A

Sa

b

a

ARRA

KADFLRRT

1

ri(oagepa8dno

0d

Field Crops Research 119 (2010) 215–224

Contents lists available at ScienceDirect

Field Crops Research

journa l homepage: www.e lsev ier .com/ locate / fc r

gro-economic evaluation of fertilizer recommendations for rainfed lowland rice

.M. Haefelea,∗, N. Sipaseuthb, V. Phengsouvannab, K. Dounphadyb, S. Vongsouthib

International Rice Research Institute, DAPO Box 7777, Metro Manila, PhilippinesNational Agricultural and Forestry Research Institute (NAFRI), Vientiane, Laos

r t i c l e i n f o

rticle history:eceived 15 January 2010eceived in revised form 1 July 2010ccepted 1 July 2010

eywords:gro-economicsroughtertilizer managementao People’s Democratic Republicainfed lowlandsiceoposequence

a b s t r a c t

Soils used for cropping of rainfed lowland rice are frequently rather poor, and nutrient limitation is knownto be a major constraint for grain yields. However, actual fertilizer use by most rainfed rice farmers isstill very low even if input-responsive rice varieties are widely adopted. To address this contradiction,this study, conducted in central and southern Laos, intended to test existing fertilizer recommendationsin farmers’ fields and under their crop management, to evaluate the effect of topographic field posi-tion on fertilizer response, and to conduct an economic evaluation of fertilizer use based on farmers’input and output prices. Trials were conducted for two seasons in three different agroecological envi-ronments (rainfed rice in lower fields and in middle/upper fields, and irrigated rice in lower fields).Tested were five different fertilizer treatments in four farmers’ fields per site and season. Commercialorganic fertilizer increased yields but resulted in negative economic returns in almost all cases becauseof its unfavorable nutrient/price ratio. In lower fields, farmyard manure and inorganic fertilizer gavegood yield responses and positive economic returns but the average value/cost ratios of 1.3–1.6 were atthe lowest level required for widespread adoption of fertilizer use. Water limitation reduced attainableyields in middle/upper fields and in this environment good fertilizer response was observed only for

farmyard manure and half the recommended rate of inorganic fertilizer. The actual benefit of fertilizeruse depended strongly on the input/output price ratios at each site. Thus, site-specific fertilizer man-agement options, including economic parameters, are required to make fertilizer use in rainfed lowlandrice more profitable. The necessary information for such guidelines is available but needs to be preparedin the form of decision tools to help farmers choose the optimal nutrient management option for theiragroecological and socioeconomic production environment.

. Introduction

Agriculture is the main livelihood of approximately 95% of theural households in Laos, and rice (Oryza sativa L.) is the single mostmportant crop. Rice provides about 64% of the total calorie supplyFAO STAT, 2010a) and constitutes around 52% of the total valuef agricultural production (FAO STAT, 2010b). Given this situation,ny increase in rice production and productivity contributes to aeneral improvement in the welfare of rural households (Schillert al., 2000). Demand for increased production comes also from fastopulation growth. The population of Laos is growing at 2.5% pernnum and is expected to increase from 5.3 million in 2000 to about

.8 million in 2020. Another factor contributing to growing riceemand is the expected increase in per-capita consumption con-ected to the fast development process in Laos. The major sharef the necessary production increase can come only from rice-

∗ Corresponding author.E-mail address: s.haefele@cgiar.org (S.M. Haefele).

378-4290/$ – see front matter © 2010 Elsevier B.V. All rights reserved.oi:10.1016/j.fcr.2010.07.002

© 2010 Elsevier B.V. All rights reserved.

based lowland systems, which currently supply 88% of the totalrice production. Although varying from year to year, the majorityof lowland rice is produced in rainfed systems (about 80%), whereasirrigated lowlands contribute about 20% of lowland rice production(e.g., MAF, 2002).

Modern high-yielding and input-responsive varieties for thelowlands were introduced from 1993 onwards. They spread fastand replaced the mostly traditional varieties grown until then.By 2005, more than 70% of the rainfed and irrigated wet-seasonrice cultivation was based on improved varieties, and, becausetraditional germplasm is photosensitive and cannot be used inthe dry season, 100% of the dry-season rice came from improvedvarieties (Inthapanya et al., 2006). Simultaneously, dry-season irri-gated area increased from 12,000 ha in 1990 to 102,000 ha in 2001.These developments were complemented by intensive research on

improved crop management techniques to accompany the newvarieties and irrigated rice cultivation. Although this research cov-ered many aspects of crop management, a major emphasis was onimproved nutrient management because of the dominance of poorsoils in the region (Linquist and Sengxua, 2001). The same authors

2 ps Research 119 (2010) 215–224

dird1ui1Sluanafohiulscavrarfinh

gyccddrftttefecfttn

TO

16 S.M. Haefele et al. / Field Cro

ocumented more than 250 individual multi-location trials on var-ous aspects of nutrient management between 1990 and 2000 inainfed and irrigated lowlands, and the data were used to developetailed nutrient management recommendations (Linquist et al.,998; Linquist and Sengxua, 2001, 2003). However, actual fertilizerse by farmers remained low. Fertilizer imports, which are the only

norganic fertilizer source in Laos, did increase continuously from990 onwards and reached more than 12,000 tons in 2001 (FAOTAT, 2010c; the most recent data from 2002 indicated a slightlyower value). But, even when assuming that all this fertilizer wassed for lowland rice, and based on the mean annual lowland ricerea of around 0.5 million ha, the total average fertilizer rate wouldot have been above 24 kg per ha. This estimate is confirmed bystudy of Pandey (2001), whose survey of 700 rainfed lowland

armers in 1996 found that 66% of them used inorganic fertilizern 48% of their rice-growing area, with an average rate of 18 kga−1 of inorganic (NPK) fertilizer. A later survey of 240 households

n the lowlands by Shrestha (2004) found that 93% of the farmerssed inorganic fertilizer but did not report the quantity used. This

ow fertilizer rate seems surprising, considering the generally lowoil fertility in the lowland rice environment of Laos, which is typi-al for large areas of rainfed lowland rice in southeast Asia (Haefelend Hijmans, 2007). And as mentioned above, responsive improvedarieties are widespread, and acceptable fertilizer responses wereepeatedly reported (Linquist and Sengxua, 2001, 2003; Linquist etl., 2007). However, such low fertilizer rates are typical for manyainfed lowland systems: data from 2005 to 2007 covering 1215elds in Cambodia, India, Indonesia, Laos, Nepal, Thailand, and Viet-am, showed an average inorganic fertilizer use of only 47 kg perectare (unpublished, provided by S. Pandey).

So why do rice farmers in rainfed lowlands not use more inor-anic fertilizer? Is it that researchers overestimated the averageield response because they conducted their trials under carefullyontrolled conditions not achievable by farmers? Did researchershoose sites not representative for the system targeted, or did theyisregard the economic conditions farmers were operating in? Oroes the dissemination and adoption of adjusted crop and naturalesource management technologies just take more time becausearmers need to acquire the necessary knowledge and build uphe confidence for successful inorganic fertilizer use? Addressinghese questions for the specific case of lowland rice in Laos washe objective of this study. In particular, we intended to test thexisting fertilizer recommendations for lowland rice in Laos inarmers’ fields and under their crop management, to evaluate theffect of topographic field position on fertilizer response, and to

onduct a simple economic evaluation of fertilizer use based onarmers’ input and output prices. The results were meant to con-ribute to further improving the understanding of fertilizer use inhe target system and to adjust the existing recommendations, ifecessary.

able 1verview of all environments, sites, and experimental seasons. The number of participati

Year Seasona Province

Low-lying fields, rainfed, low drought risk2003 + 2004 WS Vientiane2003 + 2004 WS Savannakhet2003 + 2004 WS ChampassakMiddle/upper fields, rainfed, medium drought risk2004 + 2005 WS Vientiane2004 WS Savannakhet2004 + 2005 WS ChampassakLow-lying fields, irrigated2003/04 + 2004/05 DS Vientiane2003/04 + 2004/05 DS Savannakhet2003/04 + 2004/05 DS Champasak

a Cropping seasons were the wet season (WS) and the dry season (DS).

Fig. 1. Map of Laos showing the different regions and provinces and the location ofexperimental sites.

2. Materials and methods

2.1. Site description

The agro-climatic conditions in the central/southern lowlandregion are characterized by a mean annual rainfall of 1920 mmyear−1 (Linquist and Sengxua, 2001), but the long-term aver-age rainfall varies considerably within that region (Linquist andSengxua, 2001; Basnayake et al., 2006). Mean annual rainfallis high in Vientiane, Xaisomboun, and Bolikhamsay provinces(2000–3000 mm), relatively low in Savannakhet Province (about1400 mm), and again high in Champassak Province (about2200 mm) (Fig. 1). No rainfall data were recorded at the experimen-tal sites but annual rainfall data were available from meteorological

stations close to the Vientiane airport (2003: 1481 mm; 2004:1629 mm; 2005: 1668 mm), the rice research station at Savan-nakhet (2003: 1199 mm; 2004: 1650 mm; 2005: 1768 mm), andthe rice research station at Pakse (2003: 2029 mm; 2004: 1977 mm;2005: 1956 mm). Thus, relatively low annual rainfall occurred only

ng farmers is valid for each individual cropping season.

District Village Farmers

Xaythany Ban Hai 4Champhone Ban Nongkhouvieng 4Sanasomboune Ban Nguadeng 4

Nasaythong Ban Namkieng Tai 4Champhone Ban Nong Veng 4Sanasomboune Ban Nakhoung 3

Xaythany Ban Hai 4Champhone Ban Nakhou 4Sanasomboune Ban Nguadeng 4

ps Research 119 (2010) 215–224 217

icsrsawae

awahcCftacspcdeaudf

2

rrptnwoiCawpeco

(

(

(

iwtdsfia

cter

isti

csfr

omal

lexp

erim

enta

lplo

tsat

each

site

.For

low

erfi

eld

s,ch

arac

teri

stic

sfo

rth

era

infe

dan

dir

riga

ted

seas

onar

esi

mil

arbe

cau

seth

esa

me

site

sw

ere

use

din

mos

tca

ses.

pH

(1:1

)SO

CTS

NA

vail

P Ols

enA

vail

ZnA

vail

KEx

ch.C

aEx

ch.M

gEx

ch.K

Exch

Na

Exch

.aci

dit

yEx

ch.A

lC

ECC

lay

Silt

San

d%

%p

pm

pp

mcm

ol/k

gcm

ol/k

gcm

ol/k

gcm

ol/k

gcm

ol/k

gcm

ol/k

gcm

ol/k

gcm

ol/k

g%

%%

infe

dan

dir

riga

ted

4.7

1.61

0.16

4.4

1.4

0.16

3.43

0.85

0.12

0.07

1.58

1.34

11.2

4351

65.

80.

390.

033.

00.

30.

120.

380.

550.

070.

210.

220.

142.

97

3063

5.3

0.62

0.05

2.9

2.6

0.08

4.45

2.27

0.06

0.16

0.17

0.15

10.3

2864

9te

rrac

efie

lds,

rain

fed

4.7

0.54

0.06

3.2

0.5

0.05

0.58

0.15

0.04

0.04

0.66

0.40

2.1

1155

34–

––

––

––

––

––

––

––

–4.

50.

300.

04<D

L*0.

70.

010.

390.

91<D

L*<D

L*0.

680.

501.

85

2570

the

det

ecti

onli

mit

.

S.M. Haefele et al. / Field Cro

n Savannakhet in 2003. But even if the mean rainfall does not indi-ate water constraints can the spatial and temporal variability ofeasonal rainfall in combination with toposequence effects causeegular occurrence of drought spells or flooding during the croppingeason (Basnayake et al., 2006). Minimum and maximum temper-tures in the lowlands are no constraint for rice cultivation in theet season; however, low temperatures can damage rice seedlings

t the start of the dry season in December and January (Basnayaket al., 2006).

The soils in the rice-growing areas of central and southern Laosre dominated by Acrisols (FAO, 1998). These soils are stronglyeathered, have a clay-enriched subsoil, contain mostly low-

ctivity clay minerals, and have a low base saturation (<50%) with aigh percentage of H+ and Al3+ ions at the exchange complex. Asso-iated soils, also frequently used for lowland rice cultivation, areambisols, which are usually less weathered and generally more

ertile than neighboring Acrisols. Selected typical topsoil charac-eristics at the experimental sites are presented in Table 2. Shownre average values for the four farmers at each site, because the soilharacteristics were similar for all participating farmers at eachite. For the wet- and dry-season experiments in low landscapeositions, identical or very nearby fields were used, and the soilharacteristics shown in Table 2 are representative for both. The soilata presented confirmed the general trends described by Linquistt al. (1998) but they also indicated large variability between sitesnd within a toposequence postion. The data also showed that soilssed for the wet-season trials in middle/upper fields with mediumrought risk were usually coarser textured and generally even lessertile than soils in lower toposequence positions.

.2. Experimental setup

This study focused on the central and southern agriculturalegions of Laos (Fig. 1), where about 80% of the total lowlandice area is located. Rice in these regions is mostly grown on thelains adjacent to the Mekong River (Vientiane Municipality andhe provinces of Vientiane, Bolikhamsay, Khammouane, Savan-akhet, Saravane, and Champassak). These are also the regionshere most development of irrigation capacity occurred from 1995

nwards, particularly on the Vientiane Plain (Vientiane Municipal-ty and Vientiane Province) and in the provinces of Savannakhet andhampassak; more than 90% of irrigated lowlands in Laos are situ-ted in the central and southern regions. All trials presented hereere conducted in these rice production centers, namely, in therovinces Vientiane, Savannakhet, and Champassak (Fig. 1). Withinach province, trials were conducted in three different agroecologi-al environments which were chosen based on the visual evaluationf the local topography:

1) Fields in the lower part of the topography, usually characterizedby a higher soil fertility, finer texture, better water resources,and a limited drought risk. Mostly rainfed wet-season rice.

2) Fields on middle/upper terraces, usually characterized by alower soil fertility, coarser texture, limited water resources, anda higher drought risk. Fully rainfed wet-season rice.

3) Field characteristics as in (1), but fully irrigated dry-season rice.

All low-lying fields (categories 1 and 3) were selected in existingrrigation schemes. However, farmers rarely used irrigation in the

et season to avoid related costs, and only 2 of the 4 experimen-al fields at the site in Vientiane Province were partially irrigated

uring the wet season; farmers did not irrigate during the wet sea-on at the sites in Savannakhet and Champassak. The same or, in aew cases, neighboring fields were used for the dry-season exper-ments, and full irrigation throughout the season was necessaryt all three locations. All fields in middle/upper landscape posi- Ta

ble

2M

ean

soil

char

a

Low

erfie

lds,

raV

ien

tian

eSa

van

nak

het

Ch

amp

assa

kM

iddl

e/up

per

Vie

nti

ane

Sava

nn

akh

etC

ham

pas

sak

*<D

L=

belo

w

2 ps Res

tsssefitstm

oTict

rwawdnaotmipwtfasewo0t0e1wNlr(iNt

2

rtmtaP

18 S.M. Haefele et al. / Field Cro

ions (2nd category) were located outside of operational irrigationchemes. Table 1 gives an overview of all sites and experimentaleasons. At each site (i.e., village), four participating farmers wereelected (only three at Ban Nakhoung in both seasons), and all farm-rs tested five different nutrient management treatments in theirelds. In all cases, the different treatments were established in twoo three neighboring fields, and treatments in the same field wereeparated only by an empty row. The resulting plot size for eachreatment varied but was usually between 100 and 200 m2, thus

uch larger than in most standard fertilizer trials.In all treatments, farmers managed the crop according to their

wn practice and improved high-yielding varieties were used (TDK,SN, and PNG varieties; see Inthapanya et al., 2006). Researchersmposed the nutrient management using farmyard manure (FYM),ommercial organic fertilizer (COF), and inorganic fertilizer (IF). Thereatments were:

T1: no FYM + no COF + no IFT2: 2000 kg FYM ha−1 + no COF + no IFT3: 2000 kg FYM ha−1 + 2000 kg COF ha−1 + no IFT4: 2000 kg FYM ha−1 + no COF + 30–5–6 kg NPK ha−1

T5: 2000 kg FYM ha−1 + no COF + 60–10–12 kg NPK ha−1

Treatment T1 served as the control, T2 tested the effect of aealistic FYM rate, T3 investigated the effect of the, at the time,idely advertised commercial organic fertilizers produced in Laos,

nd T4 (half the recommended rate) and T5 (recommended rate)ere based on a simplified version of the fertilizer recommen-ation developed by Linquist and Sengxua (2001). Intentionally,one of the inorganic fertilizer treatments tested was based on thepplication of inorganic fertilizer alone, because the application ofrganic materials is highly recommended for the lowlands of Laoso avoid soil nutrient depletion, especially with respect to K and

icronutrients (Linquist and Sengxua, 2001). The actual rates inndividual farmers’ fields were not always exactly according to therotocol because some farmers applied additional organic (to thehole field i.e., all treatments) or inorganic fertilizer (to individual

reatments), because only NP compound fertilizers were availableor basal application at some sites, and because researchers appliedslightly different rate in a few cases. However, the deviation was

mall (<5 kg N ha−1) in all but two cases (+10 kg N ha−1). The nutri-nt amount applied in T2, in which only 2 t FYM ha−1 were applied,as estimated at 10 kg N ha−1, 3 kg P ha−1, and 10 kg K ha−1 based

n the average NPK concentration in FYM of 0.5% N, 0.15% P, and.5% K (Fairhurst et al., 2007). The average NPK concentration ofhe COF was determined for 12 samples (mean values of 2.38% N,.65% P, and 0.82% K), and the resulting estimate of applied nutri-nts for a rate of 2 t ha−1 was 47.5 kg N ha−1, 12.9 kg P ha−1, and6.5 kg K ha−1. Inorganic fertilizers used in treatments T4 and T5ere two different compound fertilizers (15–15–15 and 16–20–0–P2O5–K2O) and urea (46–0–0 N–P2O5–K2O), depending on their

ocal availability. Organic fertilizers (FYM and COF) were incorpo-ated into the soil at initial field preparation. Compound fertilizerNPK or NP) was incorporated into the soil just before transplant-ng, containing from 20% to 30% of the total N rate. The remaining

was applied as urea, either in one split at panicle initiation or inwo equal splits at early tillering and panicle initiation.

.3. Measurements and data analysis

In each field, a composite topsoil sample (0–0.2 m) from 5andomly sampled sub-samples was collected at the beginning of

he experiment. Soil samples were analyzed for pH (1:1 soil–H2O

ixture), total soil organic carbon (Nelson and Sommers, 1996),otal soil N (Bremner, 1996), cation exchange capacity (Sumnernd Miller, 1996), and soil texture (modified from Koehn, 1928).lant-available Zn, K (NH4O-Ac extraction), and P (Olsen P) were

earch 119 (2010) 215–224

determined according to Ponnamperuma et al. (1981), Helmke andSparks (1996), and Olsen et al. (1954), respectively.

Recorded crop management and development stages wereseeding, transplanting, 50% flowering, physiological maturity, andharvest. Grain and straw yield at harvest were measured fromone or two 5-m2 areas depending on the homogeneity of the plot.Grain moisture content was determined immediately after thresh-ing (Riceter grain moisture meter, Kett Electric Laboratory, Tokyo,Japan) and grain yields were reported at 14% moisture content.Agronomic efficiency of applied nitrogen for the treatments T4 andT5 (AEN; kg yield response per kg N applied) was calculated basedon the yield increase due to N application. The field hydrology for allexperimental fields was recorded at least once a week by measuringthe depth of the floodwater layer and by visually rating the field-water status. These observations were used to estimate the meanseasonal stress from drought and stagnating deep water, which isalso known to reduce yields (Ismail et al., 2010). For the droughtstress, the weekly observations of field hydrology (1: ponded water,2: wet soil surface, 3: dry soil surface) were used to calculate themean seasonal field hydrology, a value between 1 and 3 for everyfield. For stress from stagnating water, an average value was calcu-lated in the same way based on water depth measurements (1:1–10 cm water layer, 2: 10–30 cm water layer, 3: >30 cm waterlayer).

To enable an economic evaluation of the observed fertilizerresponse, we collected information on input prices (fertilizer, labor)and output prices (paddy) directly from the participating farm-ers in every season. This information was then used to establishvalue/cost ratios based on treatment-specific costs (average laborneeds dependent on the treatment, average labor price per site andexperimental season, average fertilizer prices per site and season,actual amount and type of fertilizer used) and profits (based onthe average paddy price per site and season but treatment-specificyield response). The exchange rate throughout the experimentalperiod was 10,000 Kip = 1 US$ and all prices and costs are thereforegiven in 0000 Kip.

Grain yield data were analyzed using IRRISTAT and SAS soft-ware. For the analysis of variance, farmer fields in the same positionand season were treated as replications, but treatment effects wereanalyzed across seasons. For the ANOVA, a mixed model analysiswas used. Standard methods were used for the correlation matrixpresented.

3. Results

Table 3 gives an overview of selected agronomic characteris-tics for all sites and seasons. In the wet season (WS), sowing of theseedbed was mostly conducted towards the end of May until mid-June. Early establishment occurred in the WS 2004 at Ban Hai, andlate establishment in the WS 2005 at Ban Namkieng Tai. Especially,the start of seeding was slightly later in middle/upper fields than inlower fields. In the dry season (DS), most farmers established thecrop between late November and mid-December, demonstratingthe relatively short turnaround time between the harvest of theWS crop (usually mid- to late October) and start of the DS crop. TheDS crop was harvested in April in all cases. Average seedling ageat transplanting across all farmers at a site varied between 23 and37 days after seeding (DAS); the most extreme cases of individualfarmers were 22 and 36 DAS in the WS and 15 and 44 DAS in theDS. Average crop duration from seeding to harvest was 126–155days and the range was smallest for middle/upper fields in the

wet season. Across seasons and field categories, crop duration wasshorter when the crop was established late, which explains the rel-atively short durations in middle/upper fields, which were fullydependent on rainfall. The average harvest indices (HI) varied con-siderably, between 0.53 and 0.32, but were not particularly low

S.M.H

aefeleet

al./FieldCrops

Research

119(2010)

215–224219

Table 3Selected agronomic characteristics for all sites and experimental seasons. The range (sowing and harvesting date) or average values for all farmers (all other parameters) at the same site are shown.

Year Season Province Village na Rain-fallb Seeding Harvest Seedlingage

Cropduration

HI Grain yieldT1

Yieldresponsec

Droughtd Water stag-nationd

mm Date Date Days Days t ha−1 t ha−1

Lower fields, mostly rainfed2003 WS Vient. B. Hai 4 1090 June 7 Oct 27–Nov 3 27 143 0.39 3.37 0.89 1.2 1.12003 WS Savan. B. Nongkhouvieng 4 926 June 10 Oct 25–26 24 138 0.52 2.74 1.11 1.3 1.02003 WS Cham. B. Nguadeng 4 1507 May 29–June 7 Oct 1–24 28 137 0.38 2.26 0.57 1.1 2.32004 WS Vient. B. Hai 4 1384 May 18–29 Oct 20–28 31 155 0.32 2.23 0.62 1.3 1.12004 WS Savan. B. Nakhou 4 1151 June 4–12 Oct 12–24 30 131 0.35 2.03 0.66 1.3 1.02004 WS Cham. B. Nguadeng 4 1687 May 30–June 10 Oct 12–20 31 135 0.47 1.23 1.71 1.0 2.3

Middle/upper fields, fully rainfed2004 WS Vient. B. Namkieng Tai 4 1163 June 2–14 Oct 12–23 32 130 0.40 2.19 0.85 1.2 1.22004 WS Savan. B. Nong Veng 4 1150 June 10–11 Oct 23–29 34 137 0.39 1.24 0.34 2.5 1.02004 WS Cham. B. Nakhoung 3 1679 June 12–15 Oct 13–Nov 8 31 130 0.44 1.19 1.06 2.2 1.02005 WS Vient. B. Namkieng Tai 4 1412 June 24–30 Oct 27–Nov 11 29 127 0.39 2.22 0.81 1.2 1.22005 WS Savan. B. Nong Veng 4 1414 – – – – – – – – –2005 WS Cham. B. Nakhoung 3 1602 June 10–20 Oct 25–30 31 135 0.40 1.55 0.80 2.2 1.0

Lower fields, fully irrigated03/04 DS Vient. B. Hai 4 238 Nov 16–21 April 1–12 24 140 0.34 2.50 0.62 1.3 1.003/04 DS Savan. B. Nakhou 4 217 Dec 8 April 26–27 35 141 0.40 1.64 0.39 1.2 1.003/04 DS Cham. B. Nguadeng 4 103 Dec 11–21 April 20–22 23 126 0.39 2.00 0.91 2.0 1.004/05 DS Vient. B. Hai 4 98 Nov 25–Dec 5 April 20–30 33 144 0.53 3.07 0.89 1.2 1.004/05 DS Savan. B. Nakhou 4 146 Nov 22–26 April 17–21 37 147 0.53 1.99 1.39 1.2 1.004/05 DS Cham. B. Nguadeng 4 68 Dec 8–18 April 22–27 24 130 0.50 1.19 1.20 1.3 1.0

a Number of farmers participating at the site.b Seasonal rainfall for the crop growth period of the trials from the nearest meteorological station.c Yield response is the average across all fertilized treatments (T2–5) and all farmers at the same village.d Mean values from seasonal field hydrology observations as described in the text, 1 indicates no stress.

220 S.M. Haefele et al. / Field Crops Research 119 (2010) 215–224

Table 4Regression analysis (r values) between selected grain yield indicators and several descriptive site parameters, including soil characteristics, mean field hydrology values, andtwo agronomic crop characteristics.

Site parameter Grain yield Grain yield response Mean AEN

Control (T1) Average(T2 + T3 + T4 + T5)/4 − T1

To organicfertilizer(T2 − T1)

To inorganicfertilizer([T4 + T5]/2 − T1)

Inorganicfertilizer([T4 + T5]/2 − T1)

Drought stress −0.43** −0.24 −0.16 −0.20 −0.11Water stagnation −0.11 0.18 0.21 0.23 0.04pH −0.02 0.22 −0.03 0.26 0.24Soil organic C 0.59** −0.14 −0.04 −0.13 0.04Total soil N 0.59** −0.16 −0.06 −0.13 0.07Olsen P 0.56** −0.05 0.05 −0.13 −0.06Avail. Zn −0.15 0.18 0.20 0.15 −0.08Avail. K 0.62** −0.04 −0.06 −0.05 0.01Exch. Ca 0.09 0.16 0.17 0.20 0.04Exch. acidity 0.27* −0.28* −0.11 −0.32** −0.13CEC 0.27* 0.04 0.12 0.04 −0.02Sand −0.20 −0.07 −0.17 −0.05 0.06Silt −0.05 0.22 0.32** 0.18 −0.10Clay 0.38** −0.09 −0.02 −0.09 −0.01Seedling age −0.22 0.05 0.03 −0.03 −0.06

iomallirfisowl

atsaoaptsomipbsbticn

TR

Crop duration 0.23 −0.01

* The analysis included 63 cases and correlations are significant at p ≤ 0.05.** The analysis included 63 cases and correlations are significant at p ≤ 0.01.

n middle/upper fields, and consistently high values were achievednly in the DS 2004–2005. Average grain yields of the control treat-ent (T1, no fertilizer applied) ranged between 1.2 and 3.4 t ha−1

nd did not differ clearly between seasons or field categories, butow T1 yields were more frequent in middle/upper fields. Simi-arly, the average fertilizer response (mean value of T2–T5) did notndicate a distribution according to season or field category andanged between 0.3 and 1.7 t ha−1. The mean values for seasonaleld hydrology did indicate considerable drought stress at threeites where trials for middle/upper fields were conducted (and atne DS site because of an irrigation water shortage), and deep field-ater levels (stagnating water up to 0.42 m) were observed in all

ower fields at Ban Nguadeng in both seasons.Correlation coefficients between selected grain yield indicators

nd several soil characteristics, mean seasonal field hydrology, andwo agronomic parameters were analyzed across all sites and sea-ons (Table 4). Grain yield in the control treatment (no fertilizerpplication, T1) was significantly reduced by drought, whereas soilrganic carbon (SOC), total soil N (TSN), plant-available P (Olsen P)nd K, clay content, CEC, and exchangeable acidity all affected yieldositively. Non-significant negative yield trends were indicated forhe effect of water stagnation, Zn availability, sand content, andeedling age at transplanting. In contrast, the different indicatorsf response to fertilizer (mean response across all fertilizer treat-ents, response to organic fertilizer [T2], and mean response to

norganic fertilizer [T4/5]) seemed hardly affected by any of thearameters: the only significant positive correlation was detectedetween the response to organic fertilizer (T2–T1) and the soililt content, the only significant negative correlation was detectedetween exchangeable acidity and the average response to fer-

ilizer ([T2 + T3 + T4 + T5]/4 − T1) as well as the mean response tonorganic fertilizer ([T4 + T5]/2 − T1) and exchangeable acidity. Inomparison with the control treatment, fertilizer response seemedot affected by drought and seedling age and crop duration did not

able 5egression analysis (r values) for the observed drought stress and selected soil characteri

pH SOC TSN Olsen Exch. Ca Exch. ac

Drought stress −0.19 −0.39** −0.32* −0.62** −0.37** −0.05

* The analysis included 63 cases and correlations are significant at p ≤ 0.05.** The analysis included 63 cases and correlations are significant at p ≤ 0.01.

−0.10 −0.05 0.22

have any relation to fertilizer response. The drought stress indica-tor itself was significantly and negatively correlated to SOC, Olsen P,available K, CEC, silt content, and clay content; the only highly sig-nificant positive correlation was with sand content (Table 5). Thissuggests that drought is generally more severe on soils with a highsand content, and that drought-prone fields have a tendency forlower SOC, available P and K, and CEC values.

Agro-economic results for all treatments are presented inTable 6. The results were analyzed across the two experimentalyears but separately for the three differentiated cropping environ-ments, i.e., rainfed rice in the wet season in lower or middle/upperfields and irrigated dry-season rice in lower fields. Lower andupper fields in the wet season were analyzed separately to detectpossible effects of topography-related characteristics such as soiltexture, plant-available water resources, or indigenous soil fertil-ity on fertilizer response. Similarly, the wet and dry season weredistinguished because of the possible effects of different climaticsituations.

General yield trends were identical for all three data sets andyields increased in the sequence T1 < T2 < T3 < T4 < T5 even thoughyield differences were not always statistically significant due to theconsiderable variability between fields (Table 6 and Fig. 2). Compar-ing the three data sets revealed that the highest average treatmentyields were achieved in lower fields during the wet season, fol-lowed by lower fields during the dry season, whereas the lowestaverage treatment yields occurred in upper fields during the wetseason. The average yield increase due to the use of 2 t FYM ha−1

(T2) was 0.5 t ha−1 in all three cropping environments. TreatmentsT3 (2 t COF ha−1) and T4 (1/2 the recommended inorganic fertil-izer rate) resulted in similar and consistent grain yield increases

of about 0.9 t ha−1 over the control treatment (T1), except in upperfields (0.7 t ha−1 for T3). Treatment T5 (full recommended inorganicfertilizer rate) achieved the highest yields and yield increases inall three environments but the yield increase in comparison to T4

stics across all sites and seasons.

idity Avail. Zn Avail. K CEC Sand Silt Clay

−0.26* −0.48** −0.47** 0.49** −0.45** −0.41**

S.M. Haefele et al. / Field Crops Research 119 (2010) 215–224 221

Table 6Average grain yield (GY), yield gain over control treatments (�GY), agronomic efficiency of N (AEN), costs, value/cost (V/C) ratios, and negative returns to fertilizer use forall treatments and the three rice environments distinguished. Shown are mean values for two seasons and n gives the number of replications (i.e., farmers).

Treatment GY at 14% MCa �GY over T1 �GY over T2 AEN Costsb,c V/C ratiod Negative returnse

t ha−1 t ha−1 t ha−1 kg kg−1 0000 Kip %

Wet season 2003 and 2004, rainfed rice in lower fields (n = 24)d

T1 control 2.31 dT2 + FYM 2.85 c 0.54 51.1 1.5 4T3 + FYM + COF 3.22 bc 0.91 0.37 263.9 0.4 83T4 + FYM + 1/2 IF 3.24 ab 0.93 0.39 12 84.0 1.4 13T5 + FYM + full IF 3.80 a 1.45 0.95 16 107.6 1.6 8Wet season 2004 and 2005, rainfed rice in middle/upper fields (n = 22)d

T1 control 1.71 dT2 + FYM 2.15 c 0.47 37.9 1.3 6T3 + FYM + COF 2.40 b 0.68 0.24 210.6 0.3 78T4 + FYM + 1/2 IF 2.59 ab 0.88 0.44 14 78.4 1.4 22T5 + FYM + full IF 2.73 a 1.02 0.58 10 105.2 1.1 17Dry season 2003/04 and 2004/05, irrigated rice on lower fields (n = 24)d

T1 control 2.07 cT2 + FYM 2.57 bc 0.50 43.5 1.4 0T3 + FYM + COF 2.93 b 0.86 0.36 332.1 0.4 100T4 + FYM + 1/2 IF 2.96 b 0.89 0.39 12 80.5 1.4 13T5 + FYM + full IF 3.41 a 1.34 0.84 14 107.2 1.5 8

a Values in a column followed by a common letter are not significantly different at p = 5% according to the DMR test.b Treatment costs were calculated based on input and average labor costs (4 labor days ha−1 for application of FYM or COF, 2 labor days ha−1 for IF application).c The exchange rate throughout the experimental period was 10,000 Kip = 1 US$.d Value/cost ratio calculated using average treatment costs and the farm-gate paddy pre Negative returns to cash costs, excluding FYM costs.

Fig. 2. Average grain yields for all treatments and the three rice environmentsdistinguished. Shown are mean values for all participating farmers and two exper-imental seasons in the wet season (WS) and the dry season (DS). Bars indicate thestandard error of the mean.

Table 7Fertilizer price, farm-gate paddy price, and labor costs during the experimental period fro

16–20–0 15–15–15 46–0–0Kip kg−1 Kip kg−1 Kip kg−1

Mean 2636 3168 3021Min 2200 2660 2000Max 2900 3600 3600Price in kg paddyMean 2.1 2.6 2.5Min 1.8 2.2 1.5Max 2.6 3.6 4.0

ice.

was relatively small in rainfed middle/upper fields. This decreasingyield response in upper (sandy) fields can also be seen in Fig. 2. But,apart from this exception, the comparison of treatments T2, T4, andT5, which differed only in their inorganic fertilizer rate, indicates alinear response to inorganic fertilizer up to the recommended rate,and the response was about 0.4 t ha−1 paddy yield per 30–5–6 kgNPK ha−1 applied. This corresponds to an agronomic efficiency ofapplied N of about 13 kg paddy per kg N applied.

Input costs were variable between sites and changed within theexperimental period (Table 7). In general, costs of inorganic fer-tilizer increased over the 3-year period at all sites but costs alsovaried between provinces. Costs of commercial organic fertilizer(COF), labor, and the farm-gate paddy price were mostly depen-dent on the location, i.e., COF was most expensive in SavannakhetProvince, labor was most expensive in Vientiane Province, and thehighest paddy price was also reported from Vientiane Province.The cost of FYM showed no trend and was usually between 150and 250 Kip kg−1, except for Savannakhet in the 2003 wet sea-son (400 Kip kg−1). These variations in input costs over time andbetween sites resulted in considerable differences in input costsrelative to the farm-gate paddy price (Table 6).

The results of the economic evaluation of all fertilizer treat-ments are shown in Table 6. The cost of the treatments increased

in the sequence T2 < T4 < T5 < T3. The analysis shows that all fertil-izer options, with the exception of the use of COF (T3), resultedin average positive returns to the investment made in fertilizer(V/C ratios > 1). Negative returns (V/C ratios < 1) for the use of COF(T3) were caused by the unfavorable price/nutrient content ratio of

m 2003 to 2005.

FYM COF Paddy LaborKip kg−1 Kip kg−1 Kip kg−1 Kip day−1

189 1031 1240 16,769150 560 1000 15,000400 3600 1400 25,000

0.2 1.0 – 13.60.1 0.4 – 11.50.4 3.6 – 17.9

2 ps Res

titaT1entev(

4

tesbstttCs(latcbtssaahaSdt2tfitaseimf(thc(salKceitf

22 S.M. Haefele et al. / Field Cro

his fertilizer; the necessary application rate for the observed yieldncrease was 2000 kg COF ha−1. Low returns were also observed forreatment T5 in upper fields, caused by the declining yield responset the higher inorganic fertilizer rate. For all other treatments (T2,4, and T5), calculated average V/C ratios were between 1.3 and.6, and no obvious difference between FYM and inorganic fertiliz-rs could be observed. Table 6 also shows the relative frequency ofegative returns to the cash investment made to buy fertilizer, i.e.,he number of cases where the profit from increased yields was notnough to pay for the COF or the inorganic fertilizer. This risk wasery low for T2 (0–6%), intermediate for the treatment T4 and T58–22%), and very high for T3 (78–100%).

. Discussion

Climatic conditions during the experimental seasons wereypical for the region, and neither severe drought nor floodingvents occurred. However, a more detailed analysis of the sea-onal weather and possible effects on the rice crop was not possibleecause daily site-specific weather data were not available. Theoil characteristics of the experimental fields were also represen-ative for the region, indicating rather poor soils at most sites. Withhe exception of the fields at Ban Hai in Vientiane Province, theopsoils (0–0.2 m) had a high sand or silt fraction, relatively lowEC values, acid soil reaction, low soil organic carbon, low totaloil N concentrations, and low plant-available P and K contentsTable 2). According to Fairhurst et al. (2007), P deficiency is highlyikely at P-Olsen values <5 ppm and K deficiency is highly likelyt values <0.15 cmol kg−1, and both indicators were below thesehresholds in almost all experimental fields. These characteristicsorrespond well to the description of lowland paddy soils in Laosy Linquist et al. (1998), who found that 80% of the soils in the cen-ral/southern region contained less than 2% organic matter (<1.2%oil organic C), 68% were coarse textured (sandy, loamy sands, andandy loams), and 87% had a pH (H2O) of less than 5.5. Other authorslso confirmed that low soil fertility (especially low N and P avail-bility) is a major factor limiting rice yields in the region, sometimesighlighted as an even more important constraint than water avail-bility (Fukai et al., 1998; Suriya-arunroj et al., 2000; Linquist andengxua, 2001). And Haefele and Hijmans have shown that thisominance of very poor soils is actually representative for most ofhe rainfed lowland rice in southeast Asia (Haefele and Hijmans,007). Furthermore, the soil characteristics from our sites seemedo indicate even poorer and coarser textured soils in middle/upperelds than in lower fields. This general trend can be explained byransport processes down the toposequence, and was describednd discussed in more detail by Haefele and Konboon (2009) for theimilar lowland system in neighboring northeast Thailand. How-ver, the variability of soils within each toposequence positions usually high, and thus lower fields are not necessarily always

ore fertile than upper fields, which prevents the formulation ofertilizer recommendations based on topographic position aloneBoling et al., 2008; Haefele and Konboon, 2009). Parallel to thisrend of increasing soil fertility down the slope, the toposequenceas, of course, a direct effect on field hydrology. Several authorsonfirmed more limited water supplies in middle/upper fieldslower depth of ponded surface water, fewer days with pondedurface water, and higher depth of the groundwater surface) andhigher risk of deep ponded water or even crop submergence in

ower fields (Tsubo et al., 2006; Boling et al., 2008; Haefele andonboon, 2009). Obviously, too much as well as too little water

an affect rice yields negatively (Boling et al., 1998, 2008; Tsubot al., 2006; Homma et al., 2007) but our data (Tables 3 and 4)ndicated that there is no clear relation between field position inhe toposequence, field hydrology, and rice yield with or withoutertilizer.

earch 119 (2010) 215–224

The effects of fertilizer treatments on grain yield (Table 6 andFig. 2) mostly confirmed results by previous studies but also devi-ated in two important aspects. The general yield level for controltreatments and fertilized treatments was comparable with that ofmost other studies conducted in this environment. Across all fer-tilizer treatments, observed yields were lowest in middle/upperfields (Table 6 and Fig. 2). Lower yields in upper fields are generallyattributed to drought stress, lower soil fertility (Fukai et al., 1998;Suzuki et al., 2003; Homma et al., 2007; Boling et al., 2008), and neg-ative interactions between water and nutrient availability (Haefeleet al., 2006). That these factors did affect crop growth and yieldsis confirmed by the regression results presented in Table 4, show-ing that the control yields (T1) tended to be higher in soils withhigher soil organic carbon, total soil N, available P, available K, andless drought stress. And the same characteristics plus exchangeableCa, CEC, silt and clay content were negatively related to drought(Table 5). And, as outlined above, the same characteristics tend tobe more favorable in lower fields (Haefele and Konboon, 2009).The slightly lower yields for all treatments in lower fields in thedry season were surprising; we expected higher yields because ofthe generally higher solar radiation in the dry season. However,Linquist and Sengxua (2001) also reported lower control yields inthe dry season, and that the fertilizer response did not seem to beaffected by the season. The result that drought and several soil fac-tors mainly affected the control yield (i.e., the indigenous nutrientsupply) whereas they were not related to yield response caused byfertilizer application (Table 4) was also reported by Haefele et al.(2009a).

The use of FYM (treatment T2) caused an average yield gainof 0.5 t ha−1 relative to the unfertilized situation (treatment T1),which is similar to results reported by Linquist and Sengxua (2001),confirming that the application of organic materials at mediumrates (±2 t ha−1) represents a good option for farmers especially ifthey have no access to inorganic fertilizer. However, these organicfertilizer rates and related yield gains represent probably the upperpossible and sustainable limit. Although average total straw yieldwas 3.7 t ha−1 in the DS and 4.1 t ha−1 in the WS, a considerablefraction of this total amount remains in the field as stubble (usually20–30% of the total straw amount; data not shown) and farmersoften have other uses for rice straw, especially as fodder for live-stock in the dry season. Interesting was that organic fertilizer had aprice at all sites, and that using this price to estimate value/costratios resulted in values similar to those for inorganic fertilizer(treatment T2 versus T4/T5, Table 6). Purchasing and applying com-mercial organic fertilizer (COF) did increase yields on average by0.2–0.4 t ha−1 but this was always associated with negative finan-cial returns (see treatment T3 in Table 6). The comparison of theyield gains achieved by using COF in addition to FYM (T3) or FYMonly (T2) indicated that COF could not maintain the yield increaseachieved with FYM alone although the fertilizer costs for COF weremore than 5 times the costs of FYM (Tables 6 and 7). Thus, thecommercial organic fertilizer used in this study cannot be rec-ommended to rice farmers in the lowland environments of Laosbecause it is too expensive relative to the average rice yield gainachieved, which confirms the results of Roder et al. (2006).

The average response to inorganic fertilizer for treatment T4(half the recommended rate) and T5 (full recommended rate) indi-cated a linear response to increasing fertilizer rate up to the fullrecommended rate in lower fields (Table 6 and Fig. 2). A linear fer-tilizer response at these rates was also reported by Linquist andSengxua (2001) for the combined analysis of 12 trials in southern

Laos and 8 trials in northern Laos. However, our study indicateddeclining rates of return to fertilizer use at the higher rate (T5) inmiddle/upper fields. This can be explained by a lower attainableyield in these fields: limited water availability causes lower cumu-lative transpiration of the crop, which is linearly related to total

ps Res

d2marTabaihaoeiPtotsnwtu2tlmiN(Nrdas

imuwea1vwmt((amivAdthbprcr

ai

S.M. Haefele et al. / Field Cro

ry matter production and yield (Harmsen, 2000; Haefele et al.,009b). Without this exception, which can be avoided by recom-ending only lower fertilizer rates in drought-prone fields, average

gronomic efficiency of applied N (AEN) in each agro-ecosystemanged between 12 and 16 kg grain per kg N applied (Table 6).his compares to the higher AEN value of 15 reported by Linquistnd Sengxua (2001) as average for 107 experiments conductedetween 1991 and 1999. Both results seem similar but Linquistnd Sengxua (2001) found AEN values closer to 20 when exclud-ng failed experiments from the analysis, but they did not clarifyow many and why these experiments failed. Similarly, Linquistnd Sengxua (2003) and Linquist et al. (2007) found AEN valuesf ≥ 20. But the study of Linquist and Sengxua (2003) explicitlyxcluded all trials with stress from drought or prolonged flood-ng, seemingly no FYM was used in these trials, and often higher

and K rates were applied. In the study of Linquist et al. (2007),he control yields were relatively low (1.1–1.7 t ha−1), higher ratesf P and K were used (13 kg P ha−1, 18–71 kg K ha−1), and in one ofwo trials 30 kg S ha−1 were applied. Thus, the higher AEN in thesetudies can be explained by less constraints from other stresses andutrients which, according to Liebscher’s “Law of the Optimum”,ill also effect N efficiency. Also, we observed relatively high con-

rol yields in our trials (means between 1.7 and 2.3 t ha−1) and wesed the treatment with FYM applied (T2; means between 2.2 and.9 t ha−1) as reference for the response to inorganic fertilizer. Thus,he high level of reference yields might have further contributed toower AEN values. In addition, the lower AEN for half the recom-

ended rate (T4) in lower fields could indicate a hidden negativenteraction with the organic fertilizer applied, perhaps caused by

immobilization. Similar results were described by Linquist et al.2007), and such an effect could be relatively more important at low

rates. Finally, fertilizer efficiency in our study was probably alsoeduced because our trials were mostly farmer managed and con-ucted at a field scale whereas Linquist and Sengxua (2001, 2003)nd Linquist et al. (2007) analyzed researcher-managed trials onmall plots with a high level of management control.

Low AE(N) values are a concern because they, together with thenput/output cost ratio, determine the economic returns to invest-

ent in fertilizer. The observed V/C ratios for inorganic fertilizerse (T4 and T5) are relatively low returns to investment (meansere ranging between 1.3 and 1.6, Table 6) given that many agro-

conomic studies have indicated that small-scale farmers acceptnd adopt technologies only if they provide at least a return of50% on their investment (V/C ratio ≥ 1.5) (CIMMYT, 1989). Unfa-orable input/output cost ratios and limited fertilizer responseere also found by Pandey (2001), who argued that this was theain reason for low fertilizer use in Laos. However, the calculated

reatment costs included the costs of labor for fertilizer applicationcosts of 67,077–120,914 KIP ha−1 [7–12 US$ ha−1]) and FYM costs364,000–860,000 Kip ha−1 [36–86 US$ ha−1]), which in the case ofsmall farmer would be part of the income produced by the treat-ent. Nevertheless, the observed profit from fertilizer (FYM and

norganic fertilizer) use in Laos was not high because of the unfa-orable input/output price ratio and the limited fertilizer response.nd the combination of varying treatment costs and yield gainsid cause a substantial number of cases with negative returns tohe use of inorganic fertilizer (treatments T4 and T5; Table 6). Thisighlights that the decision whether and which fertilizer shoulde used must be made by farmers based on the local input/outputrice ratio. Especially inorganic fertilizer use at input/output priceatios substantially higher than the mean values shown in Table 7

annot be recommended, and this is most probably also true forainfed lowlands outside of Laos.

We concluded that, at least in the case of Laos but most likelylso in other agronomic systems with similar conditions, fertil-zer use for rainfed lowland rice is probably less profitable than

earch 119 (2010) 215–224 223

assumed by many earlier agronomic studies, mainly because ofunfavorable input/output cost ratios but also because of a limitedfertilizer response in some farmers’ fields. Assuming that fertilizerprices are not likely to fall, this situation can be addressed onlyby more knowledge-intensive and site-specific crop and fertilizermanagement. Our analysis provides several elements necessary toimprove the economic viability of fertilizer use. Buying organic fer-tilizer is probably rarely beneficial and farmers should use mainlytheir own sources of organic fertilizer (e.g., crop residues, FYM, oth-ers). Inorganic fertilizers should be used only if the input/outputprice ratio is below or equal to the mean values given in Table 7.Farmers should preferably use fertilizers with high nutrient con-centrations (e.g., urea, DAP, KCl) because they are often relativelycheaper per unit nutrient content; especially compound fertiliz-ers are often expensive in this respect. If only one rice crop peryear is grown and integrated fertilizer strategies, including sub-stantial crop residue applications, are used, K application rates canprobably be low or even omitted on most rainfed lowland soilsexcept on soils with a known K deficiency. And, lower fertilizerrates should be used in drought-prone, middle/upper fields becausethe water stress reduces attainable yields. The biggest challengeahead in achieving more efficient fertilizer use and higher prof-itability in rainfed lowland rice cropping is most likely to transferthe necessary knowledge for more site-specific approaches to farm-ers, enabling them to make the best choices according to theirproduction environment.

References

Basnayake, J., Fukai, S., Sipaseuth, Schiller, J.M., Monthathip, C., 2006. Advances inagronomic research in the lowland rice environments of Laos. In: Schiller, J.M.,Chanphengxay, M., Linquist, B., Appa Rao, S. (Eds.), Rice in Laos. InternationalRice Research Institute, Los Banos (Philippines), pp. 349–369.

Boling, A., Burac, M.A., Tuong, T.P., 1998. Characterization of rainfed lowland riceecosystems in Victoria, Tarlac. In: Presented at the 14th Annual Scientific Meet-ing of the Federation of Crop Science Societies in the Philippines, Cebu City,19–24 April.

Boling, A., Tuong, T.P., Suganda, H., Konboon, Y., Harnpichitvitaya, D., Bouman,B.A.M., Franco, D.T., 2008. The effect of toposequence position on soil properties,hydrology, and yield of rainfed lowland rice in Southeast Asia. Field Crops Res.106, 22–33.

Bremner, J.M., 1996. Nitrogen – total. In: Sparks, D.L. (Ed.), Methods of Soil Analysis.Part 3, Chemical Methods. Soil Science Society of America, Inc., and AmericanSociety of Agronomy, Inc., Madison, WS, USA, pp. 1085–1121.

CIMMYT, 1989. Formulation de recommandations à partir de donnés agronomiques:manuel méthodologique d’évaluation économique. Edition totalement révisée.Mexique, p. 82.

Fairhurst, T.H., Witt, C., Buresh, R.J., Dobermann, A., 2007. A Practical Guide to Nutri-ent Management, 2nd ed. International Rice Research Institute, InternationalPlant Nutrition Institute, and the International Potash Institute, p. 89.

FAO STAT, 2010a. Food and Agricultural Organization of the United Nations, Foodconsumption pattern of main food items: Dietary Energy, period of 2003 to 2005,updated 07.08.2008. Available at: http://www.fao.org.

FAO STAT, 2010b. Food and Agricultural Organization of the United Nations, Com-modities Production, Commodities by Country, updated 2007. Available at:http://www.fao.org.

FAO STAT, 2010c. Food and Agricultural Organization of the United Nations,FAOSTAT, Resource STAT, Fertilizers Archive by Country. Available at:http://www.fao.org.

FAO, 1998. World Reference Base for Soil Resources. World Soil Resources Report84, Food and Agriculture Organization of the United Nations, Rome, Italy, p. 88.

Fukai, S., Sittisuang, P., Chanphengsay, M., 1998. Increasing production of rainfedlowland rice in drought prone environments – a case study in Thailand andLaos. Plant Prod. Sci. 1, 75–82.

Haefele, S.M., Hijmans, R.J., 2007. Soil quality in rice-based rainfed lowlands ofAsia: characterization and distribution. In: Aggarwal, P.K., Ladhamm, J.K., Singh,R.K., Devakumar, C., Hardy, B. (Eds.), Science, Technology, and Trade for Peaceand Prosperity. Proceedings of the 26th International Rice Research Confer-ence, 9–12 October 2006, New Delhi, India. International Rice Research Institute,Indian Council of Agricultural Research, and National Academy of AgriculturalSciences, Los Banos (Philippines) and New Delhi (India), pp. 297–308.

Haefele, S.M., Konboon, Y., 2009. Nutrient management for rainfed lowland rice innortheast Thailand. Field Crops Res. 114, 374–385.

Haefele, S.M., Konboon, Y., Patil, S., Mishra, V.N., Mazid, M.A., Tuong, T.P., 2009.Water by nutrient interactions in rainfed lowland rice: mechanisms and impli-cations for improved nutrient management. In: Haefele, S.M. and Ismail, A.M.(eds), Natural resource management for poverty reduction and environmental

2 ps Res

H

H

H

H

H

I

I

KL

L

L

24 S.M. Haefele et al. / Field Cro

sustainability in fragile rice-based systems. Limited Proceedings No. 15. Inter-national Rice Research Institute, Los Banos (Philippines), p. 48-58.

aefele, S.M., Naklang, K., Harnpichitvitaya, D., Jearakongman, S., Skulkhu, E.,Romyen, P., Phasopa, S., Tabtim, S., Suriya-arunroj, D., Khunthasuvon, S., Kraiso-rakul, D., Youngsuk, P., Amarante, S.T., Wade, L.J., 2006. Factors affecting riceyield and fertilizer response in rainfed lowlands of northeast Thailand. FieldCrops Res. 98, 39–51.

aefele, S.M., Siopongco, J.D.L.C., Boling, A., Bouman, B.A.M., Tuong, T.P., 2009b.Transpiration efficiency of rice (Oryza sativa L.). Field Crops Res. 111, 1–10.

armsen, K., 2000. A modified Mitscherlich equation for rainfed crop production insemi-arid areas. 1. Theory. Neth. J. Agric. Sci. 48, 237–250.

elmke, P.A., Sparks, D.L., 1996. Lithium, sodium, potassium, rubidium, and cesium.In: Bigham, J.M. (Ed.), Methods of Soil Analysis. Part 3, Chemical Methods. SoilScience Society of America and American Society of Agronomy, Madison, WS,USA, pp. 551–574.

omma, K., Horie, T., Shiaiwa, T., Supapoj, N., 2007. Evaluation of transplantingdate and nitrogen fertilizer rate adapted by farmers to toposequential variationof environmental resources in a mini-watershed (Nong) in northeast Thailand.Plant Prod. Sci. 10, 488–496.

nthapanya, P., Boualaphanh, C., Hatsadong, Schiller, J.M., 2006. The history of low-land rice variety improvement in Laos. In: Schiller, J.M., Chanphengxay, M.,Linquist, B., Appa Rao, S. (Eds.), Rice in Laos. International Rice Research Institute,Los Banos (Philippines), pp. 325–348.

smail, A.M., Thomson, M.J., Vergara, G.V., Rahman, M.A., Singh, R.K., Gregorio, G.B.,Mackill, D.J., 2010. Designing resilient rice varieties for costal deltas using mod-ern breeding tools. In: Hoanh, C.T., Szuster, B., Kam, S.P., Noble, A., Ismail, A.M.(Eds.), Tropical Deltas and Coastal Zones: Community, Environment and FoodProduction at the Land–Water Interface. CABI Publishing, p. 480.

oehn, M., 1928. Pflanzenernaehr, Bodenk, AII:50.inquist, B.A., Phengsouvanna, V., Sengxua, P., 2007. Benefits of organic residues and

chemical fertilizer to productivity of rain-fed lowland rice and to soil nutrientbalances. Nutr. Cycl. Agroecosyst. 79, 59–72.

inquist, B., Sengxua, P., 2001. Nutrient management in rainfed lowland rice inthe Lao PDR. International Rice Research Institute, Los Banos, Philippines,

p. 60.

inquist, B., Sengxua, P., Whitbread, A., Schiller, J., Lathvilayvong, P., 1998. Evaluatingnutrient deficiencies and management strategies for lowland rice in Lao PDR. In:Ladha, J.K., Wade, L., Dobermann, A., Reichhardt, W., Kirk, G.J.D., Piggin, C. (Eds.),Rainfed Lowland Rice: Advances in Nutrient Management Research. IRRI, LosBanos, Philippines, pp. 59–73.

earch 119 (2010) 215–224

Linquist, B.A., Sengxua, P., 2003. Efficient and flexible management of nitrogen forrain-fed lowland rice. Nutr. Cycl. Agroecosyst. 67, 107–115.

MAF, 2002. Official Data Collected from the Ministry of Agriculture and Forestry.Vientiane, Lao PDR.

Nelson, D.W., Sommers, L.E., 1996. Total carbon, organic carbon, and organic mat-ter. In: Sparks, D.L. (Ed.), Methods of Soil Analysis. Part 3. Chemical Methods.SSSA Book Series No. 5. Soil Science Society of America and American Society ofAgronomy, Madison, WS, USA, pp. 961–1010.

Olsen, S.R., Cole, C.V., Watanabe, F.S., Dean, L.A., 1954. Estimation of available phos-phorus in soil by extraction with sodium bicarbonate. U.S.D.A. Circ., 939.

Pandey, S., 2001. Economics of lowland rice production in Laos: opportunities andchallenges. In: Fukai, S., Basnayake, J. (Eds.), Increased Lowland Rice Productionin the Mekong Region. Proceedings of an International Workshop, 2001, ACIARProceedings No. 101, Canberra, Australia, pp. 20–30.

Ponnamperuma, F.N., Cayton, M.T.C., Lantin, R.S., 1981. Dilute hydrochloric acid asan extractant for available zinc, copper and boron in rice soils. Plant Soil 61,297–310.

Roder, W., Schvrmann, S., Chittanavanh, P., Sipaseuth, K., Fernandez, M., 2006. Soilfertility management for organic rice production in the Lao PDR. Renew. Agric.Food Syst. 21, 253–260.

Schiller, J.M., Phanthavong, S., Siphaphone, V., Sidavong, S., Erguiza, A., 2000. Impactassessment of improved rice production technologies for the rainfed lowlandenvironment in the Lao PDR. Report (unpublished). Vientiane, Lao PDR.

Shrestha, S., 2004. Lao-IRRI Project: Impact Assessment of Research and TechnologyDevelopment. Consultancy report, p. 60.

Sumner, M.E., Miller, W.P., 1996. Cation exchange capacity and exchange coeffi-cients. In: Sparks, D.L. (Ed.), Methods of Soil Analysis. Part 3, Chemical Methods.Soil Science Society of America, Inc., and American Society of Agronomy, Inc.,Madison, WS, USA, pp. 1201–1229.

Suriya-arunroj, D., Chaiyawat, P., Fukai, S., Blamey, P., 2000. Identificationof nutrients limiting rice seedling growth in soils of Northeast Thailandunder water limiting and non-limiting conditions. Kasetsart J. (Nat. Sci.) 34,323–331.

Suzuki, K., Goto, A., Mizutani, M., Sriboonlue, V., 2003. Simulation model of rainfed

rice production on sloping land in northeast Thailand. Paddy Water Environ. 1,91–97.

Tsubo, M., Basnayake, J., Fukai, S., Sihathep, V., Siyavong, P., Sipaseuth, Chan-phengsay, M., 2006. Toposequential effects on water balance and productivityin rainfed lowland rice ecosystem in Southern Laos. Field Crops Res. 97,209–220.