA ROS et al., 2005;SANGOI et al., 2003).
For the season of 2007/08 in the NT system, there was no difference in behavior between urea and urea + NBPT on the levels of N-NH 4 + in the soil (Figure 2). The difference occurred in relation to the intervals of water intake, demonstrating that the maximum concentration of N-NH 4 + , occurred on the sixth day after the application of N. The behavior of N-NO 3 fraction was similar, with gradual reductions in the delay of irrigation. Unlike the NT system, the CS, the treatments differed in the evaluation of N-NH 4 + in mineral soil. Urea + NBPT showed higher contents of mineral N-NH 4 + in soil in the first six days after application, compared with common urea. This was Mineral N in soil, plants and chlorophyll meter reading reflected in the concentration of N-NO 3 -, since the greatest levels were in urea + NBPT and they increased with the time interval until the irrigation input, indicating that N, in the form of N-NH 4 + , oxidized to NO 3 overtime (over the time).
For the season of 2008/2009, the behavior of N-NH 4 + was similar to the first season, demonstrating a decrease in the content as it delayed the start of irrigation in both cropping systems. However, there was no increase in the concentration of N-NO 3 with the reduction of N-NH 4 + , because the sources and the intervals of water intake did not differ significantly in both cropping systems.
In general, the tendency of the N-NH 4 + level in these past two years was to decrease as it delayed the water intake, because the behavior of the ammonium fertilizer tends to, when applied in soil, undergo hydrolysis and can Agronomic performance of rice to the use of urease inhibitor in two cropping systems and N-NO -3 on soil according to the fertilizer applied at the no-tillage and conventional cropping systems, during the seasons of . Santa Maria, 2011. The bars on the chart indicate the confidence interval (upper and lower limit) at 95% of probability to express the difference between the sources, and they were significantly different when, at comparison, their intervals did not overlap E. Marchesan et al. be volatilized, forming N-NH 4 + , be absorbed by plants and ultimately be converted into N-NO 3 - (CARRASCO et al., 2004). Increasing the concentration of N-NO 3 over the reduction of N-NH 4 + occurred due to the fact of the N-NH 4 + released on hydrolysis of fertilizer reaching the form of N-NO 3 by nitrification process, depending on the availability of oxygen for this transformation to occur (HOLZSCHUH et al., 2009).
Although the sources used did not express differences in the levels of mineral N in soil at the NT system, the evaluation performed in rice plants showed significant differences between treatments ( Figure 3) in 2007/08. The concentration of total N in the plant decreased in treatments with urea as it delayed the beginning of irrigation, while urea + NBPT had the highest concentration of total nitrogen, up to 12 days of delay, between the beginning of fertilization and irrigation. Unlike the NT system, in the CS, some significant differences were detected in relation to the levels of mineral nitrogen in soil, without, however, response in relation to the total N level of the rice plant, because the evaluation did not show significant differences, what is completely opposite to what was found in the NT system. Added to this, chlorophyll meter readings (Table 1) were carried out on the same day of the determination of total N content of plants (43DAE). In this evaluation, there were no significant differences between N sources and the intervals of water intake into the CS, as well as into the NT system.
In the second season, 2008/2009, the total N content, determined at panicle initiation (48 DAE) (Figure 3), in shoots of rice at NT system differed between the sources and intervals of water intake, in which urea was the best treatment in almost all time intervals. However, these differences were detected only in a first moment, because, at the flowering stage, the behavior of sources was opposite to the first evaluation. In the case of CS, there were no significant differences in both evaluations, as found in the previous harvest. However, chlorophyll meter readings (Table 1) in the NT system showed different results, indicating that the first evaluation performed at 48 DAE, urea +NBPT behaved similar to the common urea, but there : water intake at the moment of nitrogen sources application; D2: water intake 3 days after the nitrogen sources application; D3: water intake 6 days after the nitrogen sources application; D4: water intake 9 days after the nitrogen sources application; D5: water intake 12 days after the nitrogen sources application; ns Not significant in P 0.05 level; * Quadratic equations adjusted to express the response of the treatments; ** Interaction among the factors studied and significance level of 5% to Test T were no differences between intervals of water intake for the first source. For the CS, NBPT presented lower readings compared to common urea when the beginning of irrigation was delayed for more than 12 days.
In the second evaluation performed at the beginning of flowering (93 DAE), the behavior of sources in relation to water intake, at NT system, was similar to the first, with no differences between them. For the CS, no differences The bars on the chart indicate the confidence interval (upper and lower limit) at 95% probability to express the difference between sources, and they were significantly different when, at comparison, their intervals did not overlap were found between sources and between intervals of water intake. In this sense, the reading of chlorophyll meter is an indirect measure of chlorophyll and its response, in the second year, did not reflect the actual nutritional status of the plant, since it has shown differences not detected by laboratory evaluation. The chlorophyll meter readings only correlate with the concentration of nitrogen in the plant when both evaluations are conducted on the same *Mathematical models regression significant at 5% for Test F E. Marchesan et al. leaf. For this experiment, all the shoot area of the plant was used to determine the total nitrogen, including those already senescent, which should dilute the total N content, rather than the chlorophyll meter which estimates the level of N directly in the assessed leaf.

Response of irrigated rice to the sources of N
At the first cropping (2007/2008), the yield of the grains for the NT system did not show significant differences between the intervals of water intake and the sources of N used in the experiment (Figure 4). In the case of CS, there were some differences in the yield of grains between the intervals of water intake and the sources of N. When the water intake occurred right after the application of nitrogen fertilizer; the best behavior was presented by the common urea (4.5% higher). However, when the beginning of the irrigation was delayed for more than nine days, the urea + NBPT presented more grain yield.
At the second cropping (2008/2009), in the NT system, the sources and the interval of water intake had an influenced the grain yield. In this assessment, urea +NBPT presented an increase in relation to urea when the beginning of irrigation took place at the ninth day after the application of fertilizer. On the other hand, in the CS, there were no significant differences between the sources and, as water intake was delayed, there was a decrease of 12% in the yield from the first in relation to the last interval of water intake.
The response in relation to grain yield differed between the systems evaluated for two years. In the first year, there were no significant differences between intervals of water intake and sources of N. This can be attributed to minor The bars on the chart indicate the confidence interval (upper and lower limit) at 95% probability to express the difference between sources, and they were significantly different when, at comparison, their intervals did not overlap differences in N losses through volatilization between sources and the possibility of a greater immobilization of part of the available nitrogen due to the implementation of the NT system after the cultivation of ryegrass, with consequent lower initial availability of nutrients to plants (MULVANEY et al., 2010). However, in the CS, there were differences in yield between periods of water intake and sources of urea, where common urea showed the best behavior when water intake occurred soon after the application. This behavior can be explained by the fact that N is readily available, since urea has a high solubility in water (SILVA et al., 2008), whereas urea coated with urease inhibitor has its gradual release, especially in anaerobic environments (CUNHA et al., 2011) and the immediate use by the crop can be jeopardized. When the beginning of irrigation was delayed for more than nine days, the urea + NBPT showed higher productivity, demonstrating that this source is important for the handling of areas with difficulties in establishing the amount of water immediately after the application of N.
The result of productivity in the second year was different from the first, since in this season, the NT system productivity was directly influenced by the sources of urea. In this case, the differences in N losses through volatilization of N between sources were higher, but lower for urea with urease inhibitor. In CS there were no significant differences for any of the factors studied. This result can be explained by lower N losses between sources, with little influence of sources and delayed irrigation on grain yield. Thus, it is possible to consider that the response of irrigated rice to urea coated with urease inhibitor depends on the risks of loss through volatilization of NH 3 , which are associated to the soil conditions, climate and management on site. Furthermore, on average, the coefficient of variation of the analyzes were approximately 15%, giving an average precision of the results obtained according to the classification of Pimental-Gomez (1990).

CONCLUSIONS
1. The use of urea coated with NBPT slows and reduces losses by ammonia volatilization compared with urea, but the magnitude of the effectiveness of adding NBPT into urea is associated with soil and climate conditions, influenced by the year of cropping and the rice cultivation system, which are not always expressed in productivity; 2. In no-till and conventional systems, the best performance of the urease inhibitor occurs as the establishment of the amount of water for irrigation is delayed after the fertilizer application. When the interval between the nitrogen application and the irrigation is within the recommended levels, there is no advantage in adding the product compared to isolated use of urea.  ABSTRACT -Irrigated rice sowing season and red rice competition are among the main factors affecting grain yield. The objective of this work was to evaluate the sowing date of irrigated rice and moments of application of the herbicide imazapyr + imazapic to control red rice management and irrigated rice grain yield. Eight experiments were performed at the following dates (09/30, 10/19, 11/08 and 12/01) for the 2010/2011 harvest season and (09/27, 10/17, 11/08 and 12/05) for the 2011/2012 harvest season. The treatments were: application of the herbicide imazapyr + imazapic at doses of 105+35 g ha -1 in pre-emergence (PRE); 52.5+17.5 g ha -1 in pre-emergence and 52.5+17.5 g ha -1 in post-emergence (PRE + POST); and 105+35 g ha -1 in post-emergence (POST), and a control without application and no weeding. The cultivar Puitá Inta CL was used and a randomized block design with four replicates. A joint analysis of the experiments was carried out. There was less emergence of red rice and higher grain yield of the irrigated rice at the early periods (09/30/10 and 09/27/11), with 10,578 and 8,653 kg ha -1 , respectively. At the end of the season (12/01/10 and 12/05/11), there was greater reduction of the red rice seed bank. Sowing at the beginning of the recommended period provided more irrigated rice grain yield. The application of imazapyr + imazapic at a dose of 52.5+17.5 g ha -1 in PRE + 52.5+17.5 g ha -1 POST, and 105+35 g ha -1 only in PRE and POST was effective in the control of red rice.

INTRODUCTION
Rice sowing season plays a significant role in crop yield potential, since it affects the response of other management practices. According to Freitas et al. (2008), the intersection of rice flowering and grain filling time with the period of highest solar radiation availability is what determines when rice sowing season will occur, once optimal conditions of temperature and solar radiation are key to raising grain yield potential. According to Junior et al. (1995), the optimal values of air temperature and radiation around flowering are 25 o C and 475 cal cm -2 day -1 , respectively. However, low temperatures at microsporogenesis or high temperatures at anthesis can cause spikelet sterility, affecting crop yield (Farrell et al., 2006). According to Jagadish et al. (2007), temperatures above 35 o C at anthesis for more than one hour may reduce the fertility of spikelets on rice. Rang et al. (2011) have reported that rice response to elevated temperature differs according to its stage of development, with greater sensitivity in the reproductive phase.
According to Slaton et al. (2003) and Freitas et al. (2008), the highest grain yields of irrigated rice are found when sowing is carried out early in the recommended season and they tend to decrease when sowing is performed at the end of the season. Sowing time delay causes reduction in the number of panicles per square meter (Freitas et al., 2008). Moreover, when sowing time is anticipated, emergence and early establishment may be negatively impacted, mainly due to suboptimal air temperatures during this period .
Besides influencing the yield potential of rice, sowing time, if considered independently, may contribute to a more efficient control of red rice, one of the most relevant weeds found in irrigated rice areas. According to Marchesan et al. (2010) and Marchesan et al. (2011), red rice affects grain yield, the quality of the final product, and the price paid for the product (Gealy et al. 2000).
There are options for the chemical control of red rice, which include the use of cultivars that are tolerant to herbicides from the imidazolinone group. However, due to the diversity of biotypes, red rice and cultivated rice crossing, seed dormancy (Burgos et al., 2008), differences in emergence onset, and speed and duration of emergence (Shivrain et al., 2009), it is difficult to control red rice. Accordingly, sowing time may be a significant crop practice for the control of red rice, since early in the recommended season (September/October), air and soil temperatures are lower compared to late in the recommended season (November/December). Lower temperature conditions maintain red rice seed dormancy, reducing germination and plant emergence (Gianinetti & Cohn, 2008;Young-Son Cho, 2010). Shivrain et al. (2009) have reported that, in sowing conducted early in the recommended season, red rice competition is reduced when compared to sowing conducted late in the recommended season. According to Norsworthy & Oliveira (2007), at the end of the recommended season, red rice emergence rate is increased due to higher temperatures, which provide rapid seedling emergence and establishment. Thus, there is a need to know the impact of sowing time, when associated with weather conditions such as temperature and solar radiation, on the dynamics of red rice and rice grain yield, in order to have a more efficient control of red rice. There is also a need to know the best time to apply these herbicides (imazapyr and imazapic) taking into account the time of sowing, because some Effects of irrigated rice sowing season and imazapyr + imazapic ... works on this subject, for instance those of Santos et al. (2007) and Marchesan et al. (2011), have assessed herbicide application time in a single sowing period. However, it is important to assess the behavior of such application times at different sowing periods.
The hypothesis of this work is that in sowing carried out early in the recommended season, there is lower emergency of red rice and greater rice grain yield; and in sowing carried out late in the recommended season, there is great seed bank reduction. Thus, this study aimed to assess irrigated rice sowing periods and imazapic + imazapyr moments of application on the control of red rice and on irrigated rice grain yield.

MATERIAL AND METHODS
Eight experiments were conducted on growing seasons 2010/11 and 2011/12. The area was located in the physiographic region of the Central Region of the State of Rio Grande do Sul (RS), in the city of Santa Maria, where climate is characterized, according to the Köppen classification, as humid subtropical (Cfa), with no dry season, with mean temperature of the warmest month above 22 ºC (Moreno, 1961).
The experiments consisted of the following sowing dates for Puitá Inta-CL rice: (09/30, 10/19, 11/08 and 12/01) for the 2010/11 harvest, and (09/27, 10/17, 11/08 and 12/05) for the 2011/12 harvest. The sowing period, according to the agricultural zoning Puitá Inta-CL in Santa Maria, goes from September 1 to November 20. However, the choice of sowing dates in this study aimed to represent the sowing period for the state of RS, which is from early September to mid-December, varying according to the cycle of each cultivar and to the cultivation region. The treatments examined were: Kifix ® (imazapic + imazapyr) application at doses 105+35 g ha -1 in pre-emergence (PRE), 52.5 +17.5 g ha -1 in pre-emergence, and 52.5 +17.5 g ha -1 in postemergence (PRE + POS) and 105 + 35 g ha -1 in post-emergence (POS), in addition to control with no herbicide and weed free. The experimental design was randomized blocks with four replications. The PRE application was held in stage S 3 (emergence of prophyll from coleoptile) of rice, and the POS application was conducted one day before final irrigation in stage V 3 /V 4 , according to the scale of Counce et al. (2000). Also in stage S3 of irrigated rice, we applied glyphosate at 1.08 kg ha -1 in all treatments, with the purpose of controlling all the red rice plants that had emerged in the period between sowing and grain formation.
We used irrigated rice cultivar Puitá Inta-CL, which has a mean cycle of 125 days, with 90 kg ha -1 seed, sown in row spacing of 0.17 m, in minimum tillage system in the 2010/11 harvest and in direct sowing in the 2011/12 harvest. Base fertilizer was 15 kg ha -1 nitrogen (N), 45 kg ha -1 P 2 O 5 and 90 kg ha -1 K 2 O, and the N dose was divided in quantities of 15 kg ha -1 at sowing, 70 kg ha -1 at tillering (V 3 /V 4 ) and 35 kg ha -1 at panicle iniciation (R0).
The following evaluations were performed: red rice seed bank, determined before sowing to sowing times, by counting the number of red rice seeds in two subsamples of 0.2 x 0.2 m in each portion at a depth of 7 cm; number of red rice plants emerged in the range from PRE application to POS application, determined by counting emerged plants in two subsamples of 0.2 x 0.2 m in each plot, later transformed into percentage in relation to seed bank; total number of desiccated plants up to the moment of each sowing date, determined by counting the number of red rice plants in 0.5 x 0.5 m per plot of 5 m x 1.53 m.
Desiccations were conducted with glyphosate at 1.08 kg ha -1 , when red rice plants were in stage V 3 /V 4 . Seed bank reduction was determined by summing emerged plants from PRE to POS application with the total number of plants desiccated up to the moment of each sowing date, and transformed in percentage in relation to the seed bank.
Furthermore, was assessed the control of red rice by counting the number of red rice panicles in each plot, and transforming in percentage in relation to control.
Grain yield was estimated by manual harvesting of 4.16 m 2 on crop area, when they reached mean moisture of 22%. After sorting, cleaning and weighing of grain with shell, data were corrected to 13% moisture and converted to kg ha -1 . The number of panicles m -2 was determined by counting the panicles in a row meter, and in that same area we collected 15 panicles at harvest to estimate the number of grains per panicle, thousand grain weight and spikelet sterility.
The values of global solar radiation and air temperature were obtained from the meteorological station of the University of Santa Maria, available at the National Institute of Meteorology (INMET). Soil temperature was obtained using temperature sensor (model L34-108-EN) at a depth of 3 cm and recorded in data logger CR1000.
Results were submitted to analysis of variance, comparing in each crop the sowing dates of irrigated rice. Means were compared by Tukey test at 5% probability.
Data that referred to red rice seed bank reduction, red rice desiccated plants m -2 , and red rice emergence were transformed to 5 . 0 ) ( + = y yt .

2010/11 Harvest
There was interaction between sowing dates and times of herbicide application for all variables except for thousand grain weight and spikelet sterility ( Table 1). The area where the experiment was conducted showed variation in red rice seed bank from 237 to 2533 seeds m -2 , the largest seed bank noted in the area of the first and third sowing times: 09/30 and 11/08, respectively (Table 2).
At the beginning of the recommended sowing period (09/30 and 10/19) there was lower emergence of red rice when compared to late sowing period (11/08 and 12/01) (Table 2), indicating that sowing time plays a key role in the management of red rice, by having lower emergency of red rice early in the sowing period, thus, there is less competition and damage to irrigated rice. This response may be related to the seed dormancy of red rice, due to the physiological immaturity of the embryo, once it is detached from the mother plant, and it may also present impermeability to gas cover and/ or imbalance of substances that promote and inhibit germination induced during the accumulation of dry seed weight (Filio, 2005), associated with lower air temperature (average temperature around 16 o C) and soil (average temperature around 18 o C) during the beginning of the recommended period ( Figure 1A), which may have contributed to inhibit germination and emergence of large population of red rice in this area.
Young-Son Cho (2010) evaluated the effect of different temperatures in the day and night, represented by 15/10, 20/15, 25/20 and 30/25 o C, and found on average a lower percentage of red rice germination in lower temperatures, with 25/20 o C as one in which there was best germination. According to this author, the percentage and speed of germination increase with increasing temperature. Gianinetti & Cohn (2008), have observed greater red rice mean germination at 20 to 35 o C, compared to 1 to 15 o C, and optimum temperature for rice development from 25 o C to 30 o C (Yoshida, 1981). According to Marcos Filho (2005), rice seeds overcome dormancy when exposed to high temperatures, and response speed is directly related to the increase in temperature.
Temperatures lower than this range (25 to 30 o C) can cause cold stress, which is considered one of the most important abiotic stresses in rice, causing changes in germination, percentage of normal seedlings, coleoptile length, among other factors, mainly affecting the expression of isozymes, such as esterase and dehydrogenase enzyme, involved in seed germination (Mertz et al., 2009). In sowings conducted at the end of the recommended period, the emergence rate of red rice was increased due to higher temperatures, which provide rapid seedling emergence (Norsworthy & Oliveira, 2007). This emergence speed is very similar to rice, implying rapid establishment of biotopes Effects of irrigated rice sowing season and imazapyr + imazapic ... (Shivrain et al., 2009), which increases the potential for competition and damage to irrigated rice.
The greatest emergence of red rice at the end of the recommended period (11/08 and 12/01) provided high reduction of seed bank (Table 2): 54 to 91.6% in the last sowing date. This reduction of seed bank was due to longer time to perform desiccations of red rice seedlings that emerged in the area, with a higher number of seedlings desiccated: 553 and 289 seedlings m -2 at end of period (11/8 and 12/01), respectively, compared with 109 and 71 seedlings m -2 in the early period 09/30 and 10/19 respectively (Table 2).
Such sowing practice at the end of the recommended period may be an important alternative to guide decision-making when you want to use the strategy to reduce seed bank in areas infested with red rice.   Variables with values †in parentheses indicate original values †, averages not followed by the same letter, lowercase in column and uppercase in row, differ by Tukey test at 5% probability; NS not significant in row; ns not significant in column ; 1 percentage of red rice emergence in relation to seed bank in PRE to POS application; 2 number of red rice desiccated plants from 09/23/2010 to the sowing date of each season, 3 seed bank reduction in relation to red rice seed bank. * Doses of imazapic + imazapyr were: 105+35 g ha -1 (PRE), 52.5+17.5 g ha -1 and 52.5+17.5 g ha -1 (PRE+POS) and 105+35 g ha -1 (POS).
With respect to the chemical control of red rice (Table 2), the POS application and the PRE + POS application were efficient, resulting in 100% control. The PRE application alone showed lower efficiency in 11/08, with 86% control, which can be explained by the greater Effects of irrigated rice sowing season and imazapyr + imazapic ... seed bank (more seeds for the same amount of herbicide) in comparison to the areas that were performed in 10/19 and 12/01, in which seed bank was smaller. Moreover, at 11/08, the conditions of air temperature and soil were greater ( Figure 1A), which improved germination and emergence of red rice. This control efficiency of treatment in POS and PRE + POS is related to the fact that in the area where the experiment was conducted there are no biotypes resistant to the imidazolinone chemical group, plus it has perfect leveling of surface area and the proper management of irrigation. Furthermore, high efficiency in the control provided by the PRE + POS application or POS only, is due to the fact that the POS applications were performed on a day before the final irrigation. Thus, with the establishment of water depth, an anaerobic environment was formed, reducing one of the main mechanisms of dissipation of these herbicides from the imidazolinone group, which is the microbial degradation (Flint & Witt, 1997;Madani et al. 2003;Alister & Kogan 2005).  the biodegradation of herbicides due to anaerobic condition, which occurs with the establishment of irrigation after herbicide application. Another very important factor that may have contributed to greater efficiency in the control treatments that were applied in POS is auto-liming (soil chemical changes resulting from the reduction process), which occurs 15 to 20 days after the formation of water depth, raising the pH of acidic soils favoring dissociation of herbicides, making herbicides molecules available for plant uptake (Avila et al. 2006), since, with the presence of the carboxylic acid and the basic functional group of pyridine, imazapyr has three pKa values (1.9, 3.6 and 11.4), which is one of the characteristics that make the behavior of this herbicide in soil pH dependent (Firmino et al., 2008). Also, a study with imazaquin conducted by Oliveira et al. (2004) showed that with increasing pH reduction occurs Kd values (soil water partition coefficient) in all levels (2-6%) of organic matter studied, with it, the adsorption is reduced due to the dominance of their shape anionic, becoming more available in the soil solution.
However, the herbicides of the chemical group of imidazolinones, when applied in PRE, may have a reduced efficiency because they can be degraded by microorganisms in greater quantities compared to POS application, whereas the application to the soil remains PRE for a longer period of time in aerobic condition, which is favorable to biodegradation. Thus, there can be reduction in the concentration of herbicide in the soil and therefore it can provide lower efficiency of control. One of the factors considered most important -responsible for the lower efficiency of herbicides when applied in PRE -that is, in conditions of lower pH of the soil, there may be adsorption of these herbicides to soil colloids (Bresnaham et al., 2000;Madani et al. 2003;Alister & Kogan, 2005;Kraemer et al. 2009), which decreases the efficiency of control.
Corroborating the study, Marchesan et al. (2011), when assessing the control of red rice using chemical herbicides of the imidazolinone group imazethapyr (75 g L -1 ) + imazapic (25 g L -1 ) (ready mix) at the recommended dose in Brazil, found no differences between application in PRE + POS and POS application only. Villa et al. (2006) have also evaluated the control of red rice in two rice varieties using imidazolinone herbicides at a dose of 75 g L -1 imazethapyr + 25 g L -1 of imazapic, and found that the split application of herbicide ( PRE + POS) using 75% of the dose in PRE followed by 50% for POS, was the one which provided better control without causing loss of yield. Santos et al. (2007) evaluated the efficiency of control of red rice provided by the use of formulated mixture (75 g L -1 imazethapyr + 25 g L -1 of imazapic) and observed that the application of 52.5 g ha -1 imazethapyr + 17.5 g ha -1 in PRE imazapic followed by the same dose in POS, is more efficient when compared to the application of 75 g ha -1 imazethapyr + 25 g ha -1 PRE imazapic only, or only in POS.
Sowing dates have also affected the yield of irrigated rice (Table 3), with higher grain yield in the first season (09/30), ranging from 9,752 kg ha -1 , with no chemical control of red rice, to 11,258 kg ha -1 , when control was performed, and this was (09/30) the best sowing time, to obtain higher yield. The highest grain yield occurred in this sowing season may be related to the coincidence that this was the period of more responsive plant behavior (flowering and grain filling) and the period of highest solar radiation availability in the months of December and January ( Figure 1C), which was reflected in a higher number of panicles, higher thousand grain weight and lower spikelet sterility in relation to sowing dates at end of recommended period (Table 3). In sowing at the end of the recommended period (12/01), the yield was 22% lower compared to the sowing of 09/30, equivalent to a reduction of 38 kg ha -1 day -1 in the mean of treatments PRE, PRE + POS and POS.
These results support the work conducted in the state of Rio Grande do Sul by Freitas et al. (2008) in which they gained greater number of panicles and higher grain yield when rice was sown at the preferred date (11/ 02) compared to late recommended period (12/ 09). Besides the sowing date, another factor that affected grain yield was the presence of red rice. Comparing the first sowing date (09/ 30) with the other, there is a reduction of 11, Effects of irrigated rice sowing season and imazapyr + imazapic ... 52 and 22% yield when chemical control was not performed (Table 3) to the second (10/19), third (11/08) and fourth times (12/01), respectively. This occurred because of greater emergence of red rice in the first sowing date.
The largest reduction in grain yield of the third season of sowing is related to the increased presence of red rice in the area, compared to the second and fourth seasons. These results show the great potential Table 3 -Grain yield (GY) (kg ha -1 ), number of panicles (NP), number of grains per panicle (NGP), thousand grain weight (TGW) and spikelet sterility (S) as a function of sowing date and time of Kifix ® (imazapic + imazapyr) applicationfor Puitá Inta-CL. 2010/ 11 harvest. Santa Maria, RS. 2012 Averages not followed by the same letter, lowercase in column and uppercase in row, differ by Tukey test at 5% probability; NS not significant in row; ns not significant in column * Doses of imazapic + imazapyr were: 105+35 g ha -1 (PRE),,52.5+17.5 g ha -1 and 52.5+17.5 g ha -1 (PRE+POS), and 105+35 g ha -1 (POS).

640
of damage to red rice with cultivated rice, according to Fischer & Ramirez (1993), 24 panicles m -2 red rice can cause losses of 50% in grain yield.
Generally, for this harvest, sowing time affected the yield of crop, with higher grain yield when seeding is performed early in the recommended date (09/30), due to better conditions of temperature and solar radiation in the reproductive period of the culture. However, from the second (10/19) until the fourth season (1/12) there was no significant reduction in grain yield, as the average values of minimum air temperature were 25, 23 and 21 o C for the months of January, February and March (reproductive period of culture), respectively, the average maximum for the same months were 26, 24 and 23 o C ( Figure  1A), respectively, without compromising spikelet fertility and hence grain yield. Even though the lowest accumulation of solar radiation, during the reproductive period of crop in such sowing dates, may explain the lower yields when compared to the first season (09/30).

2011/12 Harvest
In this season, there was interaction between sowing dates and times of herbicide application with seed bank, emergence of red rice, number of red rice desiccated seedlings, grain yield and number of panicles m -2 (Table 1). There was decrease in seed bank of red rice (Table 4) in the area compared to the 2010/11 harvest, with variation from 62 to 1,791 seeds m -2 , and the largest seed bank were in the area of the first and third seasons.
In general, except for control with no herbicide application, the results of red rice emergence (Table 4) follow similar behavior to the 2010/11 harvest, with increased emergence of red rice with the delay of sowing time with mean emergence from 4 to 18% at the beginning of the recommended period (09/27 and 10/17), respectively, and 35 and 26% at the end of the emergency period (8/11 and 5/12), respectively. The lowest emergence of red rice early in the recommended period, compared to times at the end of the period can be related to lower temperatures at the beginning of the period ( Figure 1B) and the air temperature around 15 o C to average minimum 25 o C and the average maximum around 20 o C, the seed of red rice have been induced to secondary dormancy because in work performed by Gianinetti & Cohn (2008), when evaluating the germination of red rice under different temperature conditions at 15 o C, the population of red rice seed dormancy was induced in secondary. According to these authors, the percentage of seeds that are induced in secondary dormancy decreases with increasing temperature (15-25 o C), in the optimum germination temperature (30 o C) biotypes had minimal dormancy, explaining in part the results. The highest emergence of red rice at late period may be related to higher temperatures occurring during this period ( Figure 1B), because with increasing temperature, germination processes occur faster than inducing dormancy (Gianinetti & Cohn, 2008).
As for the control without herbicide, this season there was no difference in emergence of red rice between sowing dates, emergency occurring high even in the beginning of the sowing period recommended (09/27). This may be related to the fact that in the 2010/11 season was not held control of red rice plants in this treatment (control without herbicide and weed free), promoting feedback seedbank red rice. Moreover, this harvest, the seeds of red rice resulting from the 2010/11 season were on the road surface, because, unlike the 2010/11 season, in which the system used was the minimum tillage with tillage after harvest this crop system was used direct seeding, so there is no incorporation of the seeds in the soil profile.
Thus, red rice seeds remained on the soil surface, becoming more exposed to weather conditions such as temperature, moisture, oxygen, and other factors that may have contributed to overcoming dormancy and consequently led to the germination of seeds. Fogliatto et al. (2011), when evaluating the germination of a population of red rice stored under different conditions in the field, observed that the seeds on the soil surface showed rapid release of dormancy and germination. In the United States, Noldin et al. (2006) found a higher percentage of seed dormancy of red rice seeds buried at 12 cm depth, compared to seeds exposed to soil surface.
For the variables red rice seed bank reduction and red rice control, there was no interaction between sowing dates and times of herbicide application (Table 1). At all times sowing decreased red rice seed bank (Table 4), which was lower in sowing dates performed at the beginning of the recommended period and higher at late period. This behavior is similar to what happened in the 2010/11 harvest. This further reduction of the seed bank of red Variables with values †in parentheses indicate original values †, averages not followed by the same letter, lowercase in column and uppercase in row, differ by Tukey test at 5% probability; NS not significant in row; ns not significant in column ; 1/ percentage of red rice emergence in relation to seed bank in PRE to POS application; 2/ number of red rice desiccated plants from 9/20/2011 to the sowing date of each season, 3/ seed bank reduction in relation to red rice seed bank. * Doses of imazapic + imazapyr were: 105+35 g ha -1 (PRE), 2.5+17.5g ha -1 and 52.5+17.5 g ha -1 (PRE+POS), and 105+35 g ha -1 (POS).
rice in the late recommended period is due primarily to the longer period of time to control the seedlings that emerged during the period. In sowing dates early in the period, (09/27 and 10/17) in the average of treatments, 15 and 10 seedlings m -2 were desiccated, respectively, while in late period (11/08 and 12/05) that number was 139 and 78 seedlings m -2 , respectively, which contributed to further reduction of seed bank (Table 4).
As for chemical control (Table 4) there was no significant control difference among PRE, PRE + POS and POS applications, with control average of 99.7% in the dates of application. As in the 2010/11 harvest, this control efficiency is due to appropriate irrigation management and the fact that the red rice biotypes present in the area are not resistant to the imidazolinone group of herbicides. In addition, this season had fewer red rice seeds in the seed bank compared to the 2010/11 season for the same amount of herbicide, explaining in part the efficiency of control at the time of application only in PRE. However, in absolute terms, was observed reduced control at the time of PRE application, which may in part be related to the adsorption Averages not followed by the same letter, lowercase in column and uppercase in row, differ by Tukey test at 5% probability; NS not significant in row; ns not significant in column * Doses of imazapic + imazapyr were: 105+35 g ha -1 (PRE),,52.5+17.5 g ha -1 and 52.5+17.5 g ha -1 (PRE+POS), and 105+35 g ha -1 (POS).
portion of herbicide to the soil. Thus, part of the herbicides may have been adsorbed on the soil and is not available for uptake by plants (Oliveira et al. 2004).
Corroborating the results of the 2010/11 harvest, sowing date affected grain yield (Table 5), which was higher in the sowing of 09/27, this being the best time of sowing, and lower on 12/05, with an average grain yield of treatments PRE, PRE + POS and POS of 9,213 and 7,516 kg ha -1 , respectively -a decrease of 18.4%, equivalent to 25 kg ha -1 day -1 . These results indicate a larger number of grains per panicle, greater thousand grain weight and lower spikelet sterility occurred at 09/27 (Table 5), due to better conditions of temperature and solar radiation during the reproductive period ( Figure 1B, D) compared to the climate conditions of the last date (12/05).
In the Philippines, Yang et al. (2008) evaluated the effect of temperature and solar radiation on the weight of grain filling rate and duration of grain filling in different rice genotypes and found a linear increase in yield with increasing temperature and solar radiation accumulated. In Bangladesh, Islam & Morison (1992) also evaluated the influence of solar radiation and temperature on the yield of rice, and found a positive relationship between grain yield and solar radiation during the reproductive and maturing stages. According to , high yield is a result of high accumulation of biomass and nitrogen that occurs with high solar radiation availability. Lack et al. (2012) also associated highest yield with grain filling in appropriate temperature conditions. Thus, sowing date is one of the most important management practices in irrigated rice. However, the isolated use of this practice may not be an important strategy in controlling red rice, and one should then associate sowing time with temperature conditions that hinder the emergence of red rice at the time of sowing to the final irrigation of rice, thus enabling lower emergence of red rice, in addition to matching the reproductive period of the plant with the best weather conditions. Sowing early in the recommended season provides higher grain yield of rice. The application of imazapic + imazapyr at doses of 52.5 +17.5 g ha -1 in PRE + 52.5 +17.5 g ha -1 in POS and 105 +35 g ha -1 only in PRE and POS was effective in controlling red rice, but irrigation management must be adequate and with no presence of red rice biotypes resistant to these herbicides.
Em vista disso, o trabalho teve por objetivo avaliar o rendimento de grãos e a efi ciência do uso de água de arroz irrigado na semeadura no início e fi nal da época recomendada para a Depressão Central do Rio Grande do Sul, uma das regiões orizícolas de grande relevância deste estado.