Yield of aerobic rice in rainfed lowlands of the Philippines as affected by nitrogen management and row spacing

Field Crops Research 116 (2010) 165–174

Contents lists available at ScienceDirect

Field Crops Research journal homepage: www.elsevier.com/locate/fcr

Yield of aerobic rice in rainfed lowlands of the Philippines as affected by nitrogen management and row spacing R.M. Lampayan a,*, B.A.M. Bouman a, J.L. de Dios b, A.J. Espiritu b, J.B. Soriano c, A.T. Lactaoen d, J.E. Faronilo a, K.M. Thant a a

International Rice Research Institute, DAPO Box 7777, Metro Manila, Philippines Philippine Rice Research Institute, Science City of Mun˜oz, Nueva Ecija, Philippines Bulacan Agricultural State College, San Ildefonso, Bulacan, Philippines d National Irrigation Administration, Tarlac, Philippines b c

A R T I C L E I N F O

A B S T R A C T

Article history: Received 30 July 2009 Received in revised form 10 December 2009 Accepted 12 December 2009

Management practices need to be developed for aerobic rice, a system in which rice is grown under nonflooded conditions in nonsaturated soil. We evaluated the effects of amount and timing of fertilizer nitrogen (N) application and of row spacing on the yield of aerobic rice under rainfed conditions in the 2004 and 2005 wet seasons in 3 and 2 locations, respectively, in Central Luzon, Philippines. N timing and management were also evaluated under irrigated conditions at one location in the dry season in 2005. Yields were 3.1–4.9 t ha1 with 60–150 kg ha1 of applied N. Yields increased with N rate, up to rates of 60–150 kg ha1 depending on site and season, but at rates beyond 90 kg ha1 the risk of lodging increased, especially in the wet season. Yields were similar for different splits of N over time, and the regional practice for lowland rice of three to four splits can also be used for aerobic rice. Yields were similar for row spacings ranging from 25 to 35 cm. Although the number of panicles per square meter was significantly higher at 25-cm spacing than at 35-cm spacing, this difference was compensated for by significantly more spikelets per panicle at 35-cm spacing, while spikelet fertility and grain weight were similar for all row spacings. Lodging and bending resistance of stems were not affected by row spacing. The results suggest that a row spacing of 35 cm can be used to enable easier weeding between the rows, and allows for mechanized field operations in which tire tracks do not damage the crops. ß 2009 Elsevier B.V. All rights reserved.

Keywords: Aerobic rice Lodging Nitrogen management Rainfed Row spacing

1. Introduction Aerobic rice is a relatively new cropping system in which rice is grown under nonflooded, nonpuddled, and nonsaturated soil conditions (Bouman, 2001). Because aerobic rice needs less water at the field level than conventional lowland rice, the system is targeted at relatively water-short irrigated or rainfed lowland environments. To achieve high yields under aerobic soil conditions, new varieties are required that combine the drought-resistant characteristics of upland varieties with the high-yielding characteristics of lowland varieties (Lafitte et al., 2002). In northern (temperate) China, breeders have been breeding ‘‘aerobic rice’’ varieties since the mid-eighties and have developed varieties with an estimated yield potential of around 6 t ha1 using about half the amount of water required for lowland rice (Wang et al., 2002; Yang et al., 2005; Bouman et al., 2006). The development of aerobic rice

* Corresponding author. Tel.: +63 2 891 1236; fax: +63 2 580 5699. E-mail address: [email protected] (R.M. Lampayan). 0378-4290/$ – see front matter ß 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.fcr.2009.12.007

for the tropics is of relatively recent origin. Using improved upland rice varieties, George et al. (2002) reported yields of 1.5–7.4 t ha1 under aerobic conditions in the Philippines but with 2500– 4500 mm of annual rainfall. The yields over 6 t ha1 occurred only in the first 2 years of cultivation after non-rice crops, and most yields were in the 2–3 t ha1 range. Atlin et al. (2006) reported aerobic rice yields of 3–4 t ha1 using recently developed varieties in farmers’ fields in rainfed uplands in the Philippines. Though the amount of rainfall was not reported, the conditions of the trials were described as ‘‘well watered’’. Bouman et al. (2005) and Peng et al. (2006) quantified yield and water use of the recently released upland rice variety Apo under irrigated conditions. In the dry season, yields were 4–5.7 t ha1, with 744–924 mm of total water (rain plus irrigation); in the wet season, yields were 3.5–4.2 t ha1 with 922–1301 mm of water, compared with typical inputs for puddled transplanted rice of 1500–2000 mm in this region. Based on these early results, Atlin et al. (2006) and Bouman et al. (2005) suggested that the system of aerobic rice could be an attractive option for farmers in rainfed lowlands with limited or erratic distribution of rainfall.

166

R.M. Lampayan et al. / Field Crops Research 116 (2010) 165–174

In 2003, research began on aerobic rice in rainfed lowlands in the Philippines (Lampayan et al., 2004). The first objective was to derive management recommendations for nitrogen (N) fertilizer in terms of timing and distribution within the season. In the wet season of the Philippines, heavy winds frequently cause lodging in rice (Tabbal et al., 2002), and N fertilization should be balanced between maximizing yield and minimizing the risk of lodging. The second objective was to optimize row spacing. Rice that is not permanently flooded tends to have more weed growth and a broader weed spectrum than rice that is permanently flooded (Mortimer and Hill, 1999). To limit the use of herbicides and (laborintensive) manual weed removal in aerobic rice, mechanical interrow cultivation is a promising alternative. Wider spacing between the rows is needed to avoid damage to the crop caused by the tractor wheels. However, wide row spacing may result in yield loss compared with a narrow row spacing. Moreover, during participatory field observations with farmers, it was suggested that row spacing may have an effect on lodging resistance. Therefore, field experiments were undertaken to study the interactive effects of fertilizer-N application and row spacing on yield and lodging resistance of aerobic rice. Here, we report on the results of field experiments conducted in 2004 and 2005 in the Philippines. 2. Materials and methods We conducted two field experiments at three locations in Central Luzon. The first experiment studied the interaction of amount of N application and row spacing (N  RS), and was conducted in the wet seasons (June–October) of 2004 and 2005: (i) on a loamy sand soil (78% sand; 5% clay) at the village of Dapdap (158400 N, 1208330 E) in Tarlac Province and (ii) on a clay soil (23% sand; 58% clay) at the experimental station of the National Soil and Water Resources Research and Development Center – Bureau of Soils and Water Management at San Ildefonso (158040 N, 1208570 E) in Bulacan Province. The second experiment studied the interaction of N splits and row spacing (NS  RS). This experiment was conducted on a loam soil (37% sand; 17% clay) at the experiment ˜ oz (158390 N, 1208540 E) in Nueva Ecija station of PhilRice at Mun Province, in the wet season of 2004 and the dry season of 2005 using irrigation. The 20-year (1986–2005) average annual rainfall is 1600 mm in Tarlac, 2017 mm in Bulacan, and 2000 mm in Nueva Ecija, with 80–90% of the rain falling between May and October in all provinces. In Dapdap, the site was previously used by farmers and cropped with lowland rice in the rainy season and left fallow in the dry season. At San Ildefonso, the site was previously cropped with lowland rice in the rainy season and upland crops in the dry ˜ oz, the site was planted first with rice in the 2004 season. At Mun dry season after a long fallow for several years. 2.1. Experimental layout All experiments were laid out in a split-plot design with N as the main factor and row spacing as a sub-factor, with four replicates, and split-plot sizes of 4 m  8 m. In the N-amount by row spacing experiment at Dapdap and San Ildefonso, five N amounts were used: 0 (N0), 60 (N60), 90 (N90), 120 (N120), and 150 kg N ha1 (N150). The N fertilizer was applied as urea by side-dressing each plant row in three splits over time: 30% at 10–14 days after emergence (DAE), 40% at 30–35 DAE, and 30% at 45–50 DAE. In the ˜ oz, a total fixed rate of N-split by row spacing experiment at Mun 100 kg N ha1 was applied in five different split applications: 0-3030-30-10 (NS1), 0-20-50-30-0 (NS2), 0-20-30-50-0 (NS3), 23-2329-25-0 (NS4), and 18-0-29-43-10 (NS5) kg ha1 at 0 DAE (basal), 10–14 DAE, 30–35 DAE, 45–50 DAE, and 60 DAE. Three row spacings were used in both experiments: 25 cm (RS25), 30 cm (RS30), and 35 cm (RS35).

Table 1 Monthly rainfall (mm) in 2004 and 2005 in the experimental sites in Dapdap, San ˜ oz, Philippines. Ildefonso and Mun Month

January February March April May June July August September October November December Total a b

Dapdapa

San Ildefonsoa

˜ ozb Mun

2004

2005

2004

2005

2004

2005

1 9 9 35 191 380 261 306 187 126 110 98

0 0 3 54 92 191 128 155 412 197 44 27

4 11 77 20 231 235 491 453 203 73 270 63

2 3 34 57 197 281 247 184 248 270 90 129

1 1 1 37 208 287 447 668 223 99 157 82

0 0 18 2 106 256 95 335 190 400 49 61

1713

1302

2129

1742

2212

1513

N-amount by row spacing (N  RS) experiment. N-split by row spacing (NS  RS) experiment.

In all experiments, the land was dry ploughed and harrowed 3 weeks before the onset of the rains. All main plots were bunded to avoid movement of fertilizer-N across plots. All plots received 20 kg ha1 ZnSO4 and 60 kg ha1 of P and K in two equal splits as a basal application (broadcast before sowing) and topdressing 30–35 DAE. Dry seeds of variety Apo (PSBRc 9) were sown manually at 60 kg ha1 in slits of 2–3-cm depth created using a wooden dented harrow (locally known as lithao). The seeds were covered with soil to promote seed–soil contact and to protect them from birds. Sowing and harvest dates are given in Table 1. Weeds were controlled by the application of butachlor at the recommended rate 2 days after sowing, and followed by interrow cultivation using lithao at 15–25 DAE. The experiments in the wet season were ˜ oz in the dry season of purely rainfed, whereas the crop at Mun 2005 received flush irrigations during the growing season (amounts of about 2-cm depth during the first 4 weeks of crop growth, and of 4 cm thereafter). 2.2. Measurements Plant height was measured weekly from the four permanent sample plants in each plot. Measurement was done from the base of plant to the tip of the longest leaf or panicle of the plant. At ˜ oz, two adjacent 50-cm row length plant samples Dapdap and Mun were taken outside of the central final harvest area of each subplot at mid-tillering, panicle initiation, flowering, and physiological maturity to determine total dry crop biomass and leaf area index (LAI). Dry biomass was determined after oven-drying at 70 8C to constant weight. The LAI was determined using a Licor LI 3100 area meter. At San Ildefonso, samples were taken only at physiological maturity to determine biomass. To characterize lodging resistance, we measured stem resistance to bending at all sites at panicle initiation, flowering, and physiological maturity using the methodology described by Tabbal et al. (2002). We used a push-gauge to push a 10-cm row section of plants to a 458 angle from the vertical at 10-cm height from the base. The measured force was standardized by dividing by number of stems in the 10-cm row section. At maturity, plants from 6-m2 area in the center of each subplot were harvested for determination of yield, whereas 4 row samples of 0.5 m each located at the corners outside of the harvest area were taken for determination of final aboveground biomass, yield components (panicle density, number of spikelets per panicle, 1000-grain weight, and percentage filled spikelets). Grain yield was expressed at 14% moisture content. The percentage lodged area in each subplot was estimated visually at the time of harvest.

R.M. Lampayan et al. / Field Crops Research 116 (2010) 165–174

The hydrological conditions of our experiments were monitored to enable comparison across sites and with literature data (Bouman et al., 2005). Soil moisture tension at 15–20-cm depth was measured daily using gauged tensiometers placed in between the plant rows. At San Ildefonso and Dapdap, soil moisture tension was measured in all 25-cm row spacing subplots of N90, N120, and ˜ oz, soil moisture tension was measured in the NS3, N150. At Mun NS4 and NS5 subplots also at 25-cm row spacing. At all sites, the water table depth was measured in four fully perforated 1.75-mlong PVC pipes installed in the center of the bunds between the ˜ oz, daily readings were made main plots. At San Ildefonso and Mun manually, whereas, at Dapdap, automatic recordings were made using groundwater monitor D-Diver dataloggers (Van Essen Instruments) connected to a data logger. Rainfall, maximum and minimum temperature, and Class A pan evaporation rate were measured daily at meteorological stations at all sites.

167

3.2. Hydrology In the wet season, groundwater was deeper in 2005 than in 2004 at San Ildefonso and Dapdap sites (Fig. 1). In 2004, the ˜ oz, with an average depth groundwater table was deepest at Mun of 128 cm in the wet season. At San Ildefonso, the average depth of the groundwater in the wet season was only 17 cm in 2004 and 53 cm in 2005. At Dapdap, it was 41 cm in 2004 and 93 cm in 2005. Across sites and years, the soil moisture tension patterns (Fig. 2) mirror the groundwater patterns: deeper groundwater tables coincide with dryer soils and higher soil moisture tensions. The seasonal-average (from 15 to 100 DAE) soil moisture tension in the wet season was 4.9 kPa in 2004 and 3.4 kPa in 2005 at San Ildefonso and 3.7 kPa in 2004 and 12 kPa in 2005 at Dapdap. At ˜ oz, the seasonal-average tension was 6.4 kPa in the wet Mun season of 2004 and 2.6 kPa in the dry season of 2005 due to irrigation.

2.3. Data analysis 3.3. Crop growth Data were analyzed using the IRRISTAT mixed model (IRRI, 2003) for the analysis of variance (ANOVA) for split-plot replicated experiments. The least significant difference (LSD) test, with the level of significance set at 5%, was used to test for significant differences among treatment means. Linear regression and correlation analysis for selected variables were also done.

3. Results 3.1. Weather The total amount of rainfall was higher in 2004 than in 2005 at all sites (Table 1). Heavy rains brought by typhoons in July and August in 2004 flooded San Ildefonso and Dapdap, respectively, for 2 days. At San Ildefonso, the cumulative pan evaporation in the wet season was 64 mm higher in 2005 than in 2004 (Table 2), mostly caused by higher daily evaporation rates in July 2005 (data not shown). However, at Dapdap, this trend was the opposite. At ˜ oz, the 2004 wet-season cumulative rainfall was about Mun 1500 mm, which was much higher than at San Ildefonso (1210 mm) and Dapdap (984 mm). In Central Luzon, the total rainfall during the cropping season (from December to April) is ˜ oz, only 20 mm of rainfall was recorded in usually very low. At Mun the dry season of 2005, compared with pan evaporation of 657 mm. Temperature regimes were comparable across sites, years and seasons (Table 2). Daily mean minimum and maximum temperatures during the 2004 wet season ranged from 25.1– 31.2 8C, 33.2–24.08 and 32.5–2.7 8C at San Ildefonso, Dapdap and ˜ oz sites, respectively. Mun

N-amount  row spacing. There were no significant interactions between N-amount and row spacing for LAI, biomass, or plant height in any of the experiments. There were significant effects of N rate on LAI at Dapdap, more so at later growth stages (Fig. 3a). Higher N application rates generally resulted in higher LAI, however differences between N120 and N150 were relatively small and seldom significant. The highest values at flowering were around 4.2 in N120 and N150, whereas the lowest values were around 1.2 in N0. Averaged over fertilizer-N amounts, differences among row spacings were statistically significant though relatively small: wider row spacings had higher LAI (data not shown). For example, at flowering, the difference in average LAI between RS25 and RS35 was only 0.9 in 2004 and 0.6 in 2005. In Dapdap, differences among N rates were mostly significant at flowering and maturity, with higher N rates resulting in higher biomass accumulation (Fig. 4a). The largest differences occurred between N0 and N60, whereas additional supply of N above 60 kg ha1 had a relatively small effect. In 2005, biomass accumulation was lower than in 2004, possibly caused by the dryer soil conditions due to less rain and a deeper water table (Fig. 1 and Table 1). In 2004, there were no statistical differences in biomass accumulation among the row spacings. In 2005, the biomass in RS35 was significantly lower than in RS25 and RS30 at flowering and maturity, though, in absolute terms, this difference was very small. At other growth stages and between RS25 and RS30, there were no differences in biomass accumulation. At both sites, plant height was significantly affected by N treatment, but not by row spacing (data not shown). Plants were taller with increasing N rate, with the biggest incremental increase

Table 2 Total rainfall and evaporation, and mean maximum and minimum temperatures from seeding to harvest in San Ildefonso and Dapdap experiments during 2004 and 2005 wet ˜ oz experiments during 2004 wet season (WS) and 2005 dry season (DS). seasons (WS), and in Mun Location and season

Date of sowing

Date of harvest

Total rainfall (mm)

Total evaporation (mm)

Mean max. temp. (8C)

Mean min. temp. (8C)

15 June 21 May

6 October 13 September

1210 911

495 559

30.9 31.2

25.1 25.5

Dapdapa WS 2004 WS 2005

19 June 19 June

11 October 10 October

984 897

548 468

33.2 33.2

24.0 24.2

˜ ozb Mun WS 2004 DS 2005

15 June 16 December

10 October 15 April

1492 20

462 657

32.5 31.8

23.6 21.7

San Ildefonso WS 2004 WS2005

a b

a

N-amount by row spacing (N  RS) experiments. N-split by row spacing (NS  RS) experiments.

168

R.M. Lampayan et al. / Field Crops Research 116 (2010) 165–174

Fig. 1. Daily groundwater level fluctuations in the N-amount by row spacing (N  RS) experiments in San Ildefonso (a) and Dapdap (b) during 2004 and 2005 wet seasons, and ˜ oz (c) during 2004 wet season (WS) and 2005 dry season (DS). FL = flowering; HV = harvest. in the N-split by row spacing (NS  RS) experiments in Mun

between N0 and N60. Plants were taller at San Ildefonso (maximum of 140–150 cm at maturity) than at Dapdap (maximum of around 120 cm at maturity), possibly reflecting the wetter soil conditions at San Ildefonso. N-split  row spacing. There was no interaction between N-split and row spacing for LAI, biomass, or plant height in any of the ˜ oz. experiments at Mun The timing of N application significantly affected LAI only in the 2004 wet season (Fig. 3b). Averaged over row spacing, LAI at the mid-tillering stage (30 DAE) was highest in NS4 in both seasons. At this stage, 46% of the total N was applied as basal and at 10–14 DAE in NS4, whereas the other four N treatments had received only 18– 30% of total N. At flowering, LAI in the 2004 wet season ranged from 6.3 to 8.2, with higher values in treatments with higher N application at mid-tillering and panicle initiation (NS2 and NS3). In the 2005 dry season, LAI at flowering ranged from 4.3 to 5.5, with the highest values in NS1 and lowest values in NS3. The effect of row spacing was significant only in the 2004 wet season, and only at flowering, when LAI of RS35 was 0.7 higher than LAI of RS25. In the 2005 dry season, there were no significant differences in ˜ oz at biomass accumulation between the N-split treatments in Mun any stage (Fig. 4b). In 2004 wet season, there were differences in

vegetative stage, but not at physiological maturity. Row spacing generally had no effect on biomass in both seasons, except at physiological maturity in the 2004 wet season, when average biomass was higher at RS25 (10,000 kg ha1) than at RS35 (9200 kg ha1) (data not shown). Plant height was not significantly affected by N-split or row spacing treatments in both seasons (data not shown). Plants were taller in 2004 wet season (120 cm), than in 2005 dry season (110 cm). 3.4. Bending resistance and lodging N-amount  row spacing. There were no significant interactions between N level and row spacing for bending resistance at both San Ildefonso and Dapdap. At both sites and in both years, bending resistance increased from panicle initiation to flowering and then decreased at maturity (Fig. 5a and b). Bending resistance was generally higher at San Ildefonso than at Dapdap, especially in 2004. At maturity, there was no significant or consistent effect of N application on bending resistance, except at San Ildefonso in 2004, when N0 had the lowest bending resistance. The application of (any amount of)

R.M. Lampayan et al. / Field Crops Research 116 (2010) 165–174

169

Fig. 2. Soil moisture tension at 10–15-cm depth in the N-amount by row spacing (N  RS) experiments in San Ildefonso (a) and Dapdap (b) during 2004 and 2005 wet seasons, ˜ oz (c) during 2004 wet season (WS) and 2005 dry season (DS); FL = flowering; HV = harvest; error bars are and in the N-split by row spacing (NS  RS) experiments in Mun standard error (SE).

fertilizer-N increased bending resistance significantly over N0 at both sites in both years only at panicle initiation and flowering. Differences in bending resistance among the treatments N60, N90, N120, and N150 were relatively small and inconsistent. There were no significant differences in bending resistance among row spacing. No lodging occurred in the N0 treatments at all sites in both years (Table 3). Except for San Ildefonso in 2005 where no lodging occurred at all, lodging occurred only with fertilizer-N application and increased with increasing fertilizer-N rate. At Dapdap in both years, around half of the plants in all plots lodged in N120, whereas, in N150, up to 75% of the plants lodged. In all experiments, there was no significant effect of row spacing on lodging. There was no strong statistical correlation between lodging and bending resistance at Dapdap and San Ildefonso. The r2 between lodging and bending resistance at flowering over all three N treatments (N90, N120, and N150) was 0.01 at San Ildefonso in 2004, 0.16 at Dapdap in 2004, and 0.61 at Dapdap in 2005.

N-split  row spacing. There were no significant interactions ˜ oz. between N-split and row spacing for bending resistance at Mun Fig. 5c shows the bending resistance at panicle initiation, flowering, and physiological maturity in the 2004 wet season. Bending resistance decreased after flowering at the two highest N rates. Bending resistance at physiological maturity was significantly affected by N-split treatment. The highest bending resistance was observed when 20 kg ha1 of N was applied at 10–14 DAE, 30 kg ha1 at 30–35 DAE, and 50 kg ha1, without basal application (NS3). Bending resistance, however, was not significantly affected by row spacing. No lodging occurred at ˜ oz in both years. Mun 3.5. Final biomass, yield, and yield components N-amount  row spacing. There were no significant interactions between N rate and row spacing with respect to final biomass, grain yield, and yield components.

170

R.M. Lampayan et al. / Field Crops Research 116 (2010) 165–174

Fig. 3. Mean leaf area index (LAI) of cultivar Apo at various nitrogen amounts in the N-amount by row spacing (N  RS) experiments in Dapdap during 2004 and 2005 wet ˜ oz during 2004 wet season and 2005 dry season (b); N-amount (N) and seasons (a), and at various nitrogen splits in the N-split by row spacing (NS  RS) experiments in Mun N-split (NS) treatments are explained in the text; data presented are averaged over row spacing; error bars are standard error (SE).

At San Ildefonso, final aboveground biomass and grain yield increased with the application of fertilizer-N up to 60 kg N ha1 in 2004 and up to 90 kg N ha1 in 2005 (Table 4a). Biomass and yields were statistically the same among N90, N120, and N150 in both years

(and N60 in 2004). There were significant differences in panicle number, number of spikelets per panicle, and 1000-grain weight as evidenced by 2% levels of significance in these parameters. Despite the lodging after grain filling in N90, N120, and N150 in 2004, yields

Fig. 4. Dry biomass of cultivar Apo at various nitrogen amounts in the N-amount by row spacing (N  RS) experiment in Dapdap during 2004 and 2005 wet seasons (a), and at ˜ oz during 2004 wet season and 2005 dry season (b); N-amount (N) and N-split (NS) treatments various nitrogen splits in N-split by row spacing (NS  RS) experiments in Mun are explained in the text; error bars are standard error (SE).

R.M. Lampayan et al. / Field Crops Research 116 (2010) 165–174

171

Fig. 5. Bending resistance at 10-cm bending height under varying N amounts at different growth stages in the N-amount by row spacing (N  RS) experiments in San Ildefonso ˜ oz (c) during 2004 wet season; (a) and Dapdap (b) during 2004 and 2005 wet seasons, and at varying nitrogen splits in the N-split by row spacing (NS  RS) experiment in Mun N-amount (N) and N-split (NS) treatments are explained in the text; data presented are averaged over row spacing; error bars are standard error (SE); PI = panicle initiation, FL = flowering, PM = physiological maturity.

were similar to yields in 2005 without any lodging. At Dapdap, there was a consistent trend for increasing biomass and yield with increasing N rate up to 150 kg N ha1. However, biomass and yields were statistically the same for N120 and N150 in both years, despite the significant differences in lodging between these treatments in 2004 (Table 4a). Of the yield components, there was a trend for number of spikelets per panicle to increase with N rate, with significant differences up to 90 or 120 kg N ha1, as for biomass and grain yield, whereas percentage filled spikelets, 1000grain weight, and panicle density were not affected by N rate

Table 3 Effect of nitrogen amount on lodging at harvest in the N-amount by row spacing (N  RS) experiments at San Ildefonso and Dapdap during 2004 and 2005 wet seasons (WS). Factor

Lodging (%) San Ildefonso WS 2004

N-amounta N0 0d N60 3d N90 42 c N120 56 bc N150 62 ab

San Ildefonso WS 2005

Dapdap WS 2004

Dapdap WS 2005

0 0 0 0 0

0 0 8 46 76

0 2 4 44 67

a a a a a

c c c b a

b b b a a

Data in columns followed by the same letter are not significantly different (P = 0.05). Data presented are averaged over all row spacings. a N-amount treatments are explained in the text.

(except for significantly lower 1000-grain weight in N0). Lodging was significantly correlated with yield at Dapdap and San Ildefonso, although R2 values were relatively low (25% at San Ildefonso in 2004, 49% at Dapdap in 2004, and 57% at Dapdap in 2005). Row spacing did not affect grain yield, whereas increasing row spacing reduced final biomass (Table 4b). Row spacing did not affect percentage filled spikelets and 1000-grain weight in any of the experiments. Panicle density generally decreased with increasing row spacing, though this trend was not significant at San Ildefonso in 2005 and was inconsistent at Dapdap in 2005. The number of spikelets per panicle increased significantly with increasing row spacing at San Ildefonso in 2004 and at Dapdap in 2005, but was unaffected by row spacing in the other two experiments. ˜ oz, there were no N-split  row spacing. Generally, at Mun significant interactions between N-split and row spacing with respect to final biomass, grain yield, and yield components, except for the number of spikelets per panicle in the wet season of 2004. Final aboveground biomass and yields were always higher in the dry season than in the wet season (Table 5a). Biomass ranged from 10.8 to 12.1 t ha1 in the dry season and from 9.1 to 10 t ha1 in the wet season, whereas yields ranged from 5.4 to 6.1 t ha1 in the dry season and from 3.2 to 3.9 t ha1 in the wet season. There was a consistent trend for highest biomass and yield in NS1 and lowest in NS4 in both seasons, with significant difference between these treatments in all cases except for biomass in the WS. Biomass

R.M. Lampayan et al. / Field Crops Research 116 (2010) 165–174

172

Table 4a Effect of nitrogen amount on grain yield and yield components of cultivar Apo in the N-amount by row spacing (N  RS) experiments at San Ildefonso and Dapdap during 2004 and 2005 wet seasons (WS). Site and season

N-amounta (kg ha1)

Tillers (no. m2)

Panicles (no. m2)

Spikelets (no. panicle1)

1000-Grain weight (g)

Filled spikelets (%)

Grain yield 14% MC (t ha1)

Total dry biomass at harvest (t ha1)

San Ildefonso, WS 2004

N0 N60 N90 N120 N150

283 378 386 381 361

b a a a a

274 308 346 344 334

b ab a a a

58 115 109 101 104

b a a a a

21.5 22.8 22.7 22.7 22.4

b a a a ab

72.9 76.5 82.5 77.3 75.7

c b a b bc

2.18 4.13 4.64 4.04 3.62

b c c c c

5.52 12.94 16.22 15.23 15.11

c b a a a

San Ildefonso, WS 2005

N0 N60 N90 N120 N150

266 299 303 325 323

b ab ab a a

237 271 267 288 280

b ab ab a ab

89 106 115 108 126

c b ab b a

23.2 23.5 23.4 23.6 22.9

a a a a a

71.8 69.5 70.2 66.9 65.4

a ab ab bc c

2.52 3.77 4.11 4.28 4.28

c b a a a

6.70 9.79 10.35 10.22 11.48

c b ab ab a

Dapdap, WS 2004

N0 N60 N90 N120 N150

379 403 376 373 373

ab a ab b b

358 373 357 350 354

a a a a a

37 60 75 83 88

d c b a a

19.0 21.4 21.9 21.6 21.3

b a a a a

81.9 81.4 84.2 82.7 82.1

a a a a a

1.69 3.31 3.98 4.48 4.85

d c b a a

4.65 9.28 11.19 11.84 12.46

d c b ab b

Dapdap, WS 2005

N0 N60 N90 N120 N150

323 355 360 368 352

b a a a ab

320 331 326 331 323

a a a a a

46 80 87 92 97

c b ab a a

18.7 19.6 19.9 19.8 19.8

b a a a a

69.4 68.1 70.2 68.6 69.2

a a a a a

1.67 3.07 3.35 3.77 4.00

d c bc ab a

3.86 6.93 6.88 8.42 9.27

c b b a a

Data in columns followed by the same letter (within same site and season) are not significantly different (P = 0.05). Data presented are averaged over all row spacings. a N-amount treatments are explained in the text.

Table 4b Effect of row spacings on grain yield and yield components of cultivar Apo in the N-amount by row spacing (N  RS) experiments at San Ildefonso and Dapdap during 2004 and 2005 wet seasons (WS). Site and season

Row spacinga (kg ha1)

Tillers (no. m2)

Panicles (no. m2)

Spikelets (no. panicle1)

1000-Grain weight (g)

Filled spikelets (%)

Grain yield 14% MC (t ha1)

Total dry biomass at harvest (t ha1)

San Ildefonso, WS 2004

RS25 RS30 RS35

371 a 365 a 337 a

359 a 321 b 283 c

92 b 98 ab 102 a

22.4 a 22.7 a 22.2 a

76.6 a 76.4 a 77.9 a

3.76 a 3.72 a 3.69 a

14.22 a 12.74 b 12.04 b

San Ildefonso, WS 2005

RS25 RS30 RS35

317 a 310 a 283 a

279 a 275 a 252 a

107 a 111 a 109 a

23.5 a 23.3 a 23.3 a

69.1 a 68.6 a 68.6 a

3.85 ab 3.91 a 3.61 b

10.12 a 10.10 a 8.94 b

Dapdap, WS 2004

RS25 RS30 RS35

404 a 387 a 352 b

381 a 361 a 333 b

68 a 67 a 70 a

20.9 a 21.0 a 21.1 a

82.8 a 82.7 a 81.9 a

3.65 a 3.65 a 3.69 a

10.58 a 9.78 b 9.30 b

Dapdap, WS 2005

RS25 RS30 RS35

344 a 353 a 363 a

372 a 322 b 284 a

72 c 80 b 89 a

19.5 a 19.6 a 19.6 a

68.5 a 69.1 a 69.8 a

3.20 a 3.17 a 3.15 a

7.45 a 7.29 ab 6.48 b

Data in columns followed by the same letter (within same site and season) are not significantly different (P = 0.05). Data presented are averaged over all N-amount treatments. a Row spacing treatments are explained in the text.

Table 5a ˜ oz during 2004 wet season (WS) and Effect of nitrogen splits on grain yield and yield components of cultivar Apo in the N-split by row spacing (NS  RS) experiments in Mun 2005 dry season (DS). Site and season

N-splita

Tillers number (no. m2)

Panicles (no. m2)

Spikelets (no. panicle1)

1000-Grain weight (g)

Filled spikelets (%)

Grain yield 14% MC (t ha1)

Total dry biomass (t ha1)

˜ oz, WS 2004 Mun

NS1 NS2 NS3 NS4 NS5

319 306 304 340 317

ab b b a ab

281 250 265 307 249

ab b b a b

116 125 130 108 124

bc ab a c ab

21.1 21.1 20.7 21.5 21.0

ab ab b a ab

74.0 72.9 71.4 78.4 72.9

b b b a b

3.90 3.48 3.46 3.24 3.36

a b b b b

10.0 9.6 9.5 9.6 9.1

a a a a a

˜ oz, DS 2005 Mun

NS1 NS2 NS3 NS4 NS5

339 338 330 368 316

ab ab b a b

294 312 295 309 282

a a a a a

139 139 136 134 143

a a a a a

19.5 19.9 19.7 20.0 19.7

b a ab a ab

87.0 82.9 83.5 84.0 86.7

a c c bc ab

6.10 5.80 5.69 5.40 6.10

a ab ab b a

12.1 11.6 11.6 10.8 11.6

a ab ab b ab

Data in columns followed by the same letter (within same site and season) are not significantly different (P = 0.05). Data presented are averaged over all row spacings. a N-split treatments are explained in the text.

R.M. Lampayan et al. / Field Crops Research 116 (2010) 165–174

173

Table 5b ˜ oz during 2004 wet season (WS) and Effect of row spacings on grain yield and yield components of cultivar Apo in the N-split by row spacing (NS  RS) experiments in Mun 2005 dry season (DS). Site and season

Row spacinga

Tillers (no. m2)

Panicles (no. m2)

Spikelets (no. panicle1)

1000-Grain weight (g)

Filled spikelets (%)

Grain yield 14% MC (t ha1)

Total biomass dry weight (t ha1)

˜ oz, WS 2004 Mun

RS25 RS30 RS35

346 a 317 b 289 c

286 a 270 a 256 a

121 ab 114 b 126 a

21.3 a 21.2 a 20.7 b

73.6 a 73.8 a 74.3 a

3.49 a 3.40 a 3.58 a

10.2 a 9.3 b 9.1 b

˜ oz, DS 2005 Mun

RS25 RS30 RS35

377 a 327 b 310 b

328 a 291 b 276 b

139 a 136 a 139 a

19.8 a 19.9 a 19.5 b

84.1 a 85.0 a 85.6 a

5.79 a 5.82 a 5.84 a

11.6 a 11.6 a 11.4 a

Data in columns followed by the same letter (within same site and season) are not significantly different (P = 0.05). Data presented are averaged over all N-split treatments. a Row spacing treatments are explained in the text.

and yields in the other N-split treatments (NS2, NS3, and NS5) were not significantly different from each other and from NS1 and NS4. Except for grain weight, yield components were similar or higher in the dry season than in the wet season for respective treatments, i.e. tiller number (average of 338 m2 in the dry season and 317 m2 in the wet season), panicle density (298 m2 in the dry season and 270 m2 in the wet season), number of spikelets per panicle (138 panicle1 in the dry season and 121 panicle1 in the wet season), and percentage filled spikelets (85% in the dry season and 74% in the wet season). The effect of N-split on yield components was inconsistent, especially on the panicle density and number of spikelets per panicle. In the wet season, only a few significant differences occurred in these two yield components among N-split treatments; in the dry season, no significant differences were observed. Percentage filled spikelets and 1000grain weight, however, were significantly different among some Nsplits in both the wet and dry seasons, whereas tiller number was not significantly affected by N-split treatments. On average, NS4 had the highest tiller and panicle densities, probably caused by the relatively high N application rates in the early vegetative stage. However, NS4 had the lowest number of spikelets per panicle, probably caused by the relatively low amount of N applied in the reproductive stage. As a result, both grain yield and final biomass of NS4 were low. There was a general trend for increasing yield with increasing row spacing in both seasons, but the differences were never significant (Table 5b). Significantly higher tiller and panicle densities were obtained in narrow row spacing (RS25) than in wide row spacing (RS30 and RS35). However, RS35 had the (significantly) highest number of spikelets per panicle in the wet season. Compensation among yield components is the main reason why yields were similar among row spacings. 4. Discussion With 60–150 kg N ha1, rainfed aerobic rice yield of variety Apo in our experiments varied from 3.1 to 4.9 t ha1 during the wet season. This is about the same as was reported by Bouman et al. (2005) and Peng et al. (2006) with yield of up to 4.6 t ha1 for wet˜ os (148300 N, 121810 E), season aerobic rice experiments at Los Ban Southern Luzon, using Apo and fertilizer-N input of 70 kg N ha1. However, in their experiments, the crop was established by transplanting, whereas we established our crops by direct dry seeding. In our experiments, the crops received a comparable ˜ os amount of total water, 897–1210 mm, as in the Los Ban experiments, 922–1301 mm. However, our soils were generally ˜ os: the seasonal-average soil moisture drier than at Los Ban tensions at 15–20-cm depth varied from 3 to 12 kPa in all our experiments, while they varied from 2 to 5 kPa at 10–15-cm depth ˜ os. Our soils were and from 2 to 6 kPa at 35–40-cm depth at Los Ban very wet only in 2004 at San Ildefonso, when the groundwater table reached close to the soil surface (Fig. 1a) and the soil moisture

tension was around 3 kPa for most of the growing season (from July 5–September 7). The soil moisture tension only started to increase during the grain filling stage for a week, peaking at 15 kPa before imposition of terminal drainage (Fig. 2). The deeper groundwater and dryer soil conditions at Dapdap reflect the better internal drainage of the sandier soil at Dapdap compared with the more clayey soil at San Ildefonso. The response of aerobic rice to fertilizer-N in our experiments is comparable with that of lowland rice in Central Luzon. The lowland direct-seeded rice in Central Luzon responds to fertilizer-N application up to 150 kg ha1 during the dry season. In San Ildefonso and Dapdap, high yields were observed at 120 and 150 kg N ha1. This yield response was consistent with the observed higher LAI values (Fig. 3a) and aboveground biomass accumulation (Fig. 4a), and taller plant heights (data not shown) with increased N application rates. In PhilRice, Nueva Ecija, the relationship between total applied N and yield of wet directseeded rice is generally as follows: 150 > 180 > 120 > 90 kg ha1 (PhilRice, 2000). With regard to the best time of fertilizer-N application, the more N applied around panicle initiation (40–50% of total N applied), the higher the nitrogen-use efficiency (Malabayabas et al., 2002). Farmers in Central Luzon obtain on average yields in the wet season of about 3–3.5 t ha1 using 90–120 kg N ha1 (Bouman et al., 2002). We obtained yields of around 3.5–4.5 t ha1 with 60– 90 kg N ha1 on the clayey soil at San Ildefonso and with 90– 120 kg N ha1 on the sandy soil at Dapdap. Aerobic rice variety Apo has a lower yield potential than modern high-yielding lowland rice varieties (Peng et al., 2006) and the application of more fertilizer-N ˜ oz, the highest yields were did not result in higher yields. In Mun attained when fertilizer-N was applied proportionately during the crop growth period (NS1: 30% at the early vegetative stage, 30% at mid-tillering, 30% at panicle initiation, and 10% at flowering) (Table 5a). Under this N-split application, LAI and dry biomass were relatively comparable or even slightly higher at flowering stage (as in the case of 2005 dry season LAI data shown in Fig. 3b) than the other N-split treatments. This distribution of fertilizer-N is comparable to recommended N-management practices for flooded lowland rice. Peng et al. (1996) recommended four split applications of fertilizer-N at (or prior to) early transplanting, midtillering, panicle initiation, and early flowering for irrigated ˜ os. lowland rice in Los Ban Plant lodging is a major concern in many rainfed lowland rice areas, and is often associated with fertilizer-N (mis-)management. In Dapdap and San Ildefonso, application of more than 90 kg N ha1 in our field experiments resulted in about 50–75% lodging, with subsequent negative effects on grain yield. Good timing and splitting of fertilizer-N application may reduce lodging (Carreres et al., 2000), and increase the bending resistance of the rice crop (Singh and Takahashi, 1962). Topdressing at 20 days before heading increases bending resistance by affecting the length and diameter of internodes, dry matter accumulation in the basal

174

R.M. Lampayan et al. / Field Crops Research 116 (2010) 165–174

portions, and the breaking strength of the shoots (Singh and Takahashi, 1962). However, no information could be derived from ˜ oz because the crops there did not our N-split experiment at Mun lodge. In that experiment, a total of 100 kg ha1 of N was applied in the plots, which might be low enough to not induce lodging, as suggested by the results of the experiments at San Ildefonso and ˜ oz site Dapdap. A plant-lodging experiment conducted at the Mun during the 1994 and 1995 dry seasons (January–April) by Tabbal et al. (2002) showed significant plant lodging (ranging from 14 to 50% area of lodged rice) of direct-seeded rice with 150 kg ha1 N applications (applied in 4 splits). However, these results were under continuous flooding conditions. Tabbal et al. (2002) also reported an inverse correlation between lodging and bending resistance of the rice crop in this experiment. However, in our experiments we did not find strong statistical correlation between lodging and bending resistance. There was no significant effect of row spacing between 25 and 35 cm on yield, and in bending resistance and lodging. Rice grown in 25-, 30-, and 35-cm row spacing had similar yields at all sites. Except for LAI, row spacing did not also affect other agronomic parameters (plant height and biomass accumulation). From the ˜ oz experiments, LAI is relatively higher results in Dapdap and Mun with wider row spacings. There was no significant interaction between row spacing and either total fertilizer-N-amount or splitting of fertilizer-N for yield. The number of tillers and panicles per area were lowest in 35-cm row spacing, but this was compensated for by a higher number of spikelets per panicle. 5. Conclusions Aerobic rice yields of up to 4.9 t ha1 with 60–150 kg ha1 of applied N using Apo variety were attained in our wet-season experiments in Central Luzon, Philippines. Yields were responsive to applied nitrogen, but risk of lodging also increased especially at N rates beyond 90 kg ha1. The common practice of three to four split N applications for lowland rice can also be used for rainfed aerobic rice. Crop growth and yield of aerobic rice were insensitive to variation in row spacings between 25 and 35 cm. Because of this, farmers can be flexible in choosing row spacings between 25 and 35 cm. A spacing of 35 cm would allow for interrow cultivation since medium-sized tractor tires are less than 30 cm wide and would not cause damage to the crop. Further research on fertilizer-N optimization in aerobic rice could focus on the principles and practices of site-specific nutrient management as applied in lowland rice as described by Dobermann and White (1999), Buresh et al. (2005) and Fairhurst et al. (2007). Acknowledgments This work was supported by the Swiss Agency for Development and Cooperation (SDC) through the water-saving workgroup of the Irrigated Rice Research Consortium (IRRC), and by the CGIAR Challenge Program on Water and Food through the project

‘‘Developing a System of Temperate and Tropical Aerobic Rice in Asia (STAR)’’. References Atlin, G.N., Lafitte, H.R., Tao, D., Laza, M., Amante, M., Courtois, B., 2006. Developing rice cultivars for high-fertility upland systems in the Asian tropics. Field Crops Res. 97, 43–52. Bouman, B.A.M., 2001. Water-efficient management strategies in rice production. Int. Rice Res. Notes 16 (2), 17–22. ˜ eda, A., Bhuiyan, S.I., 2002. Nitrate and pesticide contamiBouman, B.A.M., Castan nation of groundwater under rice-based cropping systems: evidence from the Philippines. Agric. Ecosyst. Environ. 92, 185–199. Bouman, B.A.M., Peng, S., Castaneda, A.R., Visperas, R.M., 2005. Yield and water use of irrigated tropical aerobic rice systems. Agric. Water Manage. 74, 87–105. Bouman, B.A.M., Yang, X., Wang, H., Wang, Z., Zhao, J., Chen, B., 2006. Performance of temperate rice varieties under irrigated aerobic conditions in North China. Field Crops Res. 97, 53–65. Buresh, R.J., Witt, C., Ramanthan, S., Mishra, B., Chandrasekaran, B., Rajendran, R., 2005. Site-specific nutrient management: managing N, P, and K for rice. Fertil. News 50 (March (3)) pp. 25–28 & 31–37. Carreres, R., Sendra, J., Ballesteros, R., Garcia dela Guardia, J., 2000. Effects of preflood nitrogen rate and midseason nitrogen timing on flooded rice. J. Agric. Sci., Cambridge 134, 379–390. Dobermann, A., White, P.F., 1999. Strategies for nutrient management in irrigated and rainfed lowland rice systems. Nutr. Cycl. Agroecosyst. 53, 1–18. Fairhurst, T.H., Witt, C., Buresh, R.J., Dobermann, A., 2007. Rice: A Practical Guide to Nutrient Management, 2nd edition. International Rice Research Institute (IRRI), International Plant Nutrition Institute (IPNI), and International Potash Institute (IPI). ˜ a, B.S., Quiton, J., 2002. Rapid yield loss George, T., Magbanua, R., Garrity, D.P., Tuban of rice cropped successively in aerobic soil. Agron. J. 94, 981–989. International Rice Research Institute (IRRI), 2003. IRRISTAT. Version 4.4. IRRI, Los ˜ os, Philippines. Ban Lafitte, R.H., Courtois, B., Arraudeau, M., 2002. Genetic improvement of rice in aerobic systems: progress from yield to genes. Field Crops Res. 75, 171–190. Lampayan, R.M., Bouman, B.A.M., De Dios, J.L., Lactaoen, A.T., Espiritu, A.J., Norte, T.M., Quilang, E.J.P., Tabbal, D.F., Llorca, L.P., Soriano, J.B., Corpus, A.A., Malasa, R.B., Vicmudo, V.R., 2004. Adoption of Water-saving Technologies in Rice Production in the Philippines. Food and Fertilizer Technology Center Extension Bulletin 548. FFTC, Republic of China on Taiwan, 15 pp. Malabayabas, D.M., de Dios, J.L., Casimero, M.C., Espiritu, A.J., Nishida, M., 2002. Estimation of nitrogen use efficiency (NUE) in wet direct-seeded rice (WDSR) using 15N fertilizers. In: Philippine Rice R&D Highlights 2000, Philippine Rice Research Institute, Maligaya Science City of Munoz, Nueva Ecija, pp. 43–44. Mortimer, A.M., Hill, J.E., 1999. Weed species shifts in response to broad spectrum herbicides in sub-tropical and tropical crops. Brighton Crop Protect. Conf. 2, 425–437. Peng, S., Garcia, F.V., Gines, H.C., Laza, R.C., Samson, M.I., Sanico, A.L., Visperas, R.M., Cassman, K.G., 1996. Nitrogen use efficiency of irrigated tropical rice established by broadcast wet seeding and transplanting. Fertil. Res. 45, 123–134. ˜ eda, A., Nie, L., Park, H.-K., 2006. Peng, S., Bouman, B.A.M., Visperas, R.M., Castan Comparison between aerobic and flooded rice: agronomic performance in a long-term (8-season) experiment. Field Crops Res. 96, 252–259. Philippine Rice Research Institute (PhilRice), 2000. Philippine Rice R&D Highlights ˜ oz, Nueva 2000. Philippine Rice Research Institute, Maligaya Science City of Mun Ecija, p. 69. Singh, J.N., Takahashi, J., 1962. Effect of varying dates of top-dressing of nitrogen on plant characters leading to tendency to lodging in rice. Soil Sci. Plant Nutr. 8, 169–176. Tabbal, D.F., Bouman, B.A.M., Bhuiyan, S.I., Sibayan, E.B., 2002. Improving lodging resistance of direct wet-seeded rice. Phil. J. Crop Sci. 27, 55–64. Wang, H., Bouman, B.A.M., Dule, Z., Wang, C., Moya, P.F., 2002. Aerobic rice in northern China: opportunities and challenges. In: Bouman, B.A.M., Hengsdijk, H., Hardy, B., Bindraban, P.S., Tuong, T.P., Ladha, J.K. (Eds.), Water-wise Rice Production. pp. 143–154. Yang, X., Bouman, B.A.M., Wang, H., Wang, Z., Zhao, J., Chen, B., 2005. Performance of temperate aerobic rice under different water regimes in North China. Agric. Water Manage. 74, 107–122.