rice with fish culture in the semi-deep waters of the mekong delta, vietnam: interaction of rice...

Aquaculture Research, 1998, 29, 59–66

Rice with fish culture in the semi-deep waters of the

Mekong Delta, Vietnam: interaction of rice culture

and fish husbandry management on fish production

A J Rothuis1,2, D K Nhan2, C J J Richter3 & F Ollevier1

1Laboratory of Ecology and Aquaculture, Catholic University of Leuven, Leuven, Belgium, 2Mekong Delta Farming

Systems Research and Development Institute, University of Can Tho, Can Tho, Vietnam, and 3Wageningen Institute of

Animal Sciences, Agricultural University, Department of Fish Culture and Fisheries, Wageningen, The Netherlands

Correspondence: A J Rothuis, Laboratory of Ecology and Aquaculture, Zoological Institute, Catholic University of Leuven,

Naamsestraat 59, B 3000 Leuven, Belgium

Abstract

Fish husbandry and rice culture management factorsinfluencing the yield of introduced fish in ricefieldsof the Vietnamese Mekong Delta were studiedby multiple regression analysis. A significant(P , 0.001) regression model was computed inwhich feed input and duration of culture periodpositively, and ricefield area, rice seeding rate andthe year of the survey negatively affected the yield ofintroduced, as well as indigenous, fish. The negativeimpact of larger ricefields is probably the result ofthe escape of fish. This is also probably the reasonfor the year of survey since the average yield ofintroduced fish was 92.5 kg ha–1 in 1995 (becauseof an extreme flood) as compared to 164.8 kg ha–1

in 1994. A high seeding rate of rice results in adense stand which suppresses the growth of fish.Opportunities for improvement of fish productionare proper ricefield construction, reduced seedingrates, stocking fingerlings early in the dry seasonand more intensive feeding.

Introduction

In rice–fish culture management, input use(fertilizers and pesticides) are primarily aimed atmaximizing the rice production. Moderatequantities of fertilizers stimulate fish growththrough the autotrophic food web while highdoses of ammonia fertilizers can result in short-term exposure to toxic NH3 levels (Middendorp

© 1998 Blackwell Science Ltd. 59

1985). Pesticides can affect fish production directlyor through effects on fish-food organisms. Pesticideapplication in ricefields changes the structure ofthe aquatic community (Roger & Kurihara 1991)and detrimental effects on cladocerans have beenreported by Ali (1990). The impacts of pesticideson fish and their food web depend on thepersistence, chemical formulation, manner andtime of application (Cagauan & Arce 1992). Notall pesticides are equally toxic: herbicides aregenerally less toxic to fish than insecticides.

In the semi-deep waters of the VietnameseMekong Delta, rice–fish farming with hatchery-produced (introduced) fingerlings differed mainlyfrom rice monoculture by a higher fertilizer andwater requirement and less pesticide use. Riceyields were not affected by the presence of fish,and the contribution of fish to the total farmprofitability was low (Rothuis, Nhan, Richter &Ollevier 1998). As a first step towards theoptimization of fish yields, the interaction of riceculture and fish husbandry management on fishproduction was studied by multiple regressionanalysis.

Materials and methods

Study area and data sets

The study was carried out within the vicinity of thecooperative farm at Co Do which encompasses anarea of approximately 6000 ha, situated in the semi-deep water zone of the Mekong Delta, Vietnam. In

Interaction of rice culture and fish husbandry management A J Rothuis et al. Aquaculture Research, 1998, 29, 59–66

Table 1 Fish species introduced into ricefields

Fish species Stocked byfarmers (%)

Silver barb, Puntius goniotus (Bleeker) 93Common carp, Cyprinus carpio (L.) 91Tilapia, Oreochromis niloticus (L.) 27Silver carp, Hypophthalmichthys molitrix (Val.) 23Indian carps, Labeo rohita (Hamilton), 19

Cirrhinus mrigala (Hamilton)Others* 15

*Including Oxyeleotrix marmoratus (Bleeker), Trichogasterpectoralis (Regan), and hybrid of Clarias macrocephalus(Gunther) and Clarias gariepinus (Burchell).

a previous paper (Rothuis et al. 1998) the socio-economical data of farming systems practising riceculture with introduced fish (R-Intro.F), rice culturewith indigenous fish (R-Ind.F) and rice monoculture(RM) in this study area (data of 1995) werepresented. The data set on rice culture and fishhusbandry management of the present studyconcerning R-Intro.F was derived from the socio-economical survey and supplemented by a similar(unpublished) data set obtained in the same areain 1994.

Fish husbandry management and rice culture

Small fingerlings (c. 1 g) were usually stockedbetween March and June during the wet season ricecrop, and harvested about 9 months later during orafter the dry season rice crop (December–February).Between August and October the land was left fallowbecause of high water levels. The fish were rearedin polyculture predominantly consisting of silverbarb, Puntius gonionotus (Bleeker), and commoncarp, Cyprinus carpio L. (Table 1). The stockingdensity was calculated as the total number of allfish introduced per m2. Usually, fish were firstconfined to the trench for about 3 weeks in orderto prevent damage to the young rice plants by thefish. Particularly during this period and during thewet season rice crop, farmers used a variety of freshand dry feeds to supplement the natural productivityof the water, the most common being rice bran(Table 2). Since energy is usually the first nutritionalfactor limiting fish growth in systems dependent onnatural food (Hepner & Pruginin 1981), the total

60 © 1998 Blackwell Science Ltd, Aquaculture Research, 29, 59–66

Table 2 Percentage of total number of farmers usingdifferent supplementary feeds and average quantity of feedtype per farmer

Feed type Farmers Quantity(%) (kg ha –1)

Rice bran 96 292.5Fish meal 28 28.9Ricea 30 163.3Water spinach (Ipomea aquatica 25 475.7

Forskal)Sweet potatob 9 607.2Cassavab 8 89.3Crabs, snails 6 67.0

aPaddy rice, broken rice and milled rice. bTubers.

input of supplementary feeds was calculated as kcalgross energy per m2 rice–fish field. The approximatecomposition of the feeds was calculated using thetables from New (1987), and FAO (1972), andconverted into gross energy according to New(1987).

Of the rice inputs, fertilizer applications (Urea,NPK and Di-Ammonium-Phosphate) were convertedinto total nitrogen and phosphate (P2O5) andexpressed in kg/ha. Pesticides were expressed bothin quantity of active ingredients (a.i.) and as thenumber of sprayings. As not all farmers stocked fishat the same time, only those rice inputs appliedduring the time that the fish were actually presentin the field were used in the analysis. The riceseeding rate was calculated as the average seedingrate of the rice crops cultivated during the presenceof the fish.

For the fish nursery, animal manure input,inorganic fertilizer use (apart from the fertilizer usedfor rice cultivation), and Derris root application(to eradicate unwanted fish), only the presence orabsence of such a treatment was indicated (dummyvariables). The year of the survey was also includedin the regression analysis as a dummy variable.

Farmers introducing hatchery produced finger-lings made no attempt to prevent the entranceof indigenous fish. Often these were deliberatelyattracted into the ricefield by putting feed andbranches at the water gate. Consequently, theharvest consisted of both stocked and wild fish. Thelatter were predominantly snakehead (Channa striata

Aquaculture Research, 1998, 29, 59–66 Interaction of rice culture and fish husbandry management A J Rothuis et al.

Bloch), climbing perch (Anabas testudineus Bloch),and Clarias spp.

Statistical analysis

The following model of variables affecting the yieldof introduced fish in ricefields was hypothesized:

Y 5 a 1 b1X1 1 b2X2 1 ... 1 b22X22 1 u

where Y 5 yield of introduced fish (kg/ha); X1 5

ricefield area, excluding fish refuge (ha); X2 5 ratioof fish refuge to rice field area (%); X3 5 period ofstocking fish into the ricefield (month code); X4 5

duration of fish rearing period (months); X5 5 Frynursery in fish pond (dummy); X6 5 fish speciescomposition (% of silver barb); X7 5 total stockingdensity (number m–2); X8 5 feed input (kcal m–2);X9 5 use of animal manure (dummy); X10 5 use ofinorganic fertilizer (dummy); X11 5 use of Derrisroot (dummy); X12 5 total nitrogen input on therice crop (kg ha–1); X13 5 total phosphorous inputon the rice crop (kg ha–1); X14 5 total insecticideapplied on the rice crop (kg a.i. ha–1); X15 5 totalfungicide applied on the rice crop (kg a.i. ha–1);X16 5 total herbicide applied on the rice crop (kga.i. ha–1); X17 5 total number of insecticide spray-ings; X18 5 total number of fungicide sprayings;X19 5 total number of herbicide sprayings; X20 5

rice seeding rate (kg ha–1); X21 5 year of the survey(dummy); X22 5 yield of indigenous fish (kg ha–1);a 5 constant (intercept); and u 5 residual.

The frequency distribution of the dependentvariable (yield of introduced fish) differedsignificantly (P . 0.05) from normal, and wastransformed to logarithmic (base 10) values. Extremecases, out of range 3 3 standard deviation (SD),were eliminated. The final database consisted of 99cases with 23 variables.

The regression model was calculated following theprocedure by van Dam (1990). First, a correlationmatrix was calculated to determine the relationshipsbetween independent variables, followed by theactual construction of the regression model. Anindependent variable would be retained in the modelonly if its partial regression coefficient wassignificantly different from zero at the 0.05 level.Pairs of independent variables with high correlationcoefficients were not jointly included into the model.Logarithmic (base 10) transformation of independ-ent variables would be incorporated if thiscontributed to the overall significance of the model.

© 1998 Blackwell Science Ltd, Aquaculture Research, 29, 59–66 61

Residuals were analysed for linearity and systematicpatterns to verify the assumptions of linearregression (Hair, Anderson & Tatham 1990). Thepossible existence of autocorrelation was verifiedby calculation of the Durbin–Watson statistic, andcompared with significance points (Neter,Wasserman & Kutner 1990). Standardized partialregression coefficients were calculated to comparethe relative importance of independent variables.

Results and discussion

Correlation and regression analysis

The fish husbandry and rice culture characteristics(untransformed data) are summarized in Table 3.The mean fish yield was 186.4 kg ha–1, consistingof 130.5 kg ha–1 introduced fish and 55.9 kg ha–1

indigenous fish. Stocking period and duration of thefish culture period, stocking density and feed input,total nitrogen and total phosphate, and the quantityand number of insecticide applications were highlycorrelated (Table 4).

A significant (P , 0.001) regression model wascomputed with five independent variables (Table 5).Feed input and duration of the culture period hada positive effect on the yield of introduced fish, whilethe year of the survey, ricefield area, and riceseed rate negatively affected fish yield. The higheststandardized partial regression coefficients (beta)were found for the year of the survey and feedinput. Residual analysis did not indicate violation ofassumptions for multiple regression. The Durbin–Watson statistic was higher than the critical valueof du at 1%, indicating the absence of autocorrelationamong the independent variables.

Higher fish yields were obtained when theduration of the rearing period was extended. Thissuggests that feed nor fish metabolites limited fishproduction. Fish yields from concurrent rice–fishsystems stocked with hatchery-produced fingerlingsusing little or no supplementary feeding are usuallyin the order of 300 kg ha–1 (Lightfoot, Costa-Pierce, Bimbao & Dela Cruz 1992). In the presentstudy the total fish biomass (indigenous andintroduced species) did not reach this level,suggesting further scope for growth. The durationof the fish rearing period was highly negativelycorrelated to the month of stocking (R2 5 –0.72,P , 0.001; Table 4). This implies that stockingfingerlings early, between January and March (dryseason), results in higher fish yields. Probably the


Table 3 Mean, standard deviation (SD), minimum (Min), and maximum (Max) of the untransformed variables (99 cases)

Variable Dimensions Mean SD Min Max

Fish husbandryRicefield area ha. 1.73 0.93 0.22 4.50Refuge/field ratio % 13.16 8.41 1.10 48.50

Fish managementStocking period month code 6.36 2.12 2.1 11.3Duration fish cultivation months 8.69 2.06 4 15Fry nursery dummy 0.87 0.34 0 1Fish species composition % silver barb 52.17 26.25 0 100Total stocking density no./m2 2.01 1.56 0.04 9.70

Fish inputsFeed kcal m–2 115.62 131.81 0 773.3Animal manure dummy 0.37 0.49 0 1Inorganic fertilizer dummy 0.75 0.44 0 1Derris root application dummy 0.37 0.49 0 1

Rice inputsTotal nitrogen kg ha–1 180.98 67.53 70.5 363.3Total phosphorous kg ha–1 84.97 32.96 25.0 159.5Total insecticide kg a.i. ha–1 0.44 0.62 0 3.2Total fungicide kg a.i. ha–1 0.19 0.37 0 2.2Total herbicide kg a.i. ha–1 0.47 0.45 0 3.0Insecticide applications no. sprayings 1.56 1.95 0 8Fungicide applications no. sprayings 1.15 1.26 0 6Herbicide applications no. sprayings 2.02 1.57 0 7Rice seed rate kg ha–1 257.19 46.28 128.2 384.6

MiscellaneousYear of survey dummy 0.47 0.50 0 1Yield indigenous fish kg ha–1 55.91 40.02 5.13 205.5

Dependent variableYield introduced fish kg ha–1 130.47 184.86 3.08 1281.3

weather conditions favour natural food resourcesthrough an increased activity of the photosyntheticaquatic biomass in this season. Besides, this couldalso be an effect of an improved survival ratesince predatory fish are more abundant in thewet season (Duong 1994). The negative impactof larger ricefields on fish production is probablythe result of a lower recovery rate (escape). Largerfields require more investment capital and are oftenless well constructed and managed. Consequentlylarger fields are prone to fish loss. Escape of fishis probably the reason for the negative effect ofthe year of the survey on fish yield. The averageyield of introduced fish in 1995 was only92.5 kg ha–1, as compared to 164.8 kg ha–1 in


1994. The main difference between these 2 yearswas the extreme flood in 1995, which inundatedlarge parts of the study area. Chapman (1992)reported that the size of fingerlings determinestheir survival in the ricefield. The positive effectof supplementary feeds on fish yield in the presentstudy is probably the result of an increasedsurvival rate of the bigger fingerlings, since feedingwas carried out primarily during the early phaseof the fish rearing period.

The rice seeding rate had a negative impact onthe fish yield. A high seeding rate quickly resultsin a dense stand which suppresses the growth ofweeds (a major problem in direct-seeded rice(Moody 1992)) and phytoplankton due to a


Tab

le4

Cor

rela

tion

mat

rix

ofva

riab

les

use

dfo

rth

ere

gres

sion

anal

ysis

[mar

ked

corr

elat

ion

s(*

)ar

esi

gnifi

can

tat

P,

0.0

1].

Nu

mbe

rof

case

s5

99

1

Var.

rfar

rf%

stp

culp

nurs

spec

std

feed

man

fert

root

year

wfis

NP

ins

fung

herb

ispr

fspr

hspr

seed

fish

rfar

1.00

rf%

–0.6

1*1.

00st

p0.

08–0

.10

1.00

culp

–0.0

10.

05–0

.72*

1.00

nurs

0.23

–0.1

60.

18–0

.21

1.00

spec

–0.0

2–0

.01

0.15

–0.1

90.

221.

00st

d–0

.45*

0.36

*–0

.03

0.05

0.05

0.11

1.00

feed

–0.4

6*0.

42*

–0.0

90.

140.

030.

020.

71*

1.00

man

–0.0

2–0

.05

–0.1

00.

100.

180.

120.

250.

101.

00fe

rt0.

060.

02–0

.02

–0.0

6–0

.02

0.14

0.09

0.08

0.02

1.00

root

0.09

–0.0

1–0

.04

0.09

0.05

0.12

0.07

0.08

0.27

*0.

061.

00ye

ar–0

.02

0.15

–0.4

8*0.

34*

0.01

–0.0

8–0

.09

–0.0

40.

06–0

.10

–0.0

21.

00w

fis–0

.42*

0.28

*–0

.29*

0.26

–0.1

9–0

.15

0.45

*0.

45*

0.12

–0.1

20.

05–0

.01

1.00

N0.

000.

07–0

.54*

0.60

*–0

.10

–0.1

4–0

.07

–0.0

20.

130.

06–0

.17

0.34

*0.

101.

00P

0.02

0.04

–0.5

5*0.

54*

–0.1

8–0

.23

–0.1

3–0

.02

0.00

0.10

–0.1

20.

31*

0.09

0.72

*1.

00in

s0.

17–0

.13

–0.1

10.

200.

010.

01–0

.12

–0.1

40.

09–0

.04

0.04

0.36

*–0

.03

0.40

*0.

37*

1.00

fung

0.07

–0.0

7–0

.11

0.01

–0.2

10.

03–0

.14

–0.1

3–0

.06

0.13

–0.1

70.

28*

–0.1

10.

07–0

.02

0.18

1.00

herb

0.09

–0.0

8–0

.28*

0.25

–0.0

3–0

.06

–0.0

30.

010.

07–0

.04

0.10

0.27

*0.

080.

29*

0.33

*0.

32*

0.20

1.00

isp

r0.

26*

–0.1

7–0

.28*

0.37

*0.

080.

00–0

.17

–0.1

20.

09–0

.07

0.03

0.48

*–0

.01

0.40

*0.

35*

0.75

*0.

140.

37*

1.00

fsp

r0.

10–0

.03

–0.3

4*0.

16–0

.05

0.19

–0.1

3–0

.17

0.06

0.00

–0.0

90.

52*

–0.1

30.

32*

0.21

0.36

*0.

54*

0.33

*0.

37*

1.00

hsp

r0.

30*

–0.1

7–0

.38*

0.31

*0.

08–0

.16

–0.1

2–0

.13

0.02

0.14

–0.0

60.

40*

–0.0

80.

42*

0.38

*0.

34*

0.28

*0.

59*

0.53

*0.

48*

1.00

seed

0.15

–0.1

50.

30*

–0.2

30.

27*

0.19

0.05

0.02

0.23

–0.0

50.

17–0

.22

–0.2

5–0

.14

–0.1

3–0

.05

–0.2

20.

01–0

.09

–0.1

6–0

.06

1.00

fish

–0.3

2*0.

25–0

.20

0.20

–0.0

80.

070.

43*

0.53

*0.

16–0

.02

0.00

–0.2

00.

57*

0.02

–0.0

1–0

.08

–0.0

6–0

.10

–0.1

0–0

.08

–0.1

7–0

.16

1.00

1rf

ar,

rice

field

area

;rf

%,

refu

ge/fi

eld

rati

o;st

p,st

ocki

ng

peri

od;

culp

,du

rati

onfis

hcu

ltiv

atio

n;

nu

rs,

fry

nu

rser

y;sp

ec,

fish

spec

ies

com

posi

tion

;st

d,st

ocki

ng

den

sity

;m

an,

anim

alm

anu

re;

fert

,in

orga

nic

fert

ilize

r;ro

ot,

Der

ris

root

;w

fish

,yi

eld

indi

gen

ous

fish

;N

,to

tal

nit

roge

n;

P,to

tal

phos

phor

ous;

ins,

tota

lin

sect

icid

e;fu

ng,

tota

lfu

ngi

cide

;h

erb,

tota

lh

erbi

cide

;is

pr,

inse

ctic

ide

spra

ys;

fspr

,fu

ngi

cide

spra

ys;

hsp

r,h

erbi

cide

spra

ys;

seed

,ri

cese

edra

te;

fish

,lo

g(y

ield

intr

odu

ced

fish

).



Table 5 Results of multiple regression analysis

Independent variable b Standard betaerror

Feed 0.0013 0.0004 0.310**Duration fish cultivation 0.067 0.023 0.247**Year of survey –0.34 0.096 –0.312***Log ricefield area –0.42 0.194 –0.216*Rice seed rate –0.0023 0.0010 –0.192*

Constant 1.896Multiple R2 0.400Adjusted R2 0.367F-value 12.382Probability ,0.001Durbin–Watson statistic 1.76

*, P , 0.05; **, P , 0.01; ***, P , 0.001

reduced availability of nutrients and increasedshading as the rice canopy closes (Simpson, Roger,Oficial & Grant 1994). At lower seeding rates,emerging weeds (Cagauan 1995) and newly formed(soft) rice tillers can be consumed by herbivorousfish (silver barb) so that the final rice plant densityremains low. This will facilitate the access of fishinto the ricefield, and increase the feed resourcesavailable to fish. According to De Datta &Nantasomsaran (1991) 100 kg of pre-germinatedrice seed ha–1 results in good stand establishmentand weed control, while the average seeding ratein the present study was 257 kg ha–1.

Not all variables identified a priori were significantin the constructed regression model. Informationwas lost by using dummy variables (nurserymanagement), other independent variables werecorrelated (ricefield area with refuge-field ratio,duration of the fish cultivation period with stockingperiod). Rice fertilizer and pesticide input did notaffect the fish yield, contrary to van Dam (1990),who found a positive effect of nitrogen and negativeeffects of phosphorous and pesticides.

In his model for gross fish yield, van Dam (1990)found a coefficient of determination (R2) of 0.6571with a F-value of 52.013. In the present study theconstructed regression model explained only 40%of the total variance in fish production. However,van Dam analysed the results of purposely designedon-station experiments, while the data in the presentstudy originate from interviews with farmers. Inthis situation errors in the variables are likely tobe higher.


Significance of indigenous and introducedfish

Indigenous fish made up 30% of the total fishyield. The high perception of indigenous fishamong farmers in the study area was also observedby Setboonsarng (1994), and Fujisaka & Vejpas(1990) in North-east Thailand. Farmers regardedhatchery-produced fish as a secondary crop, andobjected to measures to limit the entrance of morevaluable indigenous species. This raises the questionof whether this activity is competitive orcomplementary to the production of introducedfish. In the present study the yield of indigenousfish was correlated to the yield of introduced fish(R2 5 0.57, P , 0.01; Table 4). In a previousstudy Rothuis et al. (1998) found a significantlylower indigenous fish yield from rice monoculturefields than from rice–fish fields. This suggests thatindigenous fish benefit from the modificationsmade to the ricefield (refuge trench) and thespecific rice–fish management (high water levels).Apparently, the presence of indigenous fish didnot adversely affect the yield of introduced fish.Although most indigenous fish were carnivorous,only snakehead can be regarded as a vigorouspredator, feeding on frogs, fish and small aquaticsnakes (Kok 1982). Climbing perch feeds mainlyon insects while Clarias spp. feed on a variety ofsmall organisms (insect larvae, worms, shells,shrimp, small fish), aquatic plants and detritus(Ukkatawewat 1979). Probably, the abundantsmall indigenous fish (Rasbora spp. and Esomusspp.) and shrimp (Macrobrachium lanchesteri DeMan)served as primary feed for piscivorous predatorswhereas introduced fish escaped predation becauseof their size advantage. The coexistence ofintroduced and indigenous fish production inricefields was also observed by Middendorp (1992).

In conclusion, the fish yield from rice–fishculture in the semi-deep waters of the VietnameseMekong Delta mainly depends of a combined effectof mortality and escape of stocked fingerlings fromthe ricefields. Rice and fish husbandry managementopportunities available to the farmer for theimprovement of fish production, are feeding(particularly at the beginning of the fish rearingperiod), stocking fish early in the dry season,proper field construction, and a reduced riceseeding rate. The study illustrated the use ofmultiple regression analysis as a tool to identifytechnological constraints of rice–fish culture at


farmer level. Its results should be completed withan appraisal of the management variables by thefarmers themselves, and verified at controlled on-station experiments.

Acknowledgments

This study was conducted as part of a cooperativeresearch project entitled ‘Impact analysis andimprovement of rice–fish farming systems in thesemi-deep water area of the Mekong Delta, Vietnam’.Partners in this program are the University of CanTho (Mekong Delta Farming Systems Research &Development Institute), and the Catholic Universityof Leuven (Laboratory of Ecology & Aquaculture,and Laboratory of Soil Fertility & Soil Biology). Theproject is supported by the Flemish InteruniversityCouncil (Vl.I.R.) through funds provided by theBelgian Development Cooperation (BADC).

References

Ali A.B. (1990) Some ecological aspects of fish populationsin tropical ricefields. Hydrobiologia 190, 215–222.

Cagauan A.G. & Arce R.G. (1992) Overview of pesticideuse in rice-fish farming in Southeast Asia. In: Rice-Fish Research and Development in Asia (ed. by C.R.Dela Cruz, C. Lightfoot, B.A. Costa-Pierce, V.R. Carangal& M.P. Bimbao), pp. 217–233. ICLARM ConferenceProceedings 24, ICLARM, Manila, Philippines.

Cagauan A.G. (1995) Overview of the potential roles ofpisciculture on pest and disease control and nutrientmanagement in rice fields. In: The Management ofIntegrated Freshwater Agro-Piscicultural Ecosystems inTropical Areas (ed. by J.J. Symoens & J.C. Micha), pp.203–244. Seminar Proceedings, RAOS and CTA,Wageningen, Brussels, Belgium.

Chapman G. (1992) Fry nursery techniques in the rice-fish systems of Northeast Thailand. In: Rice-FishResearch and Development in Asia (ed. by C.R. DelaCruz, C. Lightfoot, B.A. Costa-Pierce, V.R. Carangal &M.P. Bimbao), pp. 139–143. ICLARM ConferenceProceedings 24, ICLARM, Manila, Philippines.

De Datta S.K. & Nantasomsaran P. (1991) Status andprospects of direct seeded flooded rice in tropical asia.In: Direct Seeded Flooded Rice in the Tropics, pp. 1–16.International Rice Research Institute, Manila,Philippines.

Duong L.T. (1994) Deep rice-fish based farming systemsin the Mekong Delta, Vietnam. Rice Farming SystemsTechnical Exchange 3, 19–24.

FAO (1972) Food Composition Table for use in East Asia.Food and Agriculture Organization, Rome, Italy.


Fujisaka S. & Vejpas C. (1990) Capture and cultured paddyfisheries in Khu Khat, Northeast Thailand. InternationalRice Research Institute Social Science Division PaperNo. 90–27, International Rice Research Institute,Manila, Philippines.

Hair J.F., Anderson R.E. & Tatham R.L. (1990) MultivariateData Analysis with Readings (2nd edition). MacmillanPubl. Co., Singapore. 449 pp.

Hepher B. & Pruginin Y. (1981) Commercial Fish Farming,with Special Reference to Fish Culture in Israel. Wiley-Interscience Publ., New York, NY, USA. 261 pp.

Kok Leong Wee (1982) Snakeheads-their biology andculture. In: Recent Advances in Aquaculture (ed. by J.F.Muir & R.J. Roberts), pp. 181–213. Croom Helm Ltd,Kent, UK.

Lightfoot C., Costa-Pierce B.A., Bimbao M.P. & Dela CruzC.R. (1992) Introduction to rice-fish research anddevelopment in Asia. In: Rice-Fish Research andDevelopment in Asia (ed. by C.R. Dela Cruz, C. Lightfoot,B.A. Costa-Pierce, V.R. Carangal & M.P. Bimbao), pp.1–10. ICLARM Conference Proceedings 24, ICLARM,Manila, Philippines.

Middendorp A.J. (1985) Integrated rice/fish culture: practicesand prospects. Working Group on Small Scale IntegratedAquaculture, International Agriculture Centre,Wageningen, the Netherlands. 33 pp.

Middendorp A.J. (1992) Contribution of stocked andwild fish in ricefields to fish production and farmernutrition in Northeast Thailand. Asian Fisheries Science5, 145–161.

Moody K. (1992) Weed management in wet-seeded ricein tropical Asia. Food & Fertilizer Technology CenterExtension Bulletin, 364, 9 pp.

Neter J., Wasserman W. & Kutner M.H. (1990) AppliedLinear Statistical models. Regression, Analysis of Variance,and Experimental Designs (3rd edition). Irwin, BurrRidge, IL, USA. 1181 pp.

New M.B. (1987) Feed and feeding of fish and shrimp. Amanual on the preparation and presentation of compoundfeeds for shrimp and fish in aquaculture. ADCP/REP/87/26, 275 pp. Food and Agriculture Organization, UnitedNations Development Program, Rome, Italy.

Roger P.A. & Kurihara Y. (1991) Floodwater biology oftropical wetland ricefields. In: Soil management forsustainable rice production in the tropics (ed. by S.Panichapong & H. Wada), pp. 211–233. IBSRAMMonograph No. 2. International Board for Soil Researchand Management, Bangkok, Thailand.

Rothuis A.J., Nhan D.K., Richter C.J.J. & Ollevier F.(1998) Rice with fish culture in the semi-deep watersof the Mekong Delta, Vietnam: a socio-economicalsurvey. Aquaculture Research 29, 49–59.

Setboonsarng S. (1994) Farmers’perception towards wildfish in ricefields: ‘product not predator’-an experiencein rice-fish development in Northeast Thailand. In:Role of Fish in Enhancing Ricefield Ecology and Integrated


Pest Management (ed. by C.R. Dela Cruz), pp. 43–44.ICLARM Conference Proceedings 43, ICLARM, Manila,Philippines.

Simpson I.C., Roger P.A., Oficial R. & Grant I.F. (1994)Effects of nitrogen fertilizer and pesticide managementon floodwater ecology in a wetland ricefield. I.Experimental design and dynamics of thephotosynthetic biomass. Biology and Fertility of Soils17, 129–137.


Ukkatawewat S. (1979) The taxonomic characteristics andbiology of some important freshwater fishes in Thailand.Technical Paper No. 3, National Inland Fisheries Inst.,Dep. Fisheries, Min. Agriculture, Bangkok, Thailand.55 pp.

Van Dam A.A. (1990) Multiple regression analysis ofaccumulated data from aquaculture experiments: arice-fish example. Aquaculture and Fisheries Management21, 1–15.

rice with fish culture in the semi-deep waters of the mekong delta, vietnam: interaction of rice...

Documents