Aquaculture pond experiments, like agricultural crop trials, are often designed according to the statistical rules of replication and randomization: several treatments are applied to a number of experimental units (in this case: ponds) after which a certain characteristic (e.g., yield) is measured in every pond. Other factors with a possible effect on the measured characteristic are held at the same constant level as much as possible so as not to disturb treatment effects. Analysis of variance (ANOVA) is used to compare the treatments.