Agroforestry is an approach to land-use which offers much promise in addres
sing many of the inter-related problems associated with rural development a
nd land degradation (Scherr, 1992; Nair, 1993). This research incorporates
remotely sensed agroforestry landscape information within a computerized mo
del that relates spatial structure at the field and landscape scale to maiz
e production. Remote sensing, cartographic modeling as well as landscape ec
ological theory and methods of analysis figure prominently in the work. The
study area is in the coffee zone on the southern footslopes of Mount Kenya
. The research tests the hypotheses that: there is a correlation between la
ndscape, field structure and crop production. Specifically, it is hypothesi
zed that spatial configuration at the field scale, (e.g. patch size, shape
and adjacency relationships etc.) as well as at the landscape scale (densit
y of trees and hedgerows) can be significantly correlated with productivity
. The central hypothesis is that spatial variability across the landscape,
at scales both at and beyond that of the field, can explain a significant a
mount of the variability in yields. In order to carry out the work, a recen
t innovation in remote sensing, the Compact Airborne Spectrographic Imager
(CASI), has been used in the project. The CASI was used to acquire a detail
ed digital image of the site at a high spatial resolution (1.75 m pixel siz
e). Six spectrally narrow width bands were acquired with the purpose of det
ermining chlorophyll density as well as crop type and vigor. Findings indic
ate that the spatial measures that were undertaken at the landscape scale d
o not explain enough crop variability to be statistically significant. Rath
er, it is far better explained at the individual field level. Here it was d
emonstrated that the distance from shelter plantings is inversely and curvi
linearly related to crop vigor as measured by the relationship between the
Near Infra Red and Red wavelengths. Findings indicate that the positive eff
ects of trees in the study landscape are limited to within 25 m of the cano
py and may increase production as much as 32% at the optimal distance. An o
ptimum hedgerow inter-distance of 40 m is suggested in order to maximize cr
op vigor. The impact of both the existing conditions as well as two possibl
e alternatives are evaluated for their total landscape wide impacts on maiz
e yields and wood production. It was determined that removal of the existin
g hedgerows (50% of all the trees in the area) by a disturbance such as a p
est outbreak would cause a 5% reduction in maize production. A second alter
native, which identified under-utilized planting niches throughout the land
scape, is predicted to result in a modest increase in maize production and
a doubling of wood products. (C) 1998 Elsevier Science B.V. All rights rese
rved.