We are in a similar situation with the tree-crop interfaces. Before the unit
can be stratified, we need to understand more of the nature of the interfaces
-- how far do they extend, how much does this vary from year to year, are
visual estimates of their extent reliable? Some of this information can come
from other agroforestry studies, but studies designed specifically to investigate
these problems are also needed. ICRAF's tree/crop interface project
[Huxley et al., 1989b] was designed for this purpose.