Low Temperature Tolerance in Rice: The Korean …aciar.gov.au/files/node/2301/pr101chapter02.pdf · Low Temperature Tolerance in Rice: The Korean Experience ... temperatures cause

109

Increased Lowland Rice Production in the Mekong Regionedited by Shu Fukai and Jaya Basnayake

ACIAR Proceedings 101(printed version published in 2001)

Low Temperature Tolerance in Rice:The Korean Experience

Moon-Hee Lee

Abstract

In 1980 and 1993, low temperatures seriously damaged the Korean rice crop. Grain yielddecreased by 26% and 9.2%, respectively, compared with the national average of other years. Lowtemperatures cause the rice plant to suffer poor and slow vegetative growth, spikelet sterility,delayed heading and poor grain filling. Rice varieties differ significantly in their capacity totolerate low temperatures at various growth stages. Tongil types (Indica/Japonica hybrids and high-yielding rices), for example, are more susceptible to low temperatures than are Japonica varieties,needing temperatures that are 2.5

o

to 3.0

o

C higher. Reliable screening methods have beendeveloped, using a phytotron, growth chambers and low water temperatures, and significantlyimproving cold-tolerant selections. For year 2000, 57% of varieties released in Korea were highlytolerant of low temperatures. Management of cultural practices is another method for improvingcold tolerance in rice. For example, optimal application of nitrogen can maximize yields andreduce damage by low temperatures. If leaf nitrogen content is too high, then, under low tempera-tures, spikelet sterility increases significantly—by 3.5% in Tongil and 2.5% in Japonica varieties.In 1993, in cool mountainous regions, applying organic matter during low temperatures signifi-cantly increased grain yield. Deepwater (20 cm) irrigation during the reproductive stage canincrease grain yields by 10% to 14%, compared with rice growing in shallow water.

R

ICE

is Asia’s most important staple, and its con-sistent production is vital for food security. Most rice-growing countries are faced with climate-inducedstresses that significantly reduce rice productivity:droughts, floods, low temperatures and winds. Lowtemperatures comprise a major climatic problem forrice growing in 25 countries, including Korea andJapan, and even in tropical countries such as thePhilippines and Thailand (Kaneda and Beachell1974).

The Korean peninsula is located in the Far East,between latitudes 33°

06

′

and 43° 01

′

north andbetween longitudes 124° 11

′

and 131° 53

′

east, in thenorthern temperate climatic zone. Summers are hotand humid and winters severely cold. Rice is there-fore a summer crop, grown between April andOctober. In the northern, mountainous regions, therice plant can suffer from low temperatures at any

stage between germination and maturity. In years ofextreme low temperatures, all rice-growing areas aresusceptible to cold at the reproductive stage. Forexample, in 1980 and 1993, low temperaturesseriously damaged the Korean rice crop, with grainyields dropping by 26% and 9.2%, respectively,compared with the national average yield on eitherside of these years (MOAF 1994).

This paper discusses the damage caused by lowtemperatures in Korea in recent years and thedevelopment of new varieties and cultural practicesfor cold tolerance. Developing cold-tolerant varietiesand suitable cultural practices is of great concern forthe future because these will lead to consistently highyields in cold regions, particularly in the highlandsand cooler regions of the subtropics.

Geo-Climatic Conditions

The geography of the Korean peninsula is charac-teristically hilly or mountainous. Three regions can bedistinguished: (1) alpine: northern and eastern

International Rice Research Institute (IRRI), PBGB, MCPOBox 3127, 1271 Makati, Philippines. E-mail: m.lee@ cgiar.org

KEYWORDS:

Cold tolerance, Low temperatures, Low temperature damage, Rice, Sterility

110



peninsula, featuring a high and long mountain rangeknown as the Hamgyong and Taebaeg Range; (2) mid-altitudes: located mid-peninsula and extending northto south; and (3) coastal plains: mostly western andsouthern peninsula. The mountainous topographyreduces arable land to about 21.7% of the nation’stotal land area (KOIS 1993).

The Korean peninsula has four distinct seasons: acold winter, spring, a warm and humid summer andautumn. The average annual temperature is around12°C, averaging between 12° and 15°C in the south,10° and 12°C in the central region, and 5° and 10°Cin the north. Temperatures range from 0°C in thecoldest month (January) to 25°C in the warmestmonth (August) (NCES 1990).

Temperature, solar radiation and water are thethree critical requirements for growing rice. Oversummer (the rice-growing season), the mean monthlyair temperature gradually increases from 11°C inApril to 25°C in August, then declines to below 13°Cin October (Figure 1). Temperatures vary, however,according to year and region, for example, thesouthern and coastal areas usually have higher tem-peratures than do the northern and inland regions.

Average annual rainfall in Korea is about1250 mm, with some regional variation. However,70% of total rainfall occurs in the summer, fromJune to September, with 330 mm falling in July. Asare autumn and winter, spring is often a dry season,frequently delaying transplanting of rice in rainfedareas. In the autumn dry period, the crop receivessufficient solar radiation between physiologicalmaturity and harvest to permit ripening and thereforeadequate yields.

Low Temperature Damage in Korea

The Korean rice industry has suffered many climate-induced disasters, including low temperatures, strongwinds, droughts and floods. Winds cause severe andfrequent lodging (Table 1). However, low tempera-tures comprise the biggest threat to the Korean riceindustry in terms of area and degree of damage. In1980 and 1993 summers, for example, 783 000 and208 000 ha were damaged, respectively, by low tem-peratures during the rice crop’s critical vegetativeand reproductive stages.

Figure 1.

Changes in monthly air temperatures (lines) and precipitation (bars) during the rice cropping season (Kim and Kimin press).

Air

tem

pera

ture

(°C

)

Pre

cipi

tatio

n (m

m)

J J JF M MA A S O N D

Month

Min.

Avg.

Max.

350

300

250

200

150

100

50

0

35

30

25

20

15

10

5

0

−5

−10

−15

111



Source: MOAF (1999).

The daily mean air temperatures during reproduc-tive and maturing (July–September) in 1980 and 1993were lower compared with the 30-year average(Figure 2). During the 1980 season, the spell of lowtemperatures started late July and continued until lateAugust, whereas, in 1993, cool weather was prevalentbetween mid-July and early August, that is, 2 weeksearlier, but 2° to 3°C warmer.

The low temperatures in 1980 and 1993 decreasedrice production significantly, compared with Korea’snational average in other years. Yields in 1980

averaged 2.89 t ha

–1

, representing a 26% decrease,compared with the national average in the previous5 years and, in 1993, yields averaged 4.18 t ha

–1

,representing a 9.2% decrease (Figure 3; MOAF 1999).These rice production data suggest that a significantdifference in rice production existed between 1980and 1993. The improvement may have been due toimproved cold-tolerant varieties and cultural prac-tices, and to the degree and extent of low temperaturedamage. Following the cold weather in 1980, threeresearch substations were established in both thealpine and mid-altitude regions to develop low-temperature-tolerant varieties and improve culturalpractices.

Figure 3.

Average milled rice yield from 1975 to 1999,Korea (MOAF 2000).

Table 1.

Area of paddy fields damaged by unfavourableclimatic factors in Korea, 1965 to 1998.

Year Unfavourable climatic factor

Drought(10

–3

×

ha)Wind

(10

–3

×

ha)Low temp.(10

–3

×

ha)Others

(10

–3

×

ha)

1965 574 48 — —1970 — 172 — 671975 24 50 — —1980 5 105 783 71985 — 111 — 31990 — 122 — —1993 — 39 208 91995 7 84 — —1998 — 235 — 67

1975 1980 1985 1990 1995 1999

Year

Mill

ed r

ice

yiel

d (t

ha−1

)

5.5

5.0

4.5

4.0

3.5

3.0

2.5

Figure 2.

Daily mean temperatures from June to August of 1980 and 1993, and the average of the last 30 years, Korea.

Dai

ly m

ean

tem

pera

ture

(°C

)

Date

1 July 16 July 31 July 15 Aug 30 Aug 14 Sept 30 Sept

Avg. 30 y

1980

1993

28

26

24

22

20

18

16

14

112



Conditions for Low Temperature Damage

For successful rice production in Korea, air tempera-ture should average more than 13°C for 150 days.Too cool air and water temperatures cause damage atcrop establishment, while low air temperature byitself affects the rice plant at the reproductive andgrain-filling stages. Research on breeding cold-tolerant varieties and cultural practices did not beginuntil 1970. Since then, it has been strengthened, par-ticularly when the Tongil types (Indica/Japonicahybrids and high-yielding rices) were found to behighly susceptible to low temperatures.

Research findings on critical low temperaturesand symptoms of resulting damage are summarizedin Table 2 (NCES 1990). The critical temperature forrice is usually below 20°C and varies according togrowth stage, for example, for germination, thecritical temperature is 10°C and for the reproductivestage, it is 17°C. Nishiyama (1985) and Yoshida(1978) report that the critical low temperature differsaccording to variety, duration of low temperatureand the plant’s physiological development.

Source: NCES (1985).

The usual symptoms of low temperature damagein Korea are poor and delayed germination, stuntedseedling growth and leaf yellowing during earlygrowth, and inhibited rooting and tillering during thevegetative stage. The rice plant is most sensitive tolow temperatures during the reproductive stage,showing inhibited panicle initiation and develop-ment, spikelet degeneration and disturbed pollen

formation. Near maturity, low temperatures inducepoor grain filling and rapid leaf senescence.

Results of NCES research show that, under con-trolled conditions and at 10°C, germination rate wasvery low, compared with the rates at 12° and 15°C(Table 3). However, germination rates were signifi-cantly different between the Japonica (74%) andIndica/Japonica (12%) varieties, 20 days aftersowing at 10

o

C. Jun et al. (1987) also reported thatJaponica rice varieties germinated well, comparedwith Indica/Japonica crosses under the same lowtemperatures.

a

D = days after sowing.Source: NCES (1977).

Seedling growth is highly variable under differenttemperature regimes (Table 4). At the vegetativestage, temperatures lower than 15°C reduce plantheight, tillering, root growth and dry weight of therice plant. Hue (1978) and other studies suggest thatJaponica rice varieties can tolerate low temperaturesmuch better than Indica/Japonica varieties, with thelatter requiring 2.5

o

–3.0

o

C higher temperatures thando Japonica rice varieties for effective growthbetween germination and maturity.

a

Grown for 10 days at 25°C and for 20 days at 12°C.

b

Grown for 10 days at 15°C and for 20 days at 25°C.

c

Natural : Field temperature condition Source: NCES (1970).

Table 2.

Type and symptoms of cold damage in rice,Korea.

Growth stage

Critical temp. (°C)

Type and/or symptoms of cold injury

Germination 10 Poor, delayedSeedling 13 Retarded seedling growth

Leaf discolourationSeedling rot

Vegetative 15 Inhibited rooting, growth and tilleringDelayed panicle initiation

Reproduc-tive

17 Inhibited panicle developmentDegenerated spikeletsDisturbed meiosis and pollen formationDelayed heading

Heading 17 Poor panicle exsertionInhibited anther dehiscence andpollination

Maturity 14 Poor grain filling and qualityEarly leaf senescence

Table 3.

Differences of germination rates (%) betweenJaponica and Indica/Japonica rice lines according totemperature.

Line group Varieties (no.)

Temperature

a

10°C 12°C 15°C

15 D 20 D 15 D 20 D 15 D 20 D

Japonica 5 45 74 85 88 93 95Indica/Jap. 7 2 12 51 64 72 97

Table 4.

Effect of temperature on seedling growth androoting in the Indica/Japonica rice variety Milyang 23.

Temperature (

o

C)Plant height(cm)

Tillers(no. per

hill)

Roots(no. per plant)

Dry weight(g per

10 plants)

20 64 9.6 84 25.112 12 2.6 10 0.725–12

a

59 3.1 32 3.315–25

b

40 4.5 22 7.1Control (Natural)

c

36 13.2 83 17.8

113



Within the reproductive stage, booting is the stageat which rice is most sensitive to low temperatures,particularly 10 days before heading, leading topanicle degeneration and empty grains at harvest.Yoshida (1978) and Satake (1976) found that lowtemperatures most affect the young microsporestage, at about 10 or 11 days before heading. Arecent phytotron experiment compared two varieties(Tongil, i.e. Indica/Japonica type, and Jinheung, i.e.Japonica type) exposed to low temperatures at dif-ferent times before heading. The lowest filled grainratios were 30% for Tongil and 70% for Jinheung at10 days before heading (Figure 4).

Figure 4.

Filled-grain ratio for Jinheung (Japonica ricetype) and Tongil (Indica/Japonica rice type) varietiesexposed to low temperatures on different days beforeheading (after Lee et al. 1987a).

Developing Cold-Tolerant Rice Varieties

Breeders aim to develop rice germplasm that canproduce high and stable yields in regions where lowtemperatures are found. Specific objectives includeintegrating qualities of other varieties into leadingvarieties. Traits include short-cycle maturity,medium stature, multiple resistance to pests anddiseases and good grain quality.

Screening

Screening for low temperature tolerance in rice ishighly complex, because responses to low tempera-tures differ between varieties, growth stages andactual temperatures used. For effective selection, thestandard screening methods and facilities used needto be reliable in providing the required low air andwater temperatures. A list of standard screening andtesting methods for low temperature tolerancedeveloped in Korea is given in Table 5. To screeneffectively for adapted cultivars, low temperaturetreatment and duration differ according to growthstage, for example, at germination, a 7-day treatmentof air and water cooled to 13° to 15°C is imposed; atthe 3-leaf stage, a 10-day treatment of 12°/10°C(day/night air) is used; at booting, 10 days of air andwater at 18°C; and at grain filling, 20°/15°C (day/night air) for 20–30 days, using the phytotron orgreenhouse. The simultaneous use of controlledfacilities such as the growth chamber or phytotron

Days before heading

Fill

ed-g

rain

rat

io (

%)

Control 15 10 5 Heading

Jinheung

Tongil

100

80

60

40

20

0

Figure 5.

Cooperative system between three experiment stations and their five substations for screening and selecting cold-tolerant rice varieties, Korea (after NCES 1985).

Request cold screening

Return the results

On-the-spot selection

CheolweonSubstation

Crop Experiment Station(NCES)

Honam Agric.Experiment Station

(NHAES)

Cold-ScreeningNursery at Chuncheon

UnbongSubstation

YeongdeogSubstation

Yeongnam Agric.Experiment Station

(NYAES)

SangjuSubstation

JinbuSubstation

114



and mass screening in the field is required forefficient selection.

For mass screening, the cold-water screeningsystem works efficiently, especially for combinedcold tolerance throughout the entire plant growth.The screening system consists of planting lines in12-m-long rows, and irrigating them with a 7°Cgradient in water temperature over 10 m from inlet tooutlet. The paddy field is constantly irrigated withcold water from 20 days after transplanting tomaturity. This screening system allows researchersto determine the response of various agronomic traitsassociated with cold damage, and to identify varietaldifferences.

Breeding system

To solve low temperature problems in rice, theNational Crop Experiment Station (NCES) estab-lished facilities in 1978 at the Chuncheon Substation.After the extreme low temperature events in 1980,four substations were established in the highland andmid-altitude regions: Jinbu at the NCES, Unbong atthe National Honam Agricultural Experiment Station(NHAES), and Sangju and Yeongdeog at theNational Yeongnam Agricultural Experiment Station(NYAES). Research facilities consist of a fieldscreening system, using cold water, and greenhousescreening.

A model of the collaborative breeding systembetween the three experiment stations (NCES,NHAES and NYAES), involving five of their sub-stations has been developed (Figure 5). Most of thecultivars selected for low temperature tolerance havebeen tested, using the cold-water field-screeningsystem at the Chuncheon Substation. Based on pre-liminary results from Chuncheon, very early

maturing material is then sent to Jinbu Substation.Short-season cultivars, selected for their adaptabilityto mountainous areas, are sent to the CheolweonSubstation, located in the mid-altitude areas ofKorea’s central region. Selections from the earlymaturing materials from the NHAES are sent to theUnbong Substation in the mid-altitude regions in thesouthern part of the peninsula. Materials sent fromthe NYAES to the Sangju and Yeongdeog Sub-stations are for mid-altitude areas in southern andeastern coastal areas.

To maximize breeding efficiency and develophighly cold-tolerant varieties, using the researchfacilities, the shuttle breeding system was establishedbetween the stations (Figure 6). The idea was toincorporate cold tolerance and other desirable traits(multiple disease resistance) into commercialvarieties, using three-way bridge crossing or singleand back crossing at the stations. Selected F

2

popula-tions and F

3

lines are sent to the Chuncheon Sub-station to screen for low temperature tolerance at thecold-screening nursery. Selections from the F

4

andF

5

generations are conducted according to maturitytype, spikelet sterility and phenotypic acceptabilityunder natural conditions. Very early maturing linesare selected from the F

6

and F

7

generations at Jinbuand Yeongdeog. However, the F

6

and F

7

lines, whichare early maturing and possess desirable agronomictraits, are selected at the Cheolweon, Unbong andSangju Substations. Subsequently, the selected F

8

lines are tested for adaptability to local conditions atthe substations and on farm. The selected elite linesare then nominated for release as new varieties.

Rice-breeding techniques for low-temperaturetolerance have led to the release of significantlyimproved new cultivars during the last 30 years.Cold-tolerant varieties usually have short cycles, are

a

D/N = day/night; Source: NCES (1990).

Table 5.

Methods and facilities for screening cold tolerance in rice, Korea.

Types of damage Screening procedure

a

Facilities

Germinability Water and air at 13°–15°C for 7 days GerminatorSeedling

Chilling injury Water and air at 5°C for 2–3 days at 2-leaf stage Growth chamberGrowth andDiscolouration

Cool-air treatment at 12°/10°C (D/N) for 10 days at 3-leaf stage PhytotronCold-water treatment at 13°C for 10 days at 3-leaf stage Cold-water flowing test at Chuncheon

Delayed heading 18°/10°C (D/N air) for 10 days at 10–20 days after transplanting PhytotronSterility 18°/18°C (air/water) for 10 days at meiosis

10°/23°C (air/water) for 10 days at headingPhytotron and greenhousePhytotron and greenhouse

Combined type Cold water at 17°C and flowing for the whole growing periodfrom 20 days after transplanting

Cold-water irrigation screening nursery at Chuncheon

Grain filling 20°/15°C (D/N air) for 20–30 days Phytotron

115



short and have a high yield potential and good grainquality. By year 2000, 101 rice varieties have beenrecommended to farmers in Korea. Of the recom-mended varieties, 86% have short cycles and 65%have intermediate cycles; these were all identified asbeing tolerant of low temperatures (Table 6). Thecold tolerance program has identified 60% ofreleased varieties as being highly resistant.

Developing Cultural Practices

In years of low temperatures, the use of cold-tolerantvarieties and implementation of appropriate culturalpractices (e.g. fertilizer and water management) areimportant for minimizing low temperature damage.

Fertilizer application

Good crop establishment is an advantage for riceduring cold conditions. Fertilizer applications areneeded for plant maintenance as well as for nourishingand accelerating plant growth. The relationshipbetween applied fertilizer and cold damage has beenthoroughly studied and the nutritional balance iscritical to varietal tolerance. For example, duringbooting, nitrogen concentration in the plant should below while temperatures are low (Amano 1984; Lee etal. 1987b). On studying the relationship between Nconcentration in the leaf blade at panicle initiationwith the filled-grain ratio under low temperatures, the

NHAES (1985) found that the filled-grain ratiodecreased with increasing N levels (Figure 7).

Figure 7.

Effect of leaf nitrogen concentration at panicleinitiation on the filled-grain ratio under low temperatures(after NHAES 1985).

The effects of N application are well illustrated bySasaki and Wada’s findings (1975). When they usedartificial climate chambers, they found that therelationship between N and fertility differed in 1977and 1980 because of low temperatures (Figure 8).The 1980 low temperatures caused a sharp decline inthe filled-grain ratio with increased N, comparedwith the stable response from 1977. Shiga et al.(1977) reported that top-dressing with N at panicleformation increased spikelet sterility. Top dressing atthe flag-leaf stage during lower temperatures there-fore increases the risk of damage to the rice crop.

Leaf N concentration (%)

1.5 2.5 3.5

Fill

ed-g

rain

rat

io (

%)

100

80

60

40

20

0

y = −23.833x + 133.45R2 = 0.914

Figure 6.

Shuttle system for breeding cold tolerance rice, Korea (after NCES 1990).

F1

F1–F3

F4–F5

o Seeding testo Screening and

selection at theCold-ScreeningNursery

F6–F7 F6–F7

F8–F9

New cultivar

InternationalRice ColdToleranceNursery

Selection by maturity and various agronomictraits (phenotypic acceptability) atCheolweon, Unbong and Sangju Substations

Local adaptabilityTest in cold regions

Farm fields

Performance test fordesirable breeding linesat Jinbu and Yongdeog

Selection by maturity, spikelet sterility andphenotypic acceptability at Jinbu andYeogdeog Substations

Evaluation for coldtolerance in phytotronand Cold-ScreeningNursery

Chu

nche

on C

old-

Scr

eein

gN

urse

ry, c

old

gree

nhou

se,

phyt

otro

n

Late and early maturingselections

Early maturingselections

116



Figure 8.

Effect of nitrogen level on filled-grain ratio of the1977 and 1980 rice crops, Korea (RDA 1981).

Phosphorus is another important nutrient in coldregions. When sufficient phosphate is applied, lowtemperature damage is reduced. When the amount ofphosphate is increased from 100 to 200 kg ha

–1

,milled rice yield increases by 12% as the grainnumber and filled-grain ratio increase (Table 7;Chungbuk RDA 1980).

Source: Chungbuk RDA (1980).

The application of organic matter, compost, ricestraw and barnyard manure is widespread in thenorthern and mountainous rice-growing areas ofKorea. Farmers believe that applying organic matterimproves the physiological strength of rice and pre-vents low temperature damage, particularly inmountainous areas. Onodera (1936) reports that,from practical experience, applying compost andbarnyard manure reduces cool weather damage inJapan. Amano (1984) shows that applications ofcompost reduces sterility and improves root health,both in morphological and physiological terms, com-pared with plots without compost application. Themechanism providing the favourable effects of com-post on yield has not yet been fully elucidated. Theseobservations were confirmed by an experiment con-ducted at Jinbu Substation during the 1993 low tem-perature year. The optimal N–P–K-supplied plot hadlow yields, compared with the plot treated with ricestraw, animal residue and compost (Table 8). Usingthe findings of this study, we found that applyingreduced synthetic fertilizer and adding an organicsource led to improved yield at low temperatures.

Water management

When the air temperature is low enough to damagerice crops at panicle initiation or early booting, deep-water irrigation (15–20 cm) is an effective way ofprotecting panicle formation and increasing thefilled-grain ratio. A yield increase of 14% and 11%was recorded in Sangjubyeo and Yeongdeog, respec-tively (Table 9).

Table 7.

Effect of phosphate application on rice under lowtemperatures.

Phosphate(kg ha

–1

)Grains(no. in

10

–3

m

–2

)

Filled-grain ratio(%)

Milled rice yield

(t ha

–1

)

100 27.5 87.0 3.89150 28.6 89.0 4.16200 31.1 90.4 4.49

240 320160800Nitrogen level (kg ha−1)

Fill

ed-g

rain

rat

io (

%)

1977

1980

100

80

60

40

20

0

Source: RDA (1998).

Table 6.

Degree of low-temperature tolerance in recommended rice varieties, Korea.

Maturity Varieties(no.)

Degree of tolerance

High Moderately high Fair Moderately susceptible

Susceptible

Early 30 20 6 2 — 2Intermediate 38 13 12 7 1 5Late to intermediate 33 — 7 19 4 3

Total 101 33 25 28 5 10

a

50% = 50% of N–P–K optimum.

Table 8.

Effect of organic matter on rice grain yield under low temperatures at Jinbu, Korea, 1993.

Treatment

a

Heading date in August Filled-grain ratio (%) Brown rice yield (t ha

–1

)

N–P–K (optimum) 12 10.3 0.4450% + rice straw 10 58.1 2.3350% + animal residue 9 49.8 2.3650% + compost 9 52.5 2.3150% + compost + animal residue 9 54.2 2.58

117



Source: NYAES (1993).

Conclusions

Rice researchers and farmers in Korea found thattwo periods of severe low temperatures damaged ricecrops in 1980 and 1993. These experiences led to theestablishment of research facilities, five researchsubstations, a phytotron and greenhouses, and aresearch program plan to develop highly resistantvarieties and appropriate cultural practices. Of thereleased rice varieties, 57% were identified as highlyresistant to low temperatures. Appropriate culturalpractices for fertilization and water managementwere developed to minimize low temperaturedamage. The research programs will continuefocusing on low temperature resistance, both inbreeding and agronomy, to stabilize rice productionunder Korea’s unpredictable climate.

References

Amano, T. 1984. Studies on cool weather damage withspecial reference to improvement in rice cultivation tech-nique. Report of the Hokkaido Prefecture AgriculturalExperiment Station, 46, 1–67.

Chungbuk RDA (Chungbuk Province Rural DevelopmentAdministration). 1980. Annual Research Report.Cheongju, Korea.

Hue, H. 1978. Studies on physiological and ecologicalcharacteristics of Indica/Japonica rice varieties. ResearchReport of Agriculture RDA (Crops), 1–48.

Jun, B.T., Cho, S.Y. and Lee, J.H. 1987. Test of germi-nation ability of new breeding variety and line of riceunder low temperature condition. Research Report onPhytotron Experiment, II. Suwon, Korea, National CropExperiment Station, 5–18.

Kaneda, C. and Beachell, H.M. 1974. Breeding rice forcold tolerance. Saturday Seminar Paper 9. Los Baños,Philippines, International Rice Research Institute (IRRI).

Kim, J.K. and Kim, Y.S. In press. Labor-Saving Cultiva-tion Technologies of Rice in Korea: Direct Seeding andMachine Transplanting. Bangkok, Asia-Pacific Associa-tion of Agricultural Research Institutions (APAARI).

KOIS (Korean Overseas Information Service). 1993. AHandbook of Korea. Seoul, Korea.

Lee, M.H., Lee, D.J., Park, S.K., Rho, Y.D., Lee, J.H. andPark, R.K. 1987a. Varietal differences in low tempera-

ture damage at the reproductive, heading and ripeningstages of the rice plant. Research Report on PhytotronExperiment, II. Suwon, Korea, National Crop Experi-ment Station, 38–59.

Lee, M.H., Oh, Y.J., Lee, J.H. and Ham, Y.S. 1987b.Physiological relationship between nitrogen fertilizer andcold tolerance at low temperature of rice plant. ResearchReport on Phytotron Experiment, II. Suwon, Korea,National Crop Experiment Station, 60–69.

MOAF (Ministry of Agriculture and Forestry). 1994. Agri-cultural Statistics. Seoul, Korea.



NCES (National Crop Experiment Station). 1970. AnnualResearch Report (Rice). Suwon, Korea.

NCES (National Crop Experiment Station). 1977. AnnualResearch Report (Rice). Suwon, Korea.

NCES (National Crop Experiment Station). 1985. RiceVarietal Improvement in Korea. Suwon, Korea.

NCES (National Crop Experiment Station). 1990. RiceVarietal Improvement in Korea. Suwon, Korea.

NHAES (National Honam Agricultural ExperimentStation). 1982. Annual Research Report (Rice). Iksan,Korea.

NHAES (National Honam Agricultural Experiment Sta-tion). 1985. Annual Research Report (Rice). Iksan, Korea.

Nishiyama, I. 1985. Physiology of Cool Weather Damageto the Rice Plant. Sapporoo, Japan, Hokkaido UniversityPress.

NYAES (National Yeongnam Agricultural ExperimentStation). 1993. Annual Research Report (Rice). Milyang,Korea.

Onodera, I. 1936. Cool summer injury to rice plant andfertilizer application. Agriculture and Economy, 3,896–907.

RDA (Rural Development Administration). 1981. Report ofLow Temperature Damage in 1980 and Its Comprehen-sive Technological Countermeasures of Rice, Suwon,Korea.

RDA (Rural Development Administration). 1998. List ofmajor recommended crop varieties. Suwon, Korea.

Sasaki, K. and Wada, S. 1975. Effect of nitrogen, phos-phoric acid and potassium apply on the incidence ofsterility in rice plant. Proceedings of the Crop ScienceSociety of Japan, 44, 250–254.

Satake, T. 1976. Determination of the most sensitive stageto sterile-type cold injury in rice plant. Research Bulletinof the Hokkaido National Agricultural ExperimentStation, 113, 1–43.

Shiga et al. 1977. Methods of Applying Nitrogen Fertilizersfor Higher yields in the Cool Temperature Region, 2.Top-dreesing of Nitrogen at the Panicle Formation andFlag Leaf Emergence Stages of Rice Plants. ResearchBulletin of Hokkaido National Agricultural ExperimentStation, 117, 31-44.

Yoshida, S. 1978. Fundamentals of Rice Crop Science. LosBaños, Philippines, International Rice Research Institute(IRRI).

Table 9.

Effect of shallow (5 cm) and deep (15–20 cm)irrigation on grain yield of rice growing under low temper-atures at two sites, Korea.

Site Water depth (cm)

Filled-grain ratio (%)

Milled-rice yield

(t ha

–1

)

Sangjubyeo 5 59.9 4.0915–20 71.9 4.66

Yeongdeogk 5 59.3 4.1215–20 80.3 4.57

118



Response of Growth and Grain Yield to Cool Water at Different Growth Stages in Paddy Rice

Hiroyuki Shimono, Toshihiro Hasegawa* and Kazuto Iwama

Abstract

Rice growing in cool climates and under flooding can be subjected to suboptimal water temper-ature (T

w

) at any stage of the crop cycle. Although the response to T

w

depends on the stage of cropgrowth, little is understood, in quantitative terms, about this stage-dependent growth response. Wetherefore conducted field trials for three years to determine the response of biomass and grain yieldto T

w

at three different stages: vegetative, reproductive and early grain filling. Cool irrigationwater, under either two or three temperature regimes (16°–25°C), was used for 20 to 34 days ofeach period. We confirmed that grain yield was most severely reduced by low T

w

(below 20°C)during the reproductive period, as a result of low spikelet fertility. Low T

w

during the vegetativeperiod also reduced grain yield by as much as 20%. Although crop growth rate (CGR) was reducedby low T

w

in all stages, the magnitude differed according to period, being greatest during thevegetative period, followed by the reproductive and early grain-filling periods. Reduced CGRbefore heading was associated largely with decreased canopy radiation interception and limitedleaf area, whereas radiation use efficiency (RUE) was relatively unaffected by T

w

. Decreased CGRafter heading was associated with reduced RUE, although leaf area was also reduced by low T

w

.The present results can be used to quantify rice growth and grain yield as affected by T

w

.

Rice (

Oryza sativa

L.) grown under flooding con-ditions can be subject to cool water stress at any timeduring growth. Water temperature (T

w

) can affectvarious growth processes, and response to T

w

candiffer according to growth stage. During the vegeta-tive period, cool water reduces the rates of tillering,leaf emergence and leaf elongation (Enomoto 1936;Takamura et al. 1960; Matsushima et al. 1964b),which, in some cases, are accompanied by leafyellowing (Kondo and Okamura 1931). Cool waterduring the reproductive period, particularly aroundthe microspore stage, substantially decreases spikelet

fertility, thereby resulting in severe yield decline(Tanaka 1962; Matsushima et al. 1964a; Tsunoda1964). Grain yield can also be reduced by low T

w

during the vegetative period (Shimazaki et al. 1963).

Because cool water temperature has a large impacton grain yield, several studies have been conductedto identify the processes and factors that affectspikelet sterility during the reproductive period (

seereviews by

Nishiyama 1983

and

Wada 1992).Takamura et al. (1960) and Matushima et al. (1964b)have also reported on the effect of T

w

on vegetativegrowth, focusing on such processes as tillering andleaf emergence. However, understanding of theresponse of biomass to T

w

is still limited to quanti-fying crop growth under suboptimal T

w

at variousstages of the crop cycle.

Crop growth is an integrated result of variousphysiological processes, including canopy radiationcapture, photosynthesis and conversion of photo-synthate to biomass. However, a simple linear

Graduate School of Agriculture, Hokkaido University,Sapporo, Hokkaido 060-8589, Japan.*Corresponding author: Toshihiro Hasegawa, GraduateSchool of Agriculture, N9, W9, Kita-ku, Sapporo, Hokkaido,060-8589; Phones: +81-11-706-2444; +81-11-706-3878;E-mail: [email protected]

KEYWORDS:

Canopy radiation interception, Cool water temperature, Crop growth rate,

Oryza sativa

L.,Radiation use efficiency, Rice, Spikelet fertility

119



association has often been observed between canopyradiation interception and biomass in many crops(Montieth 1977), and the regression slope, known asradiation use efficiency (RUE), represents thecapacity of plants to use the intercepted solar energyin the accumulation of biomass (

see review by

Sinclair and Muchow 1999).

This simple but physiologically sound relationshiphas often been used in many crop growth models(Horie 1987; Muchow et al. 1990; Amir and Sinclair1991). It also allows us to analyse the processeslimiting crop growth under various stresses. Forexample, Boonjung and Fukai (1996) studied thereduction of dry matter production of upland riceassociated with drought stress at various growthperiods, and found that reduced radiation intercep-tion was largely responsible for reduced crop growthrate (CGR) in the early growth period, but that thegrowth thereafter was limited by RUE. For wheat(Sinclair and Amir 1992) and sunflower (Giménez etal. 1994), both radiation interception and RUE canconstrain crop growth under limited N conditions,and, for maize (Andrade et al. 1993), low air temper-ature (T

a

). However, no such analysis has been madefor rice grown under suboptimal T

w

conditions.

Many attempts have been made to quantify ricegrowth as affected by T

a

(Iwaki 1975; Horie 1987;Berge et al. 1994; Drenth et al. 1994; Hasegawa andHorie 1997), but few have taken into account theeffect of T

w

, even though its importance for cropgrowth is well known. Matsushima (1964a, b) con-ducted a pot experiment with factorial combinationsof T

a

and T

w

(16°–36°C) at various growth stages,and found that growth and yield are limited more byT

w

than by T

a

before the mid-reproductive period.Takamura et al. (1960) also found that T

w

wasrelatively more important than T

a

in leaf emergencein early growth.

In paddy fields, T

w

is different from T

a

, and thedifference is generally larger in a cooler climate.Tanaka (1962) monitored the difference between T

a

and T

w

in paddy fields at Aomori (40° 49

′

north),northern Japan, and showed that the maximum andminimum T

w

can be higher than T

a

by as much as10° and 5°C, respectively, with the differencedecreasing with canopy development. Because T

w

ishigher than T

a

in the first half of the growing period,deep-water irrigation has been used to protectpanicle development from low T

a

in a cool climate(Sakai 1949; Satake et al. 1988).

However, considering the difference between T

a

and T

w

and the large impact of T

w

on growth, theresponse of biomass and yield production to T

w

needs to be determined in relation to various growthparameters to evaluate the magnitude of the stress.

This paper therefore aims to determine the effectsof cool T

w

on dry matter production and yield duringvarious growth periods, thereby improving our quan-titative understanding of growth of field-grown riceunder suboptimal T

w

.

Materials and Methods

Field experiments were conducted in paddy fields atthe Experiment Farm of the Faculty of Agriculture,Hokkaido University, Sapporo, Japan (43° 04

′

north)in 1996, 1997 and 1998. Germinated seeds of ricecultivar Kirara 397 were sown in late April (3 seedsper pot) and seedlings raised under a polyhouse.They were transplanted in late May (Table 1). Thenumber of leaves on the main culm at the trans-planting time was 4.4 in 1996 and 5.2 in 1997 and1998. Planting density was 13

×

33 cm except for1997 (16

×

33 cm). Each plot received equalamounts of basal fertilizers (9.6, 9.6 and 7.2 g m

–2

for N, P

2

O

5

, and K

2

O, respectively).

The treatment was conducted for three growthperiods: vegetative (from 17–21 days after trans-planting [DAP] to panicle initiation [PI]), reproduc-tive (from PI to full heading), early grain filling(from full heading to 20 d after full heading). Thetrial area received cool (about 15°C) irrigated waterbetween 0600 and 1800 h at a rate of about 300 Lmin

–1

(Table 2). A temperature gradient from thewater inlet was used to set either two or three T

w

regimes for each period. The size of each plot wasbetween 64 and 72 m

2

, enclosed with plasticboarding 30 or 45 cm high.

The water level was maintained at 10 cm abovethe soil surface until about 20 days after full heading(86–95 DAP), except for the reproductive period in1996 where deep-water irrigation at 20 cm deep wasconducted.

Each treatment was assigned two letters, the firstrepresenting the period (i.e. V for vegetative, R forreproductive and G for early grain filling), and thesecond, the water temperature (L = low; M = middle).The plot with the highest T

w

for all three growthperiods was designated as ‘control’. In 1996 and1997, the plot with intermediate T

w

was treated

Table 1.

Sowing and transplanting dates and harvest atdays after transplanting for the 1996, 1997 and 1998 ricecrops, Sapporo, Japan.

Year Sowing Transplanting Harvest

1996 18 Apr 21 May 1261997 23 Apr 27 May 115–1291998 22 Apr 26 May 121–129

120



continuously for all three periods and was designatedas ‘CM’. The treatment was not replicated becausethe random assignment of the plots was difficult.

a

V = vegetative stage; R = reproductive stage; G = earlygrain filling; C = continuous treatment throughout crop’slife cycle; L = low water temperature; M = intermediatewater temperature.

To show variation of data, the standard error orpooled standard error was used. Water and soil tem-peratures were measured at the centre of each plot bythermocouples placed 5 cm below either the watersurface or soil surface (Copper-Constantan, 0.6 mmin diameter), and recorded in the data logger (IDL-3200, North Hightech, Sapporo) at 10-min intervals.The daily mean temperature was expressed as theaverage of the daily maximum and minimum tem-peratures. On-site T

a

at 1.5 m above the ground andglobal solar radiation (S) were also recorded. Photo-synthetically active radiation (PAR) was estimatedas 0.5

×

S.Plants with an average number of tillers were

sampled from each plot at seven growth stages,including from the beginning and end of each treat-ment for five (1996, 1997) or four hills (1998).Plants within two rows of the edge of each plot werenot sampled to avoid the border effect.

Leaf area was measured with an automatic areameter (AAM-7, Hayashi Denko, Tokyo), and the dryweight for each organ was determined after dryingfor more than 72 h at 80°C. Leaves were ground, thensubjected to Kjeldahl analysis for N determination.

At maturity, 20 hills (four rows

×

five hills) weresampled at four (1998) or five (1996, 1997) places ineach plot between 115 and 129 DAP, to determineyield and its components. In 1996, the number ofgrains per panicle and 1000-grain weight were notmeasured. Percentage of ripened spikelets was deter-mined with ammonium sulfate solution of 1.06specific gravity. Spikelet fertility was measured forthe panicles on the three tallest culms (1996) andthose on all culms (1997, 1998) of five (1996, 1997)or four (1998) plants in each plot. Anther length andnumber of engorged pollen grains were measured forthe third, fourth and fifth spikelets growing on thefirst, second and third primary branches on the threemain culms of three hills per plot in 1996, followingthe method of Kariya et al. (1985). Plant height, tillernumber and leaf number on the main stem weremeasured weekly for 10 hills per plot until heading(1997 and 1998).

In 1998, canopy PAR transmittance was measuredunder diffuse radiation conditions about twice aweek with a PAR sensor (LI-250, LI-COR, Lincoln,NE, USA) attached to the top of a 1-m stainless steelpole. We measured PAR below the canopy at 20–30points by moving the sensor at about 7 cm intervals,perpendicularly to the rows. Immediately afterreading the below-canopy PAR, the above-canopyPAR was determined to obtain the percentage ofPAR transmittance.

Results

Climatic conditions, water temperatures and developmental stages

Average air temperature (T

a

) during the vegetativeperiod ranged from 16.4° to 18.8°C (Table 3), andwas generally higher in 1997 than in 1996 and 1998.Air temperature (T

a

) during the reproductive periodwas also higher in 1997 than in the other 2 years, butnone of the experimental years showed, for thisperiod, T

a

below 20°C, the critical temperature fordamage from mid-season coolness (Wada 1992).During early grain filling, however, T

a

in 1997 waslower than in the other 2 years. Solar radiation (S) inall 3 years showed similar yearly variation to T

a

,where S during the vegetative and reproductiveperiod was slightly higher in 1997 than in 1996 and1998, and vice versa in early grain filling.

Average water temperature (T

w

) ranged from15.6° to 24.6°C, and differed by 3.6° to 6.7°Cbetween Treatment L and control in each periodtested. Yearly variation was observed for T

w

associ-ated with T

a

and S, where T

w

in 1997 was relativelyhigh in the vegetative and reproductive periods. Soiltemperature at 5 cm below the surface showed a

Table 2.

Periods (days after transplanting) of watertemperature treatments and developmental stages of the1996, 1997 and 1998 rice crops, Sapporo, Japan.

Year Treatment

a

Treatment period

Panicle initiation

Heading

1996 Control — — —VL 21–48 — —RL 49–74 — —GL 75–94 — —CM 21–94 — —

1997 Control 39 64VL 21–45 46 71RL 42–74 39 73GL 66–85 39 63CM 21–85 43 75

1998 Control 39 66VL 17–38 47 73VM 17–38 42 70RL 39–72 39 80RM 39–72 39 76GL 73–92 39 66CM 73–92 39 66

121



smaller diurnal change than did T

w

, and the dailymean was 1° to 2°C lower than that of T

w

. Althoughcool water was supplied during the day, water andsoil temperatures were consistently lower in L and M(data not shown).

a

V = vegetative stage; R = reproductive stage; G = earlygrain filling; C = continuous treatment throughout crop’slife cycle; L = low water temperature; M = intermediatewater temperature.

Low T

w

substantially delayed all developmentalstages (Table 2). Panicle initiation (PI) in VL wasdelayed by 7–8 days because of low T

w

during thevegetative period, while the time from PI to headingwas similar for both VL and control. Low T

w

duringthe reproductive period also had a large impact on thetime between PI and heading: the crop under RLreached heading 9–14 d later than it did under control.

Yield and its components

Grain yield was largely affected by low T

w

during allgrowth periods, but the magnitude of yield loss dif-fered considerably with the period tested (Table 4),as reported elsewhere. Yield was most severelyreduced (by almost 100%) under CM in 1996 and

under RL in all 3 years—these treatments receivedwater at temperatures below 19°C during the repro-ductive period. The CM during the reproductiveperiod in 1997, when T

w

averaged at 19.3°C, alsohad severely reduced yields (58%). The substantialdecrease in spikelet fertility suggested that low T

w

during the reproductive period apparently causes atypical sterility-type cool injury. In addition, the1000-grain weight was lower than in the control,while the number of spikelets was not affected.

A yield loss of 14%–20% was also recorded in thetreatments during the vegetative period in all 3 years.Treatments VL and VM had fewer panicles—numbers were reduced by 9% to 14%—whichaccounted for most of the yield reduction. The 1000-grain weight was slightly higher under VL and VMthan under control. The effect of low T

w

during earlygrain filling was not consistent across the years. In1996, while GL negatively affected yield, little dif-ference was observed between GL and control in1997 and 1998.

Dry matter production

The crop growth rate (CGR) was significantlyreduced by cool water treatments in all the periodstested (Table 5), and the magnitude of the effectappeared different according to the stage, the mostsevere reductions (up to 74%) occurring under treat-ments VL and VM in all the years studied. Treat-ments during the reproductive and early grain-fillingperiods resulted in similar reductions in CGR. UnderCM, CGR was reduced by 13% to 33% for the threegrowth periods, except the vegetative in 1997.

The mean leaf area index (mLAI) was reduced bylow T

w

during almost all periods. Low T

w

during thevegetative period resulted in the largest reduction (asmuch as 50%), followed by RL. Even after heading,low T

w

decreased LAI by about 10%–20%, indicatingfaster senescence under cool water conditions. Theresponse of the relative leaf growth rate (RLGR) toT

w

is illustrated in Figure 1. Note that, for the repro-ductive period, data from PI to booting were used tocalculate RLGR because LAI in some treatmentsreached the maximum value at booting. For all theperiods tested, RLGR responded linearly within thetemperature range tested. As usually occurs withother crops, RLGR decreased in the present study, butthe dependence on temperature was similar betweenthe vegetative and reproductive periods, being abouta 50% decrease in RLGR with a 5°C decrease in T

w

.Moreover, the relationship was well conserved overthe years under different S and T

a

conditions. The RLGR in the vegetative period was positively

correlated with the relative rate of tillering (rTiller)and leaf emergence rate (LER) on the main culm

Table 3.

Average water temperatures, air temperaturesand global solar radiation (± standard deviations) during thevegetative, reproductive and early grain-filling periods ofthe 1996, 1997 and 1998 rice crops, Sapporo, Japan.

Period Treatment

a

1996 1997 1998

Water temperature (°C)Vegetative Control 20.4 ± 1.7 23.2 ±1.7 21.8 ±2.5

VL 16.6 ±1.8 17.7 ±1.5 16.8 ±1.3VM — — 19.2 ±2.0CM 18.0 ±1.7 20.5 ±1.5 —

Reproductive Control 23.3 ±1.9 24.6 ±2.6 23.3 ±1.2RL 16.9 ±1.2 17.9 ±1.5 16.5 ±1.1RM — — 19.5 ±1.6CM 18.4 ±1.6 19.3 ±1.6 —

Early grain Control 22.1 ±1.5 20.3 ±2.4 21.2 ±1.1filling GL 18.5 ±1.3 16.0 ±2.0 15.6 ±0.8

GM — — 16.1 ±0.8CM 19.5 ±1.3 17.3 ±2.0 —

Air temperature (°C)Vegetative 16.4 ±1.8 18.8 ±2.1 16.7 ±2.5Reproductive 20.4 ±2.2 21.9 ±2.4 20.3 ±2.2Early grain filling 21.2 ±2.0 19.8 ±2.6 20.5 ±1.8

Solar radiation (MJ m

–2

d

–1

)Vegetative 14.7 ±6.5 16.2 ±6.9 15.1 ±8.5Reproductive 14.0 ±6.4 16.2 ±7.1 14.7 ±7.6Early grain filling 14.1 ±6.4 11.6 ±6.0 13.9 ±5.9

122



(Figure 2), indicating that the limiting effect of lowT

w

on leaf area development appeared through bothprocesses. In the reproductive period, RLGR couldno longer be related with rTiller because few newtillers appeared, but LER still had a positive influenceon RLGR. Interestingly, a single regression lineeffectively expressed the relationship between RLGRand LER in both vegetative and reproductive periods,possibly providing a reliable basis for estimating leafarea development under various T

w

. To convert LAI to the canopy gap fraction, the

radiation extinction coefficient (k) needs to be

known. We therefore ascertained if the k value varieswith Tw. Figure 3 illustrates the close associationbetween PAR transmittance (logarithm) below thecanopy and LAI measured during the vegetative andreproductive periods in 1998. We observed nosignificant differences in the slope of the regressionline between the treatments in both periods, whichallowed us to use the single and constant k value of0.4 to estimate radiation interception of the canopy.This agrees with the response of the k values of othercrops to such stresses as drought for barley (Goyneet al. 1993) and wheat (Robertson and Giunta 1994)

a

SE = pooled standard error.

Figure 1.

Relationship between relative leaf growth rate (RLGR) and water temperature (Tw) in the 1996, 1997 and 1998rice crops, Sapporo, Japan. ** = significant at 1% level.

�

, 1996;

�

, 1997;

�

, 1998.

Table 4.

Yield, and its components, of the 1996, 1997 and 1998 rice crops grown under different water temperature regimesin Sapporo, Japan.

Year Treatment

a

Yield(g m

–2

)Panicles

(m

–2

)Spikelets

per panicleSpikelets(10

3

m

–2

)1000-grainweight (g)

Fertilespikelets (%)

1996 Control 633 468 — — — 88.6VL 504 401 — — — 91.4RL 0 478 — — — 0.0GL 522 385 — — — 94.8CM 0 459 — — — 0.0SE 26 59 — — — 1.0

1997 Control 533 427 57.8 24.7 22.7 96.4VL 460 381 59.5 22.6 22.9 93.1RL 45 366 60.8 21.9 18.1 31.0GL 565 436 57.8 25.2 22.8 97.0CM 223 362 57.5 20.7 20.5 62.1SE 45 59 3.4 3.0 0.3 6.2

1998 Control 652 565 61.3 34.6 22.5 94.3VL 528 503 53.9 27.2 23.7 93.7VM 539 516 52.0 26.9 23.0 94.2RL 1 710 45.0 31.9 14.5 4.7RM 96 658 57.6 37.6 19.2 16.7GL 693 609 60.1 36.6 22.8 94.5GM 568 520 52.8 27.4 23.0 95.9SE 38 35 3.2 2.3 1.2 5.0

252015252015 252015

Vegetative Reproductive(until booting)

Early grain filling

y = 0.00917Tw − 0.101r = 0.868**

y = 0.00483Tw − 0.0520r = 0.889**

y = 0.00259Tw − 0.0580r = 0.896**

0.15

0.10

0.05

0.00

RLG

R d

−1

0.08

0.06

0.04

0.02

0.00

−0.01

−0.02

−0.03

Tw (°C)

123



and nitrogen deficiency for sunflower (Giménez etal. 1994). Canopy PAR interception (PAR

i

) in 1996and 1997 was therefore derived from PAR, measuredleaf area and the k value of 0.4.

The PAR

i

was reduced by 39% to 57% in VL and13% to 26% in RL, which results are similar to thoseobserved for CGR (Table 5). As a result, radiationuse efficiency (RUE), defined as the quotient ofCGR over PAR

i

, was mostly unaffected by low T

w

inboth periods, except for the VL in 1998 where RUEwas reduced by about 40%.

During early grain filling, poor correlation existedbetween PAR transmittance and leaf area, and PAR

i

did not change with time in any treated plot, eventhough LAI decreased by as much as 22%. Accord-ingly, because the reduced LAI during early grainfilling did not affect PAR

i

, a reduced RUE wasresponsible for the reduced CGR.

Low T

w

during any period also affected dry matterafter treatment (e.g. see 1998 data in Figure 4).Notably, lower T

w

during the vegetative period led to

a wider gap in dry weight during the reproductiveperiod than in the vegetative period, resulting in asmaller biomass during early grain filling, althoughthe longer duration for RL and RM narrowed the dif-ference at harvest.

Discussion

Despite many studies conducted on rice growth atsuboptimal temperatures (Kondo and Okumura1931; Enomoto 1936; Takamura et al. 1960; Tanaka1962; Shimazaki et al. 1963; Matsushima et al.1964a, b; Tsunoda 1964; Sato 1972a, b, 1974), fewhave reported on the response of the crop biomass towater temperature (Tw) under field conditions.

We found that cool Tw, imposed at any growthperiod, reduced CGR by at least 8% to as much as74%, even after the meristem emerged above thewater surface in mid-reproductive period (Table 5).Reduction was severest under low Tw in the vegeta-tive period, causing the largest influence on biomass

a Values in brackets are relative to those of control.b V = vegetative stage; R = reproductive stage; G = early grain filling; C = continuous treatment throughout crop’s life cycle;L = low water temperature; M = intermediate water temperature.

Table 5. Crop growth rate (CGR), mean leaf area in index (mLAI), mean daily canopy PAR interception (PARi) and radiation-use efficiency (RUE) of rice grown under different temperatures in the vegetative, reproductive and early grain-filling periods.a

Period Year Treatmenta CGR (g m–2 d–1)

mLAI PARi(MJ m–2 d–1)

RUE(g MJ–1)

Vegetative 1996 Control 4.36 0.53 1.77 2.46VL 2.00 (0.46) 0.26 (0.49) 0.86 (0.48) 2.34 (0.95)CM 3.24 (0.74) 0.40 (0.75) 1.36 (0.77) 2.38 (0.97)

1997 Control 4.13 0.42 1.68 2.47VL 2.54 (0.62) 0.28 (0.67) 1.02 (0.61) 2.49 (1.01)CM 4.36 (1.06) 0.43 (1.02) 1.65 (0.98) 2.65 (1.07)

1998 Control 4.49 0.67 1.52 2.97VL 1.18 (0.26) 0.30 (0.45) 0.65 (0.43) 1.81 (0.61)VM 2.56 (0.57) 0.46 (0.68) 0.84 (0.55) 3.05 (1.03)

Reproductive 1996 Control 15.2 2.81 4.62 3.29RL 14.0 (0.92) 2.18 (0.78) 4.01 (0.87) 3.50 (1.06)CM 13.2 (0.87) 1.95 (0.69) 3.67 (0.80) 3.58 (1.09)

1997 Control 17.8 2.31 5.45 3.27RL 12.5 (0.70) 1.75 (0.76) 4.03 (0.74) 3.09 (0.94)CM 12.0 (0.67) 1.82 (0.79) 3.61 (0.66) 3.32 (1.01)

1998 Control 17.8 2.56 5.63 3.17RL 13.0 (0.73) 2.41 (0.94) 4.68 (0.83) 2.78 (0.88)RM 14.4 (0.81) 2.62 (1.02) 4.90 (0.87) 2.95 (0.93)

Early grain filling 1996 Control 23.1 3.93 5.76 4.02GL 18.7 (0.81) 3.78 (0.96) 5.67 (0.98) 3.29 (0.82)CM 18.3 (0.79) 3.48 (0.89) 5.47 (0.95) 3.35 (0.83)

1997 Control 18.8 3.22 4.35 4.33GL 11.7 (0.62) 2.83 (0.88) 4.07 (0.94) 2.87 (0.66)CM 15.3 (0.81) 2.63 (0.82) 4.33 (1.00) 3.54 (0.82)

1998 Control 18.5 3.81 5.66 3.27GL 17.0 (0.92) 3.45 (0.91) 5.44 (0.96) 3.12 (0.96)GM 17.0 (0.92) 3.38 (0.89) 5.55 (0.98) 3.07 (0.94)

124Increased Lowland Rice Production in the Mekong Region

edited by Shu Fukai and Jaya BasnayakeACIAR Proceedings 101

(printed version published in 2001)

production after treatment. The reduced crop size inVL and VM during treatment apparently decreasedcanopy radiation interception, compared with thecontrol, after treatment. The result was an evenlarger difference in biomass between control and VLand VM during the reproductive and early grain-filling periods, although time to heading was pro-longed (Table 2).

Figure 3. Relationship between logarithm of photosyntheti-cally active radiation (PAR) transmittance below the canopy(loge (l/lo)) and leaf area index (LAI) during the vegetative(closed symbols) and reproductive (open symbols) periods inthe 1998 rice crop, Sapporo, Japan. = control; = lowwater temperature; = intermediate water temperature. I =PAR below canopy; lo = PAR above canopy; *** = signifi-cance at 0.1%; vertical bar = standard error.

This after-effect decreased not only shoot mass atharvest, but also grain sink capacity, which is knownto depend largely on crop size during the mid-reproductive period (Wada 1969). In our study, wealso observed a close linear association betweenshoot dry weight at booting and spikelet number perunit land area (r = 0.882, P < 0.01). A smaller bio-mass at this stage in VL, that is, 31%–55% smallerthan the control, reduced spikelet number between9% and 23%. As a consequence, grain yielddecreased by 3.4% with a 1°C decrease in Tw in thetemperature range of 16° to 23°C during the vegeta-tive period (Figure 5).

During the reproductive period, reductions inCGR were smaller than in the vegetative period,ranging from 8% to 30%, but grain yield was mostseverely reduced by the treatments. The finding thatgrain yield was most highly sensitive to Tw duringthe reproductive period agrees with the findings ofmany earlier studies (Enomoto 1936; Tanaka 1962;Matsushima et al. 1964a; Tsunoda 1964).

Lower temperatures during the reproductive stage(notably at the microspore stage) have been longknown to reduce anther size and the number ofengorged pollen grains (Hayase et al. 1969; Nishi-yama 1983). In our study, a substantial decrease inanther length was observed with decreasing Twduring the reproductive period, and almost noengorged pollen grains in RL (measured only in1996). This resulted in a relative grain yield responseto Tw (Figure 5) that was similar to what wasobserved in the yield—Ta relationship (NIAS 1975;Wada 1992), where yield drops sharply at tempera-tures below 20°C. Apparently, spikelet fertility is

543210LAI

loge

(l/l

o)

0.0

−0.5

−1.0

−1.5

−2.0

−25

y = −0.397xR2 = 0.975

Figure 2. Relationships between relative leaf growth rate (RLGR), relative tillering rate (rTiller) and leaf emergence rate(LER) in the vegetative (closed symbols) and reproductive (open symbols) periods of the 1997 ( ) and 1998 ( ) ricecrops, Sapporo, Japan. For the reproductive period, data from panicle initiation to booting were used. ** = significant at 1%,*** = significance at 0.1%.

1.51.00.50.01.00.50.0

0.15

0.10

0.05

0.00

0.15

0.10

0.05

0.00

RLG

R d

−1

rTiller (d−1) LER (week−1)

y = 0.123x − 0.031r = 0.964***

y = 2.26x − 0.071r = 0.954**




more sensitive to Tw than is CGR, which overridesthe Tw effect on CGR for grain yield. However, bio-mass production during the reproductive period canbe an important determinant for spikelet number andthe level of carbon reserves at flowering (Wada1969). For varieties that better tolerate sterility-typecool injury, biomass reduction with low Tw mayhave a significant impact on grain yield.

Even after heading, CGR was reduced by low Twin all years to an extent similar to that of the low Tw

treatment during the reproductive period, while theeffect on grain yield was not clear. Matsushima et al.(1964a) and Tsunoda (1964) found that the effects ofTw during grain filling on yield and its componentswere negligible, compared with those of Ta, based onthe pot experiment with factorial combinations of Twand Ta ranging from 16° to 31°C. During grainfilling, the increase in panicle grain weight dependson current assimilate supply and carbon storedbefore heading.

Figure 4. Changes in shoot dry matter accumulation for the whole growth period in the 1998 rice crop, Sapporo, Japan.Thick lines = treatment period; vertical bars = standard error. = control; � = intermediate water temperature; � = lowwater temperature.

Figure 5. Relationship between grain yield (relative to control) and water temperature (Tw) during the vegetative and repro-ductive periods of the 1996 (�), 1997 (�) and 1998 (�) rice crops, Sapporo, Japan.

ReproductiveVegetative

0 40 80 120 160 0 40 80 120 160

1600

1200

800

400

0

Days after transplanting

Sho

ot d

ry m

atte

r (g

m−2

)

Vegetative Reproductive

16 18 20 22 24 16 18 20 22 24 26

Tw (°C)

Rel

ativ

e gr

ain

yiel

d

1.0

0.9

0.8

0.7

1.0

0.8

0.6

0.4

0.2

0.0




Under water stress, in early grain filling, Kobataand Takami (1979, 1981) found that grain weight ofstressed rice increased at a similar rate to that ofcontrol rice, despite heavily reduced dry matter pro-duction. Further, Takami et al. (1990) demonstratedthat grain growth in rice was limited only where totalassimilate supply (current assimilates and storedcarbon) was below grain demand for carbon. Hence,dry matter production, restricted by low Tw in ourstudy, probably did not decrease the level of totalassimilate supply below the grain demand forcarbon.

Under low Tw before heading, any reduction inCGR was caused mostly by reduced canopy PARinterception (PARi), which resulted from limited leafarea (Table 5). Leaf growth is well known to besensitive to various environmental stresses such aslow Ta (Sato 1972a), drought (Boonjung and Fukai1996) and nitrogen deficiency (Hasegawa and Horie1997). The present study also showed that leaf areawas largely responsible for limited growth underlow Tw.

Relative leaf growth rate (RLGR) is known to beclosely associated with Ta (Miyasaka et al. 1975), afinding borne out by our study, including for grainfilling. Water temperature (Tw) is generally con-sidered to strongly influence rice growth in the firsthalf of the growing season, but our results alsosuggest a strong influence of Tw on both leaf areadevelopment and senescence, which influence wasconsistent over the years under different radiationand Ta conditions.

Leaf area growth is an integrated result of tillering,leaf emergence and leaf elongation. While severalstudies have been conducted to find the effect of Twon each process (Takamura et al. 1960; Matsushimaet al. 1964b), few have tried to relate them to eachother to give a dynamic and quantitative relationship.In the vegetative period, both rTiller and LER areapparently responsible for reduced leaf area, whilethe association of RLGR with rTiller diminished inthe reproductive period because few new tillersappeared.

Although we did not measure individual leaflengths, ample evidence exists for limited elongationof the leaf under low Tw (Matsushima et al. 1964b),which might also have reduced the RLGR in ourstudy. Even though several processes are involved inleaf area growth, we found a close associationbetween RLGR and LER across years and growthperiods. Because the interval of leaf emergence canbe easily expressed as a function of Tw (Ellis et al.1993; Sie et al. 1998), the present finding will pro-vide a solid basis for modelling leaf area, which isthe major determinant for crop growth under low Tw.

In contrast to leaf area and PARi, RUE wasrelatively unaffected by low Tw during the vegetativeand reproductive periods. Radiation use efficiency(RUE) is generally considered as a stable parameterunder various environments, but some reportsshowed that low Ta decreased RUE in maize(Andrade et al. 1993) and peanut (Bell et al. 1992).The limited response of RUE to Tw may indicate thesmall impact of Tw on photosynthetic rates. In rice,the photosynthetic rate has been reported to decreasewith Ta (Ishii et al. 1977; Huang et al. 1989; Makinoet al. 1994). Only a few studies investigated theresponse of photosynthesis to Tw, but, in tomato,Shishido and Kumakura (1994) found no apparentchange in the photosynthetic rates, with the soil tem-perature ranging from 10° to 22°C.

In our experiment, the Tw range of 16°–25°C maynot have strongly affected the photosynthetic rate inthe vegetative and reproductive periods. It shouldalso be noted that the treatments conducted in thisstudy were ‘long term’ so that the plants hadprobably acclimatized to low Tw. In fact, as a resultof limited leaf area growth, leaves became thickerduring the treatments (before heading) and leaf Ncontent on an area basis increased, which probablyreduced the negative effect on RUE. ‘Short-term’treatments may possibly lead to different responses.Changes in physiological parameters under low Twneed to be evaluated to clarify this point.

Flooding conditions of irrigated paddy fields canpromote rice growth under cool climates because thewarmer Tw in the first half of the growth period canserve as a ‘water blanket’ to protect the shoot baseand developing panicles. In addition, unlike Ta, Twcan be managed in various ways, including bywarming ponds and canals.

The present study revealed that a slight differencein Tw affects CGR in all growth periods, with themagnitude depending on the given growth stage. Thelargest reduction due to low Tw was observed in thevegetative period, followed by the reproductive andearly grain-filling periods, while grain yield wasmost severely reduced in the reproductive period, asfound elsewhere. During the vegetative and repro-ductive periods, limited leaf area and PARi were themajor reasons for reduced CGR, while RUE wasrelatively unaffected. The decrease in CGR duringearly grain filling was associated with a reducedRUE, although leaf area was also reduced by low Tw.

These responses to Tw obtained in the presentstudy will be useful for identifying the magnitude oftemperature stress and for evaluating the impact ofwater management on growth and grain yield ofirrigated paddy rice.




Acknowledgments

We thank Dr. K. Kariya at the Hokkaido NationalAgricultural Experiment Station for instruction onmeasuring anther length and pollen grain numbers.Thanks are also due to N. Moki and S. Ichikawa,Faculty of Agriculture, Hokkaido University, fortheir help with field management. The collaborationof H. Asaishi, A. Itoh, C. Kashiuchi and S. Fujimuraare gratefully acknowledged.

References

Amir, J. and Sinclair, T.R. 1991. A model of the tempera-ture and solar-radiation effects on spring wheat growthand yield. Field Crops Research, 28, 47–58.

Andrade, F.H., Uhart, S.A. and Cirilo, A. 1993. Tempera-ture affects radiation use efficiency in maize. Field CropsResearch, 32, 17–25.

Bell, M.J., Wright, G.C. and Hammer, G.L. 1992. Nighttemperature affects radiation-use efficiency in peanut.Crop Science, 32, 1329–1335.

Berge, H.F.M. ten, Wopereis, M.C.S., Riethoven, J.J.M.and Drenth, H. 1994. Description of the ORYZA—0modules (version 2.0). In: Drenth, H., Berge, H.F.M. tenand Riethoven, J.J.M. ed. ORYZA Simulation Modulesfor Potential and Nitrogen Limited Rice Production. LosBaños, Philippines, Simulation and Systems Analysis forRice Production (SARP), 43–56.

Boonjung, H. and Fukai, S. 1996. Effect of soil waterdeficit at different growth stages on rice growth andyield under upland conditions. 1. Growth during drought.Field Crops Research, 48, 37–45.

Drenth, H., Berge, H.F.M. ten and Riethoven, J.J.M. 1994.Description of the ORYZA—N modules (version 2.0). In:Drenth, H., Berge, H.F.M. ten and Riethoven, J.J.M. ed.ORYZA Simulation Modules for Potential and NitrogenLimited Rice Production. Los Baños, Philippines,Simulation and Systems Analysis for Rice Production(SARP), 7–42.

Ellis, R.H., Qi, A., Summerfield, R.J. and Roberts, E.H.1993. Rate of leaf appearance and panicle developmentin rice (Oryza sativa L.): a comparison at three tempera-tures. Agricultural and Forest Meteorology, 66, 129–138.

aEnomoto, N. 1936. Effects of cool irrigation on growthand yield. Agriculture, 662, 10–20.

Giménez, C., Connor, D.J. and Rueda, F. 1994. Canopydevelopment, photosynthesis and radiation-use effi-ciency in sunflower in response to nitrogen. Field CropsResearch, 38, 15–27.

Goyne, P.J., Milroy, S.P., Lilley, J.M. and Hare, J.M. 1993.Radiation interception, radiation use efficiency andgrowth of barley cultivars. Australian Journal of Agricul-tural Research, 44, 1351–1366.

Hasegawa, T. and Horie, H. 1997. Modeling the effect ofnitrogen on rice growth and development. In: Kropff,M.J., Teng, P.S., Aggarwal, P.K. et al. ed. Applications

of Systems Approaches at the Field Level. Dordrecht,Netherlands, Kluwer Academic, 243–257.

Hayase, H., Satake, T., Nishiyama, I. and Ito, N. 1969.Male sterility caused by cooling treatment at the meioticstage in rice plants. II. The most sensitive stage tocooling and the fertilizing ability of pistils. Proceedingsof the Crop Science Society of Japan, 38, 706–711.

Horie, T. 1987. A model for evaluating climatic productivityand water balance of irrigated rice and its application toSoutheast Asia. Southeast Asian Studies, 25, 62–74.

Huang, L.K., Wong, S.C., Terashima, I., Zhang, X., Lin,D.X. and Osmond, C.B. 1989. Chilling injury in matureleaves. I. Varietal differences in the effects of chilling oncanopy photosynthesis under simulated Dry Cold DewWind conditions experienced in South-East China. Aus-tralian Journal of Plant Physiology, 16, 321–337.

Ishii, R., Ohsugi, R. and Murata, Y. 1977. The effect oftemperature on the rates of photosynthesis, respirationand the activity of RuDP carboxylase in barley, rice andmaize leaves. Japanese Journal of Crop Science, 46,516–523.

Iwaki, H. 1975. Computer simulation of vegetative growthof rice plants. In: Murata, Y. ed. JIBP SYNTHESIS,volume 11. Tokyo, Tokyo University, 105–118.

aKariya, K., Satake, T. and Koike, S. 1985. The easymethod of measuring the engorged pollen number inrice. Japanese Journal of Crop Science, 54 (2), 230–231.

bKobata, T. and Takami, S. 1979. The effects of waterstress on the grain-filling in rice. Japanese Journal ofCrop Science, 48, 75–81.

bKobata, T. and Takami, S. 1981. Maintenance of the graingrowth in rice (Oryza sativa L.) subject to water stressduring the early grain filling. Japanese Journal of CropScience, 50, 536–545.

aKondo, M. and Okamura, T. 1931. Response of rice growthto water temperature. Agriculture and Horticulture, 6,517–530.

Makino, A., Nakano, H. and Mae, T. 1994. Effects ofgrowth temperature on the responses of ribulose-1,5-bisphosphate carboxylase, electron transport com-ponents, and sucrose synthesis enzymes to leaf nitrogenin rice, and their relationships to photosynthesis. PlantPhysiology, 105, 1231–1238.

bMatsushima, S., Tanaka, T. and Hoshino, T. 1964a.Analysis of yield-determining process and its applicationto yield-prediction and culture improvement of lowlandrice. LXX. Combined effects of air-temperatures andwater-temperatures at different stage of growth on thegrain yield and its components of lowland rice. Proceed-ings of the Crop Science Society of Japan, 33, 53–58.

bMatsushima, S., Tanaka, T. and Hoshino, T. 1964b.Analysis of yield-determining process and its applicationto yield-prediction and culture improvement of lowlandrice. LXXI. Combined effects of air-temperatures andwater-temperatures at different stage of growth on thegrowth and morphological characteristics of rice plants.Proceedings of the Crop Science Society of Japan, 33,135–140.

a In Japanese. b In Japanese, with English summary.




Miyasaka, A., Murata, Y. and Iwata, T., 1975. Leaf areadevelopment and leaf senescence in relation to climaticand other factors. In: Murata, Y. ed. JIBP SYNTHESIS,volume 11. Tokyo, Tokyo University, 72–85.

Monteith, J.L. 1977. Climate and the efficiency of cropproduction in Britain. Transactions of the Royal Societyof Philosophy (London), series B, 281, 277–294.

Muchow, R.C., Sinclair, T.R. and Bennett, J.M. 1990.Temperature and solar radiation effects on potentialmaize yield across locations. Agronomy Journal, 82,338–343.

aNIAS (National Institute of Agricultural Sciences). 1975.Climate change and crop production. Miscellaneousreport. Tokyo, Division of Meteorology, NIAS, 40 p.

aNishiyama, I. 1983. The damage in rice plants imposed bysub- and supra-optimal temperature, particularly aboutspikelets sterility. Japanese Journal of Crop Science, 52,108–117.

Robertson, M.J. and Giunta, F. 1994. Responses of springwheat exposed to pre-anthesis water stress. AustralianJournal of Agricultural Research, 45, 19–35.

aSakai, K. 1949. Effects of deep irrigation water recoveringyields decreased by unseasonable cool weather duringmeiotic stage of rice plants. Agriculture and Horticulture,24, 405–408.

Satake, T., Lee, S.Y., Koike, S. and Kariya, K. 1988. Malesterility caused by cooling treatment at the young micro-spore stage in rice plants. XXVIII. Prevention of coolinjury with the newly devised water management prac-tices—effects of the temperature and depth of waterbefore the critical stage. Japanese Journal of CropScience, 57, 234–241.

bSato, K. 1972a. Growth responses of rice plant to environ-mental conditions. I. The effects of air-temperatures onthe growth at vegetative stage. Proceedings of the CropScience Society of Japan, 41, 388–393.

bSato, K. 1972b. Growth responses of rice plant to environ-mental conditions. II. The effects of air-temperatures onthe organic and mineral composition of the plant parts.Proceedings of the Crop Science Society of Japan, 41,394–401.

bSato, K. 1974. Growth responses of rice plant to environ-mental conditions. IV. The effects of air- and soil-temperatures upon translocation and distribution of 14C.Proceedings of the Crop Science Society of Japan, 43,410–415.

bShimazaki, Y., Satake, T. and Watanabe, K. 1963. Studieson the growth and nutrient absorption of the rice plantreared with cool water irrigation during its early growthstage. Research Bulletin of the Hokkaido National Agri-cultural Experiment Station, 80, 1–12.

bShishido, Y. and Kumakura, H. 1994. Effects of root tem-perature on photosynthesis, transpiration, translocationand distribution of 14C-photoassimilates and root respira-tion in tomato. Journal of the Japanese Society of Horti-cultural Science, 63, 81–89.

Sie, M., Dingkuhn, M., Wopereis, M.C.S. and Miezan,M.K. 1998. Rice crop duration and leaf appearance ratein a variable thermal environment. I. Development of anempirically based model. Field Crops Research, 57, 1–13.

Sinclair, T.R. and Amir, J. 1992. A model to assessnitrogen limitations on the growth and yield of springwheat. Field Crops Research, 30, 63–78.

Sinclair, T.R. and Muchow, R.C. 1999. Radiation useefficiency. Advances in Agronomy, 65, 215–265.

Takami, S., Kabata, T. and van Bavel, C.H.M. 1990. Quan-titative method for analysis of grain yield in rice.Agronomy Journal, 82, 1149–1153.

bTakamura, Y., Takeuchi, S. and Hasegawa, H. 1960.Studies on the effects of soil temperature upon thegrowth of crop plants. III. Soil temperature and leafemergence of rice plant, 8: Relation of temperature ofseveral parts of rice plant to the rate of leaf emergence.Proceedings of the Crop Science Society of Japan, 29,195–198.

bTanaka, M. 1962. Studies on the growth injuries of low-land rice caused by cool water irrigation and delayheading. Aomori Agricultural Experiment StationReport, 7, 1–108.

bTsunoda, K. 1964. Studies on the effects of water-temperature on the growth and yield in rice plants.Bulletin of the National Institute of Agricultural Sciences,A11, 75–174.

bWada, G. 1969. The effect of nitrogenous nutrition on theyield-determining process of rice plant. Bulletin of theNational Institute of Agricultural Sciences, A16, 27–166.

aWada, S. 1992. Cool Damage in Rice Plants. Tokyo,Youkenndo, 261 p.

a In Japanese. b In Japanese, with English summary.

129



Temperature Constraints to Rice Production inAustralia and Laos: A Shared Problem

T.C. Farrell

a,b,c*

, K.M. Fox

b,c

, R.L. Williams

b,c

, S. Fukai

a

, R.F. Reinke

c

and L.G. Lewin

b,c

Abstract

With the transition to dry-season rice production, Laos now faces many new challenges relatingto extreme temperature problems. Temperatures are highly variable throughout Laos, because ofvariable altitudes. Provinces in northern Laos suffer from low temperatures during the rice crop’svegetative and reproductive stages. These problems are shared with temperate-climate rice-growing countries such as Australia, where temperature variation is seasonal. Provinces insouthern Laos face problems relating to high temperatures during anthesis. With the establishmentof the Cooperative Research Centre for Sustainable Rice Production, Australia now focuses onimproving the level of cold tolerance in commercial varieties at establishment and during thereproductive stage. Cultivars with superior seedling vigour and cold tolerance have already beenidentified and incorporated into the Australian breeding program. Collaborative research betweenAustralia and Laos on the management of extreme temperature variability to reduce yield loss willprove mutually beneficial to both countries.

Seasonal temperature variation is common throughoutthe world, at times causing severe food shortages.Extreme temperatures throughout the rice seasondramatically reduce yield, changing key yield com-ponents. Cooperative research into the effect of tem-perature on rice can contribute to food securityworldwide.

Important factors for grain yield potential arevegetative development (emergence to panicleinitiation [PI]), reproductive development (panicleinitiation to heading) and grain formation andripening (Boerema 1974). Although the dynamics ofrice production in Australia and Laos are at differentextremes of the production and mechanization

spectrum (Table 1), significant yield losses due totemperature variability are experienced in bothcountries.

a

Calculated from ABARE’s (2000) commodity report.

With the development of the CooperativeResearch Centre for Sustainable Rice Production(Rice CRC) in Australia, a multifaceted approachto research on low temperature has begun, with

Table 1.

Estimates of rice production in Australia andLaos in 1998.

Australia Laos

Planting method

Rice area (ha)

90% aerial sowing

139 902

100% trans-planting

650 000Total annual rice production (tons) 1.32 million 1.67 millionAverage yield (t ha

–1

) 9.42 2.7Export (%) 85 <10Percentage of cropped land (%) 0.76

a

<80Irrigated land (%) 100 8.3

a

School of Land and Food Sciences, University of Queens-land, Brisbane, Qld., 4072, Australia

b

Cooperative Research Centre for Sustainable Rice Produc-tion, Yanco Agricultural Institute, New South Wales, 2703,Australia

c

New South Wales Agriculture, Yanco Agricultural Institute*Corresponding author: E-mail: [email protected]

KEYWORDS:

Australia, Cold tolerance, Laos, Low temperature, Rice

130



progress being made in the understanding of lowtemperature problems at the protein, cell, organ andplant levels. The collaborative project between RiceCRC and the ACIAR,

Increased Crop Productionfor Lowland Rice in Laos

,

Cambodia and Australia

,has initiated research towards solving the problemscaused by extreme temperatures on rice productionin Australia and Laos.

Rice Practices

Australia

Rice growing in Australia is confined to the Riverinaregion of New South Wales (NSW), centred at 35°south and 146° east. Altitude (averaging 120 m)throughout the region varies little. About 140 000 haare sown each year, producing an average yield ashigh as 9.4 t ha

–1

, with the highest yielding cropsexceeding 13 t ha

–1

. Although rice-growing areas insouthern Riverina are 1° to 2°C cooler than innorthern Riverina, temperatures usually vary littlewithin a given season across the whole rice-growingarea. Average rainfall during the growing season is200 mm with the crop requiring full irrigation. Thegrowing season is characterized by long days withhigh levels of solar radiation. Low temperaturesduring establishment and a cool grain-filling periodrestrict the length of the growing season, while lownight temperatures during reproductive developmentcan cause catastrophic yield loss.

The crop is planted in spring and, after summergrowth, is harvested in autumn. Planting starts in lateSeptember, as soon as the risk of frost is negligible.Full-season and short-season cultivars are sown inearly October and November, respectively, toensure—as far as possible—that reproductivedevelopment coincides with the warmest night tem-peratures (late January–early February). Grain fillingoccurs in February–March when the cooler tempera-tures extend the duration of grain filling, producinggrains of high quality. Rice crops are drained andharvested in March–April, before the first frosts andwhen grain moisture content is between 16% and22%. More than 90% of Australian rice crops aresown aerially. Most of the nitrogen is applied beforepermanent flooding and, if necessary, top-dressed atPI. Average N application rates are 80–100 kg ha

–1

.

Laos

The Lao People’s Democratic Republic (Lao PDR)is located in the tropics, between 14°

and 22°

northand 100°

and 108°

east. The Lao rice area comprises650 000 ha, producing an average yield of 2.7 t ha

–1

,with the highest yields reaching 4 t ha

–1

. Lao PDR isgeographically divided into northern, central and

southern regions, each having different temperatureregimes as a result of variations in altitude andlatitude. Laos has distinct wet and dry seasons.Historically, most rice production in Laos was pro-duced during the wet season as upland and lowlandcrops. With the advent of irrigation, rice productionin the dry season has increased from 13 600 ha in1995 to 87 000 ha in 1999 (NAFRI 2000). The wet-season crop is transplanted to the field in June andharvested in October, whereas the dry-season crop istransplanted between November and January andharvested in May.

Of the rice produced in Laos in 1998, the rainfedlowlands accounted for 74%, the rainfed uplands for13% and another 13% was irrigated (IRRI 1999).The long-term aim for Lao rice production is toreduce the area of rainfed upland rice and increasethat of irrigated rice. Almost all rice in Laos is trans-planted by hand and harvested by non-mechanizedmethods.

The impact of low temperature on Lao rice pro-duction relates specifically to the dry-season crop atboth establishment and during the reproductivestage. As mentioned before, temperature variation isprincipally determined by altitude and latitude. Forexample, Champassak, a southern province (15°north), has an average altitude of 120 m, with hightemperatures during flowering in April (Figure 1f)that sometimes limit dry-season yields. In contrast,Xieng Khouang has an average altitude of 1050 mand is located at 19.5° north. It suffers from lowtemperatures at establishment and during the micro-spore stage in the dry season (Figure 1d). Tempera-ture patterns in all six provinces show that theaverage minimum temperatures slowly decreasefrom November to December throughout early estab-lishment (Figure 1).

Low Temperatures during the Vegetative Stage

Background

The vegetative stage refers to the period from germi-nation to PI and is characterized by active tillering,gradual increase in height, and leaf emergence atregular intervals. Germination starts when seeddormancy has been broken, the seed absorbsadequate water, and is exposed to a soil temperatureranging from about 10° to 40°C. Temperature has aprofound influence on germination by affecting theactivation stage and post-germination growth. Thereare clear varietal differences in seed germination atlow temperatures (Yoshida 1981). Low temperaturescan affect the rice plant’s developmental processes;and impair photosynthesis, thus reducing growth and

131



Figure 1.

The long-term maximum and minimum temperatures of Yanco, Australia, compared with six provinces in Laos:(a) Luang Namtha, (b) Oudomxay, (c) Houaphanh and (d) Xieng Khouang, all in the northern region, (e) Vientiane, centralregion, and (f) Champassak, southern region. Values in parentheses refer to altitude. = Laos dry season; = Laos wetseason; = provincial maximum temperatures; = provincial minimum temperatures; = Yanco(Australia) maximum temperatures; = Yanco minimum temperatures.

o

(a) Luang Namha (560 m) (b) Oudomxay (500 m)

(c) Houaphanh (913 m) (d) Xieng Khouang (1050 m)

(e) Vientiane (171 m) (f) Champassak (120 m)

Tem

pera

ture

(°C

)

Month

Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug

40

35

30

25

20

15

10

5

0

40

35

30

25

20

15

10

5

0

40

35

30

25

20

15

10

5

0

Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug

Sowing timein Australia Microspore Harvest

132



resulting in indirect yield loss because of less carbo-hydrate available for grain production (Smillie et al.1988). Poor establishment and vegetative growth inrice, caused by low temperatures, are commonproblems in countries such as Australia and Laos.

Temperature variability

The average minimum temperature during establish-ment (November) in Yanco, New South Wales, Aus-tralia, is 12.3°C, with the variation being higherwithin, rather than across, years. For the corre-sponding period in Laos (December), the averageminimum temperatures in Xieng Khouang andChampassak were 8.3° and 18.5°C, respectively.

Variation for temperature within and across yearswas calculated according to daily maximum andminimum temperature data. The coefficient of varia-tion (CV) of temperature across years was calculatedfrom the standard deviation of the mean temperaturefor each year divided by the mean temperature acrossall years. The CV of temperature within years wascalculated as the mean of the CV of daily data foreach year. All CVs are shown as percentages(Table 2).

Establishment of dry-season rice crops in provincessuch as Xieng Khouang (where the minimum temper-ature is 8.3

o

C) is difficult, because 10°C is probablythe critical minimum temperature for the elongationof shoots and roots (Yoshida 1981). Temperaturevariability was higher within years than across yearsin all six provinces. In Laos, the distribution of lowtemperatures appears to be related to altitude, with ricein the Vientiane (171 m) and Champassak (120 m)provinces not being affected by low temperaturesduring establishment.

Seedling vigour

Seedling vigour is important for efficient crop pro-duction. Vegetative vigour—the rapid attainment ofplant biomass—depends on the initial size of seed-lings and the rate at which they grow. A controlled

environment experiment was conducted at Yanco toexplore differences in seedling vigour among 38direct-sown cultivars from the International RiceCold Tolerance Nursery (IRCTN). Seedlings weregrown at 25°–15°C for 2 weeks before temperaturetreatments were imposed. Seedling size at 2 weekswas considered as the initial size. Temperature treat-ments comprised 7°/22°C, 10°/25°C and 13°/28°C(minimum and maximum temperatures, respec-tively), reflecting the range of conditions likely tooccur during establishment in Australia and Laos.

There was a five-fold difference in average seed-ling biomass between the low and high temperaturetreatments. Cultivars exposed to the highest tempera-tures had the greatest biomass, averaging 600 mg perseedling. Average biomass of seedlings at the inter-mediate and low temperature treatments were signifi-cantly lower, at 290 mg and 120 mg, respectively. Aselection of four cultivars, including those with thegreatest and least response to temperature is shownin Figure 2.

Figure 2.

Effect of temperature on biomass per seedling, 41days after sowing from a subset of rice cultivars displayingvariation in temperature response. Vertical bars = standarderror.

17.5 20.514.5

Average temperature (°C)

Cultivar

IR36

L202HSC55Amaroo

Bio

mas

s pe

r se

edlin

g (m

g)

800

600

400

200

0

a

Coefficient of variation across years relative to the mean.

b

Coefficient of variation within years relative to the mean.

Table 2.

Comparison of average minimum and maximum temperatures during 1 month of establishment at the YancoAgricultural Institute, Australia, and six provinces in Laos.

Month Site Country Min. °C Across

a

CV% Within

b

CV% Max. °C

Nov Yanco Australia 12.3 10.0 32.4 26.6Dec Luang Namtha Laos 15.0 8.2 16.9 24.6Dec Oudomxay Laos 11.5 16.0 26.3 24.5Dec Houaphanh Laos 11.1 11.9 28.4 20.8Dec Xieng Khouang Laos 8.3 22.1 38.6 21.6Dec Vientiane Laos 17.2 7.4 13.9 28.3Dec Champassak Laos 18.5 — — 31.2

133



The tropical cultivar IR36 had the smallestresponse to increasing temperature. Surprisingly, atemperate cultivar from California, L202, showed amodest response to temperature. HSC55, a cultivarfrom Hungary (a temperate-climate region) showedthe greatest positive response to increasing tempera-ture. There were significant differences betweencultivars for each temperature treatment, with differ-ences being greatest at the highest temperature. TheNSW cultivars Amaroo (Figure 2), Jarrah, Millin andIllabong (not shown) performed similarly and wereintermediate in their response to temperature. At thelowest temperature treatment, no cultivar had a higherbiomass than Amaroo but, at higher temperatures, sig-nificant genotypic variation was seen. In Laos, tem-perature conditions during establishment resemble the10°/25°C and 13°/28°C temperature treatments.

The potential benefits of incorporating early vigourcultivars such as HSC55 into commercial cultivars,include early leaf area display causing increasedradiation capture, improved weed competitivenessand rapid biomass accumulation during the vegetativephases (Reinke, 2000).

Nitrogen uptake

The NSW rice industry considers that N uptake bythe rice plant at PI is a key determinant of yieldpotential. Temperatures during establishment controlthe amount of N in the above-ground tissue of therice plant at PI. Mean air temperature fromNovember 1 to December 31 was calculated for1989 to 1999 at the Yanco Agricultural Institute. Foreach year, the average N uptake for all ‘Amaroo’crops sown in the first 7 days of October was calcu-lated across all rice-growing areas.

A significant correlation (

r

2

= 0.55) was foundbetween average air temperature and N uptake at PI(Figure 3a). The average temperature in 1994 was20°C and N uptake was 87 kg ha

–1

. In 1995, theaverage temperature was 23°C and N uptake was125 kg ha

–1

. A significant correlation existed betweenN uptake and yield (

r

2

= 0.45), which suggests thatgood early growth resulting in higher PI nitrogenuptake is an important factor contributing to higheryields (Figure 3b). A strong correlation betweenaverage temperature during establishment and grainyield (

r

2

= 0.73) highlights the importance of earlygrowth in contributing to Australia’s high yields(Figure 3c).

Low Temperatures during theReproductive Stage

Background

Low temperatures during reproductive development(i.e. PI to maturity) comprise a major constraint to

productivity for the NSW rice industry. Low temper-atures during late January to early February disturbthe normal development of pollen grains, causingspikelet sterility. The risk of yield reduction in a coolyear is greatly enhanced by increased N status of thecrop (Heenan 1984; Satake et al. 1987). Hayase et al.(1969) concluded that, in rice plants, the youngmicrospore stage, which is related to male sterility, isthe stage that is the most sensitive to low tempera-tures. Low temperatures during reproductive devel-opment reduce the number of engorged pollen grainsand fertilized spikelets in rice (Ito 1971). Deepirrigation water (20 cm) during the reproductiveperiod can help protect young panicles from low airtemperatures by providing a buffer and increasingpanicle temperature by as much as 7°C on a coolnight (Williams and Angus 1994).

A recent greenhouse experiment at the Universityof Queensland found a significant positive correla-tion between the total number of engorged pollengrains produced in an anther and the number ofpollen grains intercepted by the stigma (

r

2

= 0.81).This correlation suggests that 600 engorged pollengrains per anther would result in more than 40 beingintercepted on the stigma (Figure 4a). A significantnegative correlation was found between the numberof engorged pollen grains and spikelet sterility (

r

2

=0.59). The relationship suggests that 600 engorgedpollen grains will result in less than 30% sterility(Figure 4b). These correlations indicate that a largenumber of engorged pollen grains per anther is keyto successful fertilization.


The critical temperature for inducing sterility variesamong cultivars. The unpredictability of low temper-ature during the microspore stage of rice has causedsevere yield losses throughout the world. Lowtemperatures throughout Japan in 1993 led to theopening up of Japanese markets to rice imports. Anextended low temperature event in Australia in 1996during the rice crop’s reproductive stage reducedyields across the rice industry by 25%. In Laos, the1999–2000 season was the coolest since 1974,causing major shortfalls in rice production. Satake(1969) estimates that critical temperatures are 15° to17°C in a tolerant cultivar and 17° to 19°C in a sus-ceptible one.

Australian rice crops are exposed to damage fromlow temperatures in the reproductive stage from lateJanuary through early February. Long-term datashow that this period is usually the warmest, with anaverage minimum temperature of 17°C.

In Laos, however, for dry-season rice crops, thetime of reproductive development is more variable,

134



Figure 3.

(a) Average temperature from November 1 to December 31 and nitrogen uptake during panicle initiation for‘Amaroo’ rice crops sown during the first 7 days of October 1990 to 2000. (b) Nitrogen uptake plotted against grain yieldfor ‘Amaroo’ rice crops, 1989 to 2000. (c) Average temperature plotted against grain yield. The correlation of (b) and(c) does not include 1996, which experienced yield reduction due to low temperatures during the reproductive phase(late January–early February). Numbers with apostrophe refer to the year plotted.

Figure 4.

(a) Number of engorged pollen grains per anther regressed against intercepted pollen grains per stigma for ricecultivar Amaroo. (b) Number of engorged pollen grains per anther plotted against spikelet sterility (Gunawardena unpublisheddata).

(a) (b)

(c)

18 19 20 21 22 23 24 80 90 100 110 120 130 140

Average temperature (°C) N uptake (kg ha−1)

18 19 20 21 22 23 24

Average temperature (°C)

N u

ptak

e (k

g ha

−1)

Gra

in y

ield

(t h

a−1)

Gra

in y

ield

(t h

a−1)

140

130

120

110

100

90

80

10

9

8

7

10

9

8

7

r2 = 0.55y = 8.45x − 67

r2 = 0.45y = 0.02x + 6.95

r2 = 0.73y = 0.3x + 2.89

’90’99

’91

’95

’98

’92’96

’00

’94’97

’93

’97’94

’00

’93

’96

’92’98 ’95

’99’91

’90

’97

’94’00

’93

’96

’92’98’95

’99’91

’90

200 12001000800600400200 12001000800600400

Engorged pollen grains per anther Engorged pollen grains per anther

Inte

rcep

ted

polle

n gr

ains

per

stig

ma

Ste

rility

(%

)

100

80

60

40

20

0

70

60

50

40

30

20

10

0

(a) (b)

r2 = 0.81r2 = 0.59

135



occurring between early March and mid-April. Theaverage minimum temperature across five Laoprovinces increases by 3.5°C from early March toearly April (Figure 5). For example, the averageminimum temperature in Oudomxay, northern Laos,increases from 12°C in early March to 16.4°C inearly April. Delaying reproductive development bysowing later may reduce the extent of low tempera-ture damage in Laos.

Genotypic variability

Greenhouse trials

Two experiments, using 36 and 18 cultivars, respec-tively, on genotypic variation for low-temperature tol-erance were recently conducted in temperature-controlled facilities at the Yanco Agricultural Insti-tute. Three day/night temperature regimes (32°/25°C,25°/15°C and 27°/13°C) were imposed on the culti-vars from after PI to head emergence. A combinedanalysis identified seven international cultivars thatconsistently performed better than all the Australian

cultivars. These were cultivars Liman and Pavlovsky(from Russia), Plovdiv 22 (Bulgaria), Akihikari andHaenuki (Japan), HSC55 (Hungary) and M103(California, USA). Low temperatures reduced harvestindex (grain/total biomass) of these tolerant cultivarsby only 20%, compared with 50% for the majorAustralian cultivars.

Field trials

Field trials at the Yanco Agricultural Institute duringthe 1998–99 and 1999–2000 seasons aimed to con-firm the tolerance of cultivars in the field. The 1998/99 season consisted of nine sowing dates from earlyOctober to late December, with each sowing dateincluding a replicated trial of 30 genotypes. How-ever, attempts to confirm cold tolerance in the fieldwere thwarted by the occurrence of above-averagetemperatures. The 1999–2000 field trial comprisedsix sowing dates from 5 October to 30 December1999. Deep (22 cm) and shallow (5 cm) water depthtreatments were imposed throughout the critical

Figure 5.

Temperature variability during the microspore period of the rice crop compared for Yanco, Australia, with thealternative growing periods in five Laos provinces. Champassak data were not available.

Critical temperature

17°C

Early March

Late March

Early April

Late January

Min

imum

tem

pera

ture

(°C

)

25

20

15

10

5

0

Luan

gN

amth

a

Oud

omxa

y

Hou

apha

nh

Vie

ntia

ne

Xie

ngK

houa

ng

Yan

co

Rice-growing area

136



young microspore stage. Twenty-eight cultivars fromdifferent origins, with varying susceptibility to lowtemperatures, were replicated in each of the 12 bays.

Low night temperatures during the late December–early January and late January caused significantlevels of sterility in most cultivars in the shallowwater treatments. Cultivars Liman, M103 andHitomebore (Japan) had low levels of sterility,despite experiencing low temperatures during criticalstages in late January. Cultivars Sprint (Russia),Doongara (Australia) and Leng Kwang (China) hadhigh levels of sterility in the shallow water treatmentsand appeared to be susceptible to mid-season lowtemperatures.

Screening for Low Temperature Tolerance during the Reproductive Stage

Background

With more than 100 000 rice cultivars worldwide,screening for low-temperature tolerance at establish-ment and during the microspore stage in rice hasbeen targeted. Australia’s researchers have success-fully used temperature-controlled facilities and serialfield trials, while Japan’s researchers have developedfield screening, using cool water from bores.

Genetic materials

The IRCTN carries a selection of cultivars fromdifferent origins evaluated at different sites for coldtolerance. Australia has sourced cold-tolerantcultivars a result of their performance in the IRCTN,reports in the literature or recommendations by inter-national scientists. In 1995, 105 internationalcultivars from the 1991 and 1992 IRCTNs wereintroduced to Australia and grown in small plots atthe Yanco Agricultural Institute. In recent tempera-ture- controlled experiments, seven cultivars fromthis nursery were more cold tolerant than ‘Millin’,Australia’s most cold-tolerant cultivar.

Australia has successfully exchanged cold-tolerantmaterial, and continues to do so, with many countriessuch as Japan. Australia’s plant physiologists andbreeders are working closely together to improve thelevel of cold tolerance in commercial rice cultivars.

Deep cool water screening

Japanese experiment stations successfully carry outrapid screening of genotypes for cold tolerance duringthe reproductive stage of rice. Many other researchstations have dedicated small experimental bays toscreening of cold-tolerant cultivars. Rice is trans-planted into these bays and cool bore water (19°C) isintroduced after PI of the first cultivar until flowering

of the last cultivar. Water depth is maintained at20 cm for about 40 days. The rice breeder from theMiyagi Prefecture Agriculture Experiment Station,Furukawa, has successfully released cold-tolerantcultivars such as Jyoudeki, using the Station’s deepcold water screening facility (Nagano 1998). Thewater temperature from a spearhead bore at Yanco,Australia, is about 20°C, which is about 5°C warmerthan bore water in Japan and may not be suitable forcool water screening.

High Temperatures during Flowering


When high temperatures occur during flowering,spikelet sterility can sometimes be seen on the wind-ward side of Australian rice crops. Evaporativecooling can reduce canopy temperature by 7°C onhot windy days, protecting spikelets from high tem-peratures at anthesis. Historically, high-temperature-induced sterility has not been a major problem inSouth-East Asia because most rice is grown in thewet season. However, high temperatures and highhumidity during flowering are now becoming con-straints to rice production in the lowlands, particu-larly in southern Laos.

A recent report confirms that spikelet sterility underhigh temperature increases with humidity (Matsui etal. 1997). The average maximum temperatures duringMarch and April in the Champassak province, is 35.3°and 35.8°C, respectively (Figure 1f). In the Sekongprovince, the average maximum temperature wasgreater than 35°C from January to May 1998.

The breeding program that produces cultivars forsouthern Laos should aim to improve heat toleranceand attempt to induce earlier flowering.

Mechanisms

The rice plant is most sensitive to high temperaturesduring flowering. Too much heat can impair pollengermination and reduce the number of pollen grainson the stigma, thus leading to spikelet sterility(Yoshida et al. 1981). IRRI (1979) confirmed thatgenotypic variation existed by identifying 13 of 291selections that tolerated high temperature damage atflowering

.

High temperatures occurring within thehour after anthesis disturb such reproductive pro-cesses as anther dehiscence, pollen shedding, pollen-grain germination and pollen-tube elongation(Yoshida 1981). More than 10 germinated pollengrains on a stigma are needed for normal fertilization(Togari and Kashiwakura 1958, cited in Yoshida1981). Yoshida (1981) suggested that high tempera-tures on the day of flowering caused spikeletsterility. Anthesis usually occurs between 1000 and

137



1200 h, with temperatures rising in the morning andexceeding the critical temperature (35°C) by 1000 hin hot areas. Therefore, early morning anthesis ishighly desirable if high temperatures are to beavoided and sterility reduced (Yoshida 1981).

The flowers of

Oryza glaberrima

, an African culti-vated rice species, open early in the morning (IRRI1979), and this species has been a source of earlinessfor

Oryza sativa

. The 1-h earlier flower-opening timemay have a significant effect in decreasing sterility,because air temperature rises at a rate of 3°–4°C h

–1

in many tropical areas. In screening tolerantmaterials, 8-h treatments of 35° and 38°C were effec-tive in selecting heat susceptible and tolerant lines,respectively (Yoshida 1981). Two methods are there-fore possible to improve the heat tolerance of culti-vars in Laos: increasing true tolerance through theuse of cultivars that have improved pollen sheddingand pollen germination, and encouraging earlierflowering time to avoid high day temperatures.

Conclusions

Although Australia and Laos have very differentenvironments, the problems relating to the effect ofextreme temperatures on rice production are shared.Altitude accounts for a large proportion of thetemperature variability in Laos. In the northern andcentral regions, low temperature causes problemsduring establishment and the reproductive stage,whereas, in the southern lowlands, high temperaturesduring flowering causes damage.

Identifying and screening genotypes to minimizethe impact of extreme temperatures at establishment,and during the reproductive and flowering stagesmust remain a major focus target of research efforts.Collaboration between international scientists on thecommon problem of temperature constraints to riceproduction can contribute to increased productivityand worldwide food security.

References

ABARE (Australian Bureau of Agricultural Resources andEconomics). 2000. Australian commodities. Forecastsand Issues, 7(1). March Quarter.

Boerema, E.B. 1974. Climatic effects on growth and yieldof rice in the Murrumbidgee Valley of New SouthWales, Australia. Riso, 23(4), 385-397.

Hayase H., Satake T., Nishiyama I. and Ito N. 1969. Malesterility caused by cooling treatment at the meiotic stagein rice plants. II. The most sensitive stage to cooling andthe fertilizing ability of pistils. Proceedings of the CropScience Society of Japan, 38(4) 706-711.

Heenan, D.P. 1984. Low-temperature-induced floret sterilityin the rice cultivars Calrose and Inga as influenced bynitrogen supply. Australian Journal of Experimental Agri-culture and Animal Husbandry, 24(125), 255-259.

IRRI (International Rice Research Institute). 1979. Annualreport for 1978. Manila, Philippines. 137 p.

IRRI (International Rice Research Institute). 1999. 1998annual technical report. National rice research programand Lao-IRRI project. Manila, Philippines.

Ito, N. 1971. Male sterility caused by cooling treatment atthe young microspore stage in rice plant. VIII. Freeamino acids in anthers. Proceedings of Crop ScienceSociety of Japan, 41, 32-37.

Matsui, T., Omasa, K. and Horie, T. 1997. High tempera-ture induced spikelet sterility of japonica rice atflowering in relation to air temperature, humidity andwind velocity conditions. Japanese Journal of CropScience, 66(3), 449-455.

National Agriculture and Forestry Research Institute(NAFRI). 1999. Research highlights. September, 2000.

Nagano, K. 1998. Development of new breeding techniquesfor cold tolerance and breeding of new rice cultivars withhighly cold tolerance, Hitomebore and Jyoudeki. Pro-ceedings of International Workshop on Breeding andBiotechnology for Environmental Stress in Rice.Sapporo, Japan.

Reinke, R.F. 2000. Genetic improvement of seedlingvigour in temperate rice. PhD Thesis.

Satake, T. (1969). Research on cool injury of paddy riceplant in Japan. JARQ, 4(4), 5.

Satake, T., Lee S.Y., Koike S. and Kariya K. 1987. Malesterility caused by cooling treatments at the youngmicrospore stage in rice plants. XXVII. Effect of watertemperature and nitrogen application before the criticalstage on the sterility induced by cooling at the criticalstage. Japanese Journal of Crop Science, 56(3), 404-419.

Smillie, R.M., Hetherington, S.E., He, J. and Nott, R. 1988.Photoinhibition at chilling temperatures. AustralianJournal of Plant Physiology, 15, 207-222.

Williams, R.L. and Angus J.F. 1994. Deep floodwater pro-tects high-nitrogen rice crops from low-temperaturedamage. Australian Journal of Experimental Agriculture,34, 927-932.

Yoshida, S., Satake T. and Mackill D.S. 1981. High-temperature stress in rice. IRRI Research Paper SeriesNo. 67. International Rice Research Institute, Manila,Philippines, 15 p.

Yoshida, S. 1981. Fundamentals of Rice Crop Science.International Rice Research Institute (IRRI), Manila,Philippines, 81–82.

138



Response of Dry-Season Irrigated Rice to Sowing Time at Four Sites in Laos

V. Sihathep

1

, Sipaseuth

2

, C. Phothisane

2

, Amphone Thammavong

3

, Sengkeo

3

, Sonekhone Phamixay

4

, Manith Senthonghae

4

, M. Chanphengsay

5

, B. Linquist

6

and S. Fukai

7*

Abstract

Experiments were conducted on dry-season irrigated rice at four sites: Vientiane Municipalityand Provinces of Champassak, Xieng Khouang and Luang Namtha, Laos. The aims were to deter-mine optimal sowing time and varietal requirements, and to identify temperature effects on growthand yield. Results indicated that optimal sowing time in southern and central Laos is December,when winter temperatures are appropriate for rice establishment. The experiments were conductedin the 1999–2000 season, which was particularly cool. Securing enough seedlings for transplantingwas therefore key to obtaining high yields this season. In northern Laos, sowing in Novemberimproved establishment, compared with sowing in December, when the cold period began at theend of the month. Once the crop was successfully transplanted, low temperatures did not appear toseverely limit yield in any of the regions studied. The experiments need to be repeated in at leastone more year to obtain crop yield response to sowing time under more typical seasonal con-ditions. Historical records should be checked to identify the risk of occurrence of low temperaturesthat would limit production of dry-season rice in northern Laos.

R

ICE

is the single most important crop in Laos. Thearea under dry-season irrigated rice is currently about12% (87 030 ha) of the total rice area in Laos. Thisrepresents an increase of about 60%, compared withthe area planted in the 1997–98 dry season (54 000ha), and almost 500%, compared with the area plantedin the 1992–93 dry season (13 000 ha). Such expan-sion reflects governmental policy to rapidly increasethe level of national rice self-sufficiency whilereducing the year-to-year variability of productioncaused by the impact of extreme climatic conditions.

Cultivation of dry-season rice is important forfurthering the economic development of Laos. How-ever, agronomic research has not kept pace with theexpansion of irrigated areas. Some parts of Laossuffer heavy rice yield losses in the dry season,because of adverse weather conditions such as overlyhigh or low temperatures (J.M. Schiller et al. 2001,this volume). Problems associated with low tempera-tures include low germination levels (Yoshida 1981;Nishiyama 1985), poor seedling establishment,delayed flowering and spikelet sterility (Hayase et al.1969; Ito 1971; Gunawardena et al. 1999). Hightemperatures during flowering and grain filling (lateMarch and April) can also cause yield losses insouthern Laos.

In northern Laos, the common cropping patternfor dry-season irrigated rice is to sow in earlyNovember and transplant in late November to earlyDecember. In the south, where altitudes are lowerand low temperatures are not such a problem, sowingis typically carried out in December and trans-planting in January. However, sowing time often

1

Rice Research Station, Champassak Province, Laos

2

National Agriculture Research Centre, Vientiane, Laos

3

Provincial Agricultural Service, Xieng Khouang Province,Laos

4

Rice Experiment Station, Luang Namtha Province, Laos

5

National Agriculture and Forestry Research Institute,Vientiane, Laos

6

Lao–IRRI Project, Vientiane, Laos

7

School of Land and Food Sciences, The University ofQueensland, Brisbane, Qld., Australia*Corresponding author: E-mail: [email protected]

KEYWORDS:

Irrigated rice, Cold tolerance, Laos

139



depends on the availability of irrigation water. Toestimate optimal sowing time and harvesting, wet-season cropping patterns must also be considered.The wet and dry-season crops are commonlyharvested in November and June, respectively. Earlymaturing, high-yielding varieties are required fordry-season irrigated conditions to ensure that thewet-season rice can be sown at an appropriate time.Optimal sowing times and varietal requirementsdiffer throughout Laos because of temperature varia-bility (J.M. Schiller, et al. 2001, this volume).

This study aims to establish appropriate sowingtimes for different phenological groups to improvedry-season rice yields. This research will contributeto the development of a double-cropping rice system,and will identify the impact of extreme temperaturesat different stages on rice growth and yield in Laoprovinces.


Field experiments were conducted at four sites inLaos, one in each of the Vientiane Municipality(central Laos, altitude 170 m), and the Provinces ofChampassak (southern Laos, 168 m), Luang Namtha(northern Laos, 600 m) and Xieng Khouang (northernLaos, 1050 m). The 5-year average minimum temper-atures during sowing time (November–December)are 21.5°, 23.0°, 19.4° and 18.5°C, respectively.

At each site, varieties from three phenologicalgroups were evaluated. These were early (var. SK12),medium (var. RD10) and late flowering (var. TDK1)plus a local check. Check varieties were TSN1(Vientiane), PN1 (Champassak), TDK3 (LuangNamtha) and Tiane (Xieng Khouang). The experi-mental design was a split plot with three replicates.Sowing times, that is, first sowing (S1), secondsowing (S2), third sowing (S3) and forth sowing (S4),were assigned to main plots, and varieties to subplots.Plot size was 4

×

4 m. Seeds were soaked for about2 days before they were sown into nursery beds.

Seeds were usually sown at 3-week intervals,although actual sowing dates varied to some extent atdifferent sites (Table 1). Seedlings were transplanted30 days after sowing. For RD10 at S1 in the VientianeMunicipality (VTN), not enough seeds were availableand seedlings were transplanted only into one repli-cate. Transplanting of the S2 crop in Luang NamthaProvince (LNT) was delayed for 15 days because theseedlings were too small at 30 days after sowing. Fer-tilizer (90 kg ha

–1

of N and 60 of P

2

O

5

) was appliedto the seedbed at 7 days after sowing. Six seedlingswere transplanted at a spacing of 15

×

15 cm. Aftertransplanting, fertilizer was again applied, at the ratesof 90 kg ha

–1

of N, 60 of P

2

O

5

and 30 of K

2

O.Nitrogen was applied in three equal splits: at trans-planting, and 30 and 50 days after transplanting.

a

CPK = Champassak Province; VTN = Vientiane Municipality; LNT = Luang Namtha Province; XK = Xieng KhouangProvince.

b

n.a.

= not available.

Table 1.

Sowing times (S1–S4), and ranges of flowering times of dry-season irrigated rice varieties, of average maximumtemperatures at flowering and of average minimum temperatures for 10–20 days before flowering at four sites in Laos.

Site

a

Sowing cycle

Sowing date Dates of flowering time Avg. maximum temperatures(°C) at flowering (± 5 days)

Avg. minimum temps. (°C) for 10–20 days before flowering

CPKS1 15 Nov 99 5 Mar–19 Mar 35.0–35.5 21.4–23.9S2 6 Dec 99 13 Mar–27 Mar 34.2–36.0 21.3–25.1S3 27 Dec 99 25 Mar–3 Apr 33.6–36.1 24.4–25.6S4 17 Jan 00 15 Apr–26 Apr 32.1–36.4 24.9–25.9

VTNS1 15 Nov 99 10 Mar–27 Mar 34.7–36.2 19.1–20.7S2 6 Dec 99 22 Mar–3 Apr 35.0–37.0 19.3–21.9S3 27 Dec 99 31 Mar–26 Apr 32.1–36.2 20.3–22.6S4 17 Jan 00 17 Apr–20 May 29.4–31.6 22.8–24.1

LNTS1 15 Nov 99 14 Apr–26 Apr 30.1–31.7 17.8–19.0S2 6 Dec 99 5 May–18 May 26.9–31.5 20.0–20.8S3 27 Dec 99 — n.a.

b

n.a.

b

S4 21 Jan 00 22 May–31 May 27.6–31.2 20.7–21.6XK

S1 15 Nov 99 31 Mar–2 Apr 34.1–35.1 15.6–17.9S2 6 Dec 99 — n.a.

b

n.a.

b

S3 30 Dec 99 9 May–14 May 29.2–30.7 21.3–21.7S4 10 Jan 00 28 May–5 June n.a.

b

20.4–22.1

140



To protect against rice bug damage in VTN andChampassak Province (CPK), Furadan (carbofuran)was applied 30 days after transplanting, and Sevin(carbaryl) was applied at flowering. In LNT andXieng Khouang Province (XK), applying rodenticideat sowing, Sevin at flowering and Furadan at 3% at30 to 45 days after transplanting is common practice.Despite these precautions, rice bug damage was asevere problem for the early maturing varieties in S1in VTN (var. SK12) and CPK (var. PN1). In LNT,the crop planted at S4 was heavily attacked by rats,birds and rice bug and could not be harvested. Ratsalso damaged the S4 nursery in XK. Seedlings weretherefore insufficient for all three replicates, theavailable seedlings being sufficient mostly for tworeplicates.

Dry weight and height of 200 seedlings were deter-mined at transplanting, and tiller number was recorded40–45 days after transplanting. Grain yield at 14%moisture content, panicle number per m

2

, grains perpanicle, 100-grain weight and the filled-grain per-centage were recorded at maturity. The filled-grainpercentage was estimated according to the 100-grainweights of filled and unfilled grain and the weights offilled grain and unfilled grain per m

2

. Daily maximumand minimum temperatures for the experiment’sduration were collected from meteorological stations,which were each located within 10 km of the trial siteand at similar altitudes.

Results

Temperatures

Figure 1 shows the mean maximum and minimumtemperatures for 10-day periods from November1999 to May 2000 for the four sites. Temperaturedata for December in XK were not available, andwere estimated from the regression of daily tempera-tures in XK against corresponding temperatures inLNT obtained in other months.

The 1999–00 season was extremely cold, particu-larly during late December, with low minimum tem-peratures in LNT (4°C), VTN (7°C) and CPK (14°C).In many places in Laos, the late-December tempera-tures were believed to be the lowest since 1974. In allstudy sites, temperatures were higher at the begin-ning of the dry season (November) and decreased tothe lowest in the last 10 days of December. Minimumtemperatures were low in January and February butincreased gradually thereafter. Seasonal variationwas higher for minimum temperatures than formaximum temperatures. The maximum temperaturewas lowest in December and increased graduallyuntil April when it was around 35°C at all sites. Themaximum temperature dropped slightly in May at thebeginning of the wet season.

If a 10-day mean minimum temperature of below15°C constitutes low temperature stress, then thecrops at the CPK site experienced no low tempera-tures during establishment. However, in centralVTN, the minimum temperature was close tothreshold between December and mid-February,except in late December when the temperaturedropped much lower. At the two northern sites, lowtemperatures extended from December to March.

Effects of low temperatures on germination and seedling growth

During late December, in LNT and XK, low temper-atures affected germination, resulting in either nogermination or seedling death for the S2 crop in XK.In LNT, germination problems occurred for thecrops sown at S2, S3 and S4. Figure 2 shows thedaily temperatures in LNT during the early part ofthe experiment. Daily minimum temperatures wereabout 15°C during most days in November and earlyDecember, before it dropped sharply to almost 0°Cin the last 5 days of December. Minimum tempera-ture was above 10°C after January 5.

Because of the extreme low temperatures, onevariety (TDK3) did not germinate, and the number ofseedlings for the other varieties was greatly reducedin S2. Consequently, only one replicate could betransplanted with the available seedlings. Seedlinggrowth for the S2 crop in the nursery was also veryslow, and transplanting was delayed for 15 days,compared with other sowings and sites. The S3sowing, which corresponded with the lowest temper-ature period in late December, resulted in no germi-nation. Germination of the S4 crop (sown 17 Jan)also failed, when the minimum temperature droppedbelow 10°C. The S4 crop was re-sown 4 days later(21 Jan), when the minimum temperature was higherthan 10°C. The 17 Jan seeds might have germinatedhad they remained in the field longer.

Seedling weight and height

Dry weight and plant height of 200 seedlings weredetermined at transplanting. Mean seedling weightand height were regressed with mean air temperatureduring the nursery period (Figure 3). Strong relation-ships existed between temperature and seedlinggrowth. Seedling height and weight data were notavailable from CPK.

At the LNT site, at S2, 45-day-old seedlings wereused, while at all other sites and sowings, 30-day-oldseedlings were used. Seedling weight and height forthe S2 crop at the LNT site were adjusted for thedifference in number of days from sowing to trans-planting. Mean air temperature in VTN was higherthan in the north, as reflected by improved seedling

141



growth. Most sowings at the VTN site had higherseedling weights and heights than did those of theLNT and XK sites.

Flowering times and heat sum requirements

The flowering time of the four varieties differedaccording to site and sowing time (Table 1). Thehigher temperatures in CPK induced earlierflowering than in VTN, LNT or XK. Maximumtemperatures at flowering were about 35°C at CPKand VTN, while it was cooler at LNT. The maximumtemperature data in Table 1 indicated that high tem-perature problems were unlikely to occur in XK,except at S1. The crop flowered 14–24 days earlier

than did the S1 crop at LNT, and it may havesuffered male sterility problems.

The mean minimum temperature for 10–20 daysbefore flowering is also shown in Table 1. Theperiod was chosen because temperatures lower than17°C at this growth stage can cause male sterility.Except for the S1 crop at the XK site, the meanminimum temperature exceeded 17°C.

Heat sum requirements from sowing to floweringwere calculated by assuming the base temperature tobe 10°C. Results showed that heat sum requirementsfor TDK1 and RD10 were similar and were greaterthan for SK12 in CPK and VTN, while the differ-ences among varieties were much smaller in LNT

Figure 1.

Maximum (-

�

-) and minimum (- -) temperatures during November 1999 to May 2000 at sites in the VientianeMunicipality (VTN), and the Provinces of Champassak (CPK), Luang Namtha (LNT) and Xieng Khouang (XK), Laos.Sowing dates (S1–S4) for dry-season irrigated rice are shown. Dotted line indicates the critical temperature of 15°C for lowtemperature damage.

ChampassakVientiane

Luang Namtha Xieng Khouang

40

35

30

25

20

15

10

5

0

40

35

30

25

20

15

10

5

0

Tem

pera

ture

(°C

)

Nov Dec Jan Feb Mar Apr May

Month

Nov Dec Jan Feb Mar Apr May

S1 S2 S3 S4

S1 S2 S3 S4

S1 S2 S3 S4 S1 S2 S3 S4

142



and XK (Table 2). The estimated heat sum wasslightly higher at the CPK site than at the other sitesfor RD10 and TDK1, possibly because the base tem-perature was lower than 10°C. The heat sum for theS1 crop in XK was consistently lower than that forother sowings. The reason for this is not known, butit should be noted that temperature records were notavailable for December at XK and, therefore, theheat sum was estimated, using a regression based ontemperatures recorded at the LNT site.

Figure 2.

Maximum (-

�

-) and minimum (- -) tempera-tures during November 1999 to February 2000 in LuangNamtha Province, Laos. Sowing dates (S1–S4) and trans-planting dates (TP1, TP2 and TP4) for dry-season irrigatedrice are also shown.

Grain yields

Grain yield varied greatly, depending on site andsowing time, while genotypic variation was relativelysmall in most cases. Figure 4 shows the effect of siteand sowing time on mean yield across four varieties.Champassak Province and VTN showed similarresponses to sowing date. Yields were lower at S1and highest at S2, with a gradual decline towards S4.In LNT, the mean yield was over 2500 kg ha

–1

for thefirst two sowings, but the S3 and S4 crops failed toyield. The S3 crop failed because of low tempera-tures, preventing germination, whereas the S4 cropfailed because of heavy infestations of rats, birds andrice bug before harvest, because most of the S4 ricecrops near the experimental site were harvested wellbefore maturity of the S4 crop. Although the S2 cropproduced a good yield, low temperatures reducedgermination and no replicates could therefore becarried out. In XK, the S2 crop failed because of lowtemperatures at germination, whereas the mean yieldof other sowings varied between 2500 and 3500 kgha

–1

. It should be pointed out that the S4 crop did notproduce sufficient seedlings because of low tempera-tures at the nursery, and yield was produced only intwo replicates.

The three highest yields were similar at4500 kg ha

–1

, being obtained by the S2 and S3 cropsin CPK and the S2 crop in VTN. The mean yield ofthe three crops was calculated for each of the threecommon varieties to estimate the potential yield for

30

20

10

0

Tem

pera

ture

(°C

)

TP

1

TP

2

TP

4

S1 S2 S3 S4

Month

Nov Dec Jan Feb

Table 2.

Estimated heat sum accumulation for sowing to flowering of rice varieties TDK1, SK12, RD10 and local checkvarieties PN1 (Champassak Province), TSN1 (Vientiane Municipality), TDK3 (Luang Namtha Province) and Tiane (XiengKhouang Province) at four sowing times (S1–S4), Laos.

VarietyLocation

Heat sum (°C-days)

S1 S2 S3 S4 Mean

SK12Champassak 1766 1592 1584 1650 1648Vientiane 1582 1456 1425 1470 1483Luang Namtha 1733 1655 — 1673 1687Xieng Khouang 1370 — 1682 1930 1661

RD10Champassak 1964 1776 1954 1817 1878Vientiane 1618 1583 1706 1874 1695Luang Namtha 1577 1813 — 1560 1650Xieng Khouang 1370 — 1698 1913 1660

TDK1Champassak 2044 1847 2039 1857 1947Vientiane 1635 1612 1724 1892 1716Luang Namtha 1604 1863 — 1706 1724Xieng Khouang 1406 — 1745 1985 1712

PN1 (Champassak) 1865 1672 1766 1702 1751TSN1 (Vientiane) 1899 1682 1924 2057 1891TDK3 (Luang Namtha) 1546 — — 1656 1601Tiane (Xieng Khouang) 1334 — 1665 1842 1614

143



the dry season. The mean yields were higher forTDK1 (4790 kg ha

–1

) and RD10 (4800) than forSK12 (4190). These were then used to estimate yieldloss percentage at other sowing times and sites foreach variety (Table 3). The reduction was lower forSK12 than for other varieties in most combinations.

a

Only one replicate.

b

Only two replicates.

Figure 4.

Association between grain yield of dry-seasonirrigated rice and sowing date at four sites: VientianeMunicipality (-

�

-) and the Champassak (- -), LuangNamtha (- -) and Xieng Khouang (-

�

-) Provinces, Laos.

Yield and yield components of the four varietiesat the four sites are shown in Table 4. In CPK, thelower yields at S1, compared with those at S2 andS3, were related to the smaller number of grains perpanicle with low percentages of filled grain in allvarieties. In the later flowering TDK1 and RD10,tiller numbers of the S1 crop were higher than for theother sowings, whereas the panicle number wassmaller in S1. This indicates the effect of adverseweather conditions during panicle developmentperiod in the S1 crop in CPK.

Table 3.

Yield reduction in three, dry-season, irrigatedrice varieties in relation to sowing time (S1–S4) at foursites, Laos. The S2 and S3 crops in Champassak Provinceand the S2 crop in Vientiane Municipality were used toestimate reduction percentage.

SiteVariety

Yield reduction (%)

S1 S2 S3 S4

ChampassakTDK1 50 0 0 17SK12 50 0 0 4RD10 50 0 0 9

VientianeTDK1 31 0 20 37SK12 28 0 7 12RD10 66

a

0 21 44Luang Namtha

TDK1 31 70

a

100 100SK12 24 38

a

100 100RD10 35 9

a

100 100Xieng Khouang

TDK1 38 100 16 54

b

SK12 30 100 11 37

b

RD10 37 100 30 49

b

5000

4000

3000

2000

1000

0

Gra

in y

ield

(kg

ha−1

)

15 Nov 99 6 Dec 99 27 Dec 99 17 Jan 00(S4)(S3)(S2)(S1)

Sowing date

Figure 3.

Relationships between mean air tempeature (°C) during the nursery stage and (a) seedling weight and (b) seedlingheight of dry-season irrigated rice at four sowing times in Vientiane Municipality (V), and the Provinces of Luang Namtha (L)and Xieng Khouang (X), Laos. The number after letter indicates sowing time.

18

15

12

9

6

3

20

16

12

8

4

See

dlin

g he

ight

(cm

)

See

dlin

g w

eigh

t (g)

18 20 22 24 26 18 20 22 24 26

Mean air temperature (°C)

V1

V2V3

X1

X4 X3 L1L4

L2

V4

V1

V2

V3

X1X4X3 L1

L4

L2

V4

y = −30.0 + 1.95 x, r = 0.86** y = −33.6 + 2.3 x, r = 0.89**

144



The lower potential yield of SK12 in the S2 and S3crops in CPK and the S2 crop in VTN was related toa smaller number of panicles per m

2

. In VTN, thelower yields of the S1 crop, compared with those ofthe S2 crop, were related to reduced panicle densityand low percentages of filled grain. Filled-grain per-centages were particularly low for SK12 because ofrice bug infestation. In the S4 crop, the yield of late-flowering varieties that flowered in May was lower

than that of SK12, which flowered in mid-April.While maximum temperatures in April were higherthan in May, late-flowering varieties had lower per-centages of filled grain and were not suitable for VTN.

Variety TDK3 did not perform well at anysowings in LNT. In the S1 crop, the low yield of thisvariety was related to low panicle number per m

2

,while in the S2 crop, it was a result of poor germi-nation, possibly because of increased susceptibility

z

Only one replicate

Table 4.

Yield and yield components of four rice varieties sown four times (S1–S4) in (a) Champassak Province, (b)Vientiane Municipality, (c) Luang Namtha Province and (d) Xieng Khoung Province, Laos.

Sowing date Variety Yield(kg ha

–1

)Tillers per m

2

Panicles per m

2

Grains perPanicle

Filled grain (%)

100-grain weight (g)

(a) Champassak Province

15 Nov 99 (S1) TDK1SK12RD10PN1

2388 a2080 a2383 a2118 a

593 a467 b614 a524 b

326 a303 a306 a312 a

60 a59 a70 a57 a

39.5 a38.7 a22.9 b38.2 a

3.0 a2.6 b3.0 a2.7 ab

6 Dec 99 (S2) TDK1SK12RD10PN1

5097 a4281 b4880 a4428 b

475 b488 b511 ab556 a

391 a319 b331 ab345 ab

81 b111 a104 a

97 ab

72.3 a58.9 b60.1 ab70.0 a

3.0 a2.6 b2.9 a2.8 ab

27 Dec 99 (S3) TDK1SK12RD10PN1

4598 a4147 b4717 a4572 a

526 a504 a491 a494 a

343 a293 a319 a324 a

84 a91 a99 a93 a

65.0 ab68.4 a59.5 b64.8 b

3.0 a2.7 b2.9 a2.9 a

17 Jan 00 (S4) TDK1SK12RD10PN1

3971 b4010 b4367 a4153 ab

510 a483 a516 a477 a

293 a281 a324 a315 a

100 a100 a

87 a88 a

58.6 b62.5 a58.3 b62.9 a

3.0 a2.7 b3.0 a2.9 a


–1

)Tillers per m

2

Panicles per m

2

Grains perPanicle

Filled grain (%)


(b) Vientiane Municipality

15 Nov 99 (S1) TDK1SK12RD10TSN1

3277 a3000 a1642

z

3106 a

519 a424 a255

z

287 b

306 a274 ab259

z

201 b

111 a82 a—80 b

55.5 b40.2 c63.7

z

63.2 a

2.8 a2.4 b2.9

z

2.7 a

6 Dec 99 (S2) TDK1SK12RD10TSN1

4697 a4136 a4805 a4387 a

535 a403 b390 b362 b

367 a286 b275 b278 b

96 a91 a94 a91 a

72.2 a68.3 a71.6 a67.4 b

2.9 a2.5 b2.9 a2.5 b

27 Dec 99 (S3) TDK1SK12RD10TSN1

3804 a3864 a3767 a3600 a

433 a352 a329 a352 a

279 a261 a215 a199 a

97 a83 a86 a86 a

49.8 b66.5 a64.2 a63.2 a

2.8 a2.6 b2.8 a2.4 c

17 Jan 00 (S4) TDK1SK12RD10TSN1

3021 ab3666 a2643 bc2016 c

523 a458 ab376 b382 b

265 ab282 a204 ab386 b

65 a85 a68 a63 a

37.6 b71.2 a46.7 b28.3 b

2.7 b2.6 c2.9 a2.4 d

145



to cooler temperatures. In XK, the low yields ofRD10 in the S1 crop and Tiane in the S3 crop wererelated to low percentages of filled grain.

Discussion

The four sites reported in these field trials experi-enced contrasting minimum and maximum tempera-tures during crop establishment and flowering. InCPK and VTN, minimum temperatures were higherthan 15°C during most of the growth period. LNTand XK experienced extremely low temperatures(below 10°C) during crop establishment that affectedgermination and seedling vigour. M.-H. Lee (2001,in this volume) has also reported temperatures ofbelow 10°C as causing germination failure in Korea.The S2 crop in XK and the S3 crop in LNT were

affected by extremely low temperatures (below 5°C),either failing to germinate or the seedlings of all fourvarieties dying. The variety TDK3 did not germinatein the S2 crop in LNT, probably indicating highersusceptibility to low temperatures (between 10° and15°C) than the other varieties, which could germi-nate. Nishiyama (1985, 1995, 1997) also reportedvarietal differences for germination under low tem-peratures. Systematic screening of varieties maytherefore be needed to identify varieties for low tem-perature tolerance in these regions.

Poor crop establishment was the main cause ofcrop failure in XK and LNT. In addition to completefailure, in some cases, not enough seedlings wereavailable to complete transplanting to the fields (e.g.the S2 and S4 crops in LNT and XK, respectively),thus reducing grain production further.

z

Only one replicate.

y

Only two replicates.

Table 4. (Continued)


–1

)Tillers per m

2

Panicles per m

2

Filled grain(%)


(c) Luang Namtha Province

15 Nov 99 (S1) TDK1SK12RD10TDK3

3307 a3146 a3108 a942 b

409 a341 a328 a337 a

255 a120 b147 b

70 c

78.3 a59.9 b57.8 b37.2 c

3.0 a2.8 a2.8 a2.9 a

6 Dec 99 (S2) TDK1SK12RD10TDK3

1426

z

2562

z

4355

z

—

248

z

294

z

286

z

—

248

z

295

z

286

z

—

80.2

z

69.8

z

76.1

z

—

3.0

z

2.9

z

2.9

z

—

21 Jan 00 (S4) TDK1SK12RD10TDK3

432 a261 b250 b148 b


–1

)Tillers per m

2

Panicles per m

2

Filled grain(%)


(d) Xieng Khoung Province

15 Nov 99 (S1) TDK1SK12RD10Tiane

2956 a2925 a2093 b3643 a

324 a300 b312 ab300 b

266 a262 a263 a268 a

58.2 a59.2 a53.1 b59.0 a

3.0 a2.7 b2.9 a3.0 a

30 Dec 99 (S3) TDK1SK12RD10Tiane

4018 a3704 a3332 a3058 a

372 a324 b325 b332 b

319 a294 b303 a207 a

57.6 a 58.5 a58.3 a54.4 a

2.8 a2.7 b2.8 a3.0 a

10 Jan 00 (S4) TDK1SK12RD10Tiane

2192 a

y

2630 a

y

2432 a

y

941 b

y

276 a

y

228 b

y

210 c

y

216 bc

y

226 a

y

200 a

y

194 b

y

201 a

y

83.1 a

y

68.1 b

y

42.5 b

y

40.5 b

y

3.0 a

y

2.7 cy

2.9 by

3.1 ay




Data do not clarify whether extreme temperaturesduring rice growth in the main fields reduced yields.While temperatures at flowering were often high, noclear relationships between temperatures at floweringand filled-grain percentages could be seen. Yoshida(1981) and Yoshida et al. (1981) reported that hightemperatures cause pollen abortion in rice grownunder tropical conditions.

Further studies need to be conducted to confirmthe impact of high temperatures on spikelet sterilityand yield in central and southern Laos. The experi-mental results for the northern regions suggest thatthe effect of low temperatures during germinationand establishment are more crucial for crop growthin northern Laos than low temperatures during thereproductive stage (T.C. Farrell, at al., 2001 in thisvolume). Low minimum temperatures of below 17°Cat the critical panicle development stage wereobtained only in the S1 crop in XK, but filled-grainpercentages in the S1 crop were similar to thoseobtained in the S3 crop for which the minimumtemperature during panicle development was higherthan 20°C.

The 1999–00 dry season in which this experimentwas conducted was extremely cool, particularly inlate December. This experiment therefore needs tobe repeated to confirm the results of the presentexperiment. Optimal times for sowing in northernLaos therefore cannot be recommended, using thesedata alone. The early and late sowings were damagedby pests, and the impact would probably have beenreduced if larger rice areas had been used.

Further research should be carried out to identifyoptimal sowing times, taking more into accountother biotic constraints (such as pest populationbuildup) prevailing in the low-temperature areas ofnorthern Laos. For central and southern Laos,December sowing appears appropriate and yieldsclose to 5000 kg ha–1 should be expected with appro-priate varieties and crop management.

Acknowledgements

This project was funded by the Australian Centre forInternational Agricultural Research (ACIAR).

ReferencesGunawardena, T.A., Farrell, T.C., Fukai, S., Blamey,

F.P.C. and Williams, R.L. 1999. Cold tolerance researchin Australia—a snapshot in understanding cold inducedyield reduction in rice varieties as exacerbated by high Nstatus. Paper presented at the Proceedings of the 2ndInternational Temperate Rice Conference, Sacramento,CA, 13–17 June.

Hayase, H., Satake, T., Nishiyama, I. and Ito, N. 1969.Male sterility caused by cooling treatment at the meioticstage in rice plants. II. The most sensitive stage tocooling and the fertilising ability of pistils. Proceedingsof the Crop Science Society of Japan, 38: 4, 706.

Ito, N. 1971. Male sterility caused by cooling treatment atthe young microspore stage in the rice plant. VIII. Freeamino acids in anthers. Proceedings of the Crop ScienceSociety of Japan, 41: 32.

Nishiyama, I. 1985. Physiology of Cool Weather Damageto Rice Plant. Sapporo, Japan, Hokkaido UniversityPress.

Nishiyama, I. 1995. Damage due to extreme temperatures.In: Matsuo, T., Kumazawa, K., Ishii, R., Ishihara, K. andHirata, H. ed. Science of the Rice Plant, vol. 2. Physi-ology. Food and Agricultural Policy Research Centre,Tokyo, 769–812.

Nishiyama, I. 1997. Cool weather damage of the rice plant.In: Asian paddy fields: their environmental, historical,cultural and economic aspects under various physicalconditions. France, Scientific Committee on Problems ofthe Environment (SCOPE), International Council forScience (ICSU), 59–69.

Yoshida, S. 1981. Fundamentals of Rice Crop Science. LosBaños, Laguna, Philippines, International Rice ResearchInstitute (IRRI), 81–82.

Yoshida, S., Satake, T. and Mackill, D.S. 1981. High-temperature stress in rice. Research Paper Series No. 67.Los Baños, Laguna, Philippines, International RiceResearch Institute (IRRI), 15 p.

147



Finding Genetic Donors for Cold Tolerance in theINGER Gene Pool

Edwin L. Javier* and Ma. Concepción Toledo

Abstract

The International Network for Genetic Evaluation of Rice (INGER) is a collaborative effortbetween national agricultural research systems (NARS) and international agricultural researchcentres. Rice-breeding programs all over the world contribute elite breeding lines and varieties toINGER, who then organizes them into nurseries that are distributed freely to interested NARS forevaluation and use. A nursery may have 50 to more than 100 entries. Cold-tolerant genetic materialsare entered in any of the International Rice Cold Tolerance Nursery (IRCTN), the International RiceBoro Observational Nursery (IRBON), and the International Rice Temperate ObservationalNursery (IRTON). The IRCTN consists of japonica and indica genetic materials, which are usedmainly as genetic donors. Testing sites differ according to low temperature regimes. IRBON carriesmostly indica genetic materials with good yield potential and tolerance at the seedling andvegetative stages in their country of origin. IRTON carries temperate rice lines with goodagronomic characteristics. Some of these lines are adapted to tropical conditions. Results of trialsover a wide range of environments are analysed, and outstanding entries are stored in the IRRIGenebank. INGER data include the origins, pedigrees and grain characteristics of genetic materialsand the results of trials across locations. Annual nursery reports are sent to collaborators andinterested parties. Some INGER partners make seed requests based on results given in annualreports. INGER data are managed through the INGER Information System, which is integrated withthe International Rice Information System, which handles genealogical, characterization andevaluation data on rice.

T

HE

development and use of improved rice varietieshas contributed substantially in improving rice pro-duction throughout the world. The InternationalNetwork for Genetic Evaluation of Rice (INGER),formerly designated as the International Rice TestingProgram (IRTP), has been a major player in theworldwide dissemination of improved cultivars andgenetic donors since its establishment in 1975. Thiscooperative activity involving international agricul-tural research centres (IARCs) and national agri-cultural research systems (NARS) has the followingobjectives:1. To facilitate the unrestricted, safe exchange of

germplasm and information across geographicaland political boundaries worldwide.

2. To broaden the genetic diversity and genetic baseof rice varieties used by farmers.

3. To acquire, characterize and evaluate superiorrice germplasm.

4. To assess and validate important traits of superiorgermplasm, including resistance to, or toleranceof, stresses and quality characteristics.

5. Characterize and evaluate G

×

E interactions forimportant traits so that rice improvement pro-grams, particularly in the NARS, can capitalizeon general and specific adaptation.

6. To enhance the capacity of NARS to use andimprove rice germplasm.More than 21 000 breeding lines and varieties of

rice developed in countries around the world havebeen exchanged and evaluated through INGER overthe years. More than 350 INGER genetic materialshave been released as 530 varieties in 62 countries.Those countries that directly use INGER materialssave 2–5 years of research time and resources permaterial. About 6000 INGER materials have beenused as parents in more than 15 000 crosses gener-ated worldwide by various NARS since 1975. They

International Rice Research Institute (IRRI), Makati City,Philippines*Corresponding author: Edwin L. JavierE-mail: [email protected]

KEYWORDS:

Boro rice, Cold tolerance, Genetic donors, INGER

148



were used to diversify and improve locally adaptedvarieties. More than 1200 lines derived from thesecrosses have also been released.

This paper discusses the finding of genetic donorsfor cold tolerance in the INGER gene pool. Itprovides an overview of how INGER operates,describes the different nurseries with cold-tolerantgermplasm, presents test entries with cold toleranceand good phenotypic acceptability over a wide rangeof environments, and describes information exchangethrough INGER.

Overview of INGER Operations

INGER members send their outstanding rice geneticmaterials, together with their pedigrees and salientcharacteristics, to the International Rice ResearchInstitute (IRRI), usually in the form of small quantitiesof seed. Nominated entries are initially multiplied atIRRI’s post-quarantine area at Los Baños, where theIRRI Seed Health Unit and the Philippine PlantQuarantine Office monitor them for disease incidence.Further seed multiplication at the IRRI Farm is doneuntil the amounts required for the nurseries areobtained. A variety or line will take at least 2 yearsfrom being nominated to being included in a nursery.

Resulting seeds are thoroughly cleaned andchecked. Seeds that are half-filled, discoloured,diseased and off-type are discarded. Only materialswith at least 90% viability and that pass the seedhealth evaluation are included in the nurseries. Forrequests from NARS, the seeds are processedaccording to the phytosanitary requirements of theimporting NARS, although INGER always ensuresthat it distributes seeds of the highest quality.

IRRI informs INGER members of the types ofnurseries available each year. INGER membersdecide what nurseries to import and where to growthem. IRRI sends the nurseries, together with theMaterial Transfer Agreement for FAO-designatedgermplasm. After growing them, IRRI’s collaboratorssend their data to IRRI for analysis and interpretation.Outstanding test entries are identified and stored inthe IRRI Genebank. Results of analyses are sent backto the collaborators.

INGER Nurseries with Genetic Donors for Cold Tolerance

International Rice Cold Tolerance Nursery (IRCTN)

The first IRCTN was established in 1975, then dis-tributed every year from 1975 to 1993 and every otheryear after 1993. Since 1975, about 3000 varieties andbreeding lines have been tested. So far, requests forIRCTN have been received from 65 countries. A

country may request for one or more sets of IRCTNin a given year. More than 170 000 seed packets (sumof number of requests

×

number of varieties) havebeen distributed worldwide.

Entries that are nominated for the IRCTN havecold tolerance at one or more growth stages in theircountries of origin. However, they may or may nothave good agronomic characteristics. Thus, they areused mainly as genetic donors for cold tolerance.Entries are classified as indica or japonica typeaccording to the information provided by the collab-orators or to isoenzyme analyses. They are evaluatedagainst the international check varieties and the bestlocal check at each trial site. The 1999 IRCTNcarried 63 test entries, originating from 11 countriesand IRRI. The indica international checks wereChina 1039 and K39-96-1-1-1-2 (both from India)and the japonica international checks were BarkatK78-13 (from India), Stejaree 45 (former USSR) andTatsumi-mochi (Japan).

An IRCTN is a non-replicated field trial. Each entryis evaluated in 5-m long plots with 4 rows and 1 plantper hill at a spacing of 25

×

25 cm. Collaboratorsdecide the fertilizer rates, pest control measures andother cultural practices to be followed. The standardscreening procedure uses the usual irrigation water inthe station. The temperature of the irrigation water isunknown or beyond the control of the collaborator.The standard screening procedure may be modified byirrigating the trial with cold water, the temperature ofwhich is controlled by the collaborators. The modifiedscreening procedure is used only at a very few sites.Plant data collected include cold tolerance at differentstages of growth, seedling vigour, tillering ability,plant height, days to flowering, phenotypic accepta-bility, panicle exsertion and spikelet fertility. Daily airtemperature is also recorded.

The analysis of varietal performance is maderelative to the temperature regime during the trial’sexecution. The air temperature pattern is defined foreach site by considering the monthly averageminimum temperature relative to the growth stagesof the rice crop. There are eight general temperaturepatterns, based on data received over the last 5 years:1. Low temperatures throughout the growing season.2. Low temperatures at the vegetative stage.3. Low temperatures at the vegetative and flowering

stages.4. Low temperatures at the seedling/vegetative stage

and at maturity.5. Low temperatures at flowering and at maturity.6. Low temperatures at flowering.7. Low temperatures at maturity.8. Moderate temperatures throughout the growing

season (no cold stress).

149



a

IRCTN = International Rice Cold Tolerance Nursery.

b

Temperature pattern 4 = 3°–17°C, low temperatures at theseedling/vegetative stage and at maturity; 5 = 8°–16°C, lowtemperatures at flowering and at maturity.Source: INGER (1994).

a


b

IRRI = International Rice Research Institute.Sources: INGER (1994, 1996, 1999b).

Low temperatures are below 18°C, whereasmoderate temperatures are between 18° and 25°C.Sites falling within the same temperature pattern mayalso vary for other factors such as soils and bioticstresses. The top entries for each trait evaluated candiffer among sites belonging to the same temperature

pattern. Collaborators decide what materials theywant to use in their breeding program.

At IRRI, top entries with wide adaptation are storedfor some years. They are selected according to coldtolerance and phenotypic acceptability at maturityover a range of environments. In the 1993 IRCTN, sixentries with good phenotypic acceptability ratingswere selected according to results of the modified coldtolerance screening procedure used at three sites(Table 1). All came from different countries, withtwo—CN839-102-8/2 and CT6742-22-5-4-M-3-M—belonging to the indica rice group and the others tothe japonica group.

Based on mean phenotypic acceptability across12 sites, employing the standard procedure, andacross 3 sites, using the modified method, six entrieswere identified as promising in the 1993 IRCTN(Table 2).

In the 1995 and 1997 IRCTN, all collaborators usedthe standard field screening procedure and scored thetest entries for cold tolerance and phenotypic accept-ability. In 1995, 19 test entries were selected for coldtolerance in at least 3 test sites (Table 3).

a


b

Temperature pattern 1 = 10°–17°C, low temperaturesthroughout the growing season; 2 = 10°–20°C, low temper-atures at the vegetative stage; 3 = 2°–18°C, low temperaturesat the vegetative and flowering stages; 6 = 11°–18°C, lowtemperatures at flowering. Source: INGER (1996).

Table 1.

Selected entries based on phenotypic accepta-bility, origin, varietal groups, and number of sites for eachtemperature pattern where selected entries showed coldtolerance in the 1993 IRCTN.

a

Selected entry Origin Group Number of sites with

temperature pattern:

b

4 5

CN839-102-8/2 India Indica 2 1CT6742-22-5-4-M-3-M Chile Indica 2 1Milyang 93 South Korea Japonica 2 1Stejaree 45 ex-USSR Japonica 2 1Tomihikari Japan Japonica 2 1‘79004-TR4-4-2-1-1’ Turkey Japonica 2 1

Table 2.

Selected test entries based on mean phenotypicacceptability across sites in the 1993, 1995 and 1997IRCTN.

a

Year Entry Origin

b

Group

1993 Tella Hamsa India IndicaTomihikari Japan JaponicaSuweon 349 (Jinmibyeo) Korea JaponicaK39-2 India JaponicaPanda Italy JaponicaArongana 688 Madagascar Japonica

1995 Jinling 78-102 China JaponicaDalizhaoxai China JaponicaYunlen 1 China JaponicaYunlen 8 China JaponicaDiangen 8 China JaponicaGendiao 3 China JaponicaGihobyeo-M2-6-1 Korea JaponicaYunlen 16 China JaponicaYunlen 18 China JaponicaKungen 4 China Japonica

1997 IR58565-2B-10-2-2 IRRI IndicaIR59469-2B-3-2 IRRI IndicaCT6742-20-20-1-M-M-M Chile JaponicaHexi 10 China JaponicaIR62445-2B-12-1-2 IRRI JaponicaIR63352-AC-202 IRRI Japonica

Table 3.

Selected entries based on cold tolerance, origin,varietal groups, and number of sites for each temperaturepattern where selected entries showed cold tolerance in the1995 IRCTN.

a

Selected entry Origin GroupNumber of sites with temperature

pattern:

b

1 2 3 6

Chuxai China Indica 0 1 3 0Dalizhaoxai China Japonica 1 0 3 0Yunlen 13 China Japonica 0 0 3 1Yunlen 16 China Japonica 0 0 3 1Yunlen 17 China Japonica 0 1 3 0Banjaiman China Japonica 0 0 3 1Diangen 8 China Japonica 1 1 3 1Gendiao 3 China Japonica 0 1 3 0Gihobyeo-M2-6-1 S. Korea Japonica 0 0 3 0K39-96-1-1-1-2 India Indica 0 2 2 2Yunlen 12 China Japonica 2 0 2 0K457-107-3-4-1-1-2 India Indica 2 0 2 1Yungen 79-635 China Indica 0 0 2 1Yungen 20 China Indica 1 0 1 1Yunlen 1 China Indica 0 0 2 1Yunlen 9 China Indica 1 0 1 2Yunlen 8 China Indica 3 0 2 1Yunlen 7 China Indica 0 0 2 1Yunlen 20 China Indica 0 1 2 1

150



All entries originated from China, except three.Ten entries belonged to the indica group and nine tothe japonica group. Based on mean phenotypicacceptability across 17 sites, the top 10 entries wereidentified. Seven entries were common in bothcategories (Table 2). There were three entries rated astolerant of low temperatures at three sites (Table 4)and six entries with good phenotypic acceptability in4 or 5 sites in 1997 (Table 2).

a


b

Temperature pattern 1 = 6°–16°C, low temperaturesthroughout the growing season; 3 = 8°–16°C, low tempera-tures at the vegetative and flowering stages; 4 = 3°–17°C,low temperatures at the seedling/vegetative stage and atmaturity; 5 = 3°–17°C, low temperatures at flowering andat maturity. Source: INGER (1999b).

International Rice Boro Observational Nursery (IRBON)

Boro rice is a special type of rice that is traditionallycultivated in India and Bangladesh during winter.The total area grown to boro rice is about 5.5 millionhectares and is increasing every year. This isattributed to the rice’s high yield potential in a seasonconsidered to be almost risk free, provided that thereis supplementary irrigation. Low temperatures fromseedling to early vegetative stage is a major con-straint to boro rice production.

IRBON started in 1993, with entries coming fromsix countries and IRRI. It started with India andBangladesh as the only countries conducting thetrial. Over the years, countries requesting for IRBONhave increased. The Philippines, Vietnam, Myanmar,South Korea, North Korea, Nepal, Iran, Senegal andNamibia are now exploring the potential of boro rice.Around 7000 seed packets of nearly 350 materialshave been distributed.

IRBON is a yield nursery. It is laid out, using anaugmented design in Latin square with five blocks.Five check varieties are always entered in each

block. The international checks are IR72 (earlymaturing), PSBRC 2 (intermediate-maturing) andIR42 (intermediate to late maturing). Collaboratorsprovide early maturing and late-maturing varieties.Plot size is 5

×

1.2 m (7 rows) and hill spacing is20

×

20 cm, with 2 seedlings per hill.Plant data collected include percentage of germi-

nation in the seedbed, seedling vigour before trans-planting, seedling vigour 15 days after transplanting,cold tolerance at vegetative stage, yield, phenotypicacceptability, plant height and days to flowering. Theaverage minimum and maximum temperatures foreach month of the growing season are also recorded.

In IRBON, the time of occurrence and duration oflow air temperatures at seedling to early vegetativestages differ from site to site. Some genetic materialstolerant of low temperatures at one site may not betolerant at other sites. In general, the mean monthlyair temperatures range from 12°–18°C at the earlyvegetative stage. Test entries with wide adaptation inthe 1996 and 1997 IRBON (INGER 1998, 1999a)are given in Table 5. The top cold-tolerant lines hadgood seedling vigour and seedling recovery.Although the top entries for high yield were differentfrom those for cold tolerance, their levels of coldtolerance were also good.

International Rice Temperate Observational Nursery (IRTON)

This yield nursery is the newest, having been estab-lished in 2000. It has 88 genetically diverse entries,contributed by 14 countries, IRRI and the Centre deCooperation Internationale en Recherche Agro-nomique pour le Developpement (CIRAD). IRTONhas three international checks: IR50 (very earlymaturing), IR72 (early maturing) and PSBRC 2(intermediate maturing). Collaborators enter theirbest early and intermediate-maturing local varieties.Field design is similar to that of IRBON.

One of IRTON’s objectives is to identify temperaterice lines that are adapted to tropical conditions, thusIRTON is being conducted under both temperate andtropical conditions. Countries currently conductingIRTON are Bhutan, India, Sri Lanka, Nepal, Pakistan,Vietnam, Philippines, North Korea, South Korea,China, Egypt, Iran, Turkmenistan and Italy.

Information exchange

Every year, INGER sends collaborators and inter-ested parties nursery reports containing informationon the pedigrees of test entries, their origins and graincharacteristics (analyses done at IRRI), and the resultsof trials across environments. Collaborators withlimited resources for testing often make seed requests

Table 4.

Selected entries based on cold tolerance, originand varietal groups, and number of sites for each tempera-ture pattern where selected entries showed cold tolerance inthe 1997 IRCTN.

a

Selected entry Origin Group Number of sites with temperature

pattern

b

1 3 4 5

K479-2-3 India Indica 1 1 1 0CT6748-CA-17 Chile Japonica 0 0 1 2PR26390-581CRF Philippines Japonica 0 0 1 2

151



based on the information provided by the INGERannual reports.

The flow of genetic information relating to INGERmaterials is as important as the flow of germplasm.Advances in information technology and availabilityof powerful database management tools offerexciting new approaches for information exchangeamong INGER partners. INGER data are managedthrough the INGER Information System, which isintegrated with the International Rice InformationSystem (IRIS), a database system for handling genea-logical, characterization and evaluation data on ricegermplasm and variety improvement. The IRIS’sgenealogical management system can be used toquickly generate genetic information for a particularvariety such as full or partial pedigree, ancestral land-races and their specific genetic contributions, cyto-plasmic source and degree of relationship (coefficientof parentage) with other varieties. Parental selection,variety development, and varietal release will bestrengthened with the use of a comprehensive infor-mation system on rice genotypes.

References

INGER. 1994. Final report of the eighteenth InternationalRice Cold Tolerance Screening Nursery (18th IRCTN),1993. Manila, Philippines, International Rice ResearchInstitute (IRRI), 41 p.

INGER. 1996. Final report of the nineteenth InternationalRice Cold Tolerance Nursery (19th IRCTN), 1995.Manila, Philippines, International Rice Research Institute(IRRI), 57 p.

INGER. 1998. Summary report of the third InternationalRice Boro Observational Nursery (3rd IRBON), 1996.In: Summary Report of the 1996 INGER Nurseries.Manila, Philippines, International Rice Research Institute(IRRI), 57–69.

INGER. 1999a. Final report of the fourth International RiceBoro Observational Nursery (4th IRBON), 1997. Manila,Philippines, International Rice Research Institute (IRRI),27 p.

INGER. 1999b. Final report of the twentieth InternationalRice Cold Tolerance Screening Nursery (20th IRCTN),1997. Manila, Philippines, International Rice ResearchInstitute (IRRI), 43 p.

a

IRBON = International Rice Boro Observational Nursery.

b

IRRI = International Rice Research Institute.Sources: INGER (1998, 1999a).

Table 5.

Selected entries in the 1996 and 1997 IRBON.

a

Year Criteria

Cold tolerance, seedling vigour Yield

Entry Origin

b

Entry Origin

b

1996 CR544-1-7 India IR56381-155-1-2-2 IRRIIR59471-2B-20-2-1 IRRI IR61009-37-2-1-2 IRRIIR58614-B-B-2-2 IRRI IR61009-47-3-1-1 IRRIIR56394-70-1-2-2 IRRI IR25924-51-2-3 IRRIIR61355-3B-11-2 IRRI IR57259-9-2-1-3 IRRIIR72 IRRI

1997 Gautam India BR1711-7-2-4-2 BangladeshIR57257-34-1-2-1 IRRI IR57080-2B-12-2-2-1 IRRINanjing 14 China IR61336-4B-14-3 IRRIOM 987-1 Vietnam NR11 VietnamPSBRC 4 IRRI RP2439-1195-623 India

Zhi 20-5 China

155



Improving the Efficiency and Sustainability ofFertilizer Use in Drought- and Submergence-Prone

Rainfed Lowlands in South-East Asia

R.W. Bell

1,*

, C. Ros

2

and V. Seng

2

Abstract

In the rainfed lowlands of South-East Asia, rice yields are low and often respond weakly tofertilizers. Studies of soils in Cambodia and North-East Thailand suggest that a complex com-bination of factors restrict rice yield and nutrient uptake in response to loss of soil-water saturation.Two significant and closely linked constraints are variable rainfall and lack of soil nutrients. Inter-mittent flooding and drying of soils depresses availability of some nutrients, even when watersupply is adequate. Moreover, extreme fluctuations in soil-water levels may impair root activity,further restricting nutrient uptake. The resulting inefficient uptake apparently leads to weakresponses to fertilizer nitrogen and phosphorus. Developing management strategies for optimizingthe mineral nutrition of rice in drought-prone rainfed lowlands, particularly in the presence ofaluminium toxicity and potassium deficiency, thus depends on understanding the function of riceroots in nutrient uptake and their response to temporal and spatial variation in water content andsoil properties. This need is particularly relevant for the adoption of direct sowing of rice, whichresults in a root system developing initially in aerobic conditions, then being exposed to floodedconditions and, during the growing season, returning to aerobic and, in extreme cases, to droughtconditions. With the potential increase of fertilizer use in the future, and thus potential pollution ofgroundwater and eutrophication of water bodies, new management strategies also need to assessrisks of such contamination and seek ways of preventing it.

R

AINFED

lowland rice is grown in a wide diversity ofenvironments, most of which are located in South andSouth-East Asia (Wade et al. 1999). From the per-spective of crop nutrition and rice productivity, themain distinguishing characteristics of the rainfed low-lands are lack of irrigation water and non-continuousflooding of soil during crop growth (Zeigler andPuckridge 1995). While rainfed lowland rice isusually grown in relatively level, bunded fields toretain surface water, the depth and duration of fieldflooding vary greatly from year to year, within a

growing season and spatially over relatively shortdistances within a field. Rainfed lowland rice is oftenexposed to an extremely variable water regime duringgrowth.

Yields in rainfed lowlands are typically half thosein irrigated rice ecosystems (Wade et al. 1999). Theamount and timing of rainfall is considered as themajor constraint to rice productivity, followed by lowsoil fertility, as represented by a range of limitingfactors, including salinity, alkalinity, Fe toxicity,sulfide toxicity, N, P, K, and Zn deficiencies, andorganic and acid sulfate conditions. The lack of soilfertility is exacerbated by the effects of a changingsoil-water regime on nutrient forms and their availa-bility in the soil. Low rates of fertilizer or, as is oftenthe case, no fertilizer, mean that many of these con-straints continue limiting rice production in farmers’fields.

1

School of Environmental Science, Murdoch University,Western Australia

2

Soil and Water Program, Cambodian AgriculturalResearch and Development Institute (CARDI), PhnomPenh, Cambodia*Corresponding author: E-mail: [email protected]

KEYWORDS:

Fertilizer, Loss of soil water saturation, Nutrient availability, Deficiency, Nutrient uptake,Root function

156



Within the rainfed lowland ecosystem, severalsubecosystems have been recognized, based on themaximum depth of water accumulating in the fields.Within areas where water depth is <25 cm, a furthersubdivision is made according to the prevalence ofdrought and/or submergence during the growingseason (Table 1). In each country of South andSouth-East Asia, most of the subecosystems arerepresented, but the mix differs so that the mainissues for soil fertility management (and breeding)also differ in emphasis from country to country.

About 20% of rainfed lowland rice grows infavourable environments where only minor events ofdrought or submergence limit rice production. Theseareas are relatively more prevalent in Indonesia,Philippines, Vietnam, Laos and Myanmar. Incontrast, more than half of the rainfed lowlands ofIndia and Thailand occur in drought-prone environ-ments. The drought-prone environments of North-East Thailand (and Cambodia and Laos) differ fromthose in India because, while the rainy period is long,rainfall has an overall bimodal distribution and, inany given season, is erratic in amount and distribu-tion. In contrast, the drought-prone areas of north-east India are characterized by a short growingseason with an end-of-season drought beingcommon. In Cambodia, more than half of the rainfedlowlands are on lands that are susceptible to bothdrought and submergence, either in different years orpossibly the same year. Submergence is a wide-spread constraint for rainfed lowland rice, particu-larly in Vietnam, Myanmar and Bangladesh.Medium to deep water levels are most prevalent inthe rainfed lowlands of Myanmar, Indonesia, thePhilippines and Bangladesh.

In this paper, we focus on the less favourable sub-ecosystems of the rainfed lowlands in South-EastAsia and hence emphasize the soil and environmental

constraints to efficient and sustainable fertilizer usein Cambodia, Laos and Thailand. Generally, in theseareas, only one rice crop is grown per year. A longdry season follows the rice harvest and soils remaindry for several months of the year, with limited plantgrowth occurring. In Thailand and Cambodia, anearly monsoon season from May to June is followedby a short dry period in June–July before the mainrainy season (Fukai et al. 1995). Transplanting isoften delayed until there is sufficient rainfall at thestart of the main rainy season for flooding of the soilto occur.

Soils

Nutrients

The soils of the rainfed lowlands of South-East Asiausually have low levels of nutrients, especially of Nand P and, to a lesser extent, K and S. This isespecially so in Cambodia (White et al. 1997; Pheavet al. 1996), Laos (Linquist et al. 1998) and North-East Thailand (Ragland and Boonpukdee 1987).Indeed, given the low rates of fertilizer applied byfarmers, the relatively low rice yields in rainfed low-lands of each of these countries can be attributedlargely to low soil fertility. In Cambodia, rainfedlowland rice will respond to N on virtually all soils,including even the old marine floodplain soils thatreceive annual depositions of alluvial sediment, andare inherently more fertile than other rainfed ricesoils of Cambodia (Table 2). Similarly, all the soils,except the old marine floodplain soils, are low in Pand yield responses to P application on rainfed riceare expected. Many of the soils are low in K, andresponses to this nutrient have been reported,especially when the supply of N and P is improved byfertilizer application. Other deficiencies, including S,

Source: Wade et al. (1999).

Table 1.

Relative occurrence (as percentage of total area) of the main rainfed lowland rice subecosystems in South andSouth-East Asia.

Country Shallow (0–25 cm) and prone to: Medium to deep

(25–50 cm)

Total area(’000 ha)

No water stress Drought Drought + submerg. Submergence

Indonesia 58 10 0 9 23 1 404Philippines 51 16 4 9 20 1 510Myanmar 41 10 0 26 23 2 990Vietnam 38 19 0 32 11 1 744Laos 33 33 33 0 0 277Bangladesh 16 16 19 30 19 5 328India 12 51 15 10 13 14 530Cambodia 10 29 57 0 5 747Thailand 9 52 24 12 3 6 039

Total 20 36 15 16 13 35 907

157



Mg and B, have been diagnosed in pot experiments(Pheav et al. 1996), but only S deficiency has beendemonstrated in the field (CIAP 1995).

Low soil pH is prevalent in lowland soils used forrainfed rice in Cambodia, but the consequence of thisfor rice production is still unclear. In their oxidizedstate, pH (CaCl

2

) is as low as 3.9, and Al saturation70% (Seng et al. 1996). However, flooding inducesrelatively rapid increases in soil pH to values in therange of 5.5 to 6.0 after 2–3 weeks (Seng et al. 1996,1999). At these pH values, no KCl-extractable Alcan be detected; hence, Al toxicity for rice in floodedsoils can be ruled out (Seng 2000). However, as dis-cussed below, the consequences of temporary loss ofsoil-water saturation for soluble Al levels in the soiland for rice nutrition and water uptake are stillunclear.

Many of the same nutrient constraints for rainfedlowland rice production found in Cambodian soilsare shared with soils of North-East Thailand. Theoften extremely sandy nature of the latter soils (claycontents of <5%; Kawaguchi and Kyuma 1977;Mitsuchi et al. 1986; Willett and Intrawech 1988) isattributed to ferrolysis, which leads to the destructionand leaching of clays. In their oxidized state, thesandy soils are acidic with a pH (H

2

O) between 4.6and 5.0. Despite low levels of N, P, K and S, responseto inorganic fertilizer is poor (Ragland and Boon-pukdee 1987; Pairoj et al. 1996). These conclusionsare further supported by a recent international studyby Wade et al. (1999), who found that yields ofrainfed lowland rice at sites in North-East Thailandwere generally lower than at other sites in Bangla-desh, Indonesia, India and the Philippines. In addi-tion, in Thai soils, responses to N–P–K or completefertilizer were generally weaker than in other sites.

In Laos, 85% of rainfed lowland rice crops in thenorthern region and 100% of those in the central andsouthern regions responded to N–P–K fertilizer(Linquist et al. 1998). Nitrogen deficiency was themost prevalent, with 40%–50% of crops responding

to N alone and another 30% when P also wasapplied. Some evidence suggests a S deficiency forrice in Laos, and leaf analysis suggest that Mg levelsmay also be too low for rice.

Physical properties

Soil physical properties have a significant bearing onsoil-water storage and retention, nutrient storage andleaching, the timing and ease of cultivation of soilsand root growth. Soils that have been under lowlandrice cultivation for several years develop a com-pacted layer at a depth of 10–40 cm in their profile(Samson and Wade 1998). The layer aids in retainingrainwater but may also restrict root penetration. InNorth-East Thailand, Laos and Cambodia, at leasthalf of the lowland rice soils are sandy, eitherthroughout the profile or in the surface layers (Whiteet al. 1997; Linquist et al. 1998). This, coupled withthe shallow plough pan and the rice crop’s shallowroot system, limits water storage and retention in theroot zone. Even the presence of a conventionalplough pan on coarse sandy soils is not sufficient toretain water for long after the rain stops. Sharma(1992) reported that water stress is evident in rice onsandy soils from North-East Thailand within 1 weekafter the rains cease.

However, when a significant amount of water inthe profile is stored below the plough pan, mechanicalimpedance in the compacted layer may restrict rootaccess to the stored water. Ahmed et al. (1996) foundthat increasing root mass density in the 10–20 cmlayer of rainfed lowland soils of north-west Bangla-desh increased rice yields (Table 3). Deep cultivationto 20 cm or growing a deep-rooted crop like thelegume

Sesbania aculeata

before rice both decreasedpenetration resistance in the 10–20 cm layer andincreased rice yields by 0.5–1.0 t ha

–1

. Deep sandy soils in North-East Thailand and

Cambodia pose particular problems with water reten-tion that not even the conventional plough pan can

Source: White et al. (1997).

Table 2.

Chemical properties of the main soils used for cultivating rainfed lowland rice, Cambodia.

Property

Deep sandy soils

Shallow sand lying over clays

Depression soils

Black soilsof rainfed lowlands

Brown plain soils

Old marine and lacustrine

floodplain soils

Percentage of shallow-water rainfed rice crop

13 30 13 5 15 21

pH (1:1 soil:water) 5.6 5.9 5.8 5.1 5.5 5.9Olsen P (mg kg

–1

) 1.3 0.4 1.0 2.6 3.1 4.6Total N (mg g

–1

) 0.5 0.3 0.6 1.1 0.9 1.0CEC (cmol(+) kg

–1

) 1.8 1.3 6.3 6.7 18.2 13.5Exch. K (cmol(+) kg

–1

) 0.02 0.03 0.07 0.06 0.16 0.19Organic C (mg g

–1

) 4.7 2.9 6.6 10.9 8.8 9.1

158



overcome. The rapid percolation of water throughthese soils means that they can lose saturationquickly after rainfall ceases. Garrity and Vejpas(1986) showed that an impermeable plastic sheetinstalled at 40 cm deep in a sandy lowland soil atUbon Ratchathani, North-East Thailand, preventedwater percolation loss and increased rice yields. Sub-soil compaction on the same soil type was also effec-tive in reducing percolation losses of water by 88%,and decreased from 60 to 17 the number of dayswhen loss of soil-water saturation occurred, thusincreasing yields by 60%–90% (Sharma et al. 1995).

Source: Ahmed et al. (1996).

Hard-setting behaviour of sandy rainfed lowlandrice soils of Cambodia and North-East Thailand is asignificant factor limiting land preparation for trans-planting. When dry, the soils hard set and thereforebecome difficult to cultivate with draught animals.Within a few hours after cultivation, the soils againachieve high soil strength as the sand and silt grainsconsolidate, making transplanting difficult. Gener-ally, farmers cultivate only as much land as can betransplanted within the hours following. The con-sequences of the hard-setting behaviour of thesesandy soils for root growth and activity are unknown.

Nutrient–water interactions

Fluctuating water regimes comprise the definingcharacteristic of rainfed lowlands. Loss of soil-watersaturation may occur at any time from transplantingonwards and for periods of up to several weeks at atime. Generally, rainfed lowland rice soils will besaturated and flooded at transplanting because, ifpossible, farmers delay transplanting until there is

sufficient water to facilitate transplanting, and tomaximize plant survival. An example from Seng et al.(1996) illustrates the variability of the soil-waterregime for a rainfed lowland rice crop. In the 6 weeksafter transplanting, soils lost saturation at the surfaceduring three periods, each of about 4–7 days(Figure 1) and coinciding with maximum tillering.Thereafter, for 6 weeks, rainfall was adequate tomaintain a water level above the soil surface, but forthe remainder of the growing season, includingpanicle initiation, the soil was again exposed to inter-mittent loss of soil-water saturation. Surprisingly,relatively few studies report on the depth of theperched water table during the growing season so thatthe effects of loss of soil-water saturation on nutrientavailability are poorly defined when comparingexperimental results among sites and seasons. Instal-lation and monitoring of observation bores in experi-mental plots are relatively simple and may add valueto many experiments on fertilizer response in therainfed lowlands.

Flooding has mostly beneficial effects on theavailability of nutrients and their uptake by rice(Ponnamperuma 1972). Flooding, by increasing pHand P availability and decreasing levels of solubleAl, particularly and significantly benefits growth andnutrient uptake of rice growing on sandy, acidic,low-fertility, rainfed lowland soils (Ponnamperuma1972; Willett and Intrawech 1988).

Flooding also has possible negative consequences,including increased levels of Fe

2+

in some soils, lossof NO

3

−

N, sulfide toxicity, organic acid toxicityand Zn deficiency. Symptoms of iron toxicity havebeen reported in rice on some soils of Cambodia butthe yield losses associated with this disorder werenot quantified (White et al. 1997). Application ofS-containing fertilizers to sandy soils in Cambodiaand North-East Thailand have been reported to causetoxicity by sulfides forming under flooded condi-tions, especially when soils are also supplied withorganic manures (Willett and Intrawech 1988; Whiteet al. 1997). The management of N under floodedconditions is different from that on oxidized soils.Under flooded conditions, NO

3

is subject to losses byleaching, except if an impermeable hardpan exists,and by denitrification in the reduced layers offlooded soils. Thus, NO

3

present in soils at floodingmay be lost, whereas NH

4

−

N supplied and incorpo-rated into the flooded soil or released by mineraliza-tion of organic matter is relatively stable.

When flooded soils lose saturation, re-oxidationreverses the beneficial changes that occurred duringflooding. Several studies have examined the effects ofdraining a previously flooded soil on nutrient formsand availability for upland crops grown after rice(Willett and Higgins 1980; Sah and Mikkelsen 1986).

Table 3.

Effect of tillage depth on penetration resistanceof soils, root mass density and grain yield of rice in arainfed lowland soil, north-west Bangladesh.

Parameter Tillage depth

6–8 cm 12–15 cm 18–20 cm 6–8 cm with

Sesbania

as pre-rice crop

Penetration resistance (MPa)

0–10 cm 1.25 1.00 0.75 0.610–20 cm 2.20 1.25 1.25 1.1

Root mass density (kg m

–3

at flowering)

0–10 cm 3.73 3.75 3.57 3.8410–20 cm 0.07 0.15 0.19 0.18

Grain yield (t ha

–1

)

3.9 4.3 4.5 4.5

159



On acid soils, pH drops and soluble Al re-appears.Oxidation of Fe

2+

after drainage of flooded soilremoves the risk of Fe toxicity, but generates amor-phous Fe oxides that react with available forms of Pto decrease P availability. Ammonium N oxidizesreadily to NO

3

. These changes in nutrient availabilityobviously are important considerations for nutrientsupply to upland crops grown after rice. Typically,upland crops require more P after a flooded rice cropthan after upland rice (Brandon and Mikkelson 1979).This fact has implications for the growing dry-seasoncrops after rice, using stored soil water.

In the rainfed lowlands, the water regime is morecomplex and dynamic than the above cases wheresoils planted to irrigated rice are drained after harvestand planted to upland crops. Significant periods ofloss of soil-water saturation occur intermittentlythroughout the growing season (e.g. Seng et al. 1996;Trebuil et al. 1998). The implications of the tem-porary periods of loss of soil-water saturation fornutrient availability are not fully understood,although growth may be depressed as nutrient avail-ability decreases (Fukai et al. 1999). Laboratorystudies show that intermittent flooding of soils resultsin a significant loss of soil N (Patrick and Wyatt1964). Oxidation of NH

4

to NO

3

occurs during lossof soil-water saturation, and the NO

3

is then subjectto leaching as the soil water drains below the root

zone, and to denitrification once re-flooding of thesoil occurs. In rainfed lowlands, where loss of soil-water saturation may occur on several occasionsduring the critical early growth stages, includingtillering and panicle initiation (Figure 1), significantlosses of N are expected to occur.

Seng et al. (1999) tested the effect of a 3-weekperiod of loss of soil-water saturation on P uptake byrice in a pot experiment with two acid rainfed low-land soils from south-eastern Cambodia. They foundthat temporary loss of soil-water saturation led todecreased P uptake and shoot dry matter, whetherwith and without P fertilizer application. Thedecreases were attributed to the decreased availa-bility of P during the period of loss of soil-watersaturation. Subsequent field experiments on the samesoil also suggested that the period of loss of soil-water saturation also depressed P uptake by rice(Seng 2000). However, Willett and Intrawech (1988)and Seng (2000) suggested that the increase insoluble Al following re-oxidation of the soil was apossible additional factor limiting P uptake duringperiods of loss of soil-water saturation.

Improving the Efficiency of Fertilizer Use

The weak responses of rainfed lowland rice to fer-tilizers, despite low soil fertility, suggest that the

Figure 1.

Perched water levels (mm) in relation to the soil surface (at 0) in a rainfed lowland rice field, south-easternCambodia. TP = transplanting; PI = panicle initiation. (After Seng et al 1996.)

DecNovOctSeptAugJuly

TP PI

300

200

100

0

−100

−200

Wat

er le

vel (

mm

)

160



efficiency of fertilizer use can be increased, buttaking into account not only soil factors limitingnutrient uptake but also plant factors.

Root biology

Rice roots can greatly modify their rhizosphere andin doing so, improve the availability of nutrients foruptake. However, optimal benefit from these rootcharacteristics can only be captured if the roots areable to access the pools of water and nutrients avail-able in the soil by appropriate root exploration.Limitations on root exploration are a function of boththe genetics of a cultivar that determine the structureof the root system and its distribution in the soil, aswell as soil physical and chemical factors that limitroot growth. Developing management strategies tooptimize mineral nutrition of rice in drought- andsubmergence-prone rainfed lowlands depends onunderstanding the function of rice roots in nutrientand water uptake and their response to temporal andspatial variation in water content and to soil physicaland chemical properties.

Extreme fluctuations in soil-water levels mayimpair root activity, thus restricting nutrient uptakeeither temporarily or in the longer term. However,the function of the root system in nutrient uptakeunder changed water regimes is not well understood(Samson and Wade 1998). The dynamic nature ofthe soil-water regimes in rainfed lowlands may alsomean that there is a distinct advantage in rootshaving a rapid and plastic response to changingwater supply and redox potential.

Lowland rice has an unusually shallow rootsystem. Generally, 70% or more of the roots are in the0–10 cm layer, 90% in the 0–20 cm layer, and veryfew roots penetrate below 40 cm (Sharma et al. 1994;Figure 2). In contrast, upland rice can have rooting

depths between 70 and 80 cm (Morita and Abe 1996).The shallow rooting behaviour of lowland rice isclearly controlled partly by genetics and partly by thesoil’s physical and chemical properties. Sharma et al.(1994) suggest that lowland rice cultivars showgenotypic differences for root length density,although no differences were found among threecultivars for depth of maximum root penetration.

Sharma et al. (1994) also compared three cultivars,KDML 105, RD6 and IR46, at three sites representinghigh and low topographical positions, and clay andloamy sand textures. Measurable differences in rootlength density among the cultivars were most consist-ently found in the 10–30 cm layer. Cultivar differ-ences in root length density were most obvious in theloamy sand and less so in the clay soil. Somewhatsurprisingly, root density was greater in the 0–10 cmlayer in the high fields, where the perched water tablewas below the soil surface for most of the season, thanin the low fields where the soil surface was flooded.Unfortunately, the effects of root length density onnutrient uptake were not examined, although anoxicconditions can markedly alter nutrient availability inthe rhizosphere.

To grow under anoxic conditions, rice roots carryO

2

to respiring tissues through longitudinal gaschannels (known as aerenchyma) found in the cortex.Aerenchyma development and O

2

flux to root tipsincrease in response to anoxic conditions (Colmer etal. 1998). Oxygen may leak along the root axis, withexcess leaking occurring from the older basal roots,thus limiting the amount of O

2

that can be deliveredto the root tips. Hence, under flooded conditions, thedepth of root growth may be limited by the amountof O

2

that can be delivered to the root tips. Cultivarselection for decreased O

2

leakage from basal rootsunder anoxic conditions may be advantageous.

Figure 2.

Decrease, with soil depth, in root length density of rice at heading, in a rainfed lowland soil. (After Samson andWade 1998.)

6050403020100

Root length density (cm cm−3)

Soi

l dep

th (

cm)

0–5

5–10

10–15

15–20

20–25

25–30

30–35

35–40

161



Colmer et al. (1998) suggested that cultivars differin the degree of aerenchyma development. They foundtwo lowland cultivars that responded to stagnantanoxic solutions by decreasing O

2

leakage to negli-gible levels at 20–25 mm from root tips. In contrast,neither of the two upland cultivars showed any declinein O

2

leakage along the root axis in response to anoxicconditions. However, even if O

2

leakage in basal rootscan be decreased, Kirk and Du (1997) showed that asignificant proportion of O

2

leakage also occurredfrom fine lateral roots, and that these roots can prolif-erate from basal as well as apical parts of the root axis.

Kirk and Du (1997) also reported that P deficiencyincreased both root porosity and the rate of O

2

releasedper plant into the rhizosphere. Releasing O

2

into therhizosphere is beneficial under anoxic conditionsbecause it oxidizes Fe

2+

and causes the soil pH of therhizosphere to decline by 1 or 2 units. The excessuptake of cations relative to anions by rice underanoxic conditions also contributes to rhizosphereacidification. Collectively, these mechanisms increaseP availability to rice because, in the acid rhizosphere,extractable-P levels increase. Rhizosphere acidifica-tion also appears to increase Zn uptake, but decreasesuptake of NH

4

−

N and K (Kirk et al. 1994).While root nutrient and water uptake processes

under oxic or anoxic conditions is reasonably wellunderstood, very little is known about how nutrientand water uptake by rice roots responds to a changingsoil-water regime. In rainfed lowlands, roots may beexposed for most of the growing season to anoxicconditions, interspersed with periods of loss of soil-water saturation. Alternatively, the roots may bemostly exposed to oxic conditions with periods ofanoxia. The transition from oxic to anoxic conditionsappears to result in increased root porosity in onlysome cultivars (Colmer et al. 1998). Whether adap-tive responses in root structure or physiology occur asconditions change from anoxic to oxic is unclear.Neither is it clear whether roots adapted to anoxicconditions effectively absorb nutrients and waterwhen exposed to oxic conditions. Response to loss ofsoil-water saturation may require rapid initiation andgrowth of new roots adapted to these conditions. Thedynamics of root response and the functionalefficiency of roots for nutrient and water uptake underchanging soil-water regimes require further research.

The problems of root adaptation to changing soil-water regimes are compounded in direct seeding.The roots of direct-seeded rice initially developunder oxidized conditions, are then exposed toanoxic conditions and, during the growing season,return to oxic conditions, which, in extreme cases,can mean drought. Fertilizer rates and methods ofapplication developed for transplanted rice may haveto be re-examined for direct-seeded rice. Detailed

studies of root biology and physiology during thevarious transitions between oxic and anoxic soil con-ditions are needed to develop a rational basis formodifying existing fertilizer recommendations fordirect-seeded rice crops.

Cultivars

Breeding for nutrient efficiency and adaptation to thevariable growing environments of rainfed lowlandsmay be a cost-effective approach towards making amajor impact on the productivity of drought- andsubmergence-prone rainfed lowlands, provided traitsassociated with efficiency in that environment can beidentified. Ample evidence exists of variation innutrient efficiency among rice germplasm in low-fertility rainfed lowlands in Laos and Thailand(Fukai et al. 1999). However, selection of germ-plasm for such conditions is difficult (Cooper et al.1999), as is the management of water use andnutrient availability. Fortunately, nutrient efficiencyappears to be expressed whether plants are grownwith or without fertilizer application, thus simpli-fying the task of selecting because only one nutrientregime needs to be tested rather than two.

The results of field-screening studies for adapta-tion to low-fertility soils of the rainfed lowlands inLaos suggest that internal efficiency is a geneticallydetermined trait that, overall, is consistent acrossenvironments (Fukai et al. 1999). Low nutrient con-centration in the plant and higher nutrient allocationto grain were identified as potential selection criteriafor nutrient-use efficiency.

Soil physical properties

Shallow rooting depth is a major constraint toproductivity for rainfed lowland rice. Alleviating soilphysical constraints to root penetration may there-fore increase rice productivity by increasing accessto stored water and nutrients in the deeper layers ofthe soil profile. If roots can penetrate the shallowplough pan, then rice crops can extract significantamounts of N from below 20 cm (Ventura andWatanabe 1984). Kundu and Ladha (1999) reportedthat deep cultivation increases rice yields. In Korea,increasing cultivation depth from 14 to 19 cmincreased root depth from 27 to 36 cm and rice yieldfrom 4.5 to 8.1 t ha

–1

. Increasing cultivation depthfrom 15 to 40 cm on a soil with a hardpan at 15 cmincreased both N uptake and grain yield (Kundu etal. 1996). The scope for increasing cultivation depthwith draught animals is of course limited. In Cam-bodia, a pair of working animals can achieve a culti-vation depth of 7–10 cm (Rickman et al. 1997).However, increasing availability of tractors for pri-mary cultivation is making deeper cultivation pos-

162



sible on rainfed lowland soils with shallow hardpans.In 1996, for example, about 12% of primary cultiva-tion in Cambodia was by tractor (Rickman et al.1997).

In contrast to the above, studies at UbonRatchathani, North-East Thailand, suggested thatsubsoil compaction of deep sandy soils helps increasegrain yield (Trebuil et al. 1998). Compaction in the10–80 cm layer of soil was achieved by 10 passeswith a 12-t vibrating roller. On such deep sandy soils,percolation is so rapid that the soil drains veryquickly after rain. The primary benefit of compactingthis deep sandy soil was to decrease saturatedhydraulic conductivity from 38 to 12 cm day

–1

and toincrease the duration of surface flooding of soils from1 to 9 weeks. However, even on these deep sandysoils, the importance of deep root penetration wasstill evident. Deep cultivation to 20 cm followingcompaction increased yields relative to shallow culti-vation (7–10 cm) of the compacted profile.

However, apart from the practical consideration ofhow to achieve subsoil compaction at a reasonablecost to small farmers, the benefits were seasondependent. In a season where water supply wasadequate throughout, crops failed to respond to sub-soil compaction + conventional fertilizer rates,requiring higher nutrient supply to benefit fully fromthe improved water retention (Trebuil et al. 1998).The decrease in percolation rates with subsoil com-paction was insufficient to retain water in the soilprofile after the rice harvest, hence providing noopportunity for growing a post-dry-season rice cropon the residual moisture.

Soil water–nutrient interactions

Nutrient losses from the root zone and decreases inavailability of nutrients to rice roots are the two keyprocesses that need to be managed if fertilizer-useefficiency is to be increased in rainfed lowlands.

Nitrogen

Considerable scope exists for decreasing N losses.Nitrate N that accumulates during the dry season maybe substantially lost during early season rains, evenbefore the rice is transplanted (Kundu and Ladha1999). On deep sandy soils with high percolationrates, leaching losses are difficult to prevent with con-ventional practices. One strategy is to grow pre-ricecrops in the early rainy season to capture NO

3

−

Nand recycle it for the rice crop as organically boundN. The N mineralizes and is supplied as NH

4

−

N tothe rice (George et al. 1994). Pulse crops like mungbeans that mature in 65 days can be planted in Mayand harvested before the main rainy season in July–August. However, in many parts of Cambodia, crops

grown during the early rainy season wereunsuccessful because of temporary flooding of soils.

Sesbania, in contrast, tolerates waterlogging and,in 45–60 days, produces 1–18 t of aboveground freshbiomass per hectare (CIAP 1994). Incorporatingsesbania into the soil before transplanting riceincreased rice yield by 40% in on-farm trials inCambodia (Nesbitt 1997). The benefits are usuallyattributed to improved N supply for rice from N

2

fixation by sesbania. However, the capture andrecycling of accumulated soil NO

3

−

N may also be asignificant benefit of growing sesbania during theearly rainy season, because the amount of N

2

fixationis usually related to soil N supply. If the supply ishigh, then N

2

fixation is typically low and vice versa(George et al. 1994).

A second strategy, carried out in most ofCambodia, Laos and North-East Thailand, is to leavethe soils fallow after the rice harvest. Animals usuallygraze the rice stubble during the dry season. In addi-tion, volunteer weed and pasture plants may grow ifresidual soil water is sufficient, or in the early rainyseason. P.F. White (pers. commun., 1999) noted thatleguminous pasture species growing in the early rainyseason increased in biomass, using residual P fromfertilizer applied to rice. The N

2

fixed during the earlywet season may therefore be a significant spin-offbenefit to the farming system from P fertilizer appli-cation. Equally important may be the uptake byvolunteer pasture species and weeds of NO

3

-N thataccumulates in the soil during the dry season (Georgeet al. 1994). Further research is needed to improve themanagement of the annual pasture-fallow period toincrease nutrient availability to rice.

The third alternative for increasing the captureand recycling of NO

3

−

N accumulating in the soilafter the dry season is to direct sowing rice duringthe early rainy season.

Phosphorus

If P uptake is restricted during growth, particularly attransplanting, loss of soil-water saturation mayimpair growth and final yield (De Datta et al. 1990;Seng et al. 1999; Seng 2000), although the relativesensitivity of yield to P deficiency at different stagesof crop growth is poorly understood. Improvedunderstanding would permit better tactical decisionson correction during the growing season. Options forminimizing the impact of periods of loss of soil-water saturation are either to use cultivars that areefficient in P uptake and use and presumably wouldbe best able to cope with a temporary decline in Pavailability (Fukai et al. 1999); or to treat soil withstraw (Seng et al. 1999). Straw keeps the redoxpotential lower during the period of soil-water

163



saturation loss, thus apparently decreasing the extentof Fe

2+

oxidation and minimizing losses in P availa-bility due to reaction with Fe oxides. Other forms oforganic matter added to the soil at planting,including cow manure, or residues from pre-ricepulse crops or green manures like sesbania, can allhelp minimize losses of P during periods of soil-water saturation loss. The minimum amount oforganic matter required to make a difference is notknown, but Seng et al. (1999) had applied theequivalent of 5 t of straw per hectare.

Losses of P in leachate from lowland rice soilshave not been adequately quantified, although Choet al. (2000) estimate that 0.2 kg ha

–1

of P areleached. On sandy soils, however, recent findingssuggest higher rates of P leaching. Linquist et al.(2000) examined residual availability of P fertilizeron sandy lowland rice soils in Laos and found that48% to 85% of the P was not used in the first crop,probably because of significant leaching of P throughhigh percolation rates and low P sorption. Leachingof P has not usually been considered a significantissue for P nutrient management, but the prevalenceof sandy lowland rice soils in North-East Thailand,Cambodia and Laos suggest that P leaching needs tobe more thoroughly studied. The low residual effec-tiveness of P fertilizer due to P leaching suggests thatannual P rates need to be matched with expected cropdemand because any extra P would only exacerbate Plosses. Alternatively, where available, rock phosphatecould be used to minimize P leaching. In Cambodia,local rock phosphate is a potentially effective sourceof P for rainfed lowland rice (White et al. 1999) but,until the marketed product shows consistent quality,its use is not recommended (CIAP 1999).

Acidity

Aluminium toxicity has generally been ruled out as asignificant limiting factor for rice under floodedconditions. However, several possible consequencesof Al toxicity during temporary loss of soil-watersaturation may warrant further attention to acid soilsof rainfed lowlands (Willett and Intrawech 1988;Seng 2000). One consequence is to directly inhibitroot elongation, thus limiting plants’ capacity toaccess water stored deeper in the profile of a dryingsoil. A second consequence is, by limiting rootextension, the uptake of nutrients, especially P,would also be limited. Finally, soluble Al may reactwith P, thus decreasing the availability of soil P foruptake by roots.

Nursery fertilizer use

In Cambodia, farmers in the rainfed lowlands oftenapply fertilizer and manures to the seedling nurseryrather than to the main fields (Lando and Mak 1994;

Ros et al. 1998). On the low-fertility soils ofCambodia, fertilizer applied to both nursery andmain fields increased rice yield (Ros 1998; CIAP1998). Fertilizer applied to the nursery increasedseedling vigour, which generally increased sub-sequent rice yields by 5%–10%, regardless ofwhether the main field was treated with fertilizer ornot. Cow manure at 3 t ha

–1

and inorganic fertilizerat 50 kg N ha

–1

and 22 kg P ha

–1

were recommendedfor increasing seedling vigour (CIAP 1998).

Other low-cost strategies suggested by Ros et al.(1997, 2000) for increasing seedling vigour includeseed coating with crushed rock phosphate andincreasing seed P concentration in nursery-plantedseed. The strategy adopted by Cambodian farmers ofapplying most of the fertilizer to the nursery was anefficient use of nutrients because applications to themain field could not replace the need for nursery fer-tilizer to produce vigorous seedlings.

Another P management strategy in the nurseryphase, reported from India, is to dip seedlings in a Pslurry before transplanting. This localizes P directlyaround roots and, in one study, improved P-useefficiency by 50% (Katyal 1978)

Managing the variability of a soil-water regime

Unevenness in the soil surface is a significant sourceof variability in a field’s soil-water regime (CIAP1997). Elevation differences between 7 and 33 cmare not uncommon in farmers’ fields (CIAP 1997).These differences are sufficient to cause some partsof the field to experience intermittent loss of soil-water saturation during the growing season whileother parts of the field are continuously flooded.Variation in surface elevation affects decisions onwhen to transplant, seedling survival after trans-planting and weed control. In addition, such micro-variability aggravates the problem of efficientlymanaging nutrient supply and fertilizer applicationsto rainfed lowland rice. Laser levelling has beendeveloped in Cambodia, leading to increases in riceyields between 400 and 1000 kg ha

–1

in farmers’fields (CIAP 1996, 1997, 1998). While laserlevelling may not be an option for poor lowland ricefarmers, the research showed the benefits of levelfields, suggesting the need for greater attention tomanual levelling during land preparation.

Sustainability of Fertilizer Use

Cropping systems

Two major changes anticipated for the croppingsystems found in rainfed lowland ecosystems willhave a major bearing on the sustainability of fertilizeruse. The first is the increasing adoption of directsowing of rice across the rainfed lowland ecosystems

164



as the reduced availability of labour makes trans-planting a less attractive option. Direct-seeded rice isplanted earlier, under essentially upland conditions,after the early rainy season. Subsequently, soilsexperience a variable soil-water regime, with periodsof flooding interspersed with loss of soil-water satu-ration for intermittent intervals of variable duration.The variation in the soil-water regime under a direct-sowing system of crop establishment will require are-examination of optimal methods and timing of fer-tilizer use, compared with establishment by trans-planting. Secondly, the increased prevalence of dry-season cropping through direct sowing increases theintensity of crop production on these soils and, hence,the overall demand for nutrients and stored soilwater. In areas other than rainfed lowlands, increasedaccess to water for supplementary irrigation in thedry season permits the production of either a dry-season rice crop, or pulse or vegetable crops.

For rainfed lowland rice crops, the accumulation ofnutrients, especially N, from the mineralization oforganic matter during the long dry season representsa significant nutrient resource that needs to bemanaged. In areas such as Cambodia and North-EastThailand, where fields mostly lie fallow during the dryseason, significant losses of NO

3

−

N may occurduring the early rainy season, although no quantitativedata exist on the likely rates. These soils are relativelylow in organic matter, and mineralization may not bethat high, compared with soils in Ilocos Norte (Phil-ippines), where high rates of NO

3

−

N accumulate inthe dry season. Furthermore, weeds often grow upduring the early wet season before tillage, and areeither grazed or incorporated during tillage. Hence, asignificant proportion of NO

3

−

N may already berecycled in these rainfed lowland rice ecosystems. Thesame may be true for SO

4

−

S and K on sandy soils,but few data are available on this point.

In this environment, the most promising tech-nologies for better using plant-available nutrients thataccumulate during the dry season are planting a pre-rice crop or direct sowing rice early. In both cases,the aim is to encourage plant uptake of these nutrientsto prevent their loss. In the case of a pre-rice crop, themineralization of crop residues releases nutrients forrice at a time when the root system of the trans-planted rice is capable of absorbing them. In the caseof early direct-seeded rice, the nutrients absorbed areused directly for growth. In either case, appropriateadjustments in the rates of fertilizer applied at sowingand later will need to be worked out.

Nutrient budgets

Increasing concerns about the sustainability of thefertility status of rainfed lowland rice soils has

prompted several studies on nutrient budgets. On anational scale, for example, Lefroy and Konboon(1998) estimated N–P–K budgets for Thailand andconcluded that much more K was being exportedfrom rice fields in harvested grain than was beingreplaced by fertilizer. In contrast, a positive balancewas found for N and P, making the simple assump-tion that no other losses of nutrients occurred, exceptthrough harvested grain. Similarly, Lefroy andKonboon (1998) estimated nutrient budgets onregional and farm scales in North-East Thailand andconcluded that N and P were, on the whole,positively balanced for rice production but that Kwas being depleted under the current croppingregime. Negative K balances were also reported forsandy soils in central Java (Indonesia; Wihardjaka etal. 1998). The removal of stubble from rice fieldssignificantly depletes soil K. Few reports of Sbalances are available for rainfed lowlands. Low-Ssoils are relatively widespread in North-EastThailand, Cambodia and Laos and so depletion ofsoil S in harvested grain may significantly affect theproductivity of rainfed lowland rice. However, S inrain may significantly offset losses through har-vested grain (Lefroy and Hussain 1991).

The calculation of nutrient budgets as a tool formanaging nutrient supply for rainfed lowland ricehas considerable promise because of its relativesimplicity, compared with alternative approachessuch as soil and plant testing, and simulation model-ling. However, Lefroy and Konboon (1998) pointedout that many of the assumptions underlying calcula-tion of nutrient budgets need more rigorous supportfrom field research.

Environmental consequences of fertilizer use

The application of fertilizer to rice has potentialunintended consequences that are of increasing con-cern in many parts of the world (e.g. Mishama et al.1999; Xing and Zhu 2000). Negative effects on thequality of surface and groundwater comprise themost common environmental impact (Shrestha andLadha 1999). In rainfed lowlands with access tosupplementary irrigation, dry-season cropping isbecoming more common, especially where popula-tion density is high and increased output per unitarea can be most readily achieved by dry-seasoncropping, using surplus labour (Pandey 1998). Theenvironmental impact of these systems is underexamination and may be a possible precursor ofmore widespread concern for agrochemical use inrainfed lowlands.

At present, fertilizer rates in rainfed lowlands aregenerally still low, leading Crosson (1995) to sug-gest that the negative environmental impact of ferti-

165



lizer use on rice production is probably minimal.However, because nutrient deficiencies are prevalentand farmers are being advised to increase their ferti-lizer use and application rates, the possible futureconsequences of implementing these recommenda-tions need to be considered. Villagers usually accessstream water or shallow groundwater for domesticcleaning, cooking and drinking. Degraded quality ofthese water resources would be a matter of consider-able concern for public health. In addition, artificialand natural wetlands in rainfed lowlands are oftensignificant food resources for villagers. Loss ofwater quality in these wetlands likewise needs to beguarded against.

Because these problems generally do not exist yetin most rainfed lowland rice environments, now is anopportune time to set in place strategies to prevent itbecoming a concern. Periodic monitoring of waterquality and identifying areas in catchment basins thatcontribute most to nutrient enrichment of waterbodies should be implemented.

Nitrogen accumulation in surface and groundwaterhas been reported as a consequence of fertilizer appli-cations to irrigated-rice crops. Intensification of pro-duction in rainfed lowlands by growing dry-seasonvegetables, using supplementary irrigation, is alsocausing similar high losses of NO

3

-N throughleaching into groundwater (Shrestra and Ladha1999). In Ilocos Norte (Philippines), annual losses ofup to 550 kg of N ha

–1

were reported. In 50% of wellssurveyed on farms practising the rice–sweet pepperrotation, nitrate concentrations exceeded the WorldHealth Organization’s (WHO) limits for drinkingwater. The high rates of N fertilizer use in theseintensive production systems are driven by the higheconomic returns from dry-season vegetables and arenot currently typical of the drought-prone and sub-mergence-prone rainfed lowlands. However, even inCambodia, installation of wells and pumps for dry-season cropping is spreading and may eventually leadto intensive production systems like those of IlocosNorte.

In the irrigated-rice fields of central Korea,annual P losses in run-off were estimated to be4.5 kg P ha

–1

in water and 0.9 kg P ha

–1

in sedi-ments (Cho et al. 2000). Increasing P concentrationin surface and groundwater also has potential envi-ronmental consequences. Most aquatic ecosystemsare P limited, so that increases in P concentration inrun-off water and groundwater can greatly increasethe biological productivity of these systems. Eutro-phied wetland systems may generate algal bloomsthat would harm fisheries by impeding movement ofboats and killing fish—some algal species in bloomsare potentially toxic.

Adoption of Improved Nutrient Management Strategies

Traditionally, rainfed lowland farmers use little fer-tilizer, probably to avert risks. Pandey (1998), forexample, reported that rates of N–P–K fertilizer useacross Asian countries correlate with percentages ofrice crops irrigated. In Cambodia, farmers in irrigatedareas have adopted modern methods of rice produc-tion to a much greater extent than those in rainfedlowlands. Surveys by Jahn et al. (1997) showed thatonly 1.2% of rainfed rice farmers grow IRRI varieties,representing only 0.9% of the rice-growing area. Interms of inorganic fertilizer use, only 27% of rainfedrice farmers use inorganic fertilizers, compared with70% of dry-season rice farmers.

As discussed above, rainfed lowlands have severalsignificant characteristics that distinguish them fromirrigated rice-growing areas where significant gains intechnology adoption and productivity have alreadybeen achieved. These are a high degree of spatial andtemporal variability both in terms of soil type andwater availability (Zeigler and Puckridge 1995); thesoils often have poorer chemical and physicalproperties than soils in irrigated systems; and farmersin rainfed lowlands also have fewer resources forcapital expenditure and limited access to credit thando farmers in irrigated systems (Zeigler and Puckridge1995) and therefore their ability to invest in innovativetechnologies is limited. The range of options forimproved management are consequently few; with agreater chance of crop failure. When crops fail,rainfed-rice farmers have fewer options to generatesupplementary food or income. Risk avoidance, there-fore, occupies a more important position in decisionmaking for rainfed-rice farmers than it does forirrigated-rice farmers (Zeigler and Puckridge 1995).Once again, the range of technologies that rainfedlowland farmers are willing to adopt is furtherrestricted.

Traditionally, fertilizer recommendations havebeen based on average responses for particular soilsand ecosystems (Dobermann and White 1999). Thehigh degree of spatial and temporal variation of soilconditions in the rainfed lowlands raises two ques-tions: firstly, how useful are blanket recommenda-tions and generalized advice on fertilizer and,secondly, how can site-specific nutrient managementstrategies be developed (Dobermann and White1999). Current strategies need to be better targetedthan were past strategies to specific environmentsthat have relatively small recommendation domains(Pingali et al. 1998). Because many such recommen-dation domains exist in the variable rainfed low-lands, the cost of developing these strategies, using

166



normal empirical experimentation for each particularenvironment, is expensive and time consuming.

A major reason for this is the difficulty of identi-fying appropriate technologies that can feasibly beapplied by a farmer, given his or her particular set ofconstraints and expectations. Predicting the magni-tude of a response to management inputs for anygiven situation is therefore difficult unless priorknowledge exists for each individual circumstance. Inan effort to solve this problem, emphasis in much ofAsia is currently placed on improved characterizationof the environment and development of mechanisticand empirical simulation models that will predict cropperformance in a given environment (Zeigler andPuckridge 1995).

While simulation models can now make good pre-dictions of crop yield (e.g. Alagarswamy and Virmani1996), the process depends on data for calibrating themodels. Such data are frequently not available in themany local environments of the rainfed lowlandsbecause of limited research infrastructure and knowl-edge base. Indeed, even the simplest data sets, suchas daily rainfall, that are needed for simulationmodels such as APSIM (McCown et al. 1996) areunavailable for many Cambodian environments.Furthermore, the crop simulation approach takes noaccount of the difficult-to-quantify farmer circum-stances and other such constraints that significantlydetermine the adoption of improved technologies.

Management strategies, particularly for nutrientsupply by fertilizer applications, must be flexible andcapable of modification, depending on the progress ofthe season and/or the outcome of the previous season.Farmers need to maximize the advantage gained froma better-than-average season while also being able tominimize the risk and losses associated with a poorseason. Farmers need to be able to make informeddecisions about changing their strategies to suit theseasonal progress and other changed circumstances.

Farmer participatory research, which follows abottom-up approach and aims at working withfarmers to identify their problems and solutions, cangive farmers the necessary knowledge to dynami-cally manage their system. Experience, nevertheless,has shown that this type of research has difficultymoving beyond the diagnostic and design stages andfew practical solutions have been developed that canbe applied outside the project area (Pillot 1988).

Improved management strategies must alsoaccommodate the farmer’s aims, which, because ofeconomic circumstances or personal preferences,may favour management strategies or fertilizer typesthat do not necessarily maximize yields or economicreturns. For example, some Cambodian farmers arereluctant to apply inorganic fertilizer to traditionalvarieties grown for their own consumption because

they believe it degrades the flavour of cooked rice.Similarly, in areas of low population density, farmerswith large farms (2–4 ha) place little emphasis onhigher yield but stress yield stability and decreasedlabour requirements (Pandey 1998). Alternatively,farmers in remote locations may not be able to pur-chase fertilizer at the recommended times for appli-cation but need to know the likely benefits fromapplication at other times in the growing season.

These factors have been well studied by socialscientists and economists. And the modelling andintegration of social, economic and biophysical datafor specific systems has been possible, but again,application to the varied environment of the rainfedlowland rice-growing ecosystem has been difficult.

Finally and importantly, there is a need for a formaland organized mechanism whereby knowledge andexperience about soil fertility and fertilizer responsecan be stored and shared at all levels within the agri-cultural sector. The system must allow for incre-mental improvements of the technologies as lessonsare learnt, and for the entry of new technologies asthey are developed. Drawing on this background,researchers are developing a model of rice farming inCambodia that integrates information from a range ofsources, including farmer experience (Bell et al.2000). The model developed will predict the likelyoutcomes of given actions. Advisers can therefore usethe model to identify the optimal technology for theparticular circumstances of a farmer. By bettertargeting advice, the model will help facilitate theimproved adoption of technology by farmers. Thesystem will also facilitate the storage and transfer offarmers’ knowledge between localities. Finally, themodel’s outputs can be represented spatially in ageographic information system. This will contributeto improved strategic planning on a provincial anddistrict basis.

Conclusions

The rainfed lowlands comprise a complex environ-ment for managing soil fertility. Contrary to previousviews that drought was the major limiting factor forrice yield, recent research and simulation modellingsuggest that nutrients and nutrient -water limitationsare more significant. Breeding for nutrient efficiencyand adaptation to these variable growing environ-ments has been suggested as the most cost-effectiveapproach towards making a major impact on theproductivity of the drought- and submergence-pronerainfed lowlands. However, a deeper understandingof soil chemical changes due to the various tran-sitions between oxic and anoxic conditions is neededto discover mechanisms of cultivar adaptation.Similarly, to better understand nutrient and water

167



uptake by rice, an improved knowledge of root func-tion, adaptation and turnover under the variablewater regimes in rainfed lowlands is needed.

Rice production in rainfed lowlands stands at thethreshold of major changes in nutrient management.Increased fertilizer use is expected. This has potentialto generate benefits and harm. Elsewhere in theworld, fertilizer practice has been developed toachieve high production but without considering thepotential negative environmental consequences untiltoo late. The opportunity exists to learn from theseexperiences by taking a more environmentallyresponsible approach to fertilizer use and nutrientmanagement in the rainfed lowlands.

References

Ahmed, H.U., Ali, M.L., Kabir, M.A. and Miah, N.M.1996. Varietal characteristics and soil management toreduce drought stress. In: Singh, V.P., Singh, R.K.,Singh, B.B. and Zeigler, R.S. ed. Physiology of StressTolerance in Rice, Proceedings of the International Con-ference on Stress Physiology of Rice, 25 Feb–5 March1994, Lucknow, India. Manila, Philippines, InternationalRice Research Institute (IRRI), 150–167.

Alagarswamy, G. and Virmani, S.M. 1996. Risk analysis ofrainfed sorghum production at various levels of nitrogenfertiliser rate with the CERES sorghum crop simulationmodel. In: Ito, O., Johansen, C., Adu-Gyamfi, J.J.,Katayama, K., Kumar Rao, J.V.D.K. and Rego, T.J. ed.Dynamics of Roots and Nitrogen Cropping Systems ofthe Semi-Arid Tropics. Tsukuba, Ibaraki, Japan, JapanInternational Research Centre for Agricultural Sciences,603–615.

Bell, R.W., Ros, C., Cook, S. and White, P.F. 2000. Asystem to reduce risk in the adoption of new rice pro-duction technologies in Cambodia. In: IRRC 2000 RiceResearch for Food Security and Poverty Alleviation,31 March–3 April 2000, Abstracts. Los Baños, Philip-pines, International Rice Research Institute (IRRI), 210.

Brandon, A.M. and Mikkelsen, D.S. 1979. Phosphorustransformations in alternately flooded California soils. I.Cause of plant phosphorus deficiency in rice rotationcrops and correction methods. Soil Science Society ofAmerica Journal, 43, 989–994.

CIAP (Cambodia–IRRI–Australia Project). 1994. Farmingsystem and nutrient management. In: Annual Report1993. Phnom Penh, Cambodia, 128–142.

CIAP (Cambodia–IRRI–Australia Project). 1995. Inte-grated nutrient management and farming systems. In:Annual Report 1994. Phnom Penh, Cambodia, 117–138.

CIAP (Cambodia–IRRI–Australia Project). 1997. Agricul-tural engineering. In: Annual Report 1996. Phnom Penh,Cambodia, 138–151.



Cho, J.Y., Han, K.H. and Choi, J.K. 2000. Balance ofnitrogen and phosphorus in a paddy field of centralKorea. Soil Science and Plant Nutrition, 46, 343–354.

Colmer, T.D., Gibberd, M.R., Wiengweera, A. and Tinh,T.K. 1998. The barrier to radial oxygen loss from roots ofrice (

Oryza sativa

L.) is induced by growth in stagnantsolution. Journal of Experimental Botany, 49, 1431–1436.

Cooper, M., Fukai, S. and Wade, L.J. 1999. How canbreeding contribute to more productive and sustainablerainfed rice systems? Field Crops Research, 64, 199–209.

Crosson, P. 1995. Natural resource and environmental con-sequences of rice production. In: Fragile Lives in FragileEcosystems, Proceedings of the International RiceResearch Conference, 13–17 Feb 1995. Los Baños, Phil-ippines, International Rice Research Institute (IRRI),83–100.

De Datta, S.K., Biswas, T.K. and Charoenchamrapcheep, C.1990. Phosphorus requirements and management for low-land rice. In: Phosphorus Requirements for SustainableAgriculture in Asia and Oceania. Los Baños, Philippines,International Rice Research Institute (IRRI), 307–324.

Dobermann, A. and White, P.J. 1999. Strategies for nutrientmanagement in irrigated and rainfed rice systems.Nutrient Cycling in Agroecosystems, 53, 1–18.

Fukai, S., Inthapanya, P., Blamey, F.P.C. and Khunthasavon, S.1999. Genotypic variation in rice grown in low fertility soilsand drought-prone, rainfed lowland environments. FieldCrops Research 64, 121–130.

Fukai, S., Rajatsasereekul, S., Boonjung, H. and Skulkhu, E.1995. Simulation modelling to quantify the effect ofdrought for rainfed lowland rice in Northeast Thailand.In: Fragile Lives in Fragile Ecosystems, Proceedings ofthe International Rice Research Conference, 13–17 Feb1995. Los Baños, Philippines, International RiceResearch Institute (IRRI), 657–674.

Garrity, D.P. and Vejpas, C. 1986. Percolation barriersincrease and stabilise rainfed lowland rice yields on well-drained soil. In: Murty, V.V.N. and Koga, K. ed. Soil andWater Engineering for Paddy Management. Bangkok,Thailand, Asian Institute of Technology, 413–421.

George, T., Ladha, J.K., Garrity, D.P. and Buresh, R.J.1994. Legumes as nitrate catch crops during dry-to-wettransition in lowland rice cropping systems. Soil ScienceSociety America Journal, 86, 267–273.

Jahn, J.C., Pheng, S., Kiev, B. and Pol, C. 1997. Farmers’pest management practices in Cambodian lowland rice.In: Heong, K.L. ed. Management Practices of RiceFarmers in Asia. Manila, Philippines, International RiceResearch Institute (IRRI).

Katyal, J.C. 1978. Management of phosphorus in lowlandrice. Phosphorus in Agriculture, 73, 21–34.

Kawaguchi, K. and Kyuma, K. 1977. Paddy Soils inTropical Asia: Their Material, Nature and Fertility.Honolulu, HI, The University of Hawaii Press.

Kirk, G.J.D. and Du, L.V. 1997. Changes in rice root archi-tecture, porosity, and oxygen and proton release underphosphorus deficiency. New Phytologist, 135, 191–200.

Kirk, G.J.D., Solivas, J.L. and Begg, C.B.M. 1994. The riceroot–soil interface. In: Kirk, G.J.D. ed. Rice Roots:Nutrients and Water Use. Manila, Philippines, Inter-national Rice Research Institute (IRRI), 1–10.

168



Kundu, D.K. and Ladha, J.K. 1999. Sustaining productivityof lowland rice soils: issues and options related to Navailability. Nutrient Cycling in Agroecosystems, 53,19–33.

Kundu, D.K., Ladha, J.K. and Lapitan de Guzman, E. 1996.Tillage depth influence on soil nitrogen distribution andavailability in a rice lowland. Soil Science Society ofAmerica Journal, 60, 1153–1159.

Lando, R.P. and Mak, S. 1994. Rainfed Lowland Rice: ABaseline Study. Research Paper Series No. 152. LosBaños, Philippines, International Rice Research Institute(IRRI).

Lefroy, R.D.B. and Hussain, A. 1991. Sulfur in rainfall andits relevance to Malaysian agriculture. In: Khanaki,Y.M., Syed Omar, S.R. and Shamsuddin, J. ed. Develop-ments in Soil Research in Malaysia. Kuala Lumpur,Malaysia, Malaysian Society of Soil Science, 14–22.

Lefroy, R.D.B. and Konboon, Y. 1998. Studying nutrientflows to assess sustainability and identify areas ofnutrient depletion and imbalance: an example for rainfedrice systems in Northeast Thailand. In: Ladha, J.K.,Wade, L.J., Dobermann, A., Reichardt, W., Kirk, G.J.D.and Piggin, C. ed. Rainfed Lowland Rice: Advances inNutrient Management Research, Proceedings of theInternational Workshop on Nutrient ManagementResearch in Rainfed Lowlands, 12–15 October 1998,Ubon Ratchatani, Thailand. Los Baños, Philippines,International Rice Research Institute (IRRI), 77–93.

Linquist, B., Sengxua, P. and Whitbread, A. 2000. Residualphosphate availability in lowland rice ecosystems ofLaos. In: IRRC 2000 Rice Research for Food Securityand Poverty Alleviation, 31 March–3 April 2000,Abstracts. Los Baños, Philippines, International RiceResearch Institute (IRRI), 180.

Linquist, B., Sengxua, P., Whitbread, A., Schiller, J. andLathvilayvong, P. 1998. Evaluating nutrient deficienciesand management strategies for lowland rice in Lao PDR.In: Ladha, J.K., Wade, L.J., Dobermann, A., Reichardt,W., Kirk, G.J.D., and Piggin, C. ed. Rainfed LowlandRice: Advances in Nutrient Management Research, Pro-ceedings of the International Workshop on NutrientManagement Research in Rainfed Lowlands, 12–15October 1998, Ubon Ratchatani, Thailand. Los Baños,Philippines, International Rice Research Institute (IRRI),59–73.

McCowan, R.L., Hammer, G.L., Hargreaves, J.N.G.,Holzworth, D.P. and Freebairn, D.M. 1996. APSIM: anovel software system for model development, modeltesting, and simulation in agricultural systems research.Agricultural Systems, 50, 255–271.

Mishama, S., Matsumoto, N. and Oda, K. 1999. Nitrogenflow associated with agricultural practices and environ-mental risk in Japan. Soil Science and Plant Nutrition,45, 881–889.

Mitsuchi, M., Wichaidit, P. and Jeungnijnirund, S. 1986.Outline of Soils of the Northeast Plateau, Thailand. TheirCharacteristics and Constraints. Technical Paper No. 1.Khon Kaen, Thailand, Agricultural DevelopmentResearch Centre in Northeast Khon Kaen Province, 79 p.

Morita, S. and Abe, J. 1996. Development of root systemsin wheat and rice. In: Ito, O., Johansen, C., Adu-Gyamfi, J.J., Katayama, K., Kumar Rao, J.V.D.K. and

Rego, T.J. ed. Dynamics of Roots and Nitrogen in Crop-ping Systems of the Semi-Arid Tropics. Tsukuba, Japan,Japan International Research Centre for AgriculturalSciences, 199–209.

Nesbitt, H.J. 1997. Rice Production in Cambodia. Manila,Philippines, International Rice Research Institute (IRRI),106 p.

Pairoj, P., Kimura, K. and Takai, Y. 1996. Long-term effectsof fertilization on paddy soil fertility for sustainable riceproduction. In: Attanandana, T., Kheoruenromne, I.,Pongsakul, P. and Vearasilp, T. ed. Proceedings of Inter-national Conference on Maximising Sustainable RiceYields through Improved Soil and EnvironmentalManagement, 7–11 November 1996. Bangkok, Thailand,Funny Publishing, 799–804.

Pandey, S. 1998. Nutrient management technologies forrainfed rice in tomorrow’s Asia: economic andinstitutional considerations. In: Ladha, J.K., Wade, L.J.,Dobermann, A., Reichardt, W., Kirk, G.J.D. and Piggin,C. ed. Rainfed Lowland Rice: Advances in NutrientManagement Research. Proceedings of the InternationalWorkshop on Nutrient Management Research in RainfedLowlands, 12–15 October 1998, Ubon Ratchatani,Thailand. Los Baños, Philippines, International RiceResearch Institute (IRRI), 3–28.

Patrick, W.H., Jr. and Wyatt, R. 1964. Soil nitrogen loss asa result of alternate submergence and drying. SoilScience Society of America Proceedings, 28, 647–653.

Pheav, S., White, P.F. and Oberthur, T. 1996. Managementconstraints of the major rice-growing soils of Cambodia.In: Attanandana, T., Kheoruenromne, I., Pongsakul, P.and Vearasilp, T. ed. Proceedings of InternationalConference on Maximising Sustainable Rice Yieldsthrough Improved Soil and Environmental Management,7–11 November 1996, Khon Kaen, Thailand. Bangkok,Thailand, Funny Publishing.

Pillot, D. 1988. Francophone and anglophone farmingsystems research: similarities and differences. In:Farming Systems Research and Development in Thai-land. Haatyai, Thailand, Prince of Songkla University, 3–25.

Pingali, P.L., Hossain, M., Pandey, S. and Price, L.L. 1998.Economics of nutrient management in Asian ricesystems: towards increasing knowledge intensity. FieldCrops Research, 56, 157–176.

Ponnamperuma, F.N. 1972. The chemistry of submergedsoils. Advances in Agronomy, 24, 29–96.

Ragland, J. and Boonpuckdee, L. 1987. Fertiliser responsesin Northeast Thailand. 1. Literature review and rationale.Thai Journal of Soils and Fertilisers, 9, 65–79.

Rickman, J.F., Piseth, M. and Sothy, O. 1997. Farmmechanization. In: Nesbitt, H.J. ed. Rice Production inCambodia. Phnom Penh, Cambodia, Cambodia–IRRI–Australia Project.

Ros, C. 1998. Nutrient Management of Rice Seedlings inthe Rainfed Lowlands of Cambodia. PhD dissertation.School of Environmental Science, Murdoch University,Perth, WA, Australia.

Ros, C., Bell, R.W. and White, P.J. 1997. Effect of seedphosphorus and soil phosphorus applications on earlygrowth of rice (

Oryza sativa

L.) cv. IR66. Soil Scienceand Plant Nutrition, 43, 499–509.

169



Ros, C., Bell, R.W. and White, P.F. 2000. Phosphorus seedcoating and soaking for improving seedling growth of

Oryza sativa (rice) cv. IR66. Seed Science and Tech-nology, 28, 201–211.

Ros, C., White P.F. and Bell, R.W. 1998. Field survey onnursery and mainfield fertilizer management. CambodianJournal of Agriculture, 1, 22–33.

Sah, R.N. and Mikkelsen, D.S. 1986. Transformations ofinorganic phosphorus during the flooding and drainingcycles of soil. Soil Science Society of America Journal,50, 62–67.

Samson, B.K. and Wade, L.J. 1998. Soil physical con-straints affecting root growth, water extraction, andnutrient uptake in rainfed lowland rice. In: Ladha, J.K.,Wade, L.J., Dobermann, A., Reichardt, W., Kirk, G.J.D.and Piggin, C. ed. Rainfed Lowland Rice: Advances inNutrient Management Research, Proceedings of theInternational Workshop on Nutrient ManagementResearch in Rainfed Lowlands, 12–15 October 1998,Ubon Ratchathani, Thailand. Los Baños, Philippines,International Rice Research Institute (IRRI), 231–244.

Seng, V. 2000. Edaphic factors limiting lowland rainfedrice production in soils in southeast Cambodia. PhD dis-sertation. School of Environmental Science, MurdochUniversity, Perth, WA, Australia.

Seng, V., Bell, R.W., Nesbitt, H.J. and Willett, I.R. 1996.Response of rainfed rice to inorganic and organicfertilisers in southeast Cambodia. In: Attanandana, T.,Kheoruenromne, I., Pongsakul, P. and Vearasilp, T. ed.Proceedings of International Conference on MaximisingSustainable Rice Yields through Improved Soil andEnvironmental Management, 7–11 November 1996.Bangkok, Thailand, Funny Publishing, 99–112.

Seng, V., Bell, R.W., Willett, I.R. and Nesbitt, H.J. 1999.Phosphorus nutrition of rice in relation to flooding andtemporary loss of saturation in two lowland soils insouth-east Cambodia. Plant and Soil, 207, 121–132.

Sharma, P.K. 1992. In situ water conservation in sandysoils for rainfed lowland rice. Rice Research Journal(Rice Research Institute, Bangkok), 1, 59–64.

Sharma, P.K., Ingram, K.T. and Harnpitchitvitaya, D. 1995.Subsoil compaction to improve water use efficiency andyields of rainfed lowland rice in coarse textured soils.Soil Tillage Research, 36, 33–44.

Sharma, P.K., Pantuwan, G., Ingram, K.T. and De Datta,S.K. 1994. Rainfed lowland rice roots: soil and hydro-logical effects. In: Kirk, G.J.D. ed. Rice Roots: Nutrientand Water Use. Manila, Philippines, International RiceResearch Institute (IRRI), 55–66.

Shrestra, R.K. and Ladha, J.K. 1999. Nitrogen dynamics inintensive rainfed lowland rice-based cropping systemsand strategies to improve efficiency. In: Ladha, J.K.,Wade, L.J., Dobermann, A., Reichardt, W., Kirk, G.J.D.and Piggin, C. ed. Rainfed Lowland Rice: Advances inNutrient Management Research. Proceedings of theInternational Workshop on Nutrient Management

Research in Rainfed Lowlands, 12–15 October 1998,Ubon Ratchathani, Thailand. Los Baños, Philippines,International Rice Research Institute (IRRI), 95–108.

Trebuil, G., Harnpitchitvitaya, D., Tuong, T.P., Pantuwan,G. and Wade, L.J. 1998. Improved water conservationand nutrient–use efficiency via subsoil compaction andmineral fertilisation. In: Ladha, J.K., Wade, L.J.,Dobermann, A., Reichardt, W., Kirk, G.J.D. and Piggin,C. ed. Rainfed lowland Rice: Advances in NutrientManagement Research, Proceedings of the InternationalWorkshop on Nutrient Management Research in RainfedLowlands, 12–15 October 1998, Ubon Ratchatani,Thailand. Los Baños, Philippines, International RiceResearch Institute (IRRI), 3–28.

Ventura, W. and Watanabe, I. 1984. Dynamics of nitrogenavailability in lowland rice soils: the role of soil belowplow layer and effect of moisture regimes. PhilippinesJournal of Crop Science, 9, 812–821.

Wade, L.J., Fukai, S., Samson, B.K., Ali, A. and Mazid,M.A. 1999. Rainfed lowland rice: physical environ-ment and cultivar requirements. Field Crops Research,64, 3–12.

White, P.F., Oberthur, T. and Pheav, S. 1997. The SoilsUsed for Rice Production in Cambodia, A Manual fortheir Recognition and Management. Manila, Philippines,International Rice Research Institute (IRRI), 71 p.

White, P.F., Nesbitt, H.J., Ros, C., Seng, V. and Lor, B.1999. Local rock phosphate deposits are a good sourceof phosphorus fertilizer for rice production in Cambodia.Soil Science and Plant Nutrition, 45, 51–63.

Wihardjaka, A., Kirk, G.J.D., Abdulrachman, S. andMarmaril, C.P. 1998. Potassium balances in rainfedlowland rice on a light-textured soil. In: Ladha, J.K.,Wade, L.J., Dobermann, A., Reichardt, W., Kirk, G.J.D.and Piggin, C. ed. Rainfed lowland Rice: Advances inNutrient Management Research, Proceedings of theInternational Workshop on Nutrient ManagementResearch in Rainfed Lowlands, 12–15 October 1998,Ubon Ratchatani, Thailand. Los Baños, Philippines,International Rice Research Institute (IRRI), 127–137.

Willett, I.R. and Higgins, M.L. 1980. Phosphate sorptionand extractable iron during irrigated rice-upland croprotations. Australian Journal of Experimental Agricultureand Animal Husbandry, 20, 346–353.

Willett, I.R. and Intrawech, A. 1988. Preliminary Studies ofChemical Dynamics of Sandy Paddy Soils of Tung KulaRonghai, Northeast Thailand. Divisional Report No. 95.Melbourne, Australia, CSIRO.

Xing, G.X. and Zhu, Z.L. 2000. An assessment of N lossfrom agricultural fields to the environment in China.Nutrient Cycling in Agroecosystems, 57, 67–73.

Zeigler, R.S. and Puckridge, D.W. 1995. Improvingsustainable productivity in rice-based rainfed lowlandsystems of South and South-East Asia. GeoJournal, 35,307–324.

170



Nutrient Requirements of Rainfed Lowland Ricein Cambodia

Vang Seng

*

, Chhay Ros

1

, R.W. Bell

2

, P.F. White

3

and Hin Sarith

1

Abstract

In most rainfed lowlands of Cambodia, soils used for rice cultivation are low in availablenitrogen (N), phosphorus (P) and potassium (K), and have low organic matter content and lowcation-exchange capacity. Over the last 4 years, areas cultivated to rice have increased substantiallyfrom about 1.9 to 2.1 million hectares, of which rainfed lowlands comprise about 88%. The averagerice yield increased from about 1.2 to 1.8 t ha

–1

, even though nutrient deficiency remains a seriousconstraint for lowland rice production. The paper reviews current understanding of the nutrientrequirements of rainfed lowland rice in Cambodia. Field and greenhouse trials have classified wide-spread N and P responses and, on sandy soils, K and sulfur responses. Recommended fertilizer rates(in kg ha

–1

) for rice vary for the different nutrients: for N, from 20 to 120, for P, from 4 to 15, andfor K, from 0 to 33. Recommendations are made for each soil type identified in the CambodianAgronomic Soil Classification system. The need for further work on nutrient requirements of riceand other agricultural crops is also discussed.

T

HE

total area cultivated to rice in Cambodia has sub-stantially increased from about 1.9 million hectares in1995 to 2.1 million ha in 1999 (MAFF 1999). Wet-season, rainfed lowland rice covers about 88% of thetotal rice area. Demand for fertilizers for rice cultiva-tion has gradually increased. For example, in 1994,total consumption of fertilizers was 80 000 t,increasing to 87 000 t by 1995, indicating a 9% risein demand (CNP and SCI 1996). The increase wasprimarily attributed to an increase in cultivated areasand higher application rates on improved ricevarieties. Despite increased fertilizer demand and use

of improved rice varieties, the national average riceyield was only 1.5 t ha

–1

(Nesbitt 1997), which is stilllow, compared with 2–3 t ha

–1

obtained by other rice-producing countries in South-East Asia (Pandey1997). Most lowland soils of potential use for ricecultivation in Cambodia are low in available nitrogen(N), phosphorus (P), potassium (K), and have loworganic matter contents and low cation-exchangecapacity (CEC) (White et al. 1997a). Hence, nutrientdeficiencies represent a major constraint to rice pro-duction at present.

Common problems of rainfed lowland rice culti-vation in the diverse soil types of Cambodia includevaried soil-water regimes, low nutrient availabilityand the interactions between these two factors(Fischer 1998; Wade et al. 1998). For example, theavailability of N and P in soil varies strongly inrelation to soil-water content (Kirk et al. 1990; DeDatta 1995). This means that the complex problemsin the rainfed lowlands cannot be solved withoutbetter understanding site-specific nutrient manage-ment. In response to this need, the CambodianAgronomic Soil Classification (CASC) system was

1

Soil and Water Research Program, Cambodian Agricul-tural Research and Development Institute (CARDI), PhnomPenh, Cambodia

2

School of Environmental Sciences, Murdoch University,Perth, Western Australia

3

Pulses Programme, Agriculture Western Australia, Perth,Western Australia*Corresponding author: Soil and Water Research Program,Cambodian Agricultural Research and DevelopmentInstitute (CARDI), Phnom Penh, Cambodia.E-mail: [email protected]

KEYWORDS:

Rainfed lowlands, Soil types, Fertilizer recommendations, Nutrient requirements, Nitrogen,Phosphorus, Potassium, Rice

171



developed to provide better understanding of soiltypes. It forms the basis on which fertilizer rates andstrategies for rainfed lowland rice production arerecommended (White et al. 2000).

This paper reviews the work on nutrient researchand management for rainfed lowland rice-basedfarming in Cambodia. Selected examples are pre-sented to illustrate key elements of the integratednutrient management strategy being developed inCambodia. Some sections of this paper are sum-marized from White et al. (1998).

Soil Classification

The CASC system is used in soil and nutrientmanagement in Cambodia to infer information aboutthe soils’ properties by using easily observablesurrogate characters such as soil colour and texture,and stone contents, thus allowing management to betailored accordingly. The CASC has also been usedon a broad scale for applied agronomic research pro-grams in Cambodia (principally on-farm trials) but,where possible, soil chemical analysis has been usedin specific research experiments.

Discussion in this paper is based on the soilgroups as described in the CASC system (White etal. 1997b). Chemical properties of the soil groups aregiven in Table 1, but where available, soil propertiesfrom specific experiments are also provided in thetext.

Cambodian soils can be broadly divided into threegroups: (1) the Prey Khmer and Prateah Langgroups, which are soils with potential for low to

moderate yields (White et al. 1997b). These havelow nutrient reserves, low levels of organic matterand low cation-exchange capacities. Maintaining anadequate supply of nutrients for high rice yields onthese soils is difficult and farmers’ experience is thatresponse to fertilizer applications is variable. (2) Thepotential of the Bakan, Koktrap and Toul Samroungsoils is considered as higher (White et al. 1997b).These soils have relatively low levels of nutrientreserves but have higher levels of organic matter,CEC and clay. These soils are therefore more robust,responding to fertilizers more readily. (3) On thehigh-potential Krakor soils, farmers can produce asmuch as 10 t ha

–1

with dry-season rice and moder-ately high N fertilizer rates.

The decision to base much of the applied agro-nomic research and extension of fertility manage-ment in Cambodia on the CASC was made for bothsound scientific and pragmatic reasons. The alterna-tive would have been soil analysis and diagnosis,using chemical tests. However, for anything otherthan a limited research program, soil analysis anddiagnosis are unworkable in the current Cambodiancontext. The country has no operating laboratory thatcan provide reliable soil analysis. Neither have soiltests been calibrated for the country’s soils, environ-ments and farming systems, and limited expertiseexists for interpreting test results.

The CASC was established to improve soil andnutrient management by improving communicationabout soil resources and by providing a usable tool tohelp agronomists, extension agents and farmersmake decisions. Surrogate characters often help in

a

Local name as classified by White et al. (1997b). Names in parentheses refer to the

Key to Soil Taxonomy

by the SoilSurvey Staff (1994).

b

1:1, soil to water, except for values in italics, which were obtained from 1:5, soil to CaCl

2

.

c

EC = electrical conductivity; dS = decisiemen.

d

CEC = cation-exchange capacity.

Table 1.

Chemical properties of major rice soils in Cambodia.

Soil type

a

pH

b

EC

c

(dS m

–1

)Olsen

P(mg kg

–1

)

TotalN

(g kg

–1

)

OrganicC

(g kg

–1

)

Exchangeable cation [cmol(+) kg

–1

] CEC

d

[cmol(+) kg

–1

]K Na Ca Mg

Prey Khmer 5.6 0.03 1.3 0.5 4.7 0.02 0.04 0.38 0.12 1.8(Psamments)Prateah Lang

4.0

0.07 0.4 0.3 2.9 0.03 0.13 0.84 0.25 1.3(Plinthustalfs)Bakan 5.8 0.05 1.0 0.6 6.6 0.07 0.26 2.33 0.68 6.3(Alfisol/Ultisol)Koktrap

4.0

0.06 2.6 1.1 10.9 0.06 0.1 0.42 0.16 6.7(Kandic Plinthaquualt)Toul Samroung 5.5 0.08 3.1 0.9 8.8 0.16 0.39 7.81 4.31 18.2(Vertisol/Alfisol)Krakor 5.9 0.30 4.6 1.0 9.1 0.19 0.53 7.78 3.05 13.5(Entisol/Inceptisol)

172



accurately predicting soil chemical properties.Indeed, elsewhere, farmers’ soil classifications,which are similarly based on surrogate characters inthe field, allow delicate management of those soilswith which farmers have first-hand experience(Bellon and Taylor 1993). Their classifications arehighly and positively correlated with laboratorychemical and physical indicators of soil quality(Sandor and Furbee 1996; Talawar 1996). A clearrelationship has also been observed between landquality, as assessed by the farmer, and variety selec-tion (Bellon and Taylor 1993).

Currently, more than 2000 copies of the Khmer-language version of the soil manual have been dis-tributed and are used to identify soils for rice produc-tion. In 1999, seven training courses on integratednutrient management and the use of soil manualwere provided to more than 250 researchers andtechnical staff, extension workers and fertilizerdealers from various non-governmental and govern-mental organizations (CIAP 1999).

Local agronomists who use the CASC can separatesoil variations sufficiently to classify a given soil andthus identify the interaction of soil type with fertilizerrates and varieties (White et al. 2000). With theCASC, agronomists can also adequately predict dif-ferences in grain yields of rice grown on the differentsoils in on-farm trials (CIAP 1998; White et al. 2000).

The CASC can be applied beyond Cambodia and,when combined with other management tools, is stillrelevant even where sophisticated soil analyticalfacilities are available. The philosophy, applicationand accuracy of the CASC are described by Dobermanand White (1999), Oberthür et al. (2000a, b) andWhite et al. (2000).

Diagnosing Nutrient Disorders

Through nutrient omission and field trials on fertilizerresponse, the major deficiencies in Cambodian low-land soils for rice cultivation have been identified(Lor et al. 1996). Deficiencies of N, P, K and sulfur(S) were identified in pot experiments and confirmedin field trials. Many soils exhibit multiple deficienciesso that factorial field trials were needed to examineinteractions between the elements. In pot experi-ments, responses to boron (B) and magnesium (Mg)were obtained in some soils, but have yet to bedemonstrated in the field.

Responses to nitrogen, potassium and sulfur

Rice is highly variable in its response to fertilizerapplication, depending on soil type. In Koktrap soil,the highest yield was obtained when all nutrients (N,P, K and S) were applied together (Figure 1). When

N, P or S were omitted the grain yield was decreasedto 1–1.5 t ha

–1

. On this site omitting K appeared toincrease yield for reasons that are not clear (NPKS-3and NPKS-4, Figure 1b). When the nutrients wereapplied singularly, only a small or sometimes nega-tive response to fertilizer resulted, especially for N orK alone (Figures 1a, b), and rice leaves becamebronzed. An adequate supply of N, P and S musttherefore be maintained to realize the full benefits offertilizer application on this soil.

On Prateah Lang soils, rice grain yield increasedwith K fertilizer application but the effect wasadditive to that of N and P fertilizer application(Table 2). That is, grain yield increased overall by12% with the addition of K at 33 kg ha

–1

, regardlessof the N and P application rates.

a

Multiple of recommended fertilizer rate for each soil asshown in Table 3.

b

Soils: PK = Prey Khmer; PL = PrateahLang; BK = Bakan; KT = Koktrap; TS = Toul Samroung.

Rice cultivated in Bakan and Koktrap soils, whichgenerally contain relatively low levels of N, P and K(Table 1), also responded to additions of N, P and Kfertilizers, similarly to rice in Prateah Lang soil(Table 2). In contrast, the response of rice toadditions of N, P and K in Prey Khmer soil was low,probably because of climatic factors, since the PreyKhmer soil experienced a mid-season loss of soil-water saturation, followed by heavy rain (CIAP1998), resulting in high losses of nutrients throughrun-off and leaching from the deep sandy texturedprofile and low-CEC soil.

Grain yields in heavy textured soils that wererelatively rich in K did not respond to applied K. Theexchangeable K level in the Toul Samroung soil[0.16 cmol(+) kg

–1

] was higher than in other light-textured soil groups (Prey Khmer, Prateah Lang and

Table 2.

The effect of potassium (K) fertilizer on grainyield of the rice cultivar Santepheap 1 with various levelsof nitrogen (N) and phosphorus (P) fertilizer applications.

N–P rate

a

K rate

(kg ha

–1

)Grain yield (t ha

–1

)

b

PK soil

PLsoil

BK soil

KTsoil

TSsoil

0 0 1.5 2.3 2.6 1.0 3.51 0 1.7 3.0 3.6 1.4 3.32 0 1.5 3.1 3.8 1.6 4.40 33 1.6 2.5 3.0 1.0 3.51 33 1.8 3.3 4.0 1.5 3.32 33 1.5 3.6 4.2 1.9 4.1

F probabilities

NP 0.116 <0.001 <0.001 0.018 0.481K 0.444 0.008 0.006 0.038 0.833NP

×

K 0.733 0.681 0.987 0.843 0.977

173



Bakan), whose exchangeable K ranged from 0.02 to0.07 cmol(+) kg

–1

(Table 1). With long-term use of N and P fertilizers to

correct deficiencies of these elements, the soilreserves of K are likely to become depleted and Kfertilizer may be needed in the future, even on theToul Samroung soils.

Response to phosphorus

In Cambodia, more than three fifths of the soils usedfor rice production are deficient in P (Pheav et al.1996), limiting rice yields over much of the country.Cambodia relies entirely on imported compoundchemical fertilizers (e.g. di-ammonium phosphate orDAP, 16–20–0 and 15–15–15) for rice production.The expense obliges farmers to apply only smallamounts of P for rice cultivation.

Local phosphate rock ores, which could supplymuch of the demand for P fertilizer, has only justbeen introduced into the local market, but with littlesupport from research on the quantity and suitabilityof rock phosphate as a P fertilizer. However, pre-vious research showed that high-grade rock phos-phate was likely to be as effective as imported triplesuper phosphate (TSP) for rainfed lowland rice pro-duction in Cambodia, producing yields ranging from1.5 to 2.5 t ha

–1

when rock phosphate was applied at

5–10 kg P ha

–1

(White et al. 1999). However, theresponse of rice to P fertilizer additions variesbetween soil types. Some soils (e.g. Koktrap andPrateah Lang) are acidic, with low available P andexchangeable Ca levels. They remain acidic, evenafter several weeks of flooding (White and Seng1997). These factors favour rock phosphate dissolu-tion, thus increasing the availability of P for rice(Hammond et al. 1986). In other soils (e.g. PrateahLang), sulfur deficiency restricts the response of riceto P additions (Lor et al. 1996).

Low soil fertility, together with fluctuations in thesoil-water regime during crop growth, causes rice torespond inconsistently to applied P fertilizers (Whiteand Seng 1997). In acidic Koktrap clay soil andunder flooded conditions, grain yield respondsstrongly to P addition alone, but the curve levels offafter adding P at 20 kg ha

–1

or more (Figure 1). Incontrast, in the sandy Prateah Lang soil, grain yielddid not respond to P additions alone, with yields ofabout 1.4 t ha

–1

under flooded conditions (Figure 2).Loss of soil-water saturation decreased rice yieldfrom 1.4 to 0.7 t ha

–1

, but not because of water stressin rice plants. Instead, loss of soil-water saturationoxidizes the soil, enabling phosphates to react withiron oxides. This, in turn, reduces P availability andthus restricts P uptake by rice plants. Relationships

Figure 1.

Rice grain yield after fertilizer application in a Koktrap soil (acid clay soil). Curves represent the predictedresponse based on the square root quadratic function fitted to experimental data. In the NPKS treatments, the rate of oneelement (i.e. in [a] N in NPKS-1, P in NPKS-2, and in [b] K in NPKS-3 and S in NPKS-4) was varied while all the otherelements were kept constatnt. The combined rate of nutrients were NPKS-1 = 0–120N + 44P + 66K + 60S; NPKS-2 = 120N+ 0–52P + 66K + 60S; NPKS-3 = 120N + 44P + 0–100K + 60S; NPKS-4 = 120N + 44P + 66K + 0–120S (from CIAP1995).

(a) (b)

NPKS-1

NPKS-2

P only

N only

NPKS-3

S only

NPKS-4

K only

Gra

in y

ield

(t h

a−1)

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0

Fertilizer rate (kg ha−1)

0 40 80 120 0 40 80 120

174



between shoot dry-matter weights and plant P uptakeremained close under different soil-water regimeswith two soil types and under greenhouse conditions(Figure 3). The relationships suggested that changesin soil-water regimes, which affect soil redox poten-tial, influenced rice growth by controlling P availa-bility (Seng et al. 1999).

Figure 2.

Modelled response of rice grain yield to P fertilizerapplication with combined nitrogen (N) and straw (ST)additions to (a) flooded and (b) non-flooded sandy PrateahLang soils of south-east Cambodia. The levels of N addedwere in kg ha

−

1

, and of straw in t ha

−

1

(from Seng 2000).

Fertilizer Recommendations

Fertilizer application, together with the use of modernvarieties, irrigation and other improved managementpractices, has been increasing food production inSouth-East Asia over the past 25 years. However,financial or crop losses and environmental damagemay result if fertilizers are applied incorrectly.Careful consideration must be given to several factorswhen the rate and timing of fertilizer are beingdecided. Farmers implicitly consider some of thesefactors, such as risk associated with erratic rainfall,when deciding on fertilizer use.

Figure 3.

Relationships between shoot dry matter (shootDM) and P uptake in rice grown in Koktrap (KT) andPrateah Lang (PL) soils under various soil-water conditionsmaintained in the greenhouse. Plotted values are means ofthree replicates. The numbers 1, 2 and 3 represent potswhere the water regime was field capacity, and intermit-tently and continuously flooded soils, respectively (fromSeng et al. 1999).

In Cambodia, current work is focused on the bestuse of fertilizer for particular soil types at the appro-priate rates, according to the CASC (Table 3). Theserates are also formulated from local knowledge andexperience and from the results of fertilizer trialsconducted by the Cambodia–IRRI–Australia Projectduring 1992–1997 (CIAP 1998).

On the Prey Khmer, Prateah Lang and Bakan soiltypes, fertilizers should be broadcast into soil thathas been flooded for at least a week. Farmyardmanure (FYM) and inorganic fertilizer applicationsare also recommended. If FYM is to be applied,inorganic fertilizers should be broadcast several daysafter the FYM is applied. Where possible, the fieldshould be drained before transplanting and the initialfertilizer broadcast and thoroughly incorporated intothe wet soil. Further fertilizer applications shouldalso be broadcast into wet soil.

At present, however, few data exist to permit soil-specific predictions of the fertilizer requirements andresponse for rice production in Cambodia. Carefuldetailed research is still needed to determine howfarmers would adjust recommended fertilizer ratesand timing to take into account drought or sub-mergence events.

No specific type of fertilizer or combination offertilizers is currently recommended. Urea, DAP andKCl have been used as examples here, based simply

80N, 5ST

Gra

in y

ield

(t h

a−1)

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0

0N, 5ST

80N, 0ST

0N, 0ST

(a) Flooded soil

(b) Not flooded soil

80N, 5ST

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0

0N, 5ST

80N, 0ST

0N, 0ST

0 20 40 60 80

Applied phosphate (P at kg ha−1)

Sho

ot D

M (

g po

t−1)

P uptake (mg pot−1)

55

50

45

40

35

30

25

20

20 40 60 80 100

PL soily = 0.4467x + 11.784R2 = 0.9917

KT soily = 0.4394x + 10.677R2 = 0.9878

1

2

3

1

2

3

175



on cost. Other fertilizers, combined to apply thesame recommended rate of nutrients, would probablyproduce similar responses.

a

Local names as classified by White et al. (1997b); namesin parentheses refer to the

Key to Soil Taxonomy

by the SoilSurvey Staff (1994). Source: DOA (1999).

For the rice nursery, research in Cambodia has sofar shown large benefits from applying fertilizers (Ros1998). Farmers apply higher rates of fertilizer andmanures to the seedling nursery than to the main fields(Ros et al. 1998). Such applications increase seedlingvigour, which then increases subsequent rice yields by5%–10%, regardless of whether the main fields weretreated with fertilizers. Cow manure at 3.0 t ha

–1

andinorganic fertilizer (N at 50 and P at 22 kg ha

–1

) arerecommended for increasing seedling vigour (CIAP1998). Other low-cost strategies suggested by Ros etal. (2000) for increasing seedling vigour includecoating seeds in rock P powder and selecting seedswith high P concentration for nursery planting.

The Cambodian farmers’ strategy of applyingmore fertilizer to the nursery is an efficient use ofnutrients because applications to seedlings alreadytransplanted to the main fields do not adequatelyimprove their vigour. Currently, manure and inor-ganic fertilizer are recommended for application toseedbeds in all soils, except the Kbal Po and Krakorgroups (White et al. 1997b).

Recommended Work on Agricultural Soilsin Cambodia

Variety

×

soil type interactions

Cultivar response to fertilizer varies between soiltypes, although, in unfertilized plots within each soiltype, no significant differences in grain yield occurbetween farmer and recommended varieties(Figure 4). In the infertile sandy soils of the PreyKhmer group (Figure 4a), rice grain yields were

lower than those obtained from the more fertile soilsof the Prateah Lang or Bakan groups (Figures 4b, c).Adding fertilizer substantially increases grain yieldof recommended varieties growing in the PreyKhmer (from 1.8 to 2.9 t ha

–1

), Prateah Lang (2.5 to3.0 t ha

–1

) and Bakan (2.3 to 3.3 t ha

–1

) soil groups.Under these conditions, farmer varieties on the PreyKhmer and Bakan soil groups show similar yields,but levels are much lower than those of the recom-mended varieties. In contrast, in the Prateah Langsoil group, grain yield of farmer varieties decreasedby 0.2 t ha

–1

when 1.5 times the recommended rateof fertilizer was applied. The greater response ofrecommended varieties to the recommended fer-tilizer rate, compared with farmer varieties, indicatedthat joint strategies of improved cultivars andnutrient management are needed to create significantimpacts on crop yield.

Soils used for rice production in Cambodia varylargely in their physical and chemical properties. Notsurprisingly, some rice varieties released by theCambodia–IRRI–Australia Project (CIAP) are welladapted to some soils, but not to others, indicatingthat it is important to know how new varieties per-form on different soils so that appropriate recom-mendations for varieties can be made.

Long-term nutrient management

Because farmers will continue to increase their inputsof inorganic fertilizers, and because the balance ofnutrients entering and exiting Cambodian agriculturalsystems is unknown, long-term monitoring systemsshould be developed. Nutrient movement, particu-larly P and K nutrient retention in soils, and nutrientcycling through cropping, pasture or fallow rotationscan therefore be monitored.

Simulation modelling shows that soil-specific fer-tilizer recommendations result in increased yieldsand more efficient fertilizer use than does the oldrecommended rate for all soils (N at 64 and P at23 kg ha

–1

). Yields increase by 0.2–0.5 t ha

–1

whenhigher rates of fertilizer are applied to responsivesoils and lower rates to unresponsive soils (PreyKhmer and Prateah Lang). Such rates result insavings of N at rates between 14 and 36 kg and of Pat 7 and 16 kg ha

–1

(Ros et al. in press). Furtherscope exists for developing site-specific nutrientmanagement.

Improved management of the consequences ofloss of soil-water saturation for rice growth maydepend on the sensitivity of a given rice varietyaccording to development stage. In the rainfed low-lands, loss of soil-water saturation can occur at anystage. Maximum tillering and panicle initiation,which are important for grain yield development in

Table 3.

Recommended rates of nutrients for rainfed riceon major soil types of Cambodia.

Soil type

a

Recommended rate of nutrients (kg ha

–1

)

N P K

Prey Khmer (Psamments) 28 4 33Prateah Lang (Plinthustalfs) 50 10 25Bakan (Alfisol/Ultisol) 75 13 25Koktrap (Kandic Plinthaquult) 73 15 25Toul Samroung (Vertisol/Alfisol) 98 15 0Krakor (Entisol/Inceptisol) 120 11 0

176



rice, may also be sensitive periods of loss of soil-water saturation (Fukai and Cooper 1996). Identi-fying critical stages for loss of soil-water saturationwill help in the management of nutrients for rainfedlowland rice because it will indicate the growthphases when the crops most need irrigation tomaintain soil saturation. Alternatively, preventing orminimizing yield losses to short periods of loss ofsoil-water saturation may be feasible with straw ororganic-matter additions (Seng et al. 1999).

Figure 4.

Grain yield response of farmer and recom-mended rice varieties to fertilizer applications inrainfed lowland soils, Cambodia. Plotted data are means of(a) 3 sites in Prey Khmer soil, (b) 8 sites in Prateah Langsoil and (c) 18 sites in Bakan soil. The recommended fer-tilizer rates used are shown in Table 3 (from CIAP 1998).

Problem soils

In South-East Cambodia, thousands of hectares ofsoils are strongly acidic and many thousands ofhectares of soils adjacent to coastal areas are alsoaffected by sea salt, making rice cultivation risky.Research is needed to provide management recom-mendations for increasing the arable potential ofthese soils.

Rice bronzing

In Cambodia, rice bronzing occurs sporadically eachyear, especially on Koktrap, Prateah Lang and Bakansoil groups. Although loss of grain yields caused bythis problem has not been confirmed, the disorderappears to be similar to that described for rice inJapan and Nigeria (Yamauchi 1989). In Cambodia,anecdotal evidence suggests that the problem is moreprevalent and more severe with the increased use ofinorganic N and P fertilizers, but no K fertilizer.Removal of rice-straw from the field limits the returnof K in soils receiving none or low rates of appliedK. Breeding for resistant cultivars may be the mosteffective strategy for the long-term alleviation of thisdisorder.

Soil classification for agricultural crops

The realization of Cambodia’s agricultural potentialis hampered by several inherent soil problems, com-pounded by limited soil information, not only forrice but also for other crops. Although a rice soilmap is available and the current classification systemfor rice soils appears to be useful for other crops,further effort is needed to obtain soil information andmaps of arable land that are both more complete andmore sophisticated.

Acknowledgments

The authors are grateful to the Cambodia–IRRI–Australia Project (CIAP) and the Australian Agencyfor International Development (AusAID) for researchfunding.

References

Bellon, M.R. and Taylor, J.E. 1993. “Folk” soil taxonomyand the partial adoption of new seed varieties. EconomicDevelopment and Cultural Change, 41, 763–786.

CIAP (Cambodia–IRRI–Australia Project). 1995. AnnualResearch Report. Phnom Penh, Cambodia.



(a)

Gra

in y

ield

(t h

a−1)

(b)

(c)

4

3

2

1

4

3

2

1

4

3

2

1

0.0 1.0 1.5

Multiple of recommended fertilizer rate

177



CNP (Center for National Policy) and SCI (Sparks Compa-nies, Inc.). 1996. Fertilizer Procurement and Distributionin Cambodia: Issues and Options. Washington, DC.

De Datta, S.K. 1995. Nitrogen transformations in wetlandrice ecosystems. Fertiliser Research, 42, 193–203.

DOA (Department of Agronomy), Ministry of Agriculture,Forestry and Fisheries. 1999. New fertilizer recommen-dation rates in rice production. Agricultural Note no.INM 99/01. Phnom Penh, Cambodia.

Fischer, K.S. 1998. Toward increasing nutrient-useefficiency in rice cropping systems: the next generationof technology. Field Crops Research, 56(1–2), 1–6.

Fukai, S. and Cooper, M. 1996. Stress physiology inrelation to breeding for drought resistance: a case studyof rice. In: Singh, V.P., Singh, R.K., Singh, B.B. andZeigler, R.S. ed. Physiology of Stress Tolerance in Rice:Proceedings of the International Conference on StressPhysiology of Rice. Lucknow, India, International, RiceResearch Institute, 123–149.

Hammond, L.L., Chien, S.H. and Mowunye, A.U. 1986.Agronomic value of unacidulated and partially acidu-lated phosphate rocks indigenous to the tropics.Advances in Agronomy 40, 89–140.

Kirk, G.J.D., Tian-ren, Y. and Choudhury, F.A. 1990.Phosphorus chemistry in relation to water regime. In:Phosphorus Requirements for Sustainable Agriculture inAsia and Oceania. Manila, Philippines, InternationalRice Research Institute (IRRI), 211–223.

Lor, B., White, P.F. and Chan, P. 1996. Nutrient require-ments for the growth of rice on Cambodian soils. In:Attanandana, T., Kheoruenromne, I., Pongsakul, P. andVerasilp, T. ed. Maximising Sustainable Rice Yieldsthrough Improved Soil and Environmental Management.Khon Kaen, Thailand, Ministry of Agriculture andCooperatives, 45–56.

MAFF (Ministry of Agriculture, Forestry and Fisheries).1999. Agricultural Statistics. Phnom Penh, Cambodia.

Nesbitt, H.J. ed. 1997. Rice Production in Cambodia.Manila, Philippines, International Rice Research Institute(IRRI).

Oberthür, T., Dobermann, A. and White, P. 2000a. The ricesoils of Cambodia. II. Statistical discrimination of soilproperties by the Cambodian Agronomic Soil Classifica-tion system. Soil Use and Management, 16, 20–26.

Oberthür, T., Ros C. and White, P.F. 2000b. Soil map ofthe main rice growing area of Cambodia. Phnom Penh,Cambodia, Cambodia–IRRI–Australia Project (CIAP).

Pandey, S. 1997. Rainfed lowland rice research: challengedand priority for the 21st century. In: Fukai, S., Cooper,M. and Salisbury, J. ed. Breeding Strategies for RainfedLowland Rice in Drought-Prone Environments; Proceed-ings of an International Workshop held at UbonRatchanthani, Thailand. Canberra, ACIAR ProceedingsNo. 77, 1–12.

Pheav, S., White, P.F. and Oberthür, T. 1996. Managementconstraints of the major rice growing soils of Cambodia.In: Attanandana, T., Kheoruenromne, I., Pongsakul, P.and Verasilp, T. ed. Maximising Sustainable Rice Yieldsthrough Improved Soil and Environmental Management.Khon Kaen, Thailand, Ministry of Agriculture andCooperatives, 537–550.

Ros, C. 1998. Nutrient Management of Rice Seedlings inthe Rainfed Lowlands of Cambodia. PhD dissertation.Perth, Western Australia, School of EnvironmentalScience, Division of Science and Engineering, MurdochUniversity.

Ros, C., Bell, R.W. and White, P.F. 2000. Phosphorus seedcoating and soaking for improving seedling growth of

Oryza sativa

(rice) cv. IR66. Seed Science and Tech-nology, 28, 201–211.

Ros, C., Vang, S., Hin, S. and White, P. In press. Better soilmanagement better impact: the Cambodian AgronomicSoil Classification. In: Proceedings of the InternationalConference on the Impact of Agricultural Research forDevelopment in Southeast Asia. Phnom Penh, Cambodia,Cambodian Agricultural Research and DevelopmentInstitute (CARDI).

Ros, C., White, P.F. and Bell, R.W. 1998. Field survey onnursery and mainfield fertilizer management. CambodianJournal of Agriculture, 1, 22–33.

Sandor, J.A. and Furbee, L. 1996. Indigenous knowledgeand classification of soils in the Andes of southern Peru.Soil Science Society of America Journal, 60, 1502–1512.

Seng, V. 2000. Edaphic Factors Limiting Rice Responsesto Applied Inorganic Fertilizers in Rainfed LowlandSoils in Southeast Cambodia. PhD dissertation. Perth,Western Australia, School of Environmental Science,Division of Science and Engineering, Murdoch Univer-sity.

Seng, V., Bell, R.W., Willett, I.R. and Nesbitt, H.J. 1999.Phosphorus nutrition of rice in relation to flooding andtemporary loss of soil-water saturation in two lowlandsoils of Cambodia. Plant and Soil, 207, 121–132.

Soil Survey Staff, Soil Conservation Service, United StatesDepartment of Agriculture. 1994. Key to Soil Taxonomy,6th ed. Washington, DC.

Talawar, S. 1996. Local soil classification and managementpractices: bibliographic review. Athens, GA, Laboratoryof Agriculture and Natural Resources Anthropology,University of Georgia.

Wade, L.J., George, T., Ladha, J.K., Sign, U., Bhuiyan, S.I.and Pandey, S. 1998. Opportunities to manipulatenutrient-by-water interactions in rainfed lowland ricesystems. Field Crops Research, 56(1–2), 93–112.

White, P.F. and Seng Vang. 1997. Response of rainfed low-land rice to P fertilizer application in Cambodia. In:Fukai, S., Cooper, M. and Salisbury, J. ed. BreedingStrategies for Rainfed Lowland Rice in Drought-ProneEnvironments; Proceedings of an International Work-shop held at Ubon Ratchanthani, Thailand. Canberra,ACIAR Proceedings No. 77, 202–208.

White, P.F., Nesbit, H.J., Chhay Ros, Vang Seng andBunna Lor. 1999. Local rock phosphate deposits are agood source of phosphorus fertilizer for rice productionin Cambodia. Soil Science and Plant Nutrition, 45(1),51–63.

White, P.F., Oberthur, T. and Pheav, S. 1997a. Soil andrice. In: Nesbitt, H.J. ed. Rice Production in Cambodia.Manila, Philippines, International Rice Research Institute(IRRI), 21–29.

White, P.F., Oberthur, T. and Pheav, S. 1997b. The SoilsUsed for Rice Production in Cambodia: A Manual for

178



Their Identification and Management. Manila, Philip-pines, International Rice Research Institute (IRRI).

White, P.F., Dobermann, A., Oberthür, T. and Ros, C.2000. The rice soils of Cambodia. I. Soil classificationfor agronomists using the Cambodian Agronomic SoilClassification system. Soil Use and Management, 16,12–19.

White, P.F., Oberthür, T., Dobermann, A. and Ros, C.1998. Simple soil classifications are important tools forimproving fertility management of infertile rice soils ofthe rainfed lowlands. In: Ladha, J.K., Wade, L.,

Dobermann, A., Reichardt, W., Kirk, G.J.D. andPiggin, C. ed. Rainfed Lowland Rice: Advances inNutrient Management Research, Proceedings of theInternational Workshop on Nutrient Research in RainfedLowlands held at Ubon Ratchanthani, Thailand, 12–15Oct., 1998. Manila, Philippines, International RiceResearch Institute (IRRI), 43–58.

Yamauchi, M. 1989. Rice bronzing in Nigeria caused bynutrient imbalances and its control by potassium sulfateapplication. Plant and Soil, 117, 275–286.

179



Nitrogen Management for the Rainfed Lowland Rice Systems of Laos

B. Linquist

1*

and P. Sengxua

2

Abstract

Nitrogen (N) is the most limiting nutrient to rice in the rainfed lowlands of Laos. It is alsorequired in higher quantities and is more susceptible to loss than are other nutrients and thereforeefficient use of N is imperative for resource poor farmers. The current N recommendation forrainfed lowland rice is 60 kg N ha

–1

, applied in three equal splits at transplanting, active tilleringand panicle initiation. If problems due to floods, droughts and pests are eliminated, the agronomicefficiency of applied fertilizer N (AE), using this recommendation, is usually greater than 20 kggrain per kg N. The timing of N applications was evaluated in efforts to improve N-use efficiencyand it was found that under conditions of sub-optimal N supply the timing of N is flexible. Thefirst N application can be applied between 0 and 30 days after transplanting and the last N applica-tion between two weeks before and one week after panicle initiation. This is an advantage in therainfed system where environmental or economic factors often results in farmers not being able toapply N on the “recommended” day. Increasing the percentage of N applied during active tilleringand panicle initiation, when demand for N is high, increased AE, on average, by 9 kg per kg N,compared with applying N in equal splits. On-farm residues combined with inorganic fertilizersgenerally did not improve fertilizer-use efficiency in the first year of application. The yield benefitto rice following green manure (GM) incorporation was similar to that of urea-N applied at 30 to60 kg N ha

–1

. However, on coarse textured soils, GM crops required about three times more P thanrice, making them unsuitable for this environment.

R

ICE

is the single most important crop in Lao PDR,occupying a 60% share of the country’s total agricul-tural production (UNDP 1998). About 70% of the totalrice area (646 000 ha) is classified as rainfed lowland,of which more than 80% is in the south, growingprimarily on six plains adjacent to the Mekong River.In the mountainous north, lowland rice production isconfined to valleys. In this paper, southern Laosincludes the provinces of Vientiane, Borikhamxay,Khammouane, Savannakhet, Saravane, Sekong,Attapeau and Champassak (south of 19° north), whilethe north refers to the remaining provinces locatedmainly north of the 19° latitude.

The annual rainfall pattern is weakly bimodal witha minor peak in May–June and a major peak inAugust–September (Fukai et al. 1998). However,annual rainfall is erratic with the possibility ofdrought and/or flooding in any given year. Rainfedlowland rice farmers are usually poor and most ofthe rice grown is consumed at home (Pandey andSanamongkhoun 1998). Because of the riskiness ofproduction in this environment and limited capital,fertilizer use is low, being the lowest in Asia (IRRI1995), even though its use is increasing in the rainfedlowlands (Pandey and Sanamongkhoun 1998).

Nitrogen is required in higher quantities and ismore susceptible to loss than other nutrients (Schnier1995). These factors, combined with the lowland ricefarmers’ general poverty, make efficient use of N forcrop production imperative. The Lao–IRRI Projecthas been working in collaboration with the LaoNational Rice Research Program since 1991 to

1

Lao-IRRI Project, Vientiane, Laos

2

National Agriculture and Forestry Research Institute,Vientiane, Laos*Corresponding author: Bruce Linquist.E-mail: b.linquist@cgiar. org

KEYWORDS:

Rainfed lowland rice, Fertilizer management, N use efficiency, N timing, Green manure, Laos

180



develop efficient nutrient management practices forrainfed lowland rice farmers.

This paper summarizes results from experimentsthat have focused on N-management strategies for therainfed lowlands, including research on inorganic-Nfertilizers, green manures and on-farm residues.

Nitrogen Deficiencies in Laos

Although Lao soils have not yet been fully classified,chemical and physical indicators of native soil fertilityin the lowlands are known to differ between the tworegions of Laos. Analyses (0–20 cm) indicate that80% of the southern soils contain less than 2% organicmatter, 68% are coarse textured (sands, loamy sands,and sandy loams) and 87% have a pH (H

2

O) of lessthan 5.5 (Figure 1). In contrast, the northern lowlandrice soils are more fertile: 66% of the soils containmore than 2% organic matter, 80% are loams or clayloams, and only 48% have a pH of less than 5.5.

Nitrogen is the most limiting nutrient for lowlandrice, according to results from on-farm N–P–Komission trials conducted throughout Laos (Linquistet al. 1998). Nitrogen deficiencies were morecommon in southern (86% of sites evaluated) than innorthern Laos (50% of the sites evaluated). When noN was applied, average yields across sites weresignificantly higher in the north (2.91 t ha

–1

) than inthe south (2.25 t ha

–1

), reflecting the differences insoil fertility discussed in Figure 1.

A direct assessment of indigenous soil-N supplywas not possible because of a lack of soil and plant N

analyses, although soil indigenous-N fertility can beestimated from yield data in -N plots where P and Khave been added. Under such conditions, N is themost limiting nutrient and yields would indicate soilN fertility level (Dobermann and Fairhurst 2000). Forour analysis, 32 rainfed sites were available (11 in thenorth and 21 in the south). These both had -N yielddata (P and K added) and soil organic matter (SOM)data. Regression analysis indicated a significantcorrelation existed between yield from -N plots andSOM (Figure 2), suggesting that the native-N supplywas derived from mineralization of organic matter.

These data contrasted with those of Cassman et al.(1996b), who found no relationship between soilorganic carbon and indigenous-N supply in intensiveirrigated rice soils. They suggested that the reasonsfor the poor correlation were (1) inputs of N fromother sources, (2) degree of congruence betweensoil-N supply and crop demand since N mineraliza-tion is sensitive to soil drying, fallow length, croprotation and residue management and (3) differencesin soil organic matter quality with intensive croppingunder submerged soil, compared with rainfed low-land soils. In the extensive rainfed conditions ofLaos, these factors are more uniform across locationsthan in intensive irrigated systems. For example,lowland rice receives no irrigation water that wouldprovide additional sources of N; soil remains dryduring the dry season; and straw, the primary residuein these systems, is either burned or grazed by theend of the dry season.

Figure 1.

Comparison of soil texture, pH (H

2

O) and organic matter contents (OM) of lowland rice soils (0–20 cm) insouthern ( ) and northern (

�

) Laos. Coarse-textured soils include sands, loamy sands and sandy loams.

Coarse texture pH < 5.5 OM < 2.0%

Per

cent

age

of s

oils

100

90

80

70

60

50

40

30

20

10

0

181



Figure 2.

The relationship between soil organic matter andrice grain yields in minus N plots (which received P and Kfertilizers). Values are means from 11 field experiments innorthern (

�

) and 21 in southern ( ) Laos. *** = signifi-cant at

P

< 0.001.

Figure 3.

Responses of improved (circles) and traditional(squares) varieties to applied nitrogen (N), as averagedfrom five experiments in southern (solid) and four experi-ments in northern (open) Laos.

Nitrogen-Use Efficiency

The primary focus of N management research con-ducted in Laos has been to improve N-use efficiency.One of the many ways of estimating or calculatingN-use efficiency is to use the agronomic efficiency(AE) index of N use, which is calculated as follows:

AE (kg grain per kg N) =

[1]

This incremental efficiency from applied N is pro-portional to the cost-benefit ratio from investment inN inputs (Cassman et al. 1996a).

The current N recommendation for improvedvarieties in the rainfed lowland environment is N at60 kg ha

–1

, applied in three equal splits at trans-planting, active tillering (AT) and panicle initiation(PI). The N rate is based on N response experiments(Figure 3) and a realization of farmer risk. In boththe northern and southern regions, response to N canbe obtained with as much as 60 kg ha

–1

(Figure 3). Inthe north, however, no response occurs with N rateshigher than 60 kg ha

–1

. The rainfed lowlands com-prise a high-risk environment that is prone todrought, floods and pest damage. While higheryields may result from increased N inputs in thesouth, the additional cost of inputs needs to be evalu-ated against the risks of no additional response.

The AE of N applied at the recommended ratewas estimated from 107 experiments conducted inLaos between 1991 and 1999 (32 in the north and 75in the south). In all cases, improved varieties wereused and P and K were applied at transplanting toensure that these nutrients were not limiting. Sitesthat were adversely affected by drought, flood orinsect damage were not removed from the analysis,so that a lack of response to N fertilizer at a site doesnot necessarily indicate that N was limiting. Riceyields in the absence of N averaged 2.9 t ha

–1

in thenorth and 2.1 t ha

–1

in the south, similar to the resultsfrom the N–P–K omission studies mentioned above.

These experiments show a significant N response(

P

< 0.05) at 77% of the sites. The average AEacross all sites in response to N at 60 kg ha

–1

was14.9 kg kg

–1

(Table 1), and ranged from 0 to 38 kgkg

–1

. Pandey and Sanamongkhoun (1998) report that,if the AE is greater than 9 kg kg

–1

, fertilizer use iseconomically attractive to farmers. As noted above,this data set includes all experiments, including thosein which some of the crop was lost to biotic andabiotic stresses and, as such, the set reflects the risksinherent in this environment with fertilizer use. Ifcrop failures are omitted from the analysis, AE iscommonly in excess of 20 kg kg

–1

, as will be seen inthe following discussion.

Further analysis of this data set indicated no dif-ferences in the AE between northern (14.5 kg kg

–1

)and southern (15.1 kg kg

–1

) Laos (Table 1). Neitherdid soils differ in indigenous-N supply as estimatedby using yield from the -N treatments (Dobermannand Farihurst 2000). Finally, an analysis of a subsetof these data (those 31 sites for which data on

3.53.02.52.01.51.00.50.0

Soil organic matter (%)

y = 766x + 1387R2 = 0.58 ***

Gra

in y

ield

(kg

ha−1

)6000

5000

4000

3000

2000

1000

0

Applied N (kg ha−1)

0 30 60 90

Gra

in y

ield

(t h

a−1)

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0

treatment yield (kg) yield in control (kg)–amount of N applied (kg)

-----------------------------------------------------------------------------------------------------

182



significant responses to applied fertilizer N and soildata were available) indicated no relationshipbetween the AE and either clay or organic mattercontent (data not shown). A prominent feature ofLao soils is that they are coarse textured (Figure 1)and, as such, tend to have higher percolation ratesthan finer textured soils. In such cases, N is moresusceptible to leaching (Katyal et al. 1985) anddenitrification (because of a higher probability ofbeing subjected to wetting and drying cycles), and alower AE would be expected. However, under theconditions in which the N was applied (three splits),differences in soil properties such as soil texture andorganic matter or indigenous-N supply did not affectthe AE. Probably, if all the N fertilizer were appliedat one time only, such differences in soil propertieswould have a significant effect.

Improving Nitrogen-Use Efficiency

Effect of Cultivar

Until recently, 79% of the rainfed lowland rice areain southern Laos was planted to traditional ricevarieties (Pandey and Sanamongkhoun 1998). How-ever, the situation is changing rapidly, as anincreasing number of improved varieties are beinggrown. Even so, many farmers still grow traditionalvarieties on a significant portion of their land. Thismay be because of preferences in taste or other grainqualities or to spread out risk and labour demands.Given the limited capital available to purchase N, theefficiency of N applied to traditional and improvedvarieties needs to be studied.

Nine studies were conducted during the 1995 and1996 wet seasons at five sites in southern Laos andfour in northern Laos to compare the response oftraditional and improved varieties to fertilizer N. As

seen in other studies, average yields in the south werelower than those in the north by about 0.5 t ha

–1

(Figure 3). When no N was applied, yields oftraditional varieties were similar to those of improvedvarieties. In all cases, yield response to N waspositive until N was being applied at 60 kg ha

–1

. Onlyin the south, with the use of improved varieties, wasthere a continued response to additional N inputs.Response to N also varied between traditional andimproved varieties, with the improved varietiesresponding better. The AE of N use at rates of up to60 kg ha

–1

averaged 18.3 kg grain per kg N forimproved varieties, compared with 15.0 kg grain perkg N for traditional varieties. Similar responses havebeen observed elsewhere and for different crops.However, these results suggest that, for resource-poor farmers to maximize N-use efficiency, limited Nsupplies should be applied to paddy fields withimproved varieties rather than as a blanket applica-tion across all fields.

Nitrogen timing

Nitrogen is highly susceptible to loss. Matchingsupply of N with crop demand should increase N-useefficiency. To this end, the effect of applying theentire N recommendation at once or of applying it insplits was evaluated in 26 on-farm trials. Four treat-ments were used: (1) all N applied at transplanting,(2) N requirement split equally between transplantingand 50 days after transplanting (DAT), (3) N require-ment split equally between transplanting, 35 DAT and55 DAT, and (4) N requirement split equally between20, 40 and 60 DAT. In all cases, improved varietieswere used. Results indicated that applying N at therecommended rate in three or more splits was superiorto one or two splits (Table 2), significantly increasingyields by about 0.37 t ha

–1

(i.e. almost 12%) andimproving the AE by 4.1 kg grain per kg N. Theseresults are consistent with reports from Prasad and DeDatta (1979) and De Datta et al. (1988), and form thebasis for the broadly applied recommendation ofsplitting N requirements. Furthermore, for high-risk,low-input systems, splitting N requirements helpsminimize risk in those cases where crops fail duringvegetative growth. A large initial purchase of ferti-lizer is thus avoided.

The recommended times for N application are attransplanting, AT and PI. However, N application atthese times is sometimes not possible because oftemporary flooding, no standing water or no cash topurchase fertilizers. Those opportunities when N canbe applied without loss in efficiency must thereforebe established. Several experiments were conductedto discover the scope for adjusting the timing of thefirst and last N applications.

Table 1.

Agronomic efficiency (AE = grain yield increaseper unit of applied N) of rice production receiving nitrogen(N) at 60 kg ha

–1

and averaged across 107 experimentsconducted in Laos between 1991 and 1999.

Parameter Sites(no.)

AE(kg grain per kg N)

AreaEntire country 107 14.9Northern region 32 14.5Southern region 75 15.1

Yield under “-N” treatment (t ha

–1

)<1 5 13.81–2 43 15.42–3 36 14.1>3 23 15.4

183



a

DAT = days after transplanting.

Recommendations typically call for the first Napplication to be incorporated into the soil justbefore transplanting. This puts the fertilizer N into areduced soil layer and minimizes loss throughdenitrification (Obcemea et al. 1984). Farmers havenot generally adopted this practice, preferring toapply the first N application after crop establishment(Schnier 1995), a situation that is also observed inLaos. To determine whether such a recommendationwas necessary, we compared the effect of incor-porating N just before transplanting with applying None day after transplanting, in six replicated experi-ments. Rice responded well to N at all sites, withyields increasing, on average, by 63% to applied N.However, no differences in rice yields were observedbetween the different methods of N incorporation(Table 3). The current farmer practice of applying Nafter crop establishment therefore seems reasonable.

Some researchers question the need for a basalapplication, arguing that transplanted rice suffersfrom physiological shock for 10 to 14 days aftertransplanting, resulting in low initial demand for N(Schneir et al. 1987). Fertilizer N applied at this timewould not be readily taken up and would be prone toloss. Furthermore, N available from the mineraliza-tion of organic matter is greatest during crop estab-lishment (Dei and Yamasaki 1979) and should beadequate to meet crop needs during the early growthstage, at least for soils with a high N status (Schnier1995). However, Lao soils are inherently low inorganic matter (Figure 1) and the need for early Napplication must be evaluated under these conditions.

The timing of the first N application was evalu-ated in Laos across 15 sites (each site being a singlereplicate). The first N application was applied eitherbefore transplanting, or 1, 10, 20 or 30 days after-wards. In all cases, N was applied at 60 kg ha

–1

inthree equal splits (20 kg ha

–1

each) at the first appli-cation and 30 and 50 DAT. The exception was the

treatment in which the first N application wasapplied at 30 DAT. In this treatment, N was appliedin two equal splits (30 kg ha

–1

each) at 30 and 50DAT. In all cases, the popular improved variety,TDK1, was used.

*** = significant at

P

< 0.001.

Analysis across sites and treatments shows thatyields increased by 1.39 t ha

–1

in response to N,corresponding to an AE of 23 kg grain per kg N (Table4), with no significant effect of timing of the first Napplication on rice yields. Furthermore, neither SOMnor soil texture affected these results. Either adequateN was available from N mineralization to provide Nneeds early during crop growth, even at low SOM, orthe crop was able to compensate for temporary Ndeficiencies in early crop growth. The latter caseseems more probable because the crops displayedvisual N-deficiency symptoms during early growthwhen N was applied late. However, under conditionsof suboptimal N supply, early season N deficienciescan be compensated for by applying N later.

The timing of the last N application should ideallybe at PI. However, late-season drought often forcesfarmers to delay this application. To determine whenthe last N application is most efficiently used, sevenreplicated studies were conducted in the wet and dryseasons, and the last N applications were tried at 50,65 and 80 DAT (in the dry-season experiments, weincluded a 40-DAT trial). During the wet season, Nwas applied at 60 kg ha

–1

, while 90 kg ha

–1

wasapplied during the dry season. In both cases, N wasapplied in three equal splits. Cultivar TDK1 wasused in all studies. Panicle initiation for TDK1 isroughly 50 DAT, varying with location. To stand-ardize timing across sites, we assumed PI to be 25days before flowering (De Datta 1981). All applica-tion times were then made relative to PI rather thanto transplanting time. To evaluate the data acrosssites, relative yield (relative to the highest yield ateach site) was plotted against the time of the last Napplication (Figure 4). The data indicated that

Table 2.

Effect on rice grain yield when nitrogen (N) appli-cations were split. In all cases, the rate of N at 60 kg ha

–1

was applied and, when the application was split, the doseswere equal. Grain yields are averages from 26 experimentsconducted in Laos.

Treatment At:

a

Grain yield

(kg ha

–1

)

One-time application Transplanting 3130 c2 splits Transplanting + 50 DAT 3312 b3 splits Transplanting + 35 and

55 DAT3496 a

4 splits Transplanting + 20,40 and 60 DAT

3405 ab

Table 3.

Comparison of rice yields after incorporating thefirst nitrogen (N) application with applying N after trans-planting and not incorporating (based on the results of sixreplicated experiments conducted in Laos).

Treatment Grain yield (kg ha

–1

)

No N 2451 bN incorporated before transplanting 3880 aN applied 1 day after transplanting 4120 a

Site ***N treatment ***Site

×

N ns

184



delaying the last N application by more than a weekafter PI results in yield declines, but no negativeeffects appear when applying N up to 20 days beforePI (although only two points support this). If con-ditions are favourable, farmers should therefore notwait until PI to apply N, but should apply it 2 weeksbefore. If conditions are unfavourable, applicationcan wait until one week after PI, after which N-useefficiency declines.

a

Days after transplanting.

b

Applied and incorporated beforetransplanting.

Figure 4.

Relative yield (relative to the highest yield ateach site) of the rice cultivar TDK1 in relation to the timingof the last application of nitrogen (N) fertilizer. Zero on thex axis = panicle initiation (PI).

Ratio of nitrogen in each split

Increasing N-use efficiency can be achieved byincreasing the congruence between N supply andcrop demand (Cassman et al. 1996a). As notedabove, demand for N is low during early vegetativegrowth following transplanting, and N from themineralization of organic matter may be enough tomeet early crop demand. Furthermore, demand ishigh during AT and PI, suggesting that more N isrequired during this period. Therefore, a study wasconducted to compare the effectiveness of N appliedas three equal splits (the current recommendation)with a strategy whereby most of the N was applied atAT and PI. The replicated experiment was conductedat six sites during the 1998 and 1999 wet seasons.Nitrogen was either applied at 60 kg ha

–1

in threeequal splits at transplanting, AT and PI; or it wasapplied at a rate of 10, 25 and 25 kg ha

–1

at trans-planting, AT and PI, respectively.

Yields responded well to N in both treatments,increasing by 1.42 t ha

–1

, on average, in response tothe 60 kg ha

–1

rate (average AE = 24 kg grain per kgN) (Table 5). When 83% of the N requirements wasapplied at AT or PI, yields were higher (0.53 t ha

–1

)than when N was applied in equal splits. On average,the AE increased by about 9 kg grain per kg N (from19.3 to 28.1 kg kg

–1

). Examining the significantinteraction between N treatment and site indicatedthat, at half of the sites, the response to applying agreater proportion of the N later during crop growthwas greater than 0.4 t ha

–1

while, at the remainingsites, yields were similar for the two N managementstrategies. Importantly, at no site was a negativeimpact found by applying a greater proportion of Nlater in the crop cycle.

*** = significant at

P

< 0.001.

Table 4.

Effect on rice grain yields of timing of the firstapplication of nitrogen (N) on soils differing in organicmatter (OM) and clay contents. Data represent the increasein yields (kg ha

–1

) relative to a control with no N.

Timing

a

of first N application

Increase in grain yield (kg ha

–1

)

All sites(n = 15)

Organic matter at: Clay content at:

<0.8%(n = 6)

>1.0%(n = 9)

<12%(n = 4)

>15%(n = 11)

0

b

1348 1450 1291 1610 12761 1353 1320 1376 1293 1375

10 1326 1308 1337 1519 127420 1395 1350 1425 1287 143430 1530 1452 1573 1406 1564Significance: ns ns ns ns ns

40200−20

Rel

ativ

e yi

eld

(%)

100

90

80

70

60

50

Last N application in relation to PI (days)

y = −0.006x2 − 0.26x + 95.5R2 = 0.50

Table 5.

The effect on rice yields of applying nitrogen (N)in three equal splits at transplanting (TP), active tillering(AT) and panicle initiation (PI) versus applying a higherproportion of N during AT and PI. Data are the results ofseven experiments conducted over two wet seasons in Laos.

N applied (kg ha

–1

) at TP–AT–PI

Yield(kg ha

–1

)Agronomic efficiency

(kg grain per kg N)

0–0–0 2458 c —20–20–20 3615 b 19.310–25–25 4141 a 28.1

SignificanceSite ***N ***Site

×

N ***

185



Effects of residue on fertilizer-use efficiency

Reports from across the border in Northeast Thailandindicate that, on some soils, rice did not respond toinorganic fertilizers without organic amendments alsobeing applied (Ragland and Boonpuckdee 1988;Willett 1995). In a more recent study, Wade et al.(1999) found that response to inorganic fertilizers inNortheast Thailand was less than at other rainfedlocations in India, Philippines, Indonesia andBangladesh. In Laos, most farmers do not efficientlyuse on-farm residues such as manure, rice straw andrice husks (Linquist et al. 1999).

To evaluate whether organic amendments wouldimprove fertilizer-use efficiency, two studies wereconducted at each of two sites during the 1998 and1999 wet seasons. At the three sites in the south (twoin Saravane and one in Champassak), soils were lowin organic C and N and available P and K, whereasthe soil at the Vientiane site was generally morefertile (Table 6).

The objective of the 1998 study was to determinewhether residues, used in combination with N,improved N-use efficiency. The experiment was con-ducted in Vientiane and Saravane (KhongsedonDistrict) Provinces. The experimental design was asplit plot, with the treatments “with and without Nfertilizer” as the main plots and residue treatments(no residue control, manure at 2.6 and 5.2 t ha

–1

andrice husks at 1.3 t ha

–1

, dry weight basis) as subplots.Basal rates of P and K were applied to ensure thatthese nutrients were not limiting.

a

Khongsedon District.

b

Saravane District.

A similar study was conducted in 1999 in Cham-passak and Saravane (Saravane District) Provinces toevaluate the effect of residues on fertilizer-use

efficiency (as opposed to N-use efficiency). Theexperiment was a split plot design, with the treatments“with and without fertilizer” (60, 13, 18 kg ha

–1

of N,P and K, respectively) as the main plots, and residuetreatments (manure, rice husks and straw, each at 2.0 tha

–1

, dry weight basis) and a control as subplots.In the 1998 study, yields increased by 0.9 (AE =

15 kg kg

–1

) and 1.4 t ha

–1

(AE = 23 kg kg

–1

) inVientiane and Saravane, respectively, in response tofertilizer N alone (Figure 5). Yield increases fromresidues alone ranged from 12% to 35%, with theresponse to residues being greater in Vientiane thanin Saravane. On average, yields for the two sitesincreased by 0.3, 0.4 and 0.7 t ha

–1

in response to ricehusks and 2.6 and 5.2 t ha

–1

of FYM, respectively. InVientiane, the interaction between residues and N wasnot significant, suggesting that the benefits of residuesand N fertilizer were additive. However, in Saravane,a significant, but negative, interaction was found. Inthis case, if N fertilizer was already applied, thenapplying residues was of no benefit to grain yield.

In the 1999 study, applying fertilizer aloneincreased yields by 134% and 107% in Champassakand Saravane, respectively, while amendments ofresidues alone increased yields by about 50% at bothsites (Figure 6). The greater response to fertilizer andresidues in 1999, compared with 1998, is probablybecause the response was to a combination of N, Pand K versus only N in 1998. In Champassak, asignificant but negative interaction, similar to thatobserved in the 1998 Saravane study, was alsoobserved. In the 1999 Saravane study, the interactionbetween fertilizer and residue treatments waspositive. In this case, manure, applied with inorganicfertilizers, increased yields by 1.4 t ha

–1

, suggesting asynergistic benefit from manure + inorganic fertilizer.

These studies demonstrate that commonly avail-able on-farm residues applied alone and at realisticrates can result in yield increases of up to 50%. Inour study, the yield responses to residues are gener-ally higher than those reported from NortheastThailand. Supapoj et al. (1998) found that amend-ments of rice straw (6.25 to 18 t ha

–1

) and rice husks(3.13 t ha

–1

) increased rice yields by 10% to 15%(0.3 t ha

–1

, on average). In another study, Whitbreadet al. (1999a) reported small but significant yieldincreases (8%–10%) in response to returning ricestraw in two of the five years of their study. Finally,with the application of manure (6.25 t ha

–1

), Won-prasaid et al. (1996) reported that yields increased byup to 0.9 t ha

–1

. Our studies further demonstrate that modest rates

of inorganic fertilizer applied alone can result in yieldincreases of over 100%. These data contrast those ofWillett (1995) and Ragland and Boonpuckdee(1988), who reported that, on some sandy soils in

Table 6.

Soil properties of the sites for experimentsconducted in the 1998 and 1999 wet seasons, Laos, to deter-mine whether residues improve nitrogen- and fertilizer-useefficiency.

Year andSite

Soil texture

Organic C (%)

Kjeldahl N (%)

Olsen P (mg kg

–1

)

Exch. K (cmol kg

–1

)

1998Vientiane Loam 0.68 0.096 1.7 0.082Saravane

a

Silty loam

0.11 0.007 1.1 0.077

1999Champassak Sandy

loam0.13 0.028 1.1 0.035

Saravane

b

Silty loam

0.31 0.070 1.1 0.085

186



Northeast Thailand, the rice crop did not respond toinorganic fertilizers without organic amendmentsalso being applied. Furthermore, only in one case wasa positive interaction between a combined residueand fertilizer application found (Saravane, manure +fertilizer, Figure 6b). In fact, at two sites, a signifi-cant negative interaction was found (Figures 5b and6a), indicating that the immediate benefits of residueswere lost when inorganic fertilizer was applied.

Reasons for this discrepancy may be, first, fer-tilizers have been used in Thailand longer than inLaos. The use of primarily N and P fertilizers over along period, as in Thailand, may result in deficienciesof other nutrients (e.g. K or other micronutrients),which are available in organic amendments. Second,the soils evaluated in our study were coarse-texturedloams and loams, whereas Thai soils were sandy witha lower capacity for nutrient retention and buffering.However, as Table 4 and the above discussionssuggest, if N is applied as a split application, it isused relatively efficiently, regardless of soil texture.

These data indicate only the immediate effect ofresidues on rice yields. Long-term benefits ofrepeated residue applications are likely to be moresignificant. A strategy of applying only N and P overtime results in deficiencies of other nutrients not

available in these fertilizers (Dobermann et al. 1998).Long-term benefits of repeated residue applicationsmay therefore result in positive interactions if SOMincreases and the benefits from other nutrientsbesides N and P are realized. However, becauseincreasing SOM in cultivated soils, especially sandysoils, is difficult, it should not be assumed thatadding organic matter will always increase SOM.Even so, applying available on-farm residues is arecommendable strategy for enhancing the sustaina-bility of these systems, even though direct evidenceof their benefits is, at present, lacking.

Green manure

Green manure (GM) technology has often beenproposed as a means of alleviating N deficiencies inlow-input cropping systems and as a way ofimproving N conservation within the croppingsystem by capturing N that accumulates during thedry and early wet seasons (George et al. 1994).Several experiments have been conducted to evaluateGM in central and southern Laos. These are reportedelsewhere by Chanphengsay et al. (1999) andWhitbread et al. (1999b), but a summary of some oftheir work and that of others is presented here.

Figure 5.

Rice grain yield response to the application of inorganic nitrogen (N at 60 kg ha

−

1

) and on-farm organic residue(R; rice husks at 1.3 t ha

−

1

, and farmyard manure [FYM] at 2.6 or 5.2 t ha

−

1

), Laos, 1998 wet season. * = significant at

P

= 0.05; ** = significant at

P

= 0.01.

Gra

in y

ield

(kg

ha−1

)

Con

trol

(−N

, −R

)

Con

trol

(−N

, −R

)

5.2

t FY

M

2.6

t FY

M

2.6

t FY

M

5.2

t FY

M

Ric

e hu

sks

Ric

e hu

sks

Con

trol

(N a

lone

)

Con

trol

(N a

lone

)

5.2

t FY

M

2.6

t FY

M

Ric

e hu

sks

5.2

t FY

M

2.6

t FY

M

Ric

e hu

sks

5000

4500

4000

3500

3000

2500

2000

1500

1000

500

0

No N With N With NNo N

(a) Vientiane (b) Saravane (Khongsedon District)

N = *

R = **

N × R = ns

N = **

R = **

N × R = *

187



A 7-year study was conducted in Vientiane toevaluate the potential of various stem-nodulatinglegumes and grain legumes. The stem-nodulatinglegumes (

Sesbania

rostrata

,

S. aculeata andAeschynomene afraspera) performed consistentlywell over the years. The grain legumes evaluated didnot tolerate saturated soils but performed as well asthe stem-nodulating legumes when soils remainedunsaturated. The aboveground biomass of the stem-nodulating legumes averaged 2.6 t ha–1 and con-tained N at 60 to 80 kg ha–1. Becker et al. (1990)reported biomass yields of S. rostrata exceeding8 t ha–1 and N contents of over 150 kg ha–1. In Laos,however, early season drought often limits yields.

In most studies, the increase in rice yields inresponse to GM was equivalent to N supplied as ureaat 30 to 60 kg ha–1. Improved response to GM may bepossible by combining straw with the GM. Becker etal. (1994) reported that doing this slowed N mineral-ization, reducing N losses and improving N-useefficiency. In Laos, we evaluated the effect of addingstraw to S. rostrate and A. afraspera during GMincorporation on rice yields and N-recoveryefficiency (NRE = increase in N uptake per unit of Napplied). Nitrogen from GM (GM-N) provided theequivalent of 30 to 60 kg ha–1 of N as urea, which wasnot enough to meet rice requirements, as evidenced

by the significant increase in yield when N wasapplied at 90 kg ha–1 (Table 7).

a Within columns, means followed by the same letter do notdiffer significantly at the 0.05 probability level.b Straw was added at a rate of 1335 kg ha–1 to all strawtreatments. c NRE = nitrogen recovery efficiency, which is the increasein N uptake per unit of N applied.

The NRE for urea-N averaged 37%. Nitrogeninputs for both GM crops were similar and averaged

Table 7. Response of rice to inorganic and organic nitrogen(N) sources. Organic sources were Sesbania rostrata (S.r.),Aeschynomene aferaspera (A.a.) and rice straw.a

Treatmentb GM yield

Total N added

Rice yield N uptake

NREc

(%)

kg ha–1

0 N 0 1749 e 17.9 c30 N 30 2854 bc 30.0 bc 40.460 N 60 3074 b 37.6 ab 32.990 N 90 3664 a 51.6 a 37.4S.r. 1916 a 35.6 3093 b 34.3 bc 46.0S.r. + straw 1916 a 42.7 3017 b 37.0 b 44.6A.a. 1626 a 34.6 2351 cd 27.8 bc 28.6A.a. + straw 1626 a 41.7 3029 b 38.7 ab 49.9Straw 7.1 2197 de 26.6 bc 123.1

Figure 6. Rice grain yield response to on-farm residues (R; 2 t ha−1, dry weight basis), with and without inorganic fertilizer(F), Laos, 1999 wet season. The inorganic fertilizer was applied at a rate of 60-30-20 kg ha−1 of N, P2O5 and K2O, respec-tively. * = significant at P = 0.05; ** = significant at P = 0.01.

(a) Champassak (b) Saravane

Non

e(−

F, −

R)

FY

M

Str

aw

Ric

e hu

sks

Non

e

FY

M

Str

aw

Ric

e hu

sks

Non

e(−

F, −

R)

FY

M

Str

aw

Ric

e hu

sks

Non

e

FY

M

Str

aw

Ric

e hu

sks

5000

4500

4000

3500

3000

2500

2000

1500

1000

500

0

Gra

in y

ield

(kg

ha−

1 )

No fertilizer No fertilizerWith fertilizer With fertilizer

F = **

R = *

F × R = *

F = ***

R = ***

F × R = **

(F a

lone

)

(F a

lone

)




35 kg ha–1. Without straw, the NRE of S. rostrata-Nwas 46%, compared with only 29% for A. afraspera-N. Adding 1.5 t ha–1 of rice straw to A. afraspera, thebiomass before incorporation increased rice yields byabout 0.7 t ha–1 and the NRE from 29% to 50% (Table7). But this was not observed for S. rostrata. The dif-ference in the effect of straw on the two GMs is mostlikely a result of S. rostrata having a higher C:N ratiothan does A. afraspera (Becker et al. 1990). In thisstudy, N concentration (1.86%) in the S. rostrata bio-mass was lower than in A. afraspera (2.13%), whichsuggests a higher C:N ratio. Furthermore, C:N ratioswere probably higher than normal for both GMs inthis study because they were grown for 75 days,rather than for the recommended 60 days.

Optimal GM growth and biological N2 fixation isdependent on an adequate P supply. The above-ground dry weight of S. rostrata averages less than0.5 t ha–1 without P fertilizer on coarse-textured soils(Whitbread et al. 1999). Data from four experimentsconducted on coarse-textured soils indicate that P at20 kg ha–1 is required to optimize S. rostrata yieldson these soils (Figure 7). On similar soils, rice Prequirements were only 6.5 kg ha–1 (Figure 7), basedon results from a different set of four P-rate experi-ments conducted in the same region as the S. ros-trata studies. The higher P requirements of GMs,relative to rice, is also reported by Ventura andLadha (1997), and may be caused by P being lessavailable under aerobic conditions, which are morecommon during the GM growth period (start of wetseason), than under anaerobic conditions, which pre-vail during most of the rice-growing season (Mahap-atra and Patrick 1969; Willett 1986). Legumes, inany case, tend to have higher P requirements than doother crops because of the energy requirements forbiological N2 fixation. Given the high P require-ments of GMs, relative to rice on these soils, GMsmay not be economically attractive to resource-poorfarmers.

Apart from the high P requirements, farmers citeother constraints for not adopting GMs, includinghigh labour demand during peak periods such astransplanting; difficulties in land preparation beforethe heavy rains (especially for farmers who usebuffalo); and difficulties in purchasing or producingseed. These constraints are consistent with Pandeyand Sanamongkhoun’s hypothesis (1998) that tech-nologies that increase labour requirements areunlikely to be accepted in places like Laos. Further-more, in southern Laos, fertilizers are readily avail-able. While the widespread adoption of GMtechnology in southern Laos is doubtful, it may havea role in the rainfed lowland systems of northern Laoswhere fertilizers are less accessible and P deficienciesare not as prevalent.

Figure 7. Relative yield (relative to maximum yield) of thelegume Sesbania rostrata ( ) and rice (�) in response tophosphorus (P) on coarse-textured soils, Laos. Data repre-sent means of four experiments. Error bars represent onestandard deviation.

Conclusions

Despite the relative efficiency of applied N whendrought, floods and pests do not lower yields, scopestill remains for improving N-use efficiency at thefarm level. First, limited N resources should beapplied to improved varieties, which are moreresponsive to inputs. Second, timing N supply withthe crop’s demand for N should be improved toensure a greater proportion of N during the vegeta-tive stage when N demand is greatest. Third, whileapplying organic amendments does not improveinorganic fertilizer use efficiency in the short term,recycling of crop residues should, in the long-term,improve N-use efficiency, as nutrients removed inthese residues become limiting and weaken responseto N. Finally, under suboptimal N conditions, con-siderable flexibility exists for varying the timing ofN application around transplanting and PI withoutdecreasing N-use efficiency or yield. While responseto pre-rice GMs is relatively good, the relatively highP requirements of the latter make the technology lessattractive to farmers.

References

Becker, M., Ladha, J.K. and Ottow, J.C.G. 1990. Growthand N2-fixation of two stem-nodulating legumes andtheir effects as green manure on lowland rice. SoilBiology and Biochemistry, 22, 1109–1119.

25 3020151050

P application rate (kg ha−1)

Rel

ativ

e yi

eld

on y

axi

s

1.0

0.8

0.6

0.4

0.2

0.0




Becker, M., Ladha, J.K., Simpson, I.C. and Ottow, J.C.G.1994. Parameters affecting residue nitrogen mineraliza-tion in flooded soils. Soil Science Society of AmericaJournal, 58, 1666–1671.

Cassman, K.G., Gines, G.C., Dizon, M.A., Samson, M.I. andAlcantara, J.M. 1996a. Nitrogen-use efficiency in tropicallowland rice systems: contributions from indigenous andapplied nitrogen. Field Crops Research, 47, 1–12.

Cassman, K.G., Dobermann, A., Sta. Cruz, P.C., Gines,G.C., Samson, M.I., Descalsota, J.P., Alcantara, J.M.,Dizon, M.A. and Olk, D.C. 1996b. Soil organic matterand the indigenous nitrogen supply of intensive irrigatedrice systems in the tropics. Plant and Soil, 182, 267–278.

Chanphengsay, M., Whitbread, A., Schiller, J., Blair, G.,Linquist, B.A., Lathvilayvong, P. and Sengxua, P. 1999.Soil fertility decline in Lao PDR and the potential of pre-rice green manures to improve the sustainability of riceproduction systems. In: Whitbread, A.M. and Blair, G.J.ed. Integrated nutrient management in farming systemsin Southeast Asia and Australia. Vientiane, Laos,ACIAR Proceedings No. 93.

De Datta, S.K. 1981. Principles and Practices of Rice Pro-duction. John Wiley, New York.

De Datta, S.K., Buresh, R.J., Samson, M.I. and Wang, K.-R.1988. Nitrogen use efficiency and nitrogen-15 balancesin broadcast flooded and transplanted rice. Soil ScienceSociety of America Journal, 52, 849–855.

Dei, Y. and Yamasaki, S. 1979. Effect of water and cropmanagement on the nitrogen-supplying capacity of paddysoils. In: Nitrogen and Rice. Los Baños, Philippines,International Rice Research Institute (IRRI), 451–463.

Dobermann, A. and Fairhurst, T. 2000. Rice: Nutrient Dis-orders and Nutrient Management. Singapore, Potash &Phosphate Institute (PPI), Potash & Phosphate Instituteof Canada (PPIC), and International Rice ResearchInstitute (IRRI), 191 p.

Dobermann, A., Cassman K.G., Mamaril, C.P. and Sheehy,J.E. 1998. Management of phosphorus, potassium andsulfur in intensive, irrigated lowland rice. Field CropsResearch, 56, 113–138.

Fukai, S., Sittisuang P. and Chanphengsay, M. 1998.Increasing production of rainfed lowland rice in droughtprone environments: A case study in Thailand and Laos.Plant Production Science, 1, 75–82.

George, T., Ladha, J.K., Garrity, D.P. and Buresh, R.J.1994. Legumes as nitrate catch crops during the dry-to-wet transition in lowland rice cropping systems.Agronomy Journal, 86, 267–273.

IRRI (International Rice Research Institute). 1995. WorldRice Statistics, 1993–1994. Los Baños, Philippines.

Katyal, J.C., Singh, B., Vlek, P.L.G. and Craswell, E.T.1985. Fate and efficiency of nitrogen fertilizers appliedto wetland rice. II. Punjab, India. Fertilizer Research, 6,279–290.

Linquist, B.A., Sengxua, P., Whitbread, A., Schiller, J. andLathvilayvong, P. 1998. Evaluating nutrient deficienciesand management strategies for lowland rice in Lao PDR.In: Ladha, J.K., Wade, L.J., Dobermann, A., Reichardt,W., Kirk, G.J.D. and Piggin, C. ed. Rainfed Lowland

Rice: Advances in Nutrient Management Research; Pro-ceedings of the International Workshop on NutrientResearch in Rainfed Lowlands, 12–15 Oct. 1998, UbonRatchatathani, Thailand. International Rice ResearchInstitute. p 59-73.

Linquist, B.A., Sengxua, P., Whitbread, A. and Chang-pengsay. 1999. The potential of on-farm residues forimproving rainfed lowland rice productivity. In:Whitbread, A.M. and Blair, G.J. ed. Integrated NutrientManagement in Farming Systems in Southeast Asia andAustralia. Vientiane, Laos, ACIAR Proceedings No. 93.

Mahapatra, I.C. and Patrick, Jr., W.H. 1969. Inorganicphosphate transformation in waterlogged soils. SoilScience, 107, 281–288.

Obcemea, W.N., De Datta, S.K. and Broadbent, F.E. 1984.Movement and distribution of fertilizer nitrogen asaffected by depth of placement in wetland rice. FertilizerResearch, 5, 125–148.

Pandy, S. and Sanamongkhoun, M. 1998. Rainfed LowlandRice in Laos: A Socio-Economic Benchmark Study. LosBaños, Philippines, International Rice Research Institute(IRRI).

Prasad, R. and De Datta, S.K. 1979. Increasing fertilizernitrogen efficiency in wetland rice. In: Nitrogen andRice. Los Baños, Philippines, International RiceResearch Institute (IRRI), 465–484.

Ragland, J. and Boonpuckdee, L. 1988. Fertilizer responsesin NE Thailand. 3. Nitrogen use and soil acidity. ThaiJournal of Soils Fertility, 10, 67–76.

Schnier, H.F. 1995. Significance of timing and method ofN fertilizer application for the N-use efficiency inflooded tropical rice. Fertilizer Research, 42, 129–138.

Schnier, H.F., De Datta, S.K. and Mengel, K. 1987.Dynamics of 15N labelled ammonium sulfate in variousinorganic and organic soil fractions of wetland rice soils.Biology and Fertility of Soils, 4, 171–177.

Supapoj, N., Naklang, K. and Konboon, Y. 1998. Usingorganic material to improve soil productivity in ranfedlowland rice in Northeast Thailand. In: Ladha, J.K.,Wade, L.J., Dobermann, A., Reichardt, W., Kirk, G.J.D.and Piggin, C. ed. Rainfed Lowland Rice: Advances inNutrient Management Research; Proceedings of theInternational Workshop on Nutrient Research in RainfedLowlands, 12–15 Oct. 1998, Ubon Ratchathani, Thai-land. International Rice Research Institute. p 161–168.

UNDP (United Nations Development Programme). 1998.Development Cooperation Report 1997—Lao People’sDemocratic Republic. Vientiane, Laos 159 p.

Ventura, W. and Ladha, J.K. 1997. Sesbania phosphorusrequirements when used as a biofertilizer for long-termrice cultivation. Soil Science Society of America Journal,61, 1240–1244.

Wade, L.J., Amarante, S.T., Olea, A., Harnpichtvitaya, D.,Naklang, K., Wihardjaka, A., Senegar, S.S., Mazid,M.A., Songh, G. and McLaren, C.G. 1999. Nutrientrequirements in rainfed lowland rice. Field CropsResearch, 64, 91–107.

Whitbread, A., Blair, G.J., Naklang, K., Lefroy, R.,Wonprasaid, S., Konboon, Y. and Suriya-Arunroj, D.1999a. The management of rice straw, fertilizers and leaflitters in rice cropping systems in Northeast Thailand.




2. Rice yields and nutrient balances. Plant and Soil,209, 29–36.

Whitbread, A.M., Chanphengsay, M., Linquist, B.A.,Sengxua, P., Blair, G.J., Vongsouthi, S. and Phengsoun-anna, V. 1999b. Sesbania rostrata as a green manure andphosphorus management for lowland rice production inLao P.D.R. In: Whitbread, A.M. and Blair, G.J. ed. Inte-grated Nutrient Management in Farming Systems inSoutheast Asia and Australia. Vientiane, Laos, ACIARProceedings No. 93.

Willett, I.R. 1986. Phosphorus dynamics in relation toredox processes in flooded soils. In: Transactions of the13th Congress of the International Society of Soil

Science, vol. 6. Hamburg, International Society of SoilScience, 748–755.

Willett, I.R. 1995. Role of organic matter in controllingchemical properties and fertility of sandy soils used forlowland rice in Northeast Thailand. In: Lefroy, R.D.B.,Blair, G.J. and Craswell, E.T. ed. Soil organic mattermanagement for sustainable agriculture: 24–26 August1994. Ubon, Thailand, ACIAR Proceedings No. 56,163 p.

Wonprasaid, S., Khunthasuvon, S., Sittisuang, P. andFukai, S. 1996. Performance of contrasting rice cultivarsselected for rainfed lowland conditions in relation to soilfertility and water availability. Field Crops Research, 47,267–275.

191



Genotypic Performance of Rainfed Lowland Rice under Different Fertilizer Conditions in Laos

P. Inthapanya

1

, Sipaseuth

1

, P. Siyavong

2

, V. Sihathep

3

, M. Chanphengsay

4

, S. Fukai

5*

and J. Basnayake

5

Abstract

Low soil fertility is a major constraint to increasing rice productivity under rainfed lowlandconditions in Laos. Even though farmers generally apply low fertilizer rates to rainfed lowlandrice, the Lao breeding program selects genotypes under higher, although moderate, rates (60 N,30 P

2

O

5

and 20 K

2

O kg ha

–1

). A combined analysis of results from a series of fertilizerexperiments conducted over 3 years at three locations revealed that the interactions of genotype-by-location (G

×

L), genotype-by-fertilizer (G

×

F) and genotype-by-location-by-fertilizer(G

×

L

×

F) had significant effects on grain yield. However, variance components for genotypeand the G

×

L interaction were larger than those for the G

×

F and G

×

L

×

F interactions. Becausethe G

×

F interaction had a small effect, the ranking of lines across non-fertilized and fertilizedconditions was consistent over the 3 years. Early selection of better performing lines under theprogram’s current fertilizer application rates in a multi-locational testing program could producebetter cultivars for the generally poor soils of Laos. The percentages of N and P in seeds were alsofound to be negatively associated with grain yield among the lines studied. Nitrogen and P useefficiencies (g dry matter g

–1

nutrient) in lines were consistent across varying fertilizer conditions.Selecting lines for low seed N concentration and high N and P use efficiencies can thereforecomprise an alternative approach for selecting high-yielding cultivars for Laos.

T

HE

soils in which most rainfed lowland rice in Laosis cultivated are inherently of low fertility. Systemicstudies, conducted since 1991, suggest that low soilfertility is a major constraint to rice production inrainfed lowlands (Lathvilayvong et al. 1997; Linquistet al. 1999). Nitrogen and phosphorus are the mostimportant nutrients that limit grain yields in mostareas of Laos. Lathvilayvong et al. (1997) suggestthat an initial application of P at 6.5–19 kg ha

–1

isneeded for high grain yield in most areas. Currentfertilizer recommendations in Laos range from 60 to

90 kg ha

–1

for N and from 13 to 19 kg ha

–1

for P(Linquist et al. 1999).

Breeding programs for rainfed lowland riceusually use higher rates of fertilizer application thanis commonly used by Lao farmers. For example, theLao breeding and selection program uses the rates of60, 13 and 16 kg ha

–1

for N, P and K, respectively.The program also uses grain yield to select advancedmaterials, and use of the appropriate fertilizer ratemay be critical in this phase of breeding. Most workon responses to fertilizer by rainfed lowland ricecultivars used only a few cultivars, making it diffi-cult to judge the impact of the genotype-by-fertilizer(G

×

F) interaction on the efficiency of a breedingprogram in developing new cultivars suitable for dif-ferent soil fertility conditions (Romyen et al. 1998).

The varying ability of genotypes to use nutrientsmay cause variation in grain yield among genotypesunder different soil fertility conditions. It is impor-tant to identify whether the nutrient use efficiencies

1

National Agriculture Research Center, Vientiane, Laos

2

Agriculture and Forestry Office, Savannakhet, Laos

3

Agriculture and Forestry Office, Champassak, Laos

4

National Agriculture and Forestry Research Institute,Vientiane, Laos

5

The University of Queensland, Brisbane, Qld., Australia*Corresponding author: E-mail: s.fukai@mailbox. uq.edu.au

KEYWORDS:

Fertilizer, Genotype-by-fertilizer interaction, Nutrient use efficiency, Rainfed lowland rice

192



(g dry matter g

–1

nutrient) of those genotypes areconsistent across different growing conditions. Thisinformation would help in the understanding of theadaptation mechanisms of different genotypes fordifferent fertilizer environments. Morphological andphysiological differences in those cultivars can bestudied to identify the traits that have high associa-tion with nutrient use efficiency.

This work aimed to:1. Quantify the magnitude of the G

×

F interactionfor grain yield of rainfed lowland rice, using alarge number of genotypes that are typical of theLao breeding program. This should help identifythe appropriate level of fertilizer to be used by thebreeding program to maximize its efficiency indeveloping new cultivars.

2. Investigate the importance of nutrient uptake andnutrient use efficiency in determining the variationfor grain yield under different fertility conditions.

3. Investigate the effects of N and P on the grainyield of different genotypes in three locations. Because detailed results of the experiments con-

ducted in 1996 and 1997 are described in Inthapanyaet al. (2000a and 2000b), this paper only summarizesthe results for those years, and reports on the mainfeatures of the 1998 experiments.


Three sets of rainfed lowland rice experiments wereconducted on farm fields in the Vientiane, Savanna-khet and Champassak Provinces of Laos in 1996,1997 and 1998. The chemical and physical proper-ties of the soils at the three locations are presented inTable 1. First, two experiments in 1996 and 1997were conducted, using 72 and 60 genotypes, respec-tively, under fertilized and non-fertilized conditions.For the 1996 experiments, the set of 72 genotypescomprised 20 lines selected from a cross betweenRD6 and IR46331-PMI-32-2-1-1, 32 lines from across between RD6 and IR49801-UBN-7-B-1-4-1,11 promising lines from Laos, 7 promising lines andimproved cultivars from Thailand and 2 traditionalThai and Lao cultivars. Chemical fertilizer wasapplied at the rate of 60, 13 and 16 kg ha

–1

of N, Pand K, respectively. The N was applied in two split-levels at planting and 50 days after transplanting.

For the 1997 experiment, 13 high-yielding geno-types were selected from the 1996 population and 47promising lines were taken from the Lao selectionprogram. Most genotypes were of the glutinous graintype. In 1998, a set of 12 lines was taken for furtherstudy of their yield response to four levels of N andP fertilizers. The levels were zero fertilizer, N only(at 60 kg ha

–1

), P only (at 13 kg ha

–1

) and NPK (at60 N + 13 P + 16 K kg ha

–1

).

Source: Inthapanya et al. (2000a).

Plot sizes differed across the 3 years. In 1996, tworows of 3-m plots were used with two replicates,whereas, in 1997, five rows of 3-m plots were usedwith two replicates. Plot size for 1998 experimentswas 1.25

×

3 m (five rows) with three replicates.Seeds were sown in June and transplanted in July inmost of the experiments.

Free water levels, both above and below theground, were measured, using 50

×

10 cm PVC pipesinserted 40 cm deep into the field. Date of floweringwas recorded when 75% of the panicles hademerged, and plots were harvested as they matured.Plant height was recorded a few days beforeharvesting. Grain yield was determined after harvestfor the whole plot in 1996, the middle three rows in1997 and the whole plot in 1998. Moisture contentwas measured and grain yield adjusted to 14%moisture content.

In the 1997 experiment, a subset of 16 genotypeswas used to estimate the N, P and K concentration inseeds and straw separately. Samples were alsocollected from 12 genotypes in four fertilizer treat-ments for nutrient analysis in 1998. The N, P and Kanalyses were conducted as described by Inthapanyaet al. (2000b). Nutrient contents in grain and strawwere estimated from the tissue nutrient concentrationand respective dry weight of seeds and straw.

Each experiment was statistically analysedseparately to identify the genotype, fertilizer andtheir interactions in each environment. Combinedanalyses were conducted across three locations for3 years. The variance component for each source ofvariation was estimated for these experiments.

Nutrient use efficiencies (NUE) were estimatedfor seed, straw and total plant for each of N, P and K,using the grain and total dry matter (TDM) at

Table 1.

Chemical and physical properties of soils atthree locations in the Vientiane (V), Savannakhet (S) andChampassak (C) Provinces, Laos.

Soil property V S C

Soil propertypH (H

2

O)Organic matter (%)Total N (%)Available phosphorus (

µ

g g

–1

) (Bray 2)Available potassium (

µ

g K

2

O g

–1

)

5.30.5—7.5

24.0

5.40.4—9.3

44.0

5.50.50.04

17.0

108.0Soil texture

Sand (%)Silt (%)Clay (%)

66.129.2

4.5

86.29.34.6

67.427.2

5.3

193



harvest. Nitrogen-use efficiencies for the plant(N

i

UEp) and grain (N

i

UEg) are defined as:

N

i

UEp = TDM / N plantN

i

UEg = Yield / N plantwhere,

N plant = N content of plant

Similarly, P use efficiency for plant and grain(PUEp and PUEg, respectively) and K use efficiencyfor plant and grain (KUEp and KUEg, respectively)was estimated.

Results

Variance components in G

×

F interaction analysis

Results of analyses of variance, variance componentsand proportion of variance components for grain yieldin 1996, 1997 and 1998 are shown in Table 2. In1996, when yield was particularly low at Vientiane,the location (L) variance was large, accounting formore than 60% of the total variance. The fertilizervariance (F) was the second largest (17.9%), andL

×

F the third. Genotype (G) effect was significant(

P

< 0.05), and the G variance component (1.1%) wasgreater than the G

×

F component. Other two-wayinteractions and the three-way interaction were alsosignificant. The sum of variances involving geno-types (G, G

×

L, G

×

F and G

×

L

×

F) was 6.8% andthe G

×

L component was the largest (2.7%) amongthe four variances. The sum of genotypic variancesthat did not involve fertilizer (G and G

×

L) wasgreater than that involving fertilizer (G

×

F and G

×

L

×

F).In 1997, the L variance was smaller than the F

variance. Similar to 1996, however, the effects ofcomponents G, G

×

L, G

×

F and G

×

L

×

F were allsignificant, and the G variance was greater than theG

×

F variance. The sum of four variances thatinvolved genotypes (11.9%) and also the G varianceitself were greater in 1997 than in 1996, but as for1996, the sum of variances for G and G

×

L wasgreater than that involving fertilizer.

As in 1997, the L variance was smaller than theF variance in 1998. The effect of G

×

F interactionwas not significant, while the effects of G, G

×

L andG

×

L

×

F were significant. The G and G

×

Lvariances had similar effects and were greater thanthe G

×

F variance. The sum of four variances thatinvolved genotypes (19.4%) and also the G varianceitself were greater in 1997 than in 1996. In all3 years, results were consistent in that the sum ofvariances for G and G

×

L was greater than thoseinvolving fertilizer.

Combined year analyses for the 13 common geno-types in 1996 and 1997 showed that the G variance

(1.3%) was greater than the G

×

F variance (0.8%)(data not shown). Among the interaction componentsinvolving genotypes, the G

×

year interaction vari-ance (1.1%) was the largest, followed by the G

×

Land G

×

F variances.

a

Data for year 1997 were taken from Inthapanya et al.(2000a).* =

P

< 0.05; ** =

P

< 0.01; ns = not significant.

Grain yield in different experiments

Mean grain yields of all genotypes with or withoutfertilizer application at three locations in 3 years areshown in Table 3. In all nine experiments (3 years

×

3 locations), there was a positive effect of fertilizerapplication, although the effect was not significant inVientiane in 1997 when analysed as a single experi-ment. The low yield at this location in 1996 was theresult of a nutrient disorder. Yield was lower at theSavannakhet location than in Champassak over the3 years because of early season drought in 1996 andlate drought in 1997 and 1998.

Table 2.

F

-ratios and significance of each source ofvariation, estimated variance components with theirapproximate standard errors and proportion of variancecomponent from the combined analysis of variance forgrain yield (kg ha

–1

) in genotype-by-fertilizer interactionexperiments conducted with rainfed lowland rice genotypesat three locations in Laos in 1996, 1997 and 1998.

YearSource of variation

F

-ratio andthe level of significance

Variancecomponent

×

10

4

Proportion of variance component

(%)

1996Location (L) 529.78 ** 73.8

±

7.9 61.5Fertilizer (F) 838.27 ** 21.6

±

3.5 17.9L

×

F 101.33 ** 8.7

±

0.9 7.3Genotype (G) 1.76 * 1.3

±

0.7 1.1G

×

L 2.00 * 3.3

±

0.8 2.7G

×

F 1.33 * 0.7

±

0.5 0.6G

×

L

×

F 1.76 * 2.8

±

0.8 2.41997

a

Location (L) 13.46 ** 13.5

±

1.6 16.1Fertilizer (F) 65.37 ** 36.3

±

53.8 43.1L

×

F 2.95 ns 3.4

±

5.2 4.0Genotype (G) 7.47 ** 5.4

±

1.5 6.5G

×

L 1.48 * 1.5

±

0.7 1.8G

×

F 1.83 * 1.7

±

0.8 2.1G

×

L

×

F 1.26 * 1.3

±

0.8 1.51998

Location (L) 52.19 ** 8.0

±

8.9 26.4Fertilizer (F) 16.54 ** 14.2

±

12.4 47.0L

×

F 4.82 ** 2.1

±

1.6 6.9Genotype (G) 3.26 ** 2.4

±

1.6 8.2G

×

L 3.72 ** 2.4

±

1.0 8.1G

×

F 1.08 ns 0.1

± 0.3 0.2G × L × F 1.34 * 0.9 ± 0.6 2.9




When grain yields at the three locations were com-bined for the 1996 experiment, the coefficient ofdetermination for yield of genotypes between the twofertilizer conditions was 0.54 (Figure 1a). However,in 1997, the coefficient was lower (0.45) where a sig-nificant G × F interaction was obtained (Figure 1b).Some high-yielding genotypes differed in theirresponses to fertilizer application: genotypes with amean yield between 1600 and 1900 kg ha–1 undernon-fertilized conditions, had yields that varied

greatly between 2200 and 3300 kg ha–1 when fer-tilizer was applied.

The 10 highest yielding lines obtained from thethree locations and two fertilizer levels for 1996 and1997 are shown in Table 4. Some of these lines (e.g.TDK1 and IR57514-PMI-5-B-1-2) were often in thetop 10 lines for each of the six experiments. Allhigh-yielding entries in 1996 were examined in1997. Line IR68102-TDK-B-B-31-3 produced thehighest yield (2462 kg ha–1), out-yielding TDK1(2414 kg ha–1) in 1997. Other high-yielding lineswere IR68105-TDK-B-B-6-1 (2330 kg ha–1) andIRUBN-4-TDK-1-2-1. In 1997, ranking was oftensimilar between the non-fertilized and fertilized con-ditions at each location, whereas differences inranking were larger in 1996. In the same year,ranking in Vientiane differed from that of other loca-tions, because of a problem with nutrient disorder,possibly Fe toxicity in the soil.

Nutrient content and nutrient use efficiency

A subset of 16 lines was used to analyse the total N,P and K content in seeds and straw in the 1997experiment. Results of combined analyses of vari-ance for the three locations for total N and P contentsand nutrient use efficiency in grain (NUEg) andplant (NUEp) for N and P are shown in Table 5(Inthapanya et al. 2000b). A highly significant effectof genotype (G) was obtained for total contents of Nand P. However, the G × F interaction effect wassignificant only for N content.

Genotypic variation in grain yield for a given totalN or P content was found. Thus, the ratios of grainyield and total dry matter to total N and P contents(NiUEg, NiUEp, PUEg and PUEp) differed signifi-cantly among genotypes (Table 5). While no signifi-cant interactions of G × F were found for theseattributes, G × L × F interaction was significant forPUEg and PUEp. While high-yielding lines such as

Table 3. Average grain yield (kg ha–1) of rice lines grown at three locations in the Vientiane (V), Savannakhet (S) andChampassak (C) Provinces, Laos, under non-fertilized (NF) and fertilized (F) conditions, 1996, 1997 and 1998.

Province Year

1996 1997 1998

NF F Mean NF F Mean NF F Mean

V 488 823 655 1405 1957 1681 1119 1937 1528S 1205 1811 1508 1217 2091 1619 1348 2061 1705C 1845 3008 2426 1774 2966 2370 1445 2664 2055

Mean 1179 1880 1530 1465 2314 1890 1304 2206 1755

LSD 5% (location) 426 420 436LSD 5% (fertilizer) 427 421 441

3500

3000

2500

2000

1500

1000

3500

3000

2500

2000

1500

1000

Yield under non-fertilized conditions (kg ha−1)

Yie

ld u

nder

fert

ilize

d co

nditi

ons

(kg

ha−1

)

(a) 1996

(b) 1997

600 800 1000 1200 1400 1600 1800 2000

R2 = 0.54y = 667.9 + 1.03x

R2 = 0.45y = 793.5 + 1.05x

Figure 1. Yield performance of rainfed lowland rice in(a) 1996, with 72 lines, and (b) 1997, with 60 lines, undernon-fertilized and fertilized conditions in Laos. Each pointrepresents the mean across three locations (after Inthapanyaet al. 2000a).




IR68102-TDK-B-B-31-3, TDK1, IRUBN-4-TDK-1-2-1, IR57514-PMI-5-B-1-2 and IR57514-SRN-299-2-1-1 generally had high NiUEg and PUEg, they didnot have high NiUEp and PUEp, except for PUEp inTDK1. These high-yielding lines generally had lowgrain N concentration.

The effect of fertilizer application on nutrientconcentration was generally small, except forP concentration in straw. The N concentration in

grain was 2–3 times greater than in straw, whilethe ratio was higher for P (data not shown). Incontrast, K concentration in straw was muchhigher than in grain. Nitrogren concentrationin grain was negatively related to NiUEg atthe Savannakhet and Champassak locations(Figure 2). At Vientiane, NiUEg was generallyhigh, even in genotypes with high N concentra-tion in grain.

Source: Inthapanya et al. (2000a).

Source: Inthapanya et al. (2000b).

Table 4. Ranking by yield of 10 rainfed lowland rice genotypes that produced the highest mean yields for all experimentsin three Lao provinces under non-fertilized (NF) and fertilized (F) conditions in 1996 and 1997.

YearLine

Province Grain yield (kg ha–1)

Vientiane Savannakhet Champassak

NF F NF F NF F

1996IR57514-PMI-5-B-1-2NSG19IR64345-TDK-148-9-2TDK1IR66368-CPA-32-P1-3RIR64906-TDK-249-10-11IR66369-CPA-51-P1-3RIR64906-TDK-15-B-1-1IR66369-CPA-39-P1-3RIR68104-CPS-6-1-2

4371457

318

92

3055

17262462

537

8114554

548

269

1834381249

53

2112

501120

857

19

2510131223

811

5

1268

474

143323

5

2187190118371827182317441741173417191687

1997IR68102-TDK-B-B-31-3TDK1IR68105-TDK-B-B-6-1IRUBN-4-TDK-1-2-1IR66488-TDK-25-1-1-1IR57514-PMI-5-B-1-2IR57514-SRN-299-2-1-1IR68102-TDK-B-B-28-1Dok-mai IR68105-TDK-B-B-27-1

762

1043

123826

8

2112713173823

81410

2111

81210

317

69

14

127589

20112619

123

15294517

119

644

1423569

131611

2462241423302313224022282213218321702153

Table 5. Degrees of freedom (DF), F-ratio and the level of significance for total N and P content and N and P use efficiencies(NiUE and PUE, respectively) of grain (g) and plant (p) of 16 rainfed lowland rice lines for the combined analysis of varianceacross three locations in Laos.

Source of variation

DF Total N content

Total P content

NiUEg NiUEp PUEg PUEp

Location (L)Fertilizer (F)

L × F

2, 61, 62, 6

12.38 **28.76 **

5.88 *

9.19 **16.79 **

3.83 ns

8.01 **8.97 **2.63 ns

14.15 **3.47 ns2.79 ns

26.70 **16.01 **

8.17 **

9.97 **7.44 **8.23 **

Genotype (G)G × LG × FG × L × F

15, 3030, 3015, 3030, 192

4.35 **1.37 ns2.29 *1.00 ns

3.25 **1.00 ns1.34 ns1.27 ns

5.98 **1.96 *1.06 ns1.01 ns

3.07 **1.15 ns1.66 ns1.19 ns

3.85 **1.20 ns2.00 ns2.12 **

2.25 *1.12 ns1.00 ns2.24 **




Figure 2. Relationship between grain nitrogen concentra-tion and nitrogen-use efficiency (NiUE) in grain for 16rainfed lowland rice lines grown across fertilized and non-fertilized conditions at three locations in Laos. Note thatonly the Savannakhet and Champassak data were used forthe regression (after Inthapanya et al. 2000b).

Because a significant G × L × F interaction forPUEg existed, the relationship of grain yield withtotal P content at each location and fertilizer level wasexamined for each line. Results for six contrastinglines in 1997 are shown in Figure 3 (Inthapanya et al.2000b). The linear regression shown for each line has

a positive intercept, indicating PUEg generallydecreases with increase in P content in the plant. Theslope of the line connecting each location/fertilizerpoint and the origin is the PUEg.

In Vientiane, some lines had low PUEg underfertilized conditions while others did not, causing theG × L × F interaction for PUEg. TDK1 had highPUEg, except in Vientiane under fertilized con-ditions (+F). PUEg was also lower in Vientiane +Ffor IR57514-PMI-5-B-1-2, but other lines did nothave such decreased levels of PUEg in Vientiane +F.Lines IR66368-CPA-11-P1-3R-0, IR66369-CPA-63-P1-3R-0 and IR58821/IR52561/CA-11 had consist-ently high, medium and low PUEg, respectively.Line IR68102-TDK-B-B-31-3 had lower PUEg thanTDK 1, but total P content was higher for the sametreatment.

Genotype response to nitrogen and phosphorus applications

Mean grain yield, total dry matter, harvest index andtotal N, P and K content under four nutrient treat-ments at the three locations in 1998 are shown inTable 6. In all locations, grain yield increased withapplications of N and P. The effect of N was greaterthan that of P in Vientiane and Champassak, whereasthe effect of N and P were similar in Savannakhet.Applications of N, P and K, however, resulted in the

100

90

80

70

60

50

40

Grain N concentration (%)

NiU

E (

g gr

ain

g−1N

)

0.70 1.100.75 0.80 0.85 0.90 0.95 1.00 1.05

Vientiane Savannakhet Champassak

R2 = 0.37y = 396.3 − 692.7x + 361.5x2

*Values followed by the same letter are not significantly different at the 5% probability level.

Table 6. Mean yield, total dry matter (TDM), harvest index and total N, P and K contents of rainfed lowland rice underfour nutrient treatments in the Champassak, Savannakhet and Vientiane Provinces, Laos, 1998.

LocationTreatment

Yield(kg ha–1)

TDM(kg ha–1)

Harvestindex

Total N(kg ha–1)

Total P(kg ha–1)

Total K(kg ha–1)

VientianeNil 1119 d* 4094 d 0.27 20.5 b 2.49 c 3.57 c+N 1754 b 6268 b 0.28 34.3 a 2.54 c 5.83 a+P 1386 c 5523 c 0.25 24.6 b 3.37 b 4.71 b+NPK 1937 a 7062 a 0.27 36.1 a 4.12 a 5.85 a

Mean 1549 5737 0.27 28.9 3.13 4.99LSD 5% 142 529 ns 5.09 0.61 0.70Savannakhet

Nil 1348 c 2791 c 0.48 a 14.81 c 2.10 c 2.87 c+N 1662 b 3965 b 0.42 bc 21.66 b 2.29 c 4.23 b+P 1737 b 3901 b 0.44 b 20.82 b 3.87 b 4.00 b+NPK 2061 a 5092 a 0.41 c 28.25 a 4.88 a 5.50 a

Mean 1702 3937 0.44 21.38 3.28 4.15LSD 5% 204 528 0.02 3.21 0.68 0.67Champassak

Nil 1445 d 3847 d 0.38 b 21.17 d 2.73 c 1.03 d+N 2429 b 6406 b 0.38 b 35.62 b 4.24 b 2.32 a+P 1966 c 4881 c 0.40 a 27.47 c 4.23 b 1.65 c+NPK 2664 a 6973 a 0.38 b 40.09 a 5.80 a 1.94 b

Mean 2126 5527 0.38 31.09 4.25 1.73LSD 5% 125 378 0.01 2.48 0.51 0.26




highest yield (P < 0.05) at all locations. Total N con-tent followed the same pattern as grain yield. How-ever, total P content responded more to P applicationand less to N application, except in Champassak,where N application promoted P uptake.

Grain yield and total dry matter were stronglyrelated to total N, P and K content (Figure 4). The R2

for total N, P and K content were higher for total drymatter than grain yield. Generally, grain yield andtotal dry matter were more strongly associated withtotal N content than with total P and K contents.

Discussion

The G × F interaction was often significant with asmall contribution to the total variation for grainyield in each experiment in 1996 and 1997. How-ever, the combined analysis showed that the variancecomponent of G was greater than the G × F variancecomponent. The sum of the G and G × L variancecomponents was greater than that of the G × F andG × L × F components. This was particularlyobvious in 1997 when adverse soil effects inVientiane was not as severe as in 1996. Thus, in

Figure 3. Relationship between grain yield and total P content for six rainfed lowland rice lines grown under non-fertilizedand fertilized conditions at three locations in Laos, 1997. V = Vientiane; S = Savannakhet; C = Champassak; + = fertilizedconditions; − = non-fertilized conditions. (After Inthapanya et al. 2000b.)

TDK1 IR57514-PMI-5-B-1-2

IR66368-CPA-11-P1-3R-0 IR66369-CPA-63-P1-3R-0

IR58821/IR52561/CA-11 IR68102-TDK-B-B-31-3

4000

3000

2000

1000

0

Gra

in y

ield

(kg

ha−1

)

4000

3000

2000

1000

0

250 5 10 15 20 250 5 10 15 20

4000

3000

2000

1000

0

Total P content (kg ha−1)

C+

V+

S+

C−

V−S−

C+

V+

S+

C− V−S−

C+

V+S+

C−V−

S−

C+

V+

S+C−

V−S−

C+V+S+C−V−S−

C+

V+S+

C−V−

S−




Figure 4. Relationships between total N, P and K contents and grain yield (GY) and total dry matter (TDM) in rainfed low-land rice at three locations, Laos, under treatments of no fertilizer ( ); with P at 13 kg ha−1 (�); with N at 60 kg ha−1 ( );with NPK at 60, 13, 16 kg ha−1 (�), 1998.

Total N content (kg ha−1)

(a)

(b)

(c)

15 20 25 30 35 40

TD

M a

nd G

Y (

t ha−1

)

TDM

GY

TDM

GY

TDM

GY

R2 = 0.80 (GY)R2 = 0.89 (TDM)

Total P content (kg ha−1)1 2 3 4 5 6

Total K content (kg ha−1)1 2 3 4 5 6

R2 = 0.47 (GY)R2 = 0.57 (TDM)

R2 = 0.51 (GY)R2 = 0.84 (TDM)

8

6

4

2

0

8

6

4

2

0

8

6

4

2

0




1997, the G variance was about the same as the sumof the all interaction variances.

The rather large G variance component may beattributed to the fact that a similar type of droughtpattern developed in all locations in 1997, preju-dicing late-maturing lines. These findings corrobo-rated those for similar situations in Thailand(Rajatasereekul et al. 1997). Moreover, this effectwas similar under both fertilized and non-fertilizedconditions. Growing conditions at the three locationswere more variable in 1996 when early seasondrought developed in Savannakhet and Vientiane butnot in Champassak, and adverse soil conditionscaused a large yield reduction only at Vientiane.

A series of experiments in Thailand also showed alarge G × L variance, compared with G variance(Fukai and Cooper 1998). The results of the com-bined analysis for the 13 entries in 1996 and 1997showed that G × year variance component wasgreater than the G × F variance component. Thelargest interaction variance component involvinggenotype was G × L in 1996 and 1998.

These results suggest that the genotypic inter-action involving fertilizer does not exert a verystrong influence on the overall genotypic perform-ance and, hence, would not significantly affect theselection of genotypes in a breeding program. Theresults of the 1996 experiments suggest that, judgingfrom the higher proportion of genetic variation tototal phenotypic variation (Inthapanya et al. 2000a),selection under fertilized conditions would be betterthan under non-fertilized conditions, although thisproportion was similar between the two conditions in1997. Coefficient of variation was often smallerunder fertilized conditions (Inthapanya et al. 2000a),providing more opportunity to differentiate genotypeperformance statistically. It should also be stated thatfertilizer use has increased rapidly in recent years inat least two provinces of Laos (Pandey and Sana-mongkhoun 1998). This trend is likely to continue inthe near future. The breeding program should there-fore apply fertilizer rather than not apply it.

The Lao breeding program is rather small and theuse of two fertilizer levels is not justified. Thecurrent rate of 60–30–20 kg ha–1 for N, P2O5 andK2O, respectively, used by the program may beconsidered moderate as rice yield often responds toeven higher N rates in Laos. It should be stated,however, that some genotypes did relatively wellunder non-fertilized conditions, suggesting that on-farm trials, conducted just before new cultivars arereleased, should be carried out under the fertilizerconditions of the targeted environments. In any case,the breeding program should not always be con-ducted in areas of high soil fertility, and the selectionprogram should avoid relying heavily on those

research stations whose soils are more fertile thanthose of the average farm.

The selection of locations appears important, par-ticularly considering the rather large G × L varianceobtained for all 3 years. If advanced genotypes con-sistently perform well in one location and not inother locations, those genotypes may be consideredfor release only to that area where they performedwell. However, the location should be representativeof farmers’ fields in terms of soil fertility anddrought occurrence.

These conclusions are similar to those obtainedfrom our earlier work where 35 cultivars andadvanced lines mostly from Thailand were comparedat two locations in the Vientiane and ChampassakProvinces for 3 years (Inthapanya et al. 1997).

Tirol-Padre et al. (1996) suggest that nutrient useefficiency for grain (NiUEg and PUEg), which didnot show any significant G × F interaction effect,may be more consistent across environments thantotal nutrient content, and may be a more usefulcharacter in developing new cultivars adapted to lowsoil fertility. In Vientiane, the different behaviour ofPUEg with fertilizer was probably caused by nutrienttoxicity problems observed in the field.

The analysis presented here indicates that NiUEgwas negatively associated with N concentration ingrains. This contrasts with the results of Tirol-Padreet al. (1996) who found larger genotypic variation instem rather than in grain N concentration, andsuggested the importance of low stem N concentra-tion for yield under low soil fertility conditions. Sahuet al. (1998), at Raipur, India, reported a significantvariation for yield among five lines at low N applica-tion rates (30 and 60 kg ha–1), and that this variationwas due to variation in total N content at the low Nrates usually applied in lowland conditions.

Similarly, in the present experiments, genotypicvariation for yield response to applied fertilizer wasclosely related to the variation for total N contentresponse. However, the ability to take up N appearsto be affected by soil conditions and water availa-bility (Fukai et al. 1999) and, in the present experi-ments, a significant G × F interaction was observedfor total N content at maturity. Poor correlation wasalso observed between years for total N content andgrain yield in the 2-year study by Tirol-Padre et al.(1996), confirming the results of our 3-year fertilizerexperiments. Selecting lines with consistently high Ncontents across different soil N environments wouldtherefore be difficult, a view supported by Ladha etal. (1998). However, low percentages of seed N andhigher N and P use efficiencies could be an alterna-tive approach for selecting lines for rainfed lowlandenvironments.




The results at the Vientiane and Champassaklocations suggested that N fertilizer could potentiallyincrease grain yield more than would the P fertilizertreatment. Yield response to P fertilizer was similarto the response to N fertilizer in Savannakhet. The Peffects observed in this experiment were similar tothe results reported by Lathvilayvon et al. (1997),where P application could increase grain yield underrainfed lowland conditions. However, the highestgrain yield was produced with the NPK fertilizertreatment. Nitrogen was the most important nutrientfor yield, with 80% of variation in yield among linesbeing accounted for by the total N content in theplant.

Conclusions

For grain yield, the G × F interaction effect wassmaller than the effects of genotype and the G × Yand G × L interactions. This indicates that currentfertilizer levels (60 N, 13 P and 16 K kg ha–1) couldbe used in the early testing programs for cultivarselection. However, because the G × L interactioneffect was strong for grain yield, the use of moreenvironments in multi-locational trials would beadvantageous in selecting lines for higher yield.

ReferencesFukai, S. and Cooper, M. 1998. Plant breeding strategies

for rainfed lowland rice in Northeast Thailand. In: WorldFood Security and Crop Production Technologies forTomorrow. Special issue of Japanese Journal of CropScience, 67, 256–257.

Fukai, S., Inthapanya, P., Blamey, F.P.C. and Khunthasuvon,S. 1999. Genotypic variation in rice grown in low fertilesoils and drought-prone, rainfed lowland environments.Field Crops Research, 64, 121–130.

Inthapanya, P., Sipaseuth, Sihathep, V., Chanphengsay, M.and Fukai, S. 1997. Drought problems and genotyperequirements for rainfed lowland rice in Lao PDR. In:Fukai, S., Cooper, M. and Salisbury, J. ed. BreedingStrategies for Rainfed Lowland Rice in Drought-ProneEnvironments, Proceedings of an International Work-shop held at Ubon Ratchathani, Thailand, 5–8 November1996. Canberra, ACIAR Proceedings No. 77, 74–81.

Inthapanya, P., Sipaseuth, Sihavong. P., Sihathep, V.,Chanhphengsay, M., Fukai, S. and Basnayake, J. 2000a.Genotypic performance under fertilized and non-

fertilized conditions in rainfed lowland rice. Field CropsResearch, 65, 1–14.

Inthapanya, P., Sipaseuth, Sihavong. P., Sihathep, V.,Chanhphengsay, M., Fukai, S. and Basnayake, J. 2000b.Genotypic differences in nutrient uptake and utilisationfor grain yield production of rainfed lowland rice underfertilized and non-fertilized conditions. Field CropsResearch, 65, 57–68.

Ladha, J.K., Kirk, G.J.D., Bennett, S., Peng, S., Reddy,C.K., Reddy, P.M. and Singh, U. 1998. Opportunities forincreased nitrogen-use efficiency from improved low-land rice germplasm. Field Crops Research, 56, 41–72.

Lathvilayvong, P., Schiller, J.M. and Phommasack, T.1997. Soil limitation for rainfed lowland rice in LaoPDR. In: Fukai, S., Cooper, M. and Salisbury, J. ed.Breeding Strategies for Rainfed Lowland Rice inDrought-Prone Environments, Proceedings of an Inter-national Workshop held at Ubon Ratchathani, Thailand,5–8 November, 1996. Canberra, ACIAR ProceedingsNo. 77, 192–201.

Linquist, B., Sengxua, P., Whitbred, A., Schiller, J.M.and Lathvilayvong, P. 1999. Evaluation of nutrientdeficiencies and management strategies for lowland ricein Laos. In: Proceedings of an International Workshopon Nutrient Research in Rainfed Lowlands, 12–15October 1998, Ubon Ratchathani, Thailand.

Pandey, S. and Sanmongkoun, M. 1998. Rainfed lowlandrice in Laos: a socio-economic benchmark study. SocialSciences Division, International Rice Research Institute(IRRI), Los Baños, Philippines.

Rajatasereekul, S., Sriwisut, S., Porn-uraisanit, P., Mitchell,J.M. and Fukai, S. 1997. Phenology requirement forrainfed lowland rice in Thailand and Lao PDR. In: Fukai,S., Cooper, M. and Salisbury, J. ed. Breeding Strategiesfor Rainfed Lowland Rice in Drought-Prone Environ-ments, Proceedings of an International Workshop held atUbon Ratchathani, Thailand, 5–8 November, 1996.Canberra, ACIAR Proceedings No. 77, 97–104.

Romyen, P., Hanviriyapant, P., Rajatasereekul, S.,Khunthasuvon, S., Fukai, S., Basnayake, J. andSkulkhu, E. 1998. Lowland rice improvement in Northernand Northeast Thailand, 2. Genotype differences. FieldCrops Research, 59, 109–119.

Sahu, R.K., Tirol-Padre, A., Ladha, J.K., Singh, U., Baghel,S.S. and Srivastava, M.N. 1998. Screening genotypes fornitrogen use efficiency on a nitrogen deficient soil.Oryza, 34, 350–357.

Tirol-Padre, A., Ladha, J.K., Singh, U., Laureles, E.,Punzalan, G. and Akita, S. 1996. Yield performance ofrice genotypes at suboptimal levels of soil N as affectedby N uptake and utilization efficiency. Field CropsResearch, 46, 127–143.

201



Microbiological Interventions in Acid Sandy Rice Soils

Wolfgang Reichardt

1*

, Suntree Meepetch

2

, Olivyn Angeles

1

and Sawaeng Ruaysoongnern

3

Abstract

Declining fertility of rainfed lowland soils in drought-prone North-East Thailand poses achallenge for microbiological intervention. Results from a series of field and phytotronexperiments on Ubon, acid, sandy soil led to the development of a conceptual model of a microbe-driven turnover of labile organic-matter pools. The model provides a yardstick for assessing theshort-term effects of soil-amelioration practices. Organic amendments and liming enhance bothsoil organic nutrient pools and the remineralization potential of soil microbial biomasses.Functional profiles in the microbial turnover of C and N sources were severely reduced duringsingle-cropping seasons. Organic amendments were more effective than liming in recovering thesefunctions, whereas liming was more effective in stimulating the production of soil microbialbiomass and suppressing soluble phenols. Mutual positive responses to both liming and organicamendments (rice-straw compost) suggested that a sustainable improvement of microbe-mediatedsoil fertility might be best achieved by combining the two treatments. Functional diversity profilesin bulk soil for rainfed lowland rice were much narrower than in the rhizosphere. A selective,stable and even variety-specific composition of microbiota in association with rice roots indicateda potential for microbiological intervention. In particular, the finding that metabolically different,ammonium-oxidizing, bacterial suppliers of nitrate associate with different drought-tolerant ricevarieties suggests promising clues for improving N-use efficiencies in rice-cropping systems foundin drought- and flood-prone rainfed lowlands.

R

AINFED

lowland rice farming in North-East Thailandaccounts for about 40% of the nation’s total rice pro-duction. However, it faces the challenge of sustainingand increasing yields on drought-prone, sandy soilswith extremely low cation-exchange capacity (CEC)and organic-matter contents. Acidification is pro-gressing with the continued chemical degradation ofdeforested Acrisol-type soils (Noble et al. 2000). Thefertility of these soils is steadily declining, or evenaccelerated by current socioeconomic trends thatfavour off-farm work for rural populations. Becausethe major soil problems are associated with drought,acidity, low CEC, low organic-matter content andslowed-down nutrient cycling, possible remedies

would include improved supplies and more efficientuse of water and nutrients by the rice crop.

In principle, such approaches to recover, sustain,or improve soil fertility can be either soil or plantmediated. Because of the key role played by biocata-lytic processes in nutrient cycling, soil microbiotaprovide promising targets in both soil and plant-directed fertility management. Hence, soil microbiotamanagement is indispensable for reversing, in a sus-tainable way, trends of declining soil quality, but itspractice implies a detailed knowledge of microbe-driven key mechanisms and their interconnections innutrient cycling that has yet to be developed. For thetime being, simple deterministic models of soil-nutrient supply can be built on emerging soil bio-chemical clues. As producer and converter of labilesoil-organic-nutrient pools, the soil microbial bio-mass varies under the influence of stabilizing anddestabilizing environmental factors (Anderson and

1

International Rice Research Institute (IRRI), Los Baños,Philippines

2

Ubon Rice Research Centre, Ubon Ratchathani, Thailand

3

Khon Kaen University, Khon Kaen, Thailand*Corresponding author: [email protected]

KEYWORDS:

Degraded soil, Rainfed rice, Nutrients, Fertility, Microbiota, Soil organic matter, Drought

202



Domsch 1980; Wardle and Parkinson 1990; Wardle1998; Reichardt et al. 2000). Conceptual nitrogen (N)sink and source models, based on these observations(Duxbury and Nkambule 1994), may help explainsoil-N-supplying capacity in terms of N release func-tions linked to the reduction of labile organic-Npools.

The biocatalytic capacity controlling these proc-esses in soils is ultimately reflected by the functionaldiversity of soil microbial communities, which diver-sity is only now being studied (Zak et al. 1994). Whilethese assays are often useful, yet far from being com-prehensive, they can be seen as promising indicatorsof sustainability in soil nutrient cycling, including Ndynamics (Reichardt et al. 2000).

In addition to microbiological intervention in soils,the microbe-driven nutrient supply functions on rootsurfaces can also be exploited. Microflora in theimmediate vicinity of nutrient-uptake sites have thepotential to alleviate abiotic stresses in the rice plant,or to modulate and finetune its nutrient-uptake effi-ciency. Certain microbial associations with roots ofdrought- and flood-affected rice plants have proved tobe variety specific (Briones and Reichardt 2000). Thissuggests a potential for microbial biofertilizers in thefuture.


Single-season field experiments conducted in rainfedlowlands on extremely nutrient-deficient, acid (pH4.1–4.5), sandy soils with low organic-matter content(0.028% N and 0.377% C). The experiments wereconducted at the Ubon Rice Research Centre(URRC), Ubon Ratchathani, Thailand, during threecropping seasons from 1997–1998. On a 40

×

80 mfield, 3

×

5 m subplots (with three replicates and arandomized block design) were planted to the ricevariety KDML105. Treatments were untreated con-trols, additions of organic fertilizer based on chickenmanure, rice-straw compost at 15 t ha

–1

, a mixture ofrice straw and farmyard manure (FYM) at 15 t ha

–1

,mineral fertilizer, ash of rice straw at 15 t ha

–1

andlime additions at 1.2 t ha

–1

.The same soil, rice variety and treatments as in the

field were also used in phytotron experiments at theInternational Rice Research Institute (IRRI) in 1.5 Lpots at 25

o

–30

o

C and 13/11 h light/dark regime at10

2

µ

E m

–2

s

–1

.Analyses of the soil organic phase included: Total soil protein (Herbert et al. 1971); Total soluble phenols (Box 1983); Phospholipid-based soil microbial biomass (PL-biomass; Tunlid and White 1992);

Production rates of heterotrophic soil microbialbiomass based on incorporating

3

H thymidine intoDNA (Christensen and Christensen 1995); Diversity of microbial functions in sole C sourceutilization, using the commercial BIOLOG assaykit (Zak et al. 1994; Reichardt et al. 2000).

Results and Discussion

Interventions through soil management practices

Underlying our experimental attempts to identifymicrobe-mediated mechanisms was the idea of thesoil microbial biomass and other labile soil-organic-matter constituents acting as a kind of nutrient pump(Figure 1). Based on a few soil biochemical parame-ters that can be used in routine soil-quality monitoring,labile organic-matter pools, consisting of microbialbiomass and extracellular protein, are permanentlytapped by the indigenous nitrogen supply (INS) andrefilled by soil organic-matter production. Nitrogenrelease is largely driven by phenol-suppressible, res-piratory, organic-matter degradation. This remineral-ization is linked to aerobic and anaerobic processes,and is measured as ‘respiratory electron systemactivity’ or ‘ETS activity’. In their function as ‘Npumps’, labile N pools are permanently replenishedby inputs of energy (solar radiation, organic cropresidues) and fertilizers. They are further modified bysoil and crop management practices, which affect thefunctional profiles and diversity of soil microbiota(Reichardt et al. 2000).

Managing the long-term fertility of sandy, nutrient-poor, rice soils in drought-prone areas requires adetailed knowledge of differential effects on nutrientimmobilization on the one hand and nutrient releaseon the other. Sustainable management will depend onthe time frame of these effects. While improving thesoil organic phase by means of straw reincorporationtakes about 4 years to show detectable effects onyields and yield stability, soil fertility determinants ofthe N-pump model can be assessed in much shorterintervals. These short-term effects have been investi-gated in both field experiments at Ubon Ratchathaniand phytotron experiments, using soil from thesefields, at IRRI, with some illustrated below.

Short-term effects of soil management

Protein

In a phytotron experiment with the rice varietyKDML105, protein pools were initially boosted bymineral fertilizer, organic fertilizer (chicken manure)or liming but afterwards declined during the vegeta-tive part of the crop’s growth cycle (Figure 2).Liming contributed to the strongest decline in soilprotein concentrations. To conclude, both organicamendments and liming caused soil protein pools to

203



fluctuate at greater amplitudes, suggesting an intensi-fication of nutrient-pump functions, either by energyinput as organic matter or by improving growth con-ditions for less acidophilic soil microbiota.

Phospholipid-based soil microbial biomass

Similar effects of soil-amelioration practices on soilorganic-nutrient pools were also indicated forPL-biomass in the field (Figure 3). These and otherdata from field and phytotron experiments confirmthe plasticity of soil microbial biomass pools under

changing environmental conditions (Reichardt et al.2000).

Microbial biomass production

Gross productivity of heterotrophic bacterial biomassin the soil as measured by tritiated thymidine uptakewas strongly enhanced by liming, whereas organicamendments failed to show such increases(Figure 4). Closed-system artefacts may have con-tributed to a general increase in bacterial production

Figure 1.

Conceptual model of labile soil-organic-matter functioning as a nutrient pump driving indigenous soil nutrientsupply in rice soils. ETS = respiratory electron system activity; INS = indigenous nitrogen supply; OM = organic matter.

Figure 2.

The rice variety KDML 105 in a phytotron experiment with Ubon soil. Samples of total soil protein concentrationswere taken at key crop growth stages: AT = 25 days after transplanting (DAT); MT = maximum tillering (43 DAT);PI = panicle initiation (64 DAT); PM = physiological maturity (136 DAT). Treatments were 1 = control; 2 = mineralfertilizer; 3 = organic manure; 4 = liming. Error bars indicate SD (n = 4).

Solar radiation

LabileOM

Temperatures

Crop residues

Fertilizers

Pesticides

INS

Microbial

biomass

Protein

Phototrophic

HeterotrophicETS

Phenols

Remineralization

Inhibiting

Increasing

Modifying composition

Increasing

Increasing

Increasing

AT MT PI PM

Crop growth stage

Soi

l pro

tein

con

cent

ratio

n (m

g g−1

)

5.0

4.5

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.01 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4

204



rates. The initial boost caused by raising the pHreflects an immediate response of soil microbialcommunities to improved growth conditions. Thisalso confirms the stimulatory effects of limingobserved in earlier field experiments at Ubon (Rei-chardt et al. 1998).

‘Respiratory electron system’ activity

‘Respiratory electron system’ activities reflectedmainly short-term responses. Organic amendmentsand, to a lesser extent, liming raised the reminerali-zation capacity, as indicated by ETS assays, signifi-cantly above those of sole mineral fertilizertreatments (Figure 5). Because respiratory activitiesare based on energy supply and organic substrate,organic amendments prove the most sustainable inmaintaining elevated activity levels.

Phenols

Soluble phenolic compounds in the soil are known toinhibit ETS activities (Reichardt et al. 2000). Ele-vated phenol concentrations in untreated plots in thefield experiment at Ubon were reduced most effec-tively by liming during the initial stages of cropgrowth (Figure 6). Hence, the positive response ofETS activities to liming could have resulted fromreduced phenol levels at an elevated pH. A linkbetween organic amendments and accumulation ofphenolic compounds can therefore be expected(Tsutsuki and Ponnamperuma 1987).

Indicators of functional microbial diversity in soil

Microbial communities in intensively cropped, irri-gated, rice soils may not necessarily show majorshifts in their functional diversity (Reichardt et al.

Figure 3.

Rainfed lowland field experiment at Ubon, Thailand. Change in soil microbial biomass as total phospholipid con-centration between crop growth stages of transplanting (white) and panicle initiation (black) for the following treatments:Con = control; Min = mineral fertilizer; CM = chicken manure; CS = rice-straw compost; AS = ash of rice straw. Error barsindicate SD (n = 4).

Figure 4.

Phytotron experiment with Ubon soil, planted to the rice variety KDML 105. Gross heterotrophic bacterial bio-mass production was measured as short-term uptake of tritiated thymidine at key crop growth stages: AT = 25 days aftertransplanting (DAT); MT = maximum tillering (43 DAT); PI = panicle initiation (64 DAT). Treatments were 1 = control;2 = mineral fertilizer; 3 = organic manure; 4 = liming. Error bars indicate SD (n = 4).

Con Min CM CS

Treatment

Tot

al p

hosp

holip

id

16

12

8

4

0

conc

entr

atio

n (n

mol

g−1

)

AS

AT MT PI

Crop growth stage

Trit

iate

d th

ymid

ine

(nm

ol h

−1 c

m−3

)

4.5

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.01 2 3 4 1 2 3 4 1 2 3 4

205



2000). However, BIOLOG assays of the infertile,acid, sandy, Ubon soils, show that biocatalytic func-tions decline steeply as the crop grows (Reichardt etal. 1998). Single organic amendments, for example,rice straw at 15 t ha

–1

per cropping season provedinsufficient to reverse this trend of declining func-tional diversity.

A subsequent phytotron experiment with Ubonsoil was designed to compare the relative effects oforganic manure, liming and mineral fertilizer on soilmicrobial diversity indices (Figure 7). Most probably

because of closed-system effects, functional richnessand Shannon diversity indices were roughly halved atthe panicle initiation stage, but recovered partially atharvest. Nevertheless, the contribution of organicmatter amendments to functional diversity during thevegetative growth phase proved significantly higherthan for liming. Increasing substrate evenness atadvanced growth stages (Figure 7) means enhancedequitability of activities across all used substrates(Zak et al. 1994). This reflects the common trend thatlower activity plateaus are reached as the crop grows.

Figure 5.

Field experiment with the rice variety KDML 105 at Ubon, Thailand, in 2000. Initial ETS activities attransplanting (AT) and maximum tillering (MT), with the following treatments: 1 = control; 2 = mineral fertilizer;3 = chicken manure; 4 = rice-straw compost; 5 = ash of rice straw; 6 = liming. Error bars indicate SD (n = 4).

Figure 6.

Concentrations of soil phenolic compounds in a phytotron experiment with Ubon soil, planted to the rice cultivarKDML 105, at key crop growth stages: AT = 25 days after transplanting (DAT); MT = maximum tillering (43 DAT);PI = panicle initiation (64 DAT); PM = physiological maturity (136 DAT). Treatments were: 1 = control; 2 = mineralfertilizer; 3 = organic manure; 4 = liming. Error bars indicate SD (n = 4).

AT MT

Crop growth stage

1 2 3 4 5 6 1 2 3 4 5 6

Initi

al E

TS

act

iviti

es (

nmol

h−1

cm

−3)

100

80

60

40

20

0

AT

Crop growth stage

Phe

nol c

once

ntra

tion

(µg

g−1)

80

70

60

50

40

30

20

10

01 2 3 4

MT

1 2 3 4

PI

1 2 3 4

PM

1 2 3 4

206



Figure 7.

Phytotron experiment with Ubon soil planted to the rice variety KDML 105. Functional diversity patterns of soilmicrobial community at key crop growth stages: AT = 25 days after transplanting (DAT); MT = maximum tillering(43 DAT); PI = panicle initiation (64 DAT); PM = physiological maturity (136 DAT). Treatments were: 1 = control;2 = mineral fertilizer; 3 = organic manure; 4 = liming. Error bars indicate SD (n = 4).

(a) Richness (%)

100

90

80

70

60

50

40

4.5

4.3

4.1

3.9

3.7

3.5

2.30

2.25

2.20

2.15

2.10

2.05

2.00

AT MT PI PM

1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4

(b) Shannon diversity index

AT MT PI PM

1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4

(c) Evenness

AT MT PI PM

1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4

Crop growth staqge

207



As far as trends sustained during the entirecropping season can be derived from three annualfield trials at Ubon and two supplementary phytotronexperiments, these have been summarized inTable 1. Organic amendments, as well as liming, canimprove soil fertility by enhancing the build-up oflabile organic-nutrient pools and their remineral-ization capacity. Liming has the additional advantageof reducing the concentrations of soluble phenolsthat suppress respiratory remineralization processes,while organic amendments proved superior in sus-taining functional diversity. Hence, a combination ofboth liming and repeated organic amendments wouldappear optimal for reaching the goal of sustainable,microbe-mediated, soil fertility in these soils.

a

farmyard manure, rice straw compost.

b

n.d. = insufficient data.

References

Anderson, J.P.E. and Domsch, K.H. 1980. Quantities ofplant nutrients in the microbial biomass of selected soil.Soil Science, 130, 211–216.

Box, J.D. 1983. Investigation of the Folin-Ciocalteauphenol reagent for the determination of polyphenolic sub-stances in natural waters. Water Research, 17, 511–525.

Briones, A.M. and Reichardt, W. 2000. Ammonium oxida-tion in the rhizosphere of drought-tolerant RL varieties:potential for microbial interventions to improve N useefficiencies. In: Thematic Workshop on: ImprovingTolerance to Abiotic Stresses in Rainfed Lowland Rice.Los Baños, Philippines, International Rice ResearchInstitute (IRRI), Abstract.

Christensen, H. and Christensen. S. 1995. [

3

H] Thymidineincorporation technique to determine soil bacterialgrowth rate. In: Alef, K. and Nannipieri, P. ed. Methodsin Applied Soil Microbiology and Biochemistry.London, Academic Press, 258–270.

Duxbury, J.M. and Nkambule, S.V. 1994. Assessment andsignificance of biologically active soil organic nitrogen.In: Doran, J.W., Coleman, D.C., Bezdicek, D.F. andStewart, B.A. ed. Defining Soil Quality for SustainableEnvironments. Soil Science Society of America (SSSA)Special Publication 35, 125–145.

Herbert, D., Phipps, P.J. and Strange, R.E. 1971. Chemicalanalysis of microbial cells. Determination of protein:The Folin Ciocalteu reagent. In: Norris, J.R., and Rib-bons, D.W. ed. Methods of Microbiology, vol. 5B, 249–252.

Noble, A.D., Gillman, G.P. and Ruaysoongnern, S. 2000. Acation exchange index for assessing degradation of acidsoil by further acidification under permanent agriculturein the tropics. European Journal of Soil Science, 51,233–243.

Reichardt, W., Meepetch, S., Briones, A.M., Ruaysoong-nern, S. and Naklang, K. 1998. Microbial biomass, itsassessment, functions and management in drought-prone,nutrient-poor, rainfed lowland soils. In: Ladha, J.K.,Wade, L., Dobermann, A., Reichardt, W., Kirk, G.J.D.and Piggin, C. ed. Rainfed Lowland Rice: Advances inNutrient Management Research. Manila, Philippines,International Rice Research Institute (IRRI), 141–159.

Reichardt, W., Inubushi, K. and Tiedje, J. 2000. Microbialprocesses in C and N dynamics. In: Kirk, G.J.D. andOlk, D.C. ed. Carbon and Nitrogen Dynamics in FloodedSoils. Makati City, Philippines, International RiceResearch Institute (IRRI), 101–146.

Tunlid, A. and White, D.C. 1992. Biochemical analysis ofbiomass, community structure, nutritional status andmetabolic activity of microbial communities in soil. SoilBiochemistry, 7, 229–262.

Tsutsuki, K. and Ponnamperuma, F.N. 1987. Behavior ofanaerobic decomposition products in submerged soils.Effects of organic material amendment, soil properties,and temperature. Soil Science Plant Nutrition, 33, 13–33.

Wardle, D.A. and Parkinson, D. 1990. Interactions betweenmicroclimatic variables and the soil microbial biomass.Biology and Fertility of Soils, 9, 273–280.

Wardle, D.A. 1998. Controls of temporal variability of thesoil microbial biomass: a global scale synthesis. SoilBiology and Biochemistry, 30, 1627–1637.

Zak, C.J., Willig, M.R., Moorhead, D.L. and Wildman, H.G.1994. Functional diversity of microbial communities: aquantitative approach. Soil Biology and Biochemistry,26, 1101–1108.

Table 1.

Relative trends of sustained positive and negativeresponses of soil biochemical parameters to soil-amelioration practices in acid, sandy, Ubon soil fromseveral field and phytotron experiments.

Effect of treatment on: Organic amendments

a

Liming

Soil protein + –Microbial biomass + n.d.

b

Microbial biomass production rate + +ETS activity + –Phenolic compounds + –Functional diversity (BIOLOG) + –

Documents

Low Temperature Tolerance in Rice: The Korean …aciar.gov.au/files/node/2301/pr101chapter02.pdf · Low Temperature Tolerance in Rice: The Korean Experience ... temperatures cause