ABSTRACT
In all natural systems, particularly biological systems, space and time are interrelated. Currently, the analytical techniques available to biologists do not deal adequately with spatial correlation. A great many studies of foraging and population dynamics are carried out on some sort of grid, regular or irregular, such as Latin squares. These are designed so that any treatment results are orthogonal, thus removing the effects posed by the grid geometry. Another standard technique for removing grid bias is to enter a dummy treatment effect both for rows and columns of the experiment to see whether this makes any difference to the statistical model being tested. If the geometry is irregular or if there is some interpolation or extrapolation to be done, then these spatial biases present a much more complex situation. Dealing with these biases as the experiment evolves through time becomes a major statistical difficulty.
