quadrats each season so that, over the seasons, all parts of the unit will be
represented. However, if the objective is to measure changes with time, then
there is an advantage in keeping the same quadrats. This ensures that the
change in crop yield from one season to the next is not due to positional
effects of quadrats.
The relative advantages in keeping or changing quadrats over time depend
on how much the quadrat-to-quadrat variation is consistent over the seasons.
Some sources of variation will not be correlated over time (e.g., attack by
pests), some will continue over a few seasons (e.g., variation caused by
uneven manuring) and others will be permanent (e.g., soil depth). The size
and nature of the tree/crop interface will probably change from season to
season, so if a fixed sample is used we will lose some of the efficiency of
stratification.