A total of 47 variables (please refer to column 2 of Table 1) were
used for the analysis. Data gathered were mostly binary (yes/no)
except for months of stocking or harvest, mangrove to pond area ratio
(MPR), proximity to the sea or road (km), pond size (ha), water level
inside the pond (cm), and number of croppings/year. No scaling was
used in the case of the non-binary data. Data analysis was done in two
steps: variable reduction to increase model stability and binary
logistic regression to identify the protective and risk factors. WSSV
incidence was used as the dependent variable that needs to be
explained. The procedure was applied to the entire dataset and to two
sub-sets of the data: farms into monoculture, and farms into
polyculture. All the statistical analyses were done with SPS