Coastal Shrimp Aquaculture in Thailand: Key Issues for Research

!In Search of Variables Contributing toProduction of Shrimp and IdentifyingShrimp Farming Provinces in ThailandThan Pe and Paul T. Smith*AbstractIn this paper, the 1995 survey data from 451 shrimp farms in Thailand were analysedusing logistic regression and multiple analysis methods. Variables that could identifythe 10 provinces practising prawn farming were determined by the method of logisticregression. The variables which play an important part in identifying the provinces are:dominant soil type; number of farming years; source of farm water; depth of productionponds; number of other aquaculture farms within 3 km of the farm; number ofaquaculture farms sharing the water supply; number of farms discharging effluent intothe water supply canal; preparation of an environmental impact assessment; site selectionto avoid impacts of other users; design of a separate water supply/drainage system;retention of a mangrove buffer zone; use of an effluent treatment pond; existence of awater or sediment related problem; and other problems with salinity, feed quality andseed quality. The high proportion of variables relating to environmental aspects ofshrimp farming is an important and interesting finding because it provides strong evidenceto support the hypothesis that environmental problems are the cause of reducedproductivity and incidence of disease in shrimp farms. The variables which contributeto the level of shrimp production (kg/ha/yr) were determined by the method of multipleregression and the key variables were: total overheads; total estimated fertiliser cost;total labour cost; total feed cost; seed cost; and total other input costs. It is recommendedthat in future, the design of surveys and format of the questionnaires could beimproved to enable deeper analysis of data, especially with respect to environmentalproblems. Also it may be better to use ‘pond’ rather than ‘farm’ as the sample unit.IntroductionThe preliminary analysis of the survey data has been discussed elsewhere (see Smith1999a,b) and in this chapter we have tried to extend the analysis by using advancedstatistical methods as well as the application of the geographical information system(GIS). However, the analysis was limited by problems in the questionnaire, particularlywith respect to missing values. To counter this, multiple regression analysis wasused to estimate the missing values of some continuous economic variables using therelationship with other variables. Then a linear multiple regression model for shrimpproduction was developed. For categorical variables, general loglinear analysis wasapplied to study the relationship among the variables, in particular provinces andproblem variables.* University of Western Sydney Macarthur, PO Box 555, Campbelltown, New South Wales 2560,Australia.%(