Thermal time in sprinkler-irrigated lowland rice

The objective of this work was to evaluate methods of thermal time calculation and the duration of the development stages of lowland rice (Oryza sativa) irrigated by sprinkling. The experiment was conducted during three growing seasons (2010/2011, 2011/2012, and 2014/2015), with five irrigation water depths, six cultivars, and four replicates. Six methods of thermal time calculation were tested: two using the minimum basal temperature; two using the minimum and optimum temperatures; and two using the minimum, optimum, and maximum basal temperatures. For the thermal time calculation, the crop development cycle was divided into the vegetative, reproductive, and grain-filling phases. The methods that used the three cardinal temperatures showed the lowest coefficients of variation for most of the developmental phases. Both irrigation water depths and rice cultivars affected the thermal time of the development stages. The greater the water availability, the lower the duration of the development cycle. Thermal time values depend on the calculation method.


Introduction
Rice (Oryza sativa L.) cultivation in the western region, known as Fronteira Oeste of Rio Grande do Sul state, Brazil, is carried out predominantly by flooded irrigation.In this system, water demand is high, as maintaining a permanent water depth on the soil surface requires water saturation, besides the replacement of the water lost by crop evapotranspiration, percolation and lateral flow in the crop area and in irrigation channels (Reunião…, 2010).Sprinkler irrigation is an alternative to this system, in which water is provided only to supply crop evapotranspiration, which reduces other losses.Therefore, some producers of Fronteira Oeste have chosen to cultivate sprinkler-irrigated rice to reduce water consumption and facilitate crop rotation.
However, sprinkler-irrigated lowland rice cultivation may alter the basic growth and development processes already known under flooded-irrigation cultivation.In a study under flooded and nonflooded environments, Bosco et al. (2009) showed that growth and development of lowland rice are affected by the interaction between genetic and environmental factors.Thus, its necessary Pesq. agropec. bras., Brasília, v.52, n.7, p.475-484, jul. 2017 DOI: 10.1590/S0100-204X2017000700001 to investigate the effect of the irrigation system on basic processes of growth and development of rice cultivars traditionally grown in flooded soil.
Air temperature (Streck et al., 2008) and water availability (Cho & Oky, 2012) are the two abiotic factors that most affect plant growth and development.The temperature effects on these processes have been evaluated by using thermal calculation values in degree-days (Renato et al., 2013).This method is based on the fact that plants require the summation of thermal units to complete their developmental stages.
Thermal time can be calculated by several methods, from which the simplest one is that which measures the thermal accumulation above a lower threshold value (lower basal temperature).However, from a biological point of view, plant response to temperature is best represented by the use of three cardinal temperatures, that is, the lower basal, optimum, and upper basal temperatures (Erpen et al., 2013).
The duration of plant development stages is important for defining crop management practices, such as the choice of sowing dates and application of fertilizers and other chemical products (Freitas et al., 2006), calibration of growth, and development simulation models (Rosa et al., 2015), as well as for the irrigation management.
The objective of this work was to evaluate methods of thermal time calculation and the duration of the development stages of lowland rice irrigated by sprinkling.

Materials and Methods
The study was carried out at the experimental area of Universidade Federal do Pampa (Unipampa), located in the municipality of Itaqui, in the state of Rio Grande do Sul, Brazil (29°09'22"S, 56°33'03"W, at 74 m altitude).The climate of the region is a Cfasubtropical without a defined dry season -according to Köppen-Geiger's classification.The soil is classified as a Plinthudult (Santos et al., 2013).
Irrigation requirement was determined from the ETc obtained by multiplying the reference evapotranspiration (ETo) by the crop coefficient (Kc).ETo was estimated by the Penman-Monteith equation (Allen et al., 1998) based on data of temperature, air humidity, wind speed, and solar radiation, obtained at an automatic weather station located 200 m from the cultivation site.Crop coefficient (Kc) values were: 1.05, up to 20 days after emergence (DAE); 1.125, from 21 to 40 DAE; 1.2, from 41 to 95 DAE; and 0.9, from 95 DAE on (Allen et al., 1998).
Irrigation was performed with a fixed irrigation shift of two days.Irrigation volume was calculated by the sum of ETc, divided by the irrigation system efficiency (70%).When the precipitation volume was above the accumulated ETc since the last irrigation, one day was not accounted for ETc, or two days when the volume was ≥50 mm.
Fertilization was carried out according to the recommendation for rice cultivation by the Comissão de Química e Fertilidade do Solo, for soils of Rio Grande do Sul and Santa Catarina (Manual…, 2004), for an expected grain yield of 12 Mg ha -1 .The control of weeds, insects, and diseases, as well as other crop management practices, were carried out whenever necessary, according to technical recommendations for irrigated rice cultivation in Southern Brazil (Reunião…, 2010) for maintain plants free of stress.
Phenology was determined according to Counce et al. (2000) scale.The emergence (EM) was considered when 50% of plants were visible above the ground.The other development stages -panicle differentiation (R1), anthesis (R4), and physiological maturity (R9)were monitored daily.In the 2010/2011 and 2011/2012 growing seasons, the development stages were Pesq. agropec. bras., Brasília, v.52, n.7, p.475-484, jul. 2017 DOI: 10.1590/S0100-204X2017000700001 monitored in ten randomly selected plants per subplot.The beginning of each developmental stage was considered as the date at which 50% of plants were at the same phenological stage.In the 2014/2015 growing season, four plants were monitored per subplot right after their emergence, and the date of the beginning of each stage was registered.
The efficiency of TTa methods was evaluated by the coefficients of variation (CV) of cultivars and irrigation water depths, in the three growing seasons (2010/2011, 2011/2012, and 2014/2015).The statistical analysis was performed for all development stages only in the 2014/2015 growing season, as the dates at which the development stages occurred in the other growing seasons were considered when the averages of all replicates, in each treatment, were at each respective development stages.Therefore, in those growing seasons, the date of each replicate was not registered separately.The data of the 2014/2015 growing season were subjected to the analysis of variance, and treatments were compared by the Tukey's test, at 5% probability.
(2) Lago et al. (2009). (3)Steinmetz ( 2004).  of rainfall (340.0 mm) were observed, with values lower than the half of the expected volume for the period, according to climatologic normal for the place (Wrege et al., 2011).Temperature (26.0°C) and relative humidity (63.1%) above and below the climatologic normal (24.9°C and 75.0%, respectively) were also registered; this can be explained by the effect of La Niña phenomenon during the period (NOAA, 2016).The 2014/2015 growing season had intermediate conditions, with 300 mm rainfall during the vegetative phase, which is above the climatologic normal, and 120 mm during the reproductive phase, which is below the historical monthly mean (Wrege et al., 2011).
A tendency of reducing CV values was observed with water increasing availability, and with the use of three cardinal temperatures for calculating thermal time (Table 2).These results agree with those found by Rosa et al. (2009) in a study on wheat cultivars.These authors reported that the standard deviation of values obtained by methods with three cardinal temperatures is lower than that obtained with other methods, since the former better represents the interaction between plant development and air temperature.
In the three evaluated growing seasons, a tendency of a reduced development cycle was observed as the water availability increased (Figure 3), but with a considerably value variability among years.By using the method 1.1, thermal time of the entire plant cycle (EM-R9), at a water depth of 100% ETc, ranged from 1,696.2 to 2,000.8°Cday.In addition, in the three growing seasons, plants with water depths of 0 and 50% ETc senesced before completing the reproductive and grain-filling stages, respectively.
In the 2014/2015 growing season, irrigation water depths affected the TTa values of vegetative and reproductive phases (Figure 3).For the vegetative phase (EM-R1), a difference of 30°C day was observed between values obtained with the water depths of 0 and 200% ETc for all TTa methods.Boonjung & Fukai (1996) stated that the vegetative phase duration depends on the intensity and duration of water restriction for rice.Despite the high precipitation observed over the period (Figure 1 C), data found in the present work corroborate those found by Boonjung & Fukai (1996) and Farooq et al. (2009), who stated that a reduction of water availability reduces the development rate, which results in a longer duration of the vegetative phase.
As rice plants subjeted to dry conditions (0% ETc) did not complete the reproductive phase (R1-R4) in the 2014/2015 growing season, the statistical analysis was performed only with water depths from 50 to 200% ETc.The difference between them was 42.8°C day more for TTa at the R1-R4 phase, in the comparison between the water depth of 50% ETc and that of 200% ETc.This difference may be related to canopy temperature, as water deficit causes its rise, which reduces the photosynthetic capacity and transpiration rate of plants (Yan et al., 2012), reflecting on the duration of the phenological phases.
For the grain-filling phase (R4-R9), in the 2014/2015 growing season, no significant difference was observed in TTa values, considering the different irrigation water depths, regardless of method employed for the thermal time calculation.In a study with upland rice cultivars, with different water availability, Lilley & Fukai (1994) also observed that the duration of the R4-R9 phase responded little to water availability.
Values of the coefficient of variation for the different groups of cultivars (Table 2) were similar between methods for each developmental phase.However, in general, the lowest values were obtained using the method 3.1.Streck et al. (2007a) calculated the thermal time by using this method, and observed a TTa variation of approximately 100°C day for the duration of the entire cycle (EM-R9) of irrigated rice.In the present study, this variation was 269.1°C day.The difference between these results may be attributed to the cultivation under sprinkler irrigation, in the present work.In the 2014/2015 growing season, a significant difference was observed for TTa values in the vegetative and reproductive phases for all calculation methods.For the grain-filling phase, difference was observed for the methods 2.1, 3.1, and 3.2 (Figure 4).The difference for the vegetative phase (EM-R1) is due to the genetic characteristics of cultivars.According to Lago et al. (2009), cultivars from hybridization show a lower minimum temperature (8.0ºC) than the other cultivars (9.3ºC).Therefore, a greater thermal accumulation can be observed, even if these cultivars are classified as medium-cycle ones.
For the reproductive phase (R1-R4), the cultivar IRGA 429 showed the highest TTa values (415.2 to 613.8°C day).In a study with lowland rice cultivars, Freitas et al. (2006) observed a variation in the duration of developmental phases according to the cultivar cycle.IRGA 429, classified as a long-cycle cultivar, showed a higher TTa in this period than the other cultivars of medium cycle (IRGA 428 and XP 102 CL).
In the grain-filling phase (R4-R9), the methods 2.1, 3.1, and 3.2 differed significantly according to cultivar cycle, with higher values of TTa for the cultivars with long cycle (IRGA 429), and lower  values for medium-cycle cultivars (IRGA428 and XP 102 CL).Probably the methods that use only one cardinal temperature (Tb) cannot differentiate the cycle duration of cultivars in short periods, such as that at R4-R9 stage.This probably hides differences in the developmental cycle duration, identified by methods that consider more than one cardinal temperature (Tb, Topt, and TB).
When the entire developmental cycle (EM-R9) was considered, in the three growing seasons, the TTa values obtained with the method 3.1 ranged from 1,245.3 to 1,669.8°Cday.Using the same method, Streck et al. (2007a) found that medium-and longcycle lowland rice cultivars required 1,750 to 2,100°C day to complete the EM-R9 cycle.Therefore, the sprinkler irrigation system provided lower TTa values than those found in flooded systems by these authors.This difference can be explained by the presence of water above the soil in the flood irrigation system, which can reduce the thermal amplitude in the canopy.Thus, as a consequence of the absence of this excess of water in the sprinkler irrigation system, temperatures can reach higher values and lead to higher rates of development.Besides, the study of Streck et al. (2007a) was carried out using different sowing dates, with higher TTa values at earlier sowings, which may also have contributed to the differences between values in the two studies.
In the three growing seasons, for the different irrigation water depths and rice cultivars, TTa values were higher with the use of methods 1.1 and 1.2, intermediate with methods 2.1 and 2.2, and lower with the methods 3.1 and 3.2.According to Rosa et al. (2009), this occurs when the mean daily temperature is above the optimum temperature for plants.
In general, values of coefficients of variation for the complete crop developmental cycle were low in methods that consider the three cardinal temperatures.However, analyzing the values at each phase separately allows the identification of high values, which affects the accuracy on the determination of the duration of each development phase.This is probably due to the water availability at these stages.However, for this possibility to be evaluated, it is necessary to adapt and calibrate the linear and nonlinear development models, in order to improve the precision of the duration estimate of developmental stages under sprinkler irrigation.

Conclusions
1.The use of different methods provide different values of thermal time, considering the same development stages of sprinkler-irrigated lowland rice (Oryza sativa); and methods that consider more than one cardinal temperature show less variation between growing seasons.
2. The lower the water depth provided by sprinkler irrigation, the longer the lowland rice cycle.

Figure 1 .
Figure 1.Rainfall and irrigation for 100% water depth of the crop evapotranspiration (ETc), during the developmental cycle of sprinkler-irrigated lowland rice in three growing seasons: A, 2010/2011; B, 2011/2012; and C, 2014/2015.Dashed lines correspond to the average dates of occurrence of panicle differentiation (R1), anthesis (R4), and physiological maturity (R9) for lowland rice cultivars using a water depth of 100% ETc.

Table 1 .
Cardinal temperatures estimated by different authors for different developmental phases of lowland rice (Oryza sativa) cultivars.(1)