DIVISION S-6—SOIL AND WATERMANAGEMENT AND CONSERVATION

Conservation Bench Terraces for Rice in a Subhumid Climate1

L. S. BHUSHAN2

ABSTRACTRain-fed crops in northern India produce less than the

potential maximum due to moisture stress at crucial stages ofplant development. Research was initiated on a conservationbench terrace (CBT) system to evaluate its potential to in-crease water supply for increasing and stabilizing rice produc-tion on 2.5% sloping land in the subhumid climate. The widthratios of the' contributing area to the collection area (terracebenches) were 6:1, 1:1, 2:1, and 3:1. The contributing area wascropped with corn planted on the contour.

Six years of research have shown that levelled benches re-ceived 0.25, 0.50, and 0.76 m"/m2 of estimated runoff fromthe 1:1, 2:1, and 3:1 ratios, respectively. Consequently, measuredsoil water in the level benches was greater than the storage inlevel benches without a contributing area at the tune of panicleemergence. The rice yields in the level benches with contribut-ing areas were 21, 47, and 88%, respectively, greater on threeratios than yields on the level check without a contributingarea. The 84% increase in rice grain yield on the 3:1 CBTwas attributed to increased soil water content at panicleemergence.

A major advantage of the CBT system was that chances ofcrop failure were reduced. The probability of obtaining riceyields of < 1,400 kg/ha was decreased from 50% on levelbench to 16% in level benches with a 3:1 ratio. Similarly, theprobability of achieving yields > 3,500 kg/ha was 50% in the3:1 CBT system and 0% in the level bench (NW). Total drymatter production for the 3:1 CBT system was 1.7 times greaterthan the level check.

Additional Index Words: rain-fed farming, direct-sown rice,watershed, water harvesting, growing season rainfall, runoff.

Bhushan, L. S. 1979. Conservation bench terraces for rice in asubhumid climate. Soil Sci. Soc. Am. J. 43:754-758.

THE LOW Himalayan hill region of northern Indiais characterized by sloping lands and untimely

and uneven distribution of rainfall. Water is oftenthe limiting factor for optimum crop production inthis rain-fed farming region. Lack of available waterin the root zone during certain stages of crop growthresults in crop yields lower than the potential maxi-mum. Rainfall is frequently of high intensity andmuch of it is lost in runoff. As a result, rice (Oryzasativa L.), crop with greatest yield potential of theregion, suffers from moisture stress even with 100 cmof growing season rainfall (GSR), and the crop yieldis low-ranging from 800 to 1,000 kg/ha (14).

In 1959, Zingg and Hauser (16) described a terrace1 Authorized for publication by the Director, as Paper 47-18/

78-S&A of the Central Soil & Water Conserv. Res. & TrainingInstitute, Dehra Dun- 248195, India. Received 2 Oct. 1978.Approved 4 Apr. 1979.

"Soil Physicist. Now located at Central Soil 8c Water Conserv.Res. Inst., Agra-282004, India.

system designed to reduce water erosion and to con-serve and distribute runoff water for crop use. Work-ing with crops like sorghum, barley, and wheat,scientists observed encouraging results with conserva-tion bench terrace (CBT) systems in the regions re-ceiving < 50 cm annual precipitation (1), (3), (4),(5), (8), (10), (11), (12). Their research has revealedthe possibility of increasing and stabilizing productionwith the CBT system.

However, research information on the performanceof the CBT for rice cultivation in a subhumid climateis lacking. Therefore, in a subhumid climate, theCBT system merited evaluating in its potential forincreasing grain yield and stabilizing production ofrice under rain-fed conditions. This paper reports 6years (1971-1976) of research results on CBTs for riceproduction in the subhumid region of Dehra Dun inthe outer Himalayas. Experimental objectives were(i) to evaluate effects of the CBT on rain-fed rice pro-duction in a subhumid climate, (ii) to determine theoptimum ratio of contributing area to level bencharea, and (iii) to determine the relation between soilmoisture and rice yields.

EXPERIMENTAL PROCEDUREThe research was conducted on Dhoolkot silty loam soil at

the Research Station (30°19'N, 78°02'E and 683 m above meansea level) of the Central Soil and Water Conservation Researchand Training Institute, Dehra Dun. Water retention variedwith soil depth from 25-26% at 1/3 bar, and from 11-13% at15 bars on an oven-dry basis. The land slope is 2-3%. Thesoil is deep (> 2 m) and the texture varied from silty loam atthe top to silty clay loam at the bottom. Hard pan, whichcould seriously restrict the water movement in the soil profilewas not observed.

The conservation bench terrace (CBT) system included acontributing area (watershed) immediately above a levelbench. Each CBT was 24 by 8 m with a 25-cm high earthenberm constructed around it to prevent water from coming intothe area from other than its contributing area, and to storethe runoff water in the levelled bench. The length of thebenches and the sloping contributing area was varied to ob-tain the ratio of runoff area to run-on area (terrace benches)of 1:1, 2:1, and 3:1. A level bench (24 by 8m) without acontributing area, referred to as level bench, was also con-structed with an earthen dike on the sides to serve as a levelcheck. In this case, the only water available for crop produc-tion was the rainfall that fell on the bench. One plot on slopingland was also maintained without a level bench as a slopingcheck. The plot layout for one of the four replications and across sectional view of the 3:1 CBT system are shown in Fig. 1.

The whole experimental area was double cropped. The firstcrops raised during the rainy season were rice on the levelbenches and corn on the contributing areas. Wheat was thesecond crop which was taken in the winter season on the wholearea. The effect of runoff on rice yields was evaluated bycomparing the production in benches having different contribut-ing watershed ratios with that on the level check. Similarly,the effect on corn yield of water harvesting was evaluated bycomparing yields of corn on the sloping check without a levelbench.

Run-on to the CBTs or runoff from the CBTs was not

754

BHUSHAN: CONSERVATION BENCH TERRACES FOR RICE IN A SUBHUMID CLIMATE 755

CONTRIBUTING AREA

CO0:1

CONSERVATION BENCH

3:1 car SYSTEM

fa-CONTRIBUTIN6 AREA CONSERVATION

Fig. 1—Plot layout and cross section of a 3:1 CBT system.

measured directly. However, on the basis of earlier research onrunoff from corn at Dehra Dun under similar situations ofsoil and rainfall (15), the runoff from the contributing area wasestimated. While this is an approximation, it affords the bestestimate possible in the absence of measurement. The harvestedwater was expressed in cubic meters of water per square meterof receiving area (ma/m2).

Rainfall—Average annual precipitation at the experimentalsite was 160 cm and varied from a maximum of 216 cm in 1956to a minimum of 126 cm in 1972 (14). Approximately 75% ofthe annual precipitation normally occurs from mid-June tomid-September, which is the growing season for rice. The mon-soon season begins by June 19 and is very active during Julyand August when 100 cm falls, before tapering off by September10. The rainfall distribution for the period of research is givenin Fig. 2. Rain storms of high intensity on rolling topographycause considerable amounts of runoff during July and August,leaving the soil profile unsaturated. During 1972 and 1974 therewas a very uneven distribution of rain with a long rainlessperiod.

Crop Production—Normal tillage, consisting of ploughing,

Table 1—Growing season rainfall and runoff in the CBT.Ratio of watershed area to CBT area

Years

197119721973197419751976Average

Rainfallcm11190

109100119128110

1:1

0.260.210.270.220.240.310.25

2:1

-Runoff— mVm1-0.520.420.540.440.480.620.50

3:1

0.780.630.810.660.720.930.76

harrowing, and planking, was used on all the fields. Cornwas planted across the slope in 60-cm row spacing with a plantdensity of approximately 55,000/ha. Rice seed was sown in 25-cm row spacing at 100 kg/ha. Corn was fertilized at 50 kg N +12 kg P/ha in 1971 and 1972, while the rates were increased to100 kg N + 22 kg P/ha in subsequent years. Rice was fertilizedwith 40 kg N + 10 kg P/ha during 1971 and 1972, and 80 kgN + 22 kg P/ha in other years. In 1975, however, P could notbe added. Adequate plant protection measures were taken toprevent yield reduction due to diseases and pests.

Since the panicle development of rice is the most criticalperiod for water stress (9), the soil moisture during this periodwas measured to a depth of 100 cm in 20-cm increments. Sam-ples were taken from all of the four replications.

RESULTS AND DISCUSSIONRainfall, Runoff Harvesting, and Soil-Water

StorageThe amount of water harvested averaged 0.76 m3/

m2 of bench area in 3:1 CBT system and varied from0.91 in 1976 to 0.6 m3/rn2 in 1972 (Table 1). Amountof run-on in the 2:1 and 1:1 benches averaged 0.5 and0.25 m3 water/m2 of bench area, respectively. Runoffimpondment increased soil water storage in level

60 70 80 90 100DAYS AFTER SOWING OF RICE

Fig. 2—Daily rainfall during rice crop season. Arrow indicates head initiation.

756 SOIL SCI. SOC. AM. J., VOL. 43, 1979

J8 eoMOISTURE CONTENT, 00 (BY WT.)

84 6 8 K> 12 14 16 18 16 17 19 21 23 23 14 16 18 20 24 26

I00 L .——-LEVEL BENCH (NW) «——«> CBT (III ) »——« CBT (2:i) 4——6 CBT <3:l>

Fig. 3—Soil moisture profile in conservation benches and level bench (NW) at heading stage.

benches throughout the growing season. As a result,standing water was often noticed on CBT.

During 1971, 1973, 1975, and 1976, a few stormsduring the panicle development stage in rice producedrunoff. The soil profile remained near saturation evenafter panicle emergence in 1973 and 1976. In 1972,there were two drought periods; the first began 15 daysafter sowing and lasted about 22 days, and the secondperiod began during panicle development and con-tinued thereafter. In 1974, no rainfall occurred beyond60 days after sowing. As a result, the soil moistureat the time of panicle initiation was near the wiltingpercentage in the top 60 cm of the level bench (NW).However, the CBTs had water available below the top60 cm to sustain the crop (Fig. 3). Even in the normalrainfall years of 1975 and 1976, there was less waterin the 0 to 20-cm soil layer of the level check than inthe level benches with a contributing area. Total wa-ter volume in the 60- and 100-cm soil profile at panicleemergence in rice is given in Table 2. The 1:1 CBTscontained 6.9% more soil moisture than the levelcheck, and the 2:1 and 3:1 CBTs held 16.3 and 25.2%more, respectively. Increased soil water at panicle de-velopment is important because adequate soil waterduring panicle development until the seed setting stageis required for good grain yield.

Crop YieldsThere was no significant difference between corn

yield due to water harvesting in the CBT system, andthe corn yield on the sloping check without the level

Table 2—Total water content (cm) in soil profile t______at heading stage in CBT and level check.______

Ratio ofcontributing 1973 1974 1975 1976

area to ——————— ——————— ——————— ———————CBT area 0-60 0-100 0-60 0-100 0-60 0-100 0-60 0-100

0:11:12:13:1

19.020.2t§

33.336.5t§

8.110.411.814.0

20.020.822.725.1

15.816.016.818.5

28.429.430.833.0

16.918.219.720.5

30.031.834.336.0

TO-60 cm and 0-100 cm soil profile held 23 and 38 cm water at fieldcapacity and 10.8 and 18.1 cm of water at wilting percentage, respective-ly-

J Standing water about 5 cm.§ Standing water about 8 cm.

bench (Table 3). On the other hand, rice grain yieldswere significantly increased by water harvesting (Ta-ble 4). Rice grain yields for the 1:1, 2:1, and 3:1 CBTsystems were 21, 47, 88% greater than the level check,respectively. Among the three CBTs, maximum yieldswere harvested in the 3:1 system. Haas et al. (7) alsoobserved increased wheat yield and soil-water storage,when the ratio of the contributing area to the levelbench area was > 0.1. My results are contrary to theresults from the Northern Great Plains of the U.S.,where the use of a contributing area did not producesignificantly higher yields than bench levelling alone,Haas and Willis (5, 6). This contradiction may arisebecause rainfall-runoff from the grassland was very lowunder the conditions of their research.

The CBT system increased rice production in allthe years. In 1972 and 1974, total rainfall was belownormal and unevenly distributed. As a result, the cropexperienced severe moisture stress at the heading andpollination stages. The crop, therefore, had a lowgrain yield in 1972, and average dry matter yields inboth years. In 1974, the grain yields were also low.Nevertheless, the data illustrates the significance of theCBT system. The grain yields were highest in bencheswith 3:1 ratio.

The average rice yields with the short growing sea-sons in 1971 and 1972 could have been greater withthe early maturity variety 'Tellahamsa' which wasplanted after 1972. Grain yields in 1971 and 1972 werethose of the rice variety 'Jaya' which has a growingseason of 120 days. Nevertheless, total dry matter pro-duction for the 3:1 CBT system was 1.7 times greaterthan on the level check. Six years of research have

Table 3—Summary of corn gram yields in the contributing area.Treatments (Contributing area/CBT ratio)

Years 1:0 1:1 2:1 3:1

———————— yields, kg/ha ————————197119721973197419751976

1,3083,0542,0953,0702,512bf2,540b

1,0333,0162,9803,2402.810C2,810c

1,0733,5762,6013,2812,365a2,259a

1,1333,1812,3863,5282,593b2,547b

t Means within years not followed by same letter are significantly differ-ent at 5% leveli. Data without letters are not statistically different at5% level

t According to F test (2).

BHUSHAN: CONSERVATION BENCH TERRACES FOR RICE IN A SUBHUMID CLIMATE 757Table 4—Rice grain yield, dry matter production, seedlingemergence, and ear bearing tillers on CBT and level check.

TreatmentsRatio of watershed to terrace bench area

Years 0:1 1:1 2:1 3:1

Grain yield, kg/ha1971197219731974197519766-year avg.Highest 5 years avg.Highest 4-year avg.

1,346 af61 a

2,907 a605 a

2,158 a2,396 a1,579 a1,882 a2,202 a

l,755ab67 a

3,337 b1,006 ab2,233 a3,039 b1,906 ab2,274 ab2,591 b

2,429 be78 a

3,912 c1,076 ab2,429 a4,012 c2,323 b2,772 b3,195 c

2,725 c257 b

4,602 d1,592 b2,727 b4,879 d2,964 c3,505 c3,983 d

Total dry matter, kg/ha197119721973197419751976

4,538 a3,288 a7,902 a4,282 a5,453 a4,896 a

4,789 a4,272 b7,933 a5,444 b6,109 b7,392 b

5,879 b6,219 c

7,425 c6,880 c

9,207 b 10,940 c6,181 c6,086 b7,768 b

7,566 d6,805 c9,372 c

Plant density, no./m row length

1973197419751976

61551859

60 6355 5619 1858 66

60531666

Ear-bearing tillers, no./m row length1973197419751976

127 a8a

106 a130 a

128 a 164 b30b 40b

112 ab 118 b141 b 185 c

196 c81 c

142 c206 d

t Data within year not followed by same letter are significantly differentat 5% level according to 'F' test (2). Data without letters are statisticallynot different at 5% level.

shown that a contributing area three times the areaof the bench resulted in the highest grain yield andtotal dry matter production. It appeared that totaldry matter production might have been a better cri-terion than grain yields for evaluating the CBT system,because yield potential of rice grain is frequently af-fected by uneven distribution of rain water. Somedry matter was produced every year, in spite of un-timely distribution of rain during the crop growingseason. In 1972, total dry matter produced in the 3:1CBT system was more than twice that in the levelbenches (NW).

Crop yields stability is very important in rain-fedfarming systems. Rice grain yields on the 3:1 CBTsystem were < 1,400 kg/ha in only 1 of the 6 years(1972). In the 1:1 and 2:1 CBT systems, yields were< 1,400 kg/ha in 2 of the 6 years. Conventional levelchecks on the other hand, produced < 1,400 kg/hain 3 of the 6 years. Thus, the probability of producinglow yields (< 1,400 kg/ha) was reduced from 50% onlevel checks to 16% on the 3:1 CBT system. Similarly,the probability of producing high yields (> 3,500 kg/ha) was 50% on the 3:1 CBT system compared to 0%on level checks. Thus, extra water stored in the CBTsfrom water-shed runoff resulted in greater stabilizedrice production than in conventional level checks.These results agreed with those of earlier researcherers(8), (11), who recorded the beneficial effect of theCBT on available soil water, and yield of grain sor-ghum and barley in low-rainfall, dry farming areas.

Seedling emergence was not significantly affectedby the ratios of contributing area to area levelledbench area. However, the number of tillers per meter

Y-2O0.6X-3345.5r-0.93

22 26 3O 34SOIL WATER CONTENT, cm

36

Fig. 4—Grain yield of rice vs. stored water in a 1-meter soilprofile at heading stage.

did significantly vary with ratios of contributing areasto levelled-bench area. Tillering was favorably in-fluenced by runoff-water collected in the CBTs. Theincreased rice yields in the CBT system may be attri-buted to a greater number of fertile tillers per meterrow.

Harvested Water RelationshipIn an attempt to develop better water conservation

practices, it is important to determine relationshipsbetween crop yield and soil water. The effect of soilwater at panicle development on rice yield is shownin Fig. 4. Each cm of soil water in excess of 8 cm (avail-able at the panicle development stage) increased thegrain yields about 200 kg/ha. The r2 value indicatesthat 84% of rice yield increase was attributable to soil-water content at the panicle emergence stage.

SUMMARYRunoff farming on the CBT was tested for rice

production in the subhumid climate of Dehra Dun,India, under rain-fed conditions. The ratios of runoffarea (which was cropped with corn) to collection areawere 0:1, 1:1, 2:1, and 3:1.

Six years of results showed that harvested runoffwater on CBTs greatly increased rice production.Available soil water increased at the reproductive stageof rice in CBTs with water harvesting. Rice grainyield was increased by 21, 47, and 88% in CBTs with1:1, 2:1, and 3:1 area ratio, respectively, over that ofbench levelling alone.

A major advantage of the CBT system over levelchecks, was that the probability of receiving low yields(< 1,400 kg/ha) of rice was reduced from 50 to 16%by the 3:1 ratio. Similarly, the probability of produc-ing high yields (>3,500 kg/ha) increased from 0 to50%.

In a subhumid climate receiving crop-growing-seasonrainfall of > 100 cm, the CBT system can effectivelycontrol runoff and erosion, increase soil water storage,and can increase rice yields and stabilize production.

ACKNOWLEDGEMENTThe author wishes to acknowledge (i) Dr. K. G Tejwani,

Director, and Shri Gurmel Singh, Chief Scientist, for providingphysical facilities and advice, (ii) Mr. D. T. Anderson, Prin-

758 SOIL SCI. SOC. AM. J., VOL. 43, 1979

cipal Canadian Adviser, AICRP for Dryland Agriculture, Hy-derabad; Shri D. C. Das, Dy. Commissioner (Soil Cons. Engg.),Govt. of India, and Dr. S. Chandra for going through the manu-script critically with valuable comments and suggestions, (iii)With appreciation for the help rendered by Mr. Inder Pal indata collection.