Environmental impact and economic benefits of site-specific nutrient management (SSNM) in irrigated rice systems

AGRICULTURAL

www.elsevier.com/locate/agsy

Agricultural Systems 93 (2007) 1–24

SYSTEMS

Environmental impact and economic benefitsof site-specific nutrient management (SSNM)

in irrigated rice systems

M.F. Pampolino a, I.J. Manguiat a,1, S. Ramanathan b,H.C. Gines c, P.S. Tan d, T.T.N. Chi d, R. Rajendran e,

R.J. Buresh a,*

a Crop, Soil and Water Sciences Division, International Rice Research Institute, DAPO Box 7777,

Metro Manila, Philippinesb Tamil Nadu Agricultural University, Coimbatore 641003, Tamil Nadu, India

c Philippine Rice Research Institute, Maligaya, Science City of Munoz, 3119 Nueva Ecija, Philippinesd Cuu Long Delta Rice Research Institute, Omon, Cantho, Viet Nam

e Tamil Nadu Rice Research Institute, Aduthurai 612101, Tamil Nadu, India

Received 13 May 2005; received in revised form 24 March 2006; accepted 13 April 2006

Abstract

Site-specific nutrient management (SSNM) provides a field-specific approach for dynami-cally applying nutrients to rice as and when needed. This approach advocates optimal useof indigenous nutrients originating from soil, plant residues, manures, and irrigation water.Fertilizers are then applied in a timely fashion to overcome the deficit in nutrients betweenthe total demand by rice to achieve a yield target and the supply from indigenous sources.We estimated environmental impact of SSNM and evaluated economic benefits in farmers’fields in southern India, the Philippines, and southern Vietnam for two cropping seasons in2002–2003. On-farm research comparing SSNM and the farmers’ fertilizer practice showedincreased yield with SSNM for the three locations, even with reduced fertilizer N rates in some

0308-521X/$ - see front matter � 2006 Elsevier Ltd. All rights reserved.

doi:10.1016/j.agsy.2006.04.002

* Corresponding author. Tel.: +63 2 580 5600; fax: +63 2 580 5699.E-mail address: [email protected] (R.J. Buresh).

1 Present address: Soil Science Division, Agricultural Systems Cluster, College of Agriculture, Universityof the Philippines Los Banos, College, Laguna 4031, Philippines.

2 M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24

cases. SSNM increased partial factor productivity (kg grain kg�1 fertilizer N) when fertilizer Nuse efficiency with the farmers’ fertilizer practice was relatively low such as at locations in Viet-nam and the Philippines. Use of on-farm data with the DNDC model revealed lower percent-age of total N losses from applied fertilizers with SSNM during an annual cycle of croppingand fallows. At the location in India, SSNM showed the potential of obtaining higher yieldswith increased fertilizer N use while maintaining low N2O emissions. SSNM in the Philippinesand Vietnam showed greater yields with less fertilizer N through improved fertilizer use effi-ciency, which could reduce N2O emissions and global warming. Use of SSNM never resultedin increased emissions of N2O per unit of grain yield, and in environments where higher yieldcould be obtained with less fertilizer N, the use of SSNM could result in reduced N2O emis-sions per unit of grain yield. For the economic analysis, data were generated through focusgroup discussions (FGD) with farmers practicing SSNM and with other farmers not practicingSSNM. Based on FGD, the seasonal increase in yield of farmers solely due to use of SSNMaveraged 0.2 Mg ha�1 in southern Vietnam, 0.3 Mg ha�1 in the Philippines, and 0.8 Mg ha�1

in southern India. Farmers practicing SSNM at the study site in India used less pesticide. Theadded net annual benefit due to use of SSNM was 34 US$ ha�1 year�1 in Vietnam,106 US$ ha�1 year�1 in the Philippines, and 168 US$ ha�1 year�1 in India. The increased ben-efit with SSNM was attributed to increased yield rather than reduced costs of inputs.� 2006 Elsevier Ltd. All rights reserved.

Keywords: DNDC model; Focus group discussion; Global warming potential; Nitrogen management;Paddy rice systems; Site-specific nutrient management

1. Introduction

Site-specific nutrient management (SSNM) is a low-tech, plant need-basedapproach for optimally applying N, P, and K to rice (IRRI, 2006). It enables farmersto dynamically adjust fertilizer use to fill the deficit between the nutrient needs of ahigh-yielding crop and the nutrient supply from naturally occurring indigenoussources such as soil, crop residues, manures, and irrigation water. The SSNMapproach aims to apply nutrients at optimal rates and times to achieve high riceyields and high efficiency of nutrient use by the crop. It does not specifically aimto either reduce or increase fertilizer use (Buresh et al., 2005). It is based on scientificprinciples developed through nearly a decade of on-farm research throughout Asia.

The initial concept of SSNM was developed in the mid-1990s and then evaluatedfrom 1997 to 2000 in about 200 irrigated rice farms at eight sites in six Asian coun-tries (Dobermann et al., 2002, 2004). The SSNM practices developed and evaluatedin farmers’ fields before 2001 increased yield and profit as compared to farmers’ fer-tilizer practices (Dawe et al., 2004). Since 2001 the initial SSNM concept had beensystematically transformed through collaboration with national agriculture researchand extension systems in Asia into a simplified framework for dynamic plant-needbased management of N, P, and K for rice. Refined SSNM recommendations havebeen developed and evaluated through on-farm research involving thousands offarmers in major rice-growing areas of Bangladesh, China, India, Indonesia, Myan-mar, the Philippines, Thailand, and Vietnam.

M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24 3

The key features of the SSNM approach include (1) dynamic adjustments in fer-tilizer N, P, and K management to accommodate field- and season-specific condi-tions; (2) effective use of indigenous nutrients; (3) efficient fertilizer N managementthrough the use of the leaf colour chart (LCC), which helps ensure N is applied atthe time and in the amount needed by the rice crop; (4) use of the omission plot tech-nique to determine the requirements for P and K fertilizer, and (5) managing fertil-izer P and K to both overcome P and K deficiencies and avoid the mining of thesenutrients from the soil (Witt et al., 2002; IRRI, 2006).

The total amount of required fertilizer N can be approximated from the antici-pated crop response to fertilizer N application, which is the difference betweenattainable target yield and N-limited yield (i.e., yield with no fertilizer N and no lim-itation of other nutrients). The estimated total fertilizer N requirement by the crop isthen apportioned among multiple times of application during the growing season toensure that the supply of N matches the crop need at critical growth stages. A keyingredient for dynamic N management is a method for the rapid assessment of leafN content, which is closely related to photosynthetic rate and biomass productionand is a sensitive indicator of changes in crop N demand within a growing season(Peng et al., 1996). The LCC is an inexpensive and simple tool for monitoring therelative greenness of a rice leaf as an indicator of leaf N status, and then enablingfarmers to apply fertilizer N whenever leaves reach a critical N status determinedby their yellowish-green colour (Peng et al., 1996; Shukla et al., 2004; Alam et al.,2005; Witt et al., 2005).

In this paper, we estimate environmental impact and evaluate the economic ben-efits of refined SSNM practices at on-farm research sites in 2002–2003 in major rice-growing areas in southern India, the Philippines, and southern Vietnam. It is the firststudy to assess the farm-level economic benefits of SSNM for rice across severalAsian countries with data collected after 2001. It represents an initial effort to esti-mate – through use of a simulation model – probable environmental benefits arisingfrom use of SSNM to increased fertilizer N use efficiency in farmers’ fields. The spe-cific objectives of the study are to (1) estimate through use of a simulation model andon-farm observations of fertilizer, soil, water, and crop management the effects ofSSNM – as compared to farmers’ fertilizer practices (FFP) – on gaseous emissionsof N and global warming potential associated with emissions of nitrous oxide(N2O), a potent contributor to global warming and (2) assess the added net eco-nomic benefit of SSNM as compared to FFP in rice farming.

2. Methodology

2.1. The study sites

The study was conducted at three locations with intensive irrigated rice produc-tion systems: (1) the Cauvery Delta, Tamil Nadu, India; (2) Nueva Ecija, centralLuzon, Philippines; and (3) the Mekong River Delta, southern Vietnam. On-farmtrials to develop and evaluate SSNM started in 1997 at these locations. Beginning

4 M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24

in 2001 the evaluation of SSNM was conducted in at least 10 farmers’ fields per vil-lage or district at a location. In Tamil Nadu, the study was conducted in two dis-tricts: Aduthurai in the Old Cauvery Delta and Thanjavur in the New CauveryDelta. In Nueva Ecija, the study was conducted in three villages: Bantug and Mam-barao in Munoz Municipality and Burgos village in Santo Domingo Municipality.In the Mekong River Delta, the study was conducted in three districts: Omon inCantho Province, Chau Phu in An Giang Province, and Go Cong Tay in Tien GiangProvince. Two irrigated rice crops were grown and monitored each year at alllocations.

2.2. Treatments and data collection in the on-farm research

The on-farm trials had two treatments: SSNM and FFP. SSNM plots wereinstalled in farmers’ fields with a minimum area of 400 m2 in southern India,120 m2 in the Philippines, and 500 m2 in southern Vietnam. The adjacent farmers’fields represented the FFP treatment with a minimum area of 650 m2 in southernIndia, 2500 m2 in the Philippines, and 900 m2 in southern Vietnam. Crop manage-ment in SSNM fields other than fertilization was handled by farmers, and it wascomparable to crop management in FFP fields. In India, rice was established bytransplanting with a plant density of 50–66 plants m�2. In the Philippines, rice wasestablished by both transplanting (25 plants m�2) and direct seeding by broadcastingof pre-germinated seeds (80–120 kg seeds ha�1). In southern Vietnam, rice was directseeded by broadcasting of pre-germinated seeds at 100–120 kg seeds ha�1 in SSNMplots and 120–300 kg seeds ha�1 in FFP. At all sites, crops were generally grownwith good irrigation water supply.

Yield parameters and nutrient management practices were collected every crop-ping season from fields with SSNM and from adjacent farmers’ fields with FFP.In this study, we used results for two rice-cropping seasons in 2002–2003, which con-sisted of one dry (high-yielding) season and one wet (low-yielding) season.

2.2.1. Fertilizer management

In SSNM plots, fertilizer N was applied several times during a cropping season(Table 1) based on crop needs as determined by the leaf colour chart. Fertilizer Pwas applied only once at the early stage of the crop, within 14 days after transplant-ing (DAT) or 21 days after seeding (DAS); and fertilizer K was applied twice about50% before 14 DAT or 21 DAS and 50% at early panicle initiation. Table 1 showsthe rates of N, P, and K fertilizer applied with SSNM and FFP at each of the sites.

2.3. Assessment of environmental impact

We used the DNDC (denitrification–decomposition) model to evaluate the effectof SSNM and FFP on daily fluxes of N gases throughout an annual cycle with tworice crops and fallow periods in 2002–2003 at each location. Simulations started at 1or 2 days before transplanting of transplanted rice or sowing of direct-seeded rice,when the soil has already been flooded for about 3 weeks.

Table 1Rates of fertilizer N, P, and K and number of N split applications with farmers’ fertilizer practice (FFP)and site-specific nutrient management (SSNM) in three countries for two irrigated rice crops in 2002–2003

Country Location CEa FFP SSNM

Nsplitsb

FNc

(kg Nha�1)

FPd

(kg Pha�1)

FKe

(kg Kha�1)

Nsplits

FNc

(kg Nha�1)

FPd

(kg Pha�1)

FKe

(kg Kha�1)

Dry season rice

India Aduthurai TPR 3.6 121 21 41 4.0 118 15 50Thanjavur TPR 3.6 105 22 34 4.0 119 15 84

Philippines Nueva Ecija DSR 3.0 169 14 39 3.8 133 15 54TPR 2.4 144 22 34 3.0 105 15 54

Vietnam An Giang DSR 3.2 100 20 45 3.2 95 19 45Cantho DSR 3.0 115 21 35 3.0 102 16 41Tien Giang DSR 3.4 123 16 44 3.0 101 26 40

Wet season rice

India Aduthurai TPR 3.8 137 20 34 4.0 138 10 50Thanjavur TPR 2.8 96 15 35 3.6 123 15 80

Philippines Nueva Ecija TPR 1.8 95 17 26 4.8 108 9 42TPR 1.6 91 14 23 4.6 104 9 42

Vietnam An Giang DSR 4.2 96 18 49 3.0 78 20 30Cantho DSR 3.6 108 22 34 3.0 87 20 30Tien Giang DSR 3.0 95 20 41 3.0 87 20 30

Values shown are means of five farms per location.a Crop establishment method: TPR = transplanted rice; DSR = direct seeded rice.b Number of times fertilizer N was applied during the cropping season.c FN = Nitrogen fertilizer.d FP = Phosphorus fertilizer.e FK = Potassium fertilizer.

M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24 5

2.3.1. The DNDC model

DNDC is a simulation model for predicting trace gas emissions from agriculturalecosystems (Li, 2000). The first component of the model consists of the sub-modelson soil, climate, plant growth, and decomposition. The second component consistsof sub-models on nitrification, denitrification, and fermentation. Daily fluxes ofNH3, N2O, NO, and N2 are predicted through the second component. The simula-tion outputs generate a yearly total for the gas emissions. Hence, annual losses of Nfrom fertilizer N can be easily simulated. For our simulations, we have used version8.6 of the DNDC model with its logistic function for which we used measured grainyields as an input.

2.3.2. Parameter inputs for the DNDC model

The DNDC model requires inputs for climate, soil, crop, and farming manage-ment (Institute for the Study of Earth, Oceans and Space, 2003). In our simulations,we used daily rainfall, minimum air temperature, and maximum air temperature col-

6 M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24

lected from a centrally located meteorology station in each of the three country loca-tions. We used texture, organic C, and soil pH measured on soil samples collectedfrom on-farm research at each location; and we used soil bulk density, soil nitrate,soil ammonium, and soil temperature either determined in experiments at the loca-tions or estimated from the literature for submerged rice soils. Crop, soil, and watermanagement used as inputs in the model – such as for soil submergence, tillage,planting date, weeding, irrigation, drainage, and harvest date – were collected fromon-farm research at each location in 2002–2003. Grain yield of individual farmerswith the FFP and SSNM collected from on-farm research were specified as inputsfor the yield in simulations with the model. Simulations used rates and timing of fer-tilizer application measured for SSNM and FFP in on-farm research at each loca-tion. For all simulations, we used urea as the fertilizer N because it is commonlyused by most farmers in all the sites of this study.

2.3.3. Assessment of the DNDC model

The performance of the DNDC model was assessed by simulating N losses at twosites in the Philippines with published data on N gas losses. We used the work ofBronson et al. (1997a,b) on N2O emissions at IRRI (Los Banos) to evaluate theN2O emission as simulated by DNDC – where soil and climate data collected fromthe site during the years of the study (1993–1994) were used as inputs. We also con-ducted a simulation of N losses in a farmers’ field in Nueva Ecija in 2003 using farm-er’s fertilizer rates and timing of application and compared to the work of De Dattaet al. (1988) where losses of fertilizer N were determined in 15N balance studies con-ducted in a nearby field. Comparison was done with the farmer’s fertilizer split treat-ment, where fertilizer was applied in two splits during the crop season in 1985. Thesplitting of fertilizer N application used in the simulation was comparable to the 15Nbalance study, although the rates were different. We also evaluated the crop Nuptake predicted by the model by comparing it with published research findings.Soil, climate, and crop parameter inputs collected or measured on-farm in 2003 wereused for simulating N losses.

2.3.4. Estimating global warming potential from N2O emissions

Global warming potential (GWP) is a quantified measure of the globally averagedrelative radiative forcing impacts of a particular greenhouse gas (USEPA, 2002). It isexpressed as carbon dioxide equivalent. We calculated GWP contributed by nitrousoxide (N2O) emissions as follows (Li et al., 2004):

GWP ¼ ðN2O flux=28Þ � 44 � 310

GWP is expressed in carbon dioxide equivalents (CDE) as kg CO2 equiva-lent ha�1 year�1, whereas N2O flux is expressed in kg N ha�1 year�1. Yields and fer-tilizer N rates measured in on-farm research were used to express GWP on a yieldand fertilizer N basis.

Each simulation with the model was analyzed for total N losses through gases(NH3, N2O, NO, and N2) and leaching. The losses of N from fertilizer were calcu-lated as the difference between simulated N losses when fertilizer was added as

M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24 7

recorded in on-farm research and simulated N losses when no fertilizer N wasapplied. Emissions of N2O during crop and fallow periods were evaluated for doublecrop transplanted rice and double crop direct-seeded rice systems in selected cases.

2.3.5. Number of samples and statistical analysisFive individual (farm) simulations were conducted per treatment at each site or

crop establishment method. We compared simulations of two treatments – FFPand SSNM – with input data collected from the same farmer’s field for the samecropping period. There were two sites in India and three in southern Vietnam repre-senting three provinces of the Mekong Delta (Section 2.1). In the case of the Philip-pines, there was only one site but a distinction was made between two cropestablishment methods: (1) transplanted rice for both dry and wet seasons (TPR–TPR) and (2) direct-seeded rice in the dry season followed by transplanted rice inthe wet season (DSR–TPR).

On-farm data on grain yield and fertilizer N as well as simulated outputs werenormally distributed and homogeneity of variance was indicated (Bartlett test).The analysis of variance with mixed model using SAS (version 8.02) was appliedfor each site (or crop establishment method) with outputs from five farms as repli-cates. Mean separation for significant effects was done using least significant differ-ence (LSD) test.

2.4. Assessment of economic benefits

2.4.1. Collection of economic data through focus group discussions (FGD)

In all the locations, we collected economic data through focus group discussions(FGD). Focus groups are a form of group interview that capitalizes on communica-tion between research participants in order to generate data, and they are particu-larly suited to the study of attitudes and experiences (Kitzinger, 1995).

We conducted FGD sessions at six locations (provinces or districts) in three coun-tries, the same sites where we conducted the on-farm trial on SSNM. Two FGD ses-sions per location were held in India and Vietnam. In the Philippines, there were fourFGD sessions: two with farmers establishing rice by transplanting and two withfarmers practicing direct seeding in the dry season. For each group session, therewas one facilitator and there were two or three recorders documenting discussionproceedings. All discussions were conducted in the local language.

2.4.2. FGD groups and compositionTo assess the economic benefits derived from SSNM, we compared two groups of

farmers – SSNM and non-SSNM. Farmers who collaborated in the on-farm evalu-ation of SSNM in 2002–2003 or before comprised the SSNM group; while farmerswho had no involvement with SSNM trials belonged to the non-SSNM group.Non-SSNM farmers came from the same or nearby villages as those of the SSNMfarmers. FGD sessions for the two groups were conducted separately; each sessionconsisted of 8–12 farmers coming from two to three villages within a district. Farm-ers for both groups were those directly managing the day-to day field activities and

8 M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24

were mainly male and land owners. Farm size holding of the farmers ranged from 0.2to 4 ha in India, <0.1 to 4 ha in the Philippines, and 0.25 to 3 ha in Vietnam. Averagefarm size and crop establishment method of the two groups are shown in Table 2.For this analysis, we compared two groups of farmers having different fields ratherthan just two treatments on the same field, which was the case for the environmentalanalysis.

2.4.3. Analysis of data

For both SSNM and non-SSNM groups, data collected were specific for the cropyear 2002–2003 consisting of one high-yielding and one low-yielding season. Eachfarmer provided data on the following: (1) crop establishment operations, (2) weed-ing and herbicide use, (3) fertilizer management, (4) insecticide and fungicide use, (5)harvesting and threshing operations, and (6) grain yield production. In addition, theSSNM groups were requested to give their observations on differences betweenSSNM and non-SSNM (i.e., FFP) crops in terms of crop management, quality ofproduced grain, pest incidence, and cash returns from buyers or traders. Other datathat were not covered during the FGD such as costs of irrigation water and landpreparation were obtained from research or extension staff at each site. Prevailingprices of rice (harvest) and inputs during the period covered, obtained also fromthe groups, were used in the analysis.

Economic benefits were assessed by using the following relations:

Gross revenue = Grain yield (kg ha�1) · Farm gate rice price (US$ kg�1);Total input cost = Fertilizer cost + Seed cost + Irrigation water cost + Pesticidecost + Labour cost;Net benefit = Gross revenue � Total input cost;Gross return above fertilizer cost (GRF) = Gross revenue � Fertilizer cost.

Table 2Average farm size in 2002–2003 of farmers that participated in focus group discussions in three countries

Country District/Province Crop establishmenta Farm size (ha)

SSNM Non-SSNM

India Aduthurai TPR–TPR 2.8 3.1Thanjavur TPR–TPR 5.6 2.1

Philippines Nueva Ecija DSR–TPR 1.2 1.2TPR–TPR 0.8 1.3

Vietnam An Giang DSR–DSR 1.1 1.1Cantho DSR–DSR 0.7 1.6Tien Giang DSR–DSR 0.8 0.7

Values shown are means of 8–12 replicates (farms).a Crop establishment method: TPR–TPR = transplanted rice for both dry and wet seasons; DSR–

TPR = direct seeded rice in the dry season followed by transplanted rice in the wet season; DSR–DSR = direct seeded rice for both dry and wet seasons.

M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24 9

3. Results and discussions

3.1. Environmental assessment

The prospects for wide-scale dissemination and farmer uptake of SSNM appearencouraging given the favourable financial returns with SSNM and the existencein major Asian rice-growing areas of partnerships among research, extension, andprivate sector organizations for dissemination of SSNM. It is therefore opportuneto assess probable environmental benefits in addition to the financial benefits asso-ciated with adoption of SSNM. Fertilizer N applied to submerged rice soils is proneto large losses through mechanisms as ammonia volatilization and nitrification–deni-trification (Buresh et al., 2006). Nitrous oxide, a greenhouse gas, is one of the endproducts of nitrification–denitrification. It is quite likely that applying fertilizer Nat rates and times to better match the dynamic needs of the rice plant, as practicedwith SSNM, can lead to reduced losses of fertilizer N as gases including N2O.

3.1.1. Assessment of the DNDC model

Much information on the processes and magnitudes of N losses from lowland riceecosystems is available (Buresh et al., 2006). Before conducting simulations at thesites included in this study, we ensured that the model was generating realistic out-puts by comparing results with published magnitudes of N losses. Table 3 showsparameter inputs used for assessing the DNDC model at two locations in the Phil-ippines. The assessment conducted at IRRI (Los Banos, Philippines) for 1993–1994

Table 3Soil and climate inputs used for model assessment and on-farm simulation of N losses with DNDC

Country Location/system

Claycontent(%)

Soil pH(H2O, 1:1)

Soil org. C(g kg�1)

Bulkdensity(g cm�3)

Meanannualtemperature(�C)

Annualprecipitation(mm)

Model assessmenta

Philippines IRRI 49 6.4 20 0.8 27.1 2107Nueva Ecija 31 5.5 13 1.0 27.4 2126

On-farm simulationa

India Aduthurai 39 7.3 10 0.9 28.4 670Thanjavur 9 7.1 7 1.2 28.4 887

Philippines DSR–TPRb 29 5.6 15 1.0 27.4 2126TPR–TPRc 29 5.4 14 1.0 27.4 2126

Vietnam An Giang 44 4.5 20 0.8 27.8 1800Cantho 70 4.6 30 0.8 27.8 1800Tien Giang 49 4.6 20 0.8 27.8 1800

a For all simulations and locations, the initial soil NO3-N was 0.2 mg kg�1 and the initial soil NH4-Nwas 15 mg kg�1.

b Direct seeded rice in the dry season followed by transplanted rice in the wet season.c Transplanted rice for both dry and wet seasons.

Table 4Assessment of simulated annual N gas losses from soil and fertilizer with published data on N losses

Location Gas Simulated annual N gas losses

From soil (kg N ha�1) From fertilizer

kg N ha�1 % of fertilizer

IRRIa N2 4.4 30.2 9.4N2Ob 0.4 2.1 0.7NH3 9.8 112 35NO 0.3 1.8 0.6Total 15 146 46

Nueva Ecijac N2 3.9 7.6 3.8N2O 2.1 2.4 1.2NH3 14 67 34NO 1.1 1.7 0.9Total 21 79 40d

Simulation covered two (dry and wet) irrigated transplanted rice crops with continuous flooding duringcrop seasons in 1993 at IRRI (Los Banos) and in 2002–2003 at Nueva Ecija, Philippines.

a The total amount of applied fertilizer N for two seasons was 320 kg N ha�1 (200 kg N in the dry and120 kg N in the wet season) applied in four splits per cropping season (Bronson et al., 1997a,b).

b Bronson et al. (1997a,b) reported a total N2O emission of 2.29 kg N ha�1 for two irrigated rice cropswith continuous flooding during the crop and two fallows in 1993.

c The amount of fertilizer N was 114 kg N ha�1 in the dry season applied in three splits and85 kg N ha�1 in the wet season applied in two splits (farmer’s fertilizer practice).

d De Datta et al. (1988) reported a 37% loss from fertilizer N in transplanted rice with farmer’s fertilizersplit determined in 15N balance studies conducted on station in Nueva Ecija in 1985.

10 M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24

yielded an annual N2O emission (2.5 kg N ha�1, Table 4) very close to what wasobserved (2.29 kg N ha�1) by Bronson et al. (1997a,b) during the same time or per-iod. Simulation of total N losses conducted for a field in Nueva Ecija in 2003 withfarmers’ fertilizer practice showed a 40% total loss of N from fertilizer (Table 4),which approximated the 37% reported by De Datta et al. (1988) determined by15N balance technique conducted in 1985 in a nearby location. The N rates usedin the simulation were 114 kg N ha�1 in the dry season and 85 kg N ha�1 in thewet season as compared to 87 kg N ha�1 in the N balance study for a dry seasoncrop. The number of fertilizer splits was comparable between farmers’ fields usedin the simulation (2–3 splits per crop) and the 15N balance study (two splits percrop). Crop N uptake predicted by the model was within 15–18 kg N per ton of grainyield, which is consistent with published research findings (Witt et al., 1999, 2002).

3.1.2. Simulated N losses

In the absence of direct measurements of N gas fluxes at the research locations inthe three countries, we used the DNDC model to estimate N losses for SSNM andFFP in daily time steps for an annual crop production cycle with two rice crops andtwo fallow periods at each location. Total simulated annual losses of fertilizer Nthrough gases and leaching at the three countries are given in Table 5. Virtuallyall the simulated losses were as N gases (NH3, N2O, NO, and N2).

Table 5Simulated annual losses of N from fertilizer N applied by the farmers’ fertilizer practice (FFP) and site-specific nutrient management (SSNM) in three countries for 1 year with two irrigated rice crops in 2002–2003

Country Location Treatment Appliedfertilizer(kg N ha�1)

Simulated N losses (kg N ha�1 year�1)

N2 N2O NH3 NO Leached Total

India Aduthurai FFP 258 9.4 1.2 87 0.9 0.18 98SSNM 256 9.7 1.2 76 1.0 0.17 88Prob > ta 0.75 0.87 0.92 0.03 0.78 0.96 0.13

Thanjavur FFP 201 18.2 0.9 100 1.6 2.10 122SSNM 241 29.4 1.3 114 2.4 3.45 151Prob > t <0.001 <0.001 0.06 0.008 <0.001 <0.001 0.002

Philippinesb Nueva Ecija FFP 249 20.8 4.7 109 3.3 1.16 139SSNM 225 13.1 4.3 80 3.0 0.92 102Prob > t 0.07 0.005 0.52 0.01 0.44 0.36 0.01

Vietnam An Giang FFP 197 11.6 3.2 78 2.1 0.21 95SSNM 174 8.8 2.3 62 1.5 0.14 75Prob > t 0.006 0.03 <0.001 0.01 <0.001 <0.001 0.01

Cantho FFP 224 17.8 2.6 127 1.7 0.22 150SSNM 189 12.8 1.6 97 1.0 0.15 112Prob > t <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001

Tien Giang FFP 218 14.8 3.05 98.2 2.0 0.26 118SSNM 188 9.1 2.01 70.8 1.2 0.17 83Prob > t 0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001

Values shown are means of five farms per location.a Probability level of the difference between FFP and SSNM.b There was no crop establishment · treatment interaction; hence values shown are means of the two

crop establishment methods (n = 10).

M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24 11

The simulated losses of fertilizer N are comparable in magnitude to the losses offertilizer N directly determined in 15N balance studies where N losses are estimatedby the portion of added 15N-labeled fertilizer not recovered in the plant and soil atrice maturity. The simulated mean fertilizer N loss of 45% with SSNM at the Phil-ippines location (Table 5) is comparable to fertilizer N losses of 31% and 39% mea-sured in 1985 by the 15N balance technique for researcher’s fertilizer split in direct-seeded and transplanted rice, respectively at a site near the on-farm research (DeDatta et al., 1988).

Simulated N loss for farmers’ fertilizer practice (55%) is somewhat higher than the26% (direct-seeded rice) and 37% (transplanted rice) losses of fertilizer N measuredfor farmers’ fertilizer split in the above mentioned 15N balance research (De Dattaet al., 1988). Such difference could be due to the higher fertilizer rates (157 kg N ha�1

in the dry season and 93 kg N ha�1 in the wet season) used in the simulation of farm-ers’ fields than the N rate (87 kg N ha�1 in dry season) used by researchers in the 15Nbalance technique.

The simulated mean fertilizer N losses of 56% (FFP) and 49% (SSNM) at the Viet-nam location (Table 5) are higher than the 28% loss of fertilizer N reported by the

12 M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24

15N balance technique in Cantho Province for direct-seeded rice (Hung, 1995). Thehigher total simulated N losses than those measured by Hung (1995) in Vietnamcould be partly due to the lower seed rate used by farmers in 2002–2003 (120–154 kg ha�1) in this study compared to the 200 kg ha�1 used in 1989–1990 (Hung,1995). It can be due to overestimation of NH3 volatilization losses by the model espe-cially in Cantho Province, where NH3 losses were >50% of applied fertilizer N.

The recovery efficiency of N reported in previous SSNM studies including sites inthis study was about 40% (Dobermann et al., 2002), indicating that fertilizer N lossespredicted in this study by the DNDC model are within values that could occur in ricefields at those locations, assuming that only a relatively small fraction of fertilizer Nremains in the soil.

The much larger loss of fertilizer N as NH3 rather than N2O and N2 (Table 5) isconsistent with previous direct measurements of fertilizer N losses by NH3 volatili-zation and nitrification–denitrification in irrigated lowland rice ecosystems with pud-dled soils (Buresh and De Datta, 1990). Direct measurements of 15N-labeled N2 andN2O flux from 15N-labeled fertilizer N consistently revealed negligible to very smalllosses of fertilizer N as N2O and N2 when rice soils remained flooded (Buresh and DeDatta, 1990). Losses of N as N2O and N2 can however increase following periods ofsoil drying and wetting, especially during the fallows between rice crops (Bronsonet al., 1997a,b).

Simulations with the DNDC model indicated significantly lower losses of fertilizerN with SSNM than FFP in Vietnam and the Philippines (Table 5). These findingssuggest that N management with the LCC is effective in reducing losses of fertilizerN. In India, the total loss from fertilizer N with SSNM tended to decrease in Adu-thurai; while losses in Thanjavur increased with SSNM due to significantly lowerrates of fertilizer N applied by farmers (FFP) at that location.

Based on the simulations, the highest loss of fertilizer N was as NH3 (Table 5),which is not considered as a greenhouse gas. Except for the Thanjavur site in India,fertilizer N losses as NH3, N2O, N2, and NO were generally reduced with SSNM.The greater simulated loss of N gases for FFP than SSNM can be attributed to themarkedly higher rates of fertilizer N application by farmers, often in doses>40 kg N ha�1, during early stage of crop growth when the crop is still smalland demand for added N is small. Numerous studies directly measuring NH3 vol-atilization and determining N losses by the 15N balance technique report higher Nlosses from fertilizer broadcast during early growth stages than from fertilizerbroadcast at later stages such as mid-tillering and panicle initiation (Bureshet al., 2006).

3.1.3. Fertilizer N use efficiency

On-farm research at the locations in this study confirmed higher efficiency of fertil-izer N use with SSNM. The rates and timing of fertilizer use and grain yield were mea-sured in adjacent SSNM and FFP fields managed by farmers. Annual use of fertilizerN for two rice crops was only slightly lower for SSNM than FFP in Vietnam and thePhilippines (Table 5). In India, annual fertilizer N was the same for SSNM and FFP inAduthurai, while it was higher for SSNM than FFP in Thanjavur (Table 5). The

M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24 13

reductions in fertilizer use with SSNM averaged 10% in the Philippines, and 14% inVietnam. Measured grain yields were significantly (p < 0.01) higher in SSNM thanFFP fields at all locations in the three countries (Table 6). Yield increase (of two crops)due to SSNM ranged from 0.3 Mg ha�1 in An Giang (Vietnam) to 1.2 Mg ha�1 inThanjavur (India).

Partial factor productivity of fertilizer N (PFPN), defined as kg rice grain yield perkg of fertilizer N applied (Cassman et al., 1998), was significantly increased withSSNM in the Philippines and Vietnam (Table 6). The increase in PFPN with SSNMwas greater than the reduction in fertilizer N because of increased yields with SSNM.SSNM on average increased PFPN by 21% in the Philippines and 26% in Vietnam.An increase in PFPN could reasonably be associated with increased plant use of fer-tilizer N and reduced loss of fertilizer N. The simulated reduction in fertilizer N losswith SSNM is consistent with the measurements for fertilizer N use efficiency asPFPN in on-farm research at the locations.

Table 6The contribution of simulated N2O emission from soil and fertilizer to annual global warming potential(GWP) as influenced by nutrient management practices based on measured on-farm annual rice yields andfertilizer N application rates in three countries for two irrigated rice crops in 2002–2003

Country Location Nutrientmanagementpracticea

Annualyield(Mg ha�1)

Partialfactorproductivity(kg grain kg�1

applied N)

GWPb expressed ascarbon dioxideequivalent (CDE)

(kg Mg�1

grain yield)(kg kg�1

fertilizer N)

India Aduthurai FFP 12.4 48 61 2.8SSNM 12.9 51 58 2.9Prob > tc 0.004 0.23 0.66 0.85

Thanjavur FFP 11.6 58 81 4.7SSNM 12.8 53 89 4.7Prob > t <0.001 0.01 0.31 0.87

Philippinesd Nueva Ecija FFP 9.7 39 415 15.8SSNM 10.4 47 366 16.7Prob > t 0.004 0.02 0.22 0.38

Vietnam An Giang FFP 9.3 47 212 9.6SSNM 9.6 55 160 8.5Prob > t 0.24 <0.001 <0.001 <0.001

Cantho FFP 9.4 42 180 7.5SSNM 10.3 55 116 6.3Prob > t 0.002 <0.001 <0.001 <0.001

Tien Giang FFP 8.8 41 201 8.0SSNM 9.6 51 130 6.6Prob > t 0.003 <0.001 <0.001 0.003

a FFP is farmers’ fertilizer practice and SSNM is site-specific nutrient management.b GWP based on N2O emissions from soil and fertilizer.c Probability level of the difference between FFP and SSNM.d There was no crop establishment · treatment interaction; hence values shown are means of the two

crop establishment methods (n = 10).

14 M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24

3.1.4. Global warming potential

We limited our environmental assessment of SSNM to global warming potential(GWP) associated with N2O emissions from fertilizer and soil. The GWP expressedper unit of grain yield and unit of fertilizer N was significantly reduced by SSNM attwo sites in Vietnam (Table 6). Treatment differences in India and the Philippineswere not significant, although a reduction in GWP with SSNM was also indicatedin the Philippines. The highest GWP from simulated N2O emissions was in the Phil-ippines followed by Vietnam and India.

The simulated reduction in GWP with use of SSNM averaged 49 kg CDE Mg�1

grain in the Philippines and 62 kg CDE Mg�1 grain in Vietnam (Table 6). This cor-responded to an average reduction of 56 kg CDE Mg�1 grain across the two coun-tries or a 22% reduction in CDE per unit of rice produced. There was noreduction in GWP due to SSNM in India.

In India, the efficiency of fertilizer N (e.g. PFPN) and yields were already relativelyhigh with FFP (Table 6). The simulated N2O emissions were low (Table 5), perhapsas a reflection of the already relatively high efficiency of fertilizer N use. In such case,SSNM was able to increase yield with the same (Aduthurai) or increased (Thanjavur)rate of fertilizer N without additional N2O emission per unit of grain yield or fertil-izer used.

In the Philippines, greater yield with less fertilizer N was obtained with SSNM.The efficiency of fertilizer N use was increased while the N2O emissions per unitof grain yield slightly decreased. The simulated emissions were highest at this site,and most emissions of N2O occurred in the fallow periods between rice crops, wherea sizeable portion of the N2O emission originated from soil N, rather than added fer-tilizer N (Table 7).

In Vietnam, like the Philippines, greater yield with less fertilizer N was obtainedwith SSNM. This led to increased fertilizer N use efficiency and reduced N2O emis-sion per unit of grain yield. According to the simulation, much of the N2O emissionoccurred between rice crops and in the period following direct seeding of germinatedseed on saturated soil when the soil would undergo wetting and drying before per-manent flooding (Table 7, Fig. 1). Our research monitored only two rice cropsand simulated only two rice crops per year in the Mekong Delta, but farmers oftengrow three rice crops per year especially in Cantho and Tien Giang Provinces. Undersuch conditions with less soil drying between rice crops, the emissions of N2O coulddecrease rather than increase. We therefore report in Table 7 simulations with tworice crops per year for only the An Giang site. Our simulations suggest the meritof research to assess whether fertilizer N is lost as N2O soon after direct seedingin relatively hot dry seasons with soil drying and wetting before permanent flooding,and whether loss of N as N2O is important during soil drying before rice harvest(Fig. 1).

This research indicated relatively low emissions of N2O, which is consistent withfield measurements (Bronson et al., 1997a,b). Use of SSNM never resulted inincreased emissions of N2O per unit of grain yield, and in environments where higheryield could be obtained with less fertilizer N, the use of SSNM could result inreduced N2O emissions per unit of grain yield.

Table 7Simulated N2O emission derived from soil and fertilizer during an annual cycle of two rice crops and twofallows in 2002–2003 three countries

Country Location Crop establishmenta Period N2O emission (kg N ha�1)

Soilb Fertilizerc

FFP SSNM

India Aduthurai TPR–TPR WS rice 0.00 0.03 0.03WS–DS fallow 0.11 0.45 0.49DS rice 0.00 0.01 0.01DS–WS fallow 0.23 0.73 0.66Total 0.34 1.21 1.19

Thanjavur TPR–TPR WS rice 0.01 0.13 0.24WS–DS fallow 1.03 0.31 0.47DS rice 0.00 0.19 0.26DS–WS fallow 0.02 0.24 0.30Total 1.06 0.87 1.28

Philippines Nueva Ecija TPR–TPR DS rice 0.00 0.05 0.02DS–WS fallow 0.14 0.81 0.24WS rice 0.00 0.04 0.05WS–DS fallow 2.71 3.62 4.36Total 2.86 4.52 4.66

DSR–TPR DS rice 1.85 1.79 0.24DS–WS fallow 0.28 1.11 0.82WS rice 0.01 0.04 0.05WS–DS fallow 1.80 1.97 2.83Total 3.94 4.92 3.94

Vietnam An Giang DSR–DSR DS rice 0.51 1.58 1.52DS–WS fallow 0.00 0.02 0.02WS rice 0.01 0.21 0.12WS–DS fallow 0.18 1.37 0.68Total 0.70 3.19 2.33

a TPR–TPR = transplanted rice for both dry and wet seasons; DSR–TPR = direct seeded rice in the dryseason followed by transplanted rice in the wet season; DSR–DSR = direct seeded rice for both dry andwet seasons.

b Based on N2O emissions when fertilizer is not applied.c Calculated as the difference between total N2O emissions when fertilizer is applied and N2O emissions

when fertilizer is not applied.

M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24 15

3.1.5. Nitrous oxide emissions derived from soil and fertilizer

Nitrification–denitrification in lowland rice ecosystems can also lead to emissionsof N2O derived from soil N, which can be strongly influenced by soil drying and wet-ting especially between rice crops (Bronson et al., 1997b). In order to assess thedynamics of N2O emissions derived from soil and fertilizer we report simulatedfluxes of N2O during various periods in an annual cropping cycle.

Table 7 reports simulated emissions of N2O from soil and fertilizer at the differentlocations. In double crop transplanted rice (i.e., India, Philippines), the period ofhighest emission of soil N-derived as well as fertilizer-derived N2O was between ricecrops – from the harvest of one crop to the end of the fallow before the next crop

Sow

ing

FFP

Flooding

Harvest

SowingDrainage

Harvest

Flooding

Drainage

SSNM

N2O emission (kg N ha-1 day-1)

0.00

0.04

0.080.50

1.00

1.50

2.00 DS rice WS rice

Day1 31 59 90 120 151 181 212 243 273 304 334 365

0.00

0.04

0.080.50

1.00

1.50

2.00 WS riceDS rice

Fig. 1. Emission of N2O derived from fertilizer N as simulated by the DNDC model based on measured Napplication rates and timing for farmers’ fertilizer practice (FFP) and site-specific nutrient management(SSNM) for dry season (DS) 2002–2003 and wet season (WS) 2003 at Chau Phu District, An GiangProvince, Vietnam.

16 M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24

(Table 7). This emission of N2O apparently originated from nitrification–denitrifica-tion associated with alternate soil drying and wetting during the onset of wet seasonrains (Bronson et al., 1997b).

High peaks of N2O emissions were observed after soil flooding for land prepara-tion, especially during the land preparation for dry season rice (Table 7). Nitrate isknown to accumulate during fallows between rice crops and then undergo rapid losswhen soil is flooded (Buresh and De Datta, 1991). Soil drying is more favoured afterthe wet season crop than the dry season crop at the location in the Philippines. Inaddition, residual N was also more likely after the wet season than the dry season cropdue to the high rates of fertilizer N application in the wet season and low N use effi-ciency due to climatic factors in that season. The simulations suggest negligible emis-sions of N2O when soil is continuously flooded, which is consistent with results fordirect field measurements (Buresh and De Datta, 1990; Bronson et al., 1997a).

M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24 17

For double crop direct-seeded rice such as in Vietnam, highest N2O emissionswere observed during the dry season crop and during the fallow after the wet seasoncrop (Table 7, Fig. 1). The high emissions at the early stage of the dry season cropcould be due to alternate wetting and drying since the field is only flooded continu-ously at about 3 weeks after rice sowing. The absence of rainfall during this periodmay allow soil oxidation that could favour nitrification; while irrigation events andthe subsequent flooding of the field may favour losses of N through denitrification.Minimal emissions were observed during the fallow after the dry season and alsoduring the wet season crop (Table 7, Fig. 1), probably due to the very short fallowafter the dry season allowing limited soil oxidation hence low nitrification.

With SSNM, the timing and doses of fertilizer N are adjusted through use of theLCC to match the crop demand for added N. The reduced N2O emissions withSSNM apparently result from reduced build up of soil inorganic N during periodswhen soil undergoes alternate drying and wetting. The SSNM practices adoptedby farmers for N management should therefore ensure negligible or minimal inor-ganic soil N at the end of the rice-growing season when soil dries and N2O can beemitted through nitrification–denitrification. Some N2O emitted from lowland riceecosystems originates from soil N. Effective management of fertilizer N should alsoensure that plant-available N originating from soil, manure, and crop residues is uti-lized by the crop and does not remain in the soil at the end of the rice-growingseason.

3.2. Economic assessment

3.2.1. Yield comparison between SSNM and non-SSNM farmers

Based on the FGD, rice yields across all locations were consistently higher in thedry than the wet season, and yields were higher for farmers practicing SSNM thanfor farmers not practicing SSNM (Table 8), consistent with results obtained fromon-farm research of this study. In India, yields across the two seasons were 17%higher for SSNM than non-SSNM farmers. In the Philippines, average yields forSSNM were higher by 10% in the dry season and by 2% in the wet season. In Viet-nam, yields were only slightly higher for SSNM at locations in Cantho and TienGiang Provinces. In An Giang, SSNM farmers did not report higher yields thannon-SSNM farmers (Table 8), presumably because most non-SSNM farmers partic-ipating in the FGD were previously exposed through local extension to SSNM tech-nologies, and they had consequently already altered their nutrient managementpractices. Results from on-farm trials also showed the lowest yield increase withSSNM in An Giang compared to Cantho and Tien Giang (Table 6).

Based on the average of two crop seasons, yields were higher for SSNM farmersthan non-SSNM farmers by 0.8 Mg ha�1 in India, 0.3 Mg ha�1 in the Philippines,and 0.2 Mg ha�1 in Vietnam. The overall average yield increase with SSNM basedon FGD was 0.4 Mg ha�1 (8%), which is less than the yield increase of 0.7 Mg ha�1

observed from the on-farm study (Table 6); but in terms of percentage increase,results from both FGD and on-farm measurements were comparable (8% withFGD and 7% with on-farm). These results are also comparable with the

Table 8Rice yield for SSNM and non-SSNM farmers in three countries as determined from focus groupdiscussions for two irrigated rice crops in 2002–2003

Country District/Province Croppingseason

Grain yield

SSNMa

(Mg ha�1)Non-SSNMb

(Mg ha�1)Incrementc (%)

India Aduthurai Dry 5.5 4.9 11Wet 4.9 3.9 25

Thanjavur Dry 5.5 4.6 19Wet 4.4 3.9 13

Average 5.1 4.3 17

Philippines Nueva Ecija Dry 7.0 6.4 10Wet 5.3 5.1 2

Average 6.1 5.8 6

Vietnam Cantho Dry 6.8 6.7 1Wet 4.3 4.0 7

An Giang Dry 7.0 7.1 �1Wet 5.2 5.2 0

Tien Giang Dry 5.7 5.3 9Wet 4.5 4.3 5

Average 5.6 5.4 4

Overall average 5.6 5.2 8

a SSNM farms were using site-specific nutrient management.b Non-SSNM farms were using farmers’ fertilizer practice.c Refers to the percentage increase in yield for SSNM farmers as compared with non-SSNM farmers.

18 M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24

0.36 Mg ha�1 (7%) higher yield measured in SSNM plots than in plots with the FFPin 179 farmers’ fields across five Asian countries in 1997 to 1999 (Dobermann et al.,2002). By 2002–2003 at the time of this study, researchers had more time to improveSSNM with collaborating farmers. Hence higher yield gains with SSNM in 2002–2003 (as shown from on-farm results) than in 1997–1999 could be expected. OurFGD would however underestimate yield gains with SSNM when SSNM hadalready diffused to some ‘non-SSNM’ farmers who had consequently altered theirfertilizer practices, which appears to be the case in at least the An Giang location.

With SSNM, fertilizer N, P, and K are simultaneously adjusted to optimize thesupply of nutrients to rice. The management of all three nutrients could consequentlydiffer between SSNM and non-SSNM farmers, and it is therefore not possible toascertain which nutrient contributes to the reported yield gains with SSNM in theFGD. It is nonetheless likely that a sizable portion of the yield gain with SSNMcould arise from improved N management, which through improved timing of fer-tilizer N with the LCC is an integral component of SSNM.

3.2.2. Contribution of different resource inputs to total production costsThe relative contribution of various resource inputs to rice production costs for

SSNM and non-SSNM farmers were comparable at the India and the Philippineslocations (Table 9). Labour was the main input cost, accounting for 63–72% of

M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24 19

the total input costs at these two locations. Fertilizers were the second highest inputcost representing 14–20% of the total cost. This was followed by seed (6–9%), irriga-tion water (4–5%), and pesticides (1–6%). Input costs, particularly labour, werelower in Vietnam largely because of direct seeding rather than transplanting for cropestablishment. Pesticides represent a much larger portion of the total input cost inVietnam (23%).

Farmers in the Philippines used pesticides primarily to control weeds and snails,especially during the dry season when they practice direct seeding; while farmers inVietnam commonly used pesticides to control insects and diseases in addition to con-trolling weeds. Farmers in Vietnam used more seeds than farmers in the Philippinesbecause their method of crop establishment is direct seeding for both dry and wetseasons; while farmers in the Philippines commonly practiced direct seeding onlyin the dry season. In addition, the seed rate used in Vietnam is higher than the rateused in the Philippines. In India, the method of crop establishment is transplantingfor both dry and wet seasons.

The use of SSNM did not reduce total input costs (Table 9). In India, farmerspracticing SSNM had reduced seed and pesticide inputs; but labour costs werehigher with SSNM, and total input costs were comparable for SSNM and non-SSNM farmers. In the Philippines, the farmers practicing SSNM reduced fertilizer

Table 9Average annual cost of resource inputs for SSNM and non-SSNM farmers in three countries asdetermined from focus group discussions for two irrigated rice crops in 2002–2003

Country Resource inputs Cost (US$ ha�1 year�1) Fraction of total cost (%)

SSNMa Non-SSNMb SSNM Non-SSNM

India Fertilizers 98 90 15 14Seeds 48 58 7 9Irrigation water 32 32 5 5Pesticidesc 8 28 1 4Labour 478 454 72 68Total 664 662

Philippines Fertilizers 135 139 19 20Seeds 46 46 6 7Irrigation water 29 29 4 4Pesticides 39 39 6 6Labour 454 438 65 63Total 703 690

Vietnam Fertilizers 140 136 22 22Seeds 60 62 9 10Irrigation water 42 40 7 6Pesticides 146 144 23 23Labour 246 246 39 39Total 634 628

a SSNM farms were those practicing site-specific nutrient management.b Non-SSNM farms were those using farmers’ fertilizer practice.c Pesticides include insecticides, fungicides, herbicides, and molluscicides.

20 M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24

inputs; but labour costs were higher with SSNM, and total input costs were notreduced with SSNM. In Vietnam, the input costs were comparable for SSNM andnon-SSNM farmers, and SSNM farmers had not reduced pesticide inputs.

The rice production costs reported in our study (Table 9) are generally compara-ble to production costs reported for 1999 by Moya et al. (2004) at these locations.One difference is the lower pesticide costs reported by Moya et al. (2004) for farmersin Cantho Province, Vietnam in 1999 (44 US$ ha�1 year�1) than reported in ourstudy for farmers in three provinces in the Mekong Delta of southern Vietnam in2002–2003 (144 US$ ha�1 year�1). The high input cost reported in our study for pes-ticides in the Mekong Delta and the absence of reduced pesticide use with adoptionof SSNM suggests that an opportunity remains in the Mekong Delta for reducinguse of pesticides, especially fungicides because they represent a substantial portionof the total pesticide costs.

3.2.3. Economic performance of SSNM and non-SSNM farmers

SSNM was financially attractive in the three countries (Table 10). Gross revenueand GRF (gross return above fertilizer costs) were consistently higher for SSNMthan non-SSNM farmers across the three countries. The adoption of SSNM didnot reduce total inputs costs (Tables 9 and 10), but it provided added net benefitsof 168 US$ ha�1 year�1 in India, 106 US$ ha�1 year�1 in the Philippines, and34 US$ ha�1 year�1 in Vietnam (Table 10). The relatively low net benefits for SSNMin Vietnam probably resulted because non-SSNM farmers were already benefitingfrom improved nutrient management arising from diffusion of SSNM technologythrough local extension.

Our findings of increased GRF with SSNM from FGD based on the 2002–2003cropping season are consistent with earlier reports of higher GRF with SSNM thanFFP evaluated in adjacent plots in farmers’ field before 2001 (Dawe et al., 2004;Gines et al., 2004; Nagarajan et al., 2004; Tan et al., 2004). In our study, the addednet benefit with SSNM was clearly associated with increased revenue (Table 10) aris-ing from higher yield (Table 8) rather than reduced input costs (Table 9). Fertilizer isan essential ingredient for high yield, and the greatest opportunity for increasing netbenefit through SSNM in lowland irrigated rice ecosystems arises from increasinggrain yield rather than from reducing fertilizer costs. The greatest opportunity forreducing input costs are for pesticides in Vietnam and for labour at all locations(Table 9).

3.2.4. Qualitative differences between SSNM and non-SSNM crop performanceAfter the introduction of SSNM, farmers observed benefits associated with grain

quality, crop management, pest incidence, and cash returns from rice buyers (Table11). Some of the SSNM farmers in the FGD across the three countries indicated thatafter adoption of SSNM the grains were less chalky, more filled, more uniform insize, heavier, and brighter in colour without dark spots. Farmers in Vietnam alsoreported higher milling recovery with SSNM. With the adoption of SSNM, farmersfrom all three countries often reduced early applications of fertilizer N, applied a lar-ger portion of the fertilizer N near panicle initiation, and applied fertilizer P at earlier

Table 10Annual financial benefits for SSNM and non-SSNM farmers in three countries as determined from focusgroup discussions for two irrigated rice crops in 2002–2003

Country Performance parameter Amount (US$ ha�1 year�1)

SSNM Non-SSNM Differencea

India Gross revenue 1184 1014 170Net benefitb 520 352 168GRFc 1086 924 162

Philippines Gross revenue 1880 1762 118Net benefit 1177 1071 106GRF 1745 1623 122

Vietnam Gross revenue 1468 1428 40Net benefit 834 800 34GRF 1326 1290 36

Average Gross revenue 1511 1401 109Total input cost 667 660 7Net benefit 844 741 103GRF 1386 1279 107

a Difference = SSNM minus non-SSNM.b Net Benefit = Gross revenue � Total input cost.c GRF = gross return above fertilizer costs.

M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24 21

growth stages. Lodging was no longer reported as a problem in Vietnam after use ofSSNM because of improved use of fertilizer N and application of adequate fertilizerK. Reduced incidence of insects and diseases was also reported by farmers usingSSNM. Pesticide use was reduced by SSNM farmers in India, but SSNM farmersin the Philippines and Vietnam did not have lower costs for pesticides (Table 9)despite the purported reduction in pest incidence with SSNM. Some farmers practic-

Table 11Observations of farmers after the adoption of SSNM, as determined from focus group discussions in threecountries in 2002–2003

Features/characteristics Observation after use of SSNM

Grain quality Less chalkiness (Vietnam)Less unfilled grains (Philippines, Vietnam)Higher milling recovery (Vietnam)More uniform grain size (Vietnam)Bright seed colour without dark spots (India, Vietnam)Heavier grains (India, Philippines, Vietnam)High seed purity (Vietnam)

Crop management No more lodging problem (Vietnam)

Pest incidence Improved N management makes crop less attractiveto pests (India, Vietnam)

Cash returns from buyers Higher by about US$ 0.03 kg�1 due to perceivedhigh seed quality (Vietnam)

22 M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24

ing SSNM in Vietnam reported higher cash returns from the rice buyers because ofperceived better quality of grains with use of SSNM.

4. Conclusions

More effective N management through SSNM can increase the fertilizer N use effi-ciency leading to more grain yield per unit of fertilizer N. This can avoid accumula-tion of inorganic N during periods when crop demand for added N is low and duringperiods, such as at the end of the rice-growing season, when inorganic N remaining insoils can be prone to loss as N2O. Simulations with DNDC provided the first approx-imations for possible benefits of SSNM on reduced GWP associated with reducedemissions of N2O. Increasing the productivity of fertilizer N (yield per unit of fertilizerN applied) might also provide environmental benefits associated with producing andtransporting less fertilizer N per unit of grain produced. SSNM has shown potentialto increase grain yield; hence in assessing the impacts of SSNM, the emissions of Ngases should be expressed per unit of grain yield produced rather than per unit of landarea. Use of pesticides was reduced by farmers practicing SSNM in one (southernIndia) of three study sites. Further estimations and direct measurement of environ-mental benefits associated with wide-scale adoption of SSNM for rice are merited.

The added net benefits associated with use of SSNM as determined through FGDconfirm that SSNM merits wide-scale dissemination. The SSNM practice dissemi-nated to farmers should strive to increase profitability of rice farming throughincreased rice yields. The scope for increasing profit through reduced fertilizer costsis probably small because balanced use of nutrient inputs including potassium andmicronutrients, such as zinc, are increasingly essential for attaining high rice yieldsin Asia. The opportunities for increasing profit through reduced input costs areprobably greatest through management practices that reduce labour costs. Thereis also scope for reducing pesticide costs, where pesticide use is high such as in theMekong Delta of Vietnam.

Acknowledgements

This study was supported by the Swiss Agency for Development and Cooperation(SDC), the International Fertilizer Industry Association (IFA), the InternationalPotash Institute (IPI), and the Potash and Phosphate Institute/Potash and Phos-phate Institute Canada (PPI/PPIC). The authors acknowledge with appreciationthe assistance, suggestions, and recommendations provided by the following scien-tists: Dr. David Dawe, Ms. Fe Gascon, and Ms. Pie Moya of IRRI, and Ms. AuroraCorales of PhilRice, Philippines.

References

Alam, M.M., Ladha, J.K., Khan, S.R., Foyjunnessa, Rashid Harun-ur, Khan, A.H., Buresh, R.J., 2005.Leaf color chart for managing nitrogen fertilizer in lowland rice in Bangladesh. Agron. J. 97, 949–959.

M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24 23

Bronson, K.F., Neue, H.-U., Singh, U., Abao Jr., E.B., 1997a. Automated chamber measurements ofmethane and nitrous oxide flux in a flooded rice soil. I. Residue, nitrogen, and water management. SoilSci. Soc. Am. J. 61, 981–987.

Bronson, K.F., Singh, U., Neue, H.-U., Abao Jr., E.B., 1997b. Automated chamber measurements ofmethane and nitrous oxide flux in a flooded rice soil. II. Fallow period emissions. Soil Sci. Soc. Am. J.61, 988–993.

Buresh, R.J., De Datta, S.K., 1990. Denitrification losses from puddled rice soils in the tropics. Biol. Fert.Soils 9, 1–13.

Buresh, R.J., De Datta, S.K., 1991. Nitrogen dynamics and management in rice-legume cropping systems.Adv. Agron. 45, 2–59.

Buresh, R.J., Reddy, K.R., Van Kessel, C., 2006. Nitrogen transformations in submerged soils. In:Schepers, J.S., Raun, W.R. (Eds.), Nitrogen in Agricultural Soils. ASA, CSSA, and SSSA, Madison,WI.

Buresh, R.J., Witt, C., Ramanathan, S., Chandrasekaran, B., Rajendran, R., 2005. Site-specific nutrientmanagement: managing N, P, and K for rice. Fert. News 50 (3), 25–28, 31–37.

Cassman, K.G., Peng, S., Olk, D.C., Ladha, J.K., Reichardt, W., Dobermann, A., Singh, U., 1998.Opportunities for increased nitrogen use efficiency from improved resource management in irrigatedrice systems. Field Crops Res. 56, 7–38.

Dawe, D., Dobermann, A., Witt, C., Abdulrachman, S., Gines, H.C., Nagarajan, R., Satawathananont,S., Son, T.T., Tan, P.S., Wang, G.H., 2004. Nutrient management in the rice soils of Asia and thepotential of site-specific nutrient management. In: Dobermann, A., Witt, C., Dawe, D. (Eds.),Increasing Productivity of Intensive Rice Systems through Site-Specific Nutrient Management. SciencePublishers, Enfield, NH (USA) and International Rice Research Institute, Los Banos, Philippines, pp.337–358.

De Datta, S.K., Buresh, R.J., Samson, M.I., Wang, K.-R., 1988. Nitrogen use efficiency and nitrogen-15balances in broadcast-seeded flooded and transplanted rice. Soil Sci. Soc. Am. J. 52, 849–855.

Dobermann, A., Witt, C., Dawe, D., Abdulrachman, S., Gines, H.C., Nagarajan, R., Satawathananont,S., Son, T.T., Tan, P.S., Wang, G.H., Chien, N.V., Thoa, V.T.K., Phung, C.V., Stalin, P.,Muthukrishnan, P., Ravi, V., Babu, M., Chatuporn, S., Sookthongsa, J., Sun, Q., Fu, R., Simbahan,G.C., Adviento, M.A.A., 2002. Site-specific nutrient management for intensive rice cropping systems inAsia. Field Crops Res. 74, 37–66.

Dobermann, A., Abdulrachman, S., Gines, H.C., Nagarajan, R., Satawathananont, S., Son, T.T., Tan,P.S., Wang, G.H., Simbahan, G.C., Adviento, M.A.A., Witt, C., 2004. Agronomic performance ofsite-specific nutrient management in intensive rice-cropping systems of Asia. In: Dobermann, A., Witt,C., Dawe, D. (Eds.), Increasing Productivity of Intensive Rice Systems through Site-Specific NutrientManagement. Science Publishers, Enfield, NH (USA) and International Rice Research Institute, LosBanos, Philippines, pp. 307–336.

Gines, H.C., Redondo, G.O., Estigoy, A.P., Dobermann, A., 2004. Site-specific nutrient management inirrigated rice systems of Central Luzon, Philippines. In: Dobermann, A., Witt, C., Dawe, D. (Eds.),Increasing Productivity of Intensive Rice Systems through Site-Specific Nutrient Management. SciencePublishers, Enfield, NH (USA) and International Rice Research Institute, Los Banos, Philippines, pp.145–169.

Hung, N.N., 1995. Efficiency of urea on broadcast-seeded rice in Mekong Delta of Vietnam as affected bytime of split application and rate of seeding. M.Sc. thesis. University of the Philippines Los Banos,Laguna, Philippines..

International Rice Research Institute (IRRI). 2006. Site-specific nutrient management. Available from:<http://www.irrri.org/irrc/ssnm/>. Accessed 22 March 2006..

Kitzinger, J., 1995. Qualitative research: introducing focus groups. Brit. Med. J. 311, 299–302.Li, C., 2000. Modeling trace gas emissions from agricultural systems. Nutr. Cycling Agroecosyst. 58, 259–

276.Li, C., Mosier, A., Wassman, R., Cai, Z., Zheng, X., Huang, Y., Tsuruta, H., Boonjawat, J., Lantin, R.,

2004. Modeling greenhouse gas emissions from rice-based production systems: sensitivity andupscaling. Global Biogeochem. Cycles 18, GB1043. doi:10.1029/2003GB00204.

24 M.F. Pampolino et al. / Agricultural Systems 93 (2007) 1–24

Moya, P.F., Dawe, D., Pabale, D., Tiongco, M., Chien, N.V., Devarajan, S., Djatiharti, A., Lai, N.X.,Niyomvit, L., Ping, H.X., Redondo, G., Wardana, P., 2004. The economics of intensively irrigated ricein Asia. In: Dobermann, A., Witt, C., Dawe, D. (Eds.), Increasing Productivity of Intensive RiceSystems through Site-Specific Nutrient Management. Science Publishers, Enfield, NH (USA) andInternational Rice Research Institute, Los Banos, Philippines, pp. 29–58.

Nagarajan, R., Ramanathan, S., Muthukrishnan, P., Stalin, P., Ravi, V., Babu, M., Selvam, S.,Sivanatham, M., Dobermann, A., Witt, C., 2004. Site-specific nutrient management in irrigated ricesystems of Tamil Nadu, India. In: Dobermann, A., Witt, C., Dawe, D. (Eds.), Increasing Productivityof Intensive Rice Systems through Site-Specific Nutrient Management. Science Publishers, Enfield, NH(USA) and International Rice Research Institute, Los Banos, Philippines, pp. 101–123.

Peng, S., Garcia, F.V., Laza, R.C., Sanico, A.L., Visperas, R.M., Cassman, K.G., 1996. Increased N-useefficiency using a chlorophyll meter on high yielding irrigated rice. Field Crops Res. 47, 243–252.

SAS Institute Inc., 1999. SAS Version 8.02. Cary, NC, USA..Shukla, A.K., Ladha, J.K., Singh, V.K., Dwivedi, B.S., Balasubramanian, V., Gupta, R.K., Sharma, S.K.,

Singh, Y., Pathak, H., Pandey, P.S., Padre, A.T., Yadav, R.L., 2004. Calibrating the leaf color chartfor nitrogen management in different genotypes of rice and wheat in a systems perspective. Agron. J.96, 1606–1621.

Tan, P.S., Tuyen, T.Q., Huan, T.T.N., Khuong, T.Q., Hoai, N.T., Diep, L.N., Dung, H.T., Phung, C.V.,Lai, N.X., Dobermann, A., 2004. Site-specific nutrient management in irrigated rice systems of theMekong Delta of Vietnam. In: Dobermann, A., Witt, C., Dawe, D. (Eds.), Increasing Productivity ofIntensive Rice Systems Through Site-Specific Nutrient Management. Science Publishers, Enfield, NH(USA) and International Rice Research Institute, Los Banos, Philippines, pp. 193–215.

USEPA, 2002. Greenhouse gases and global warming potential values. US Greenhouse Gas InventoryProgram, Office of Atmospheric Programs, US Environmental Protection Agency, Washington, DC,USA.

Institute for the Study of Earth, Oceans and Space, 2003. User’s Guide for the DNDC Model (version8.2). University of New Hampshire, NH, USA..

Witt, C., Dobermann, A., Abdulrachman, S., Gines, H.C., Wang, G., Nagarajan, R., Satawatananont,T.T., Son, T.T., Tan, P.S., Tiem, L.V., Simbahan, G.C., Olk, D.C., 1999. Internal nutrient efficienciesof irrigated lowland rice in tropical and subtropical Asia. Field Crops Res. 63, 113–138.

Witt, C., Balasubramanian, V., Dobermann, A., Buresh, R.J., 2002. Nutrient management. In: Fairhurst,T., Witt, C. (Eds.), Rice – A Practical Guide to Nutrient Management. Potash and Phosphate Institute(PPI) and Potash and Phosphate Institute of Canada (PPIC) (Singapore) and International RiceResearch Institute, Los Banos, Philippines, pp. 1–45.

Witt, C., Pasuquin, J.M.C.A., Mutters, R., Buresh, R.J., 2005. New leaf color chart for effective nitrogenmanagement in rice. Better Crops 89, 36–39.