Agroforestry Systems 32: 81-96, 1995. 9 1995 Kluwer Academic Publishers. Printed in the Netherlands.
A spatial database approach for estimating areas suitable for agroforestry in subSaharan Africa: aggregation and use of agroforestry case studies J.D. U N R U H ~ and R A . L E F E B V R E 2 Department of Geography, Harvill #2, University of Arizona, Tucson AZ 857]2; 2The Woods Hole Research Center, P.O. Box 296, Woods Hole, MA 02543, USA
Key words: Africa, GIS, satellite imagery, database Abstract. The growing awareness of the importance and potential of agroforestry has resulted in an invaluable proliferation of site specific case studies. These presently exist in the necessary quantity and quality so as to facilitate their aggregate utilization in larger-scale research and application inquiry. We report here an approach for aggregating and using agroforestry case studies for agroforestry-related research at scales larger than the local site. The work presented here describes how ICRAF's agroforestry database - the primary source of case studies - can be used in combination with satellite imagery, and ancillary information, via matrix operations and a geographic information system, to produce a GIS-based agroforestry research tool for subSaharan Africa. This tool is used in a large-scale and preliminary way, to estimate the areas to which appropriate agroforestry systems in Africa might be extended.
1. I n t r o d u c t i o n
T h e f i e l d o f a g r o f o r e s t r y as a s c i e n t i f i c e n d e a v o r has p r o g r e s s e d q u i c k l y in the last s e v e r a l decades. S i g n i f i c a n t a d v a n c e s in the botanical, a g r o n o m i c , and s o c i a l a s p e c t s o f a g r o f o r e s t r y , a n d the i n t e r a c t i o n s b e t w e e n t h e s e h a v e addressed problems of food supply, rural poverty, deforestation, and land d e g r a d a t i o n [Nair, 1989a and the r e f e r e n c e s cited therein]. T h e study o f agroforestry to date has largely f o c u s e d on specific interactions in i n d i v i d u a l cases. W h i l e this has p r o d u c e d an e n o r m o u s p r o l i f e r a t i o n o f v e r y v a l u a b l e site s p e c i f i c studies, i n v e s t i g a t i o n s w h i c h e x a m i n e the e x t r a p o l a t i o n o f these to larger areas h a v e b e e n few [Nair, 1989b, 1991]. 1.1. Case studies and aggregation
B e c a u s e the n u m b e r o f c a s e studies has g r o w n large, the f i e l d o f a g r o f o r e s t r y is p r e s e n t l y at a p o i n t in its d e v e l o p m e n t w h e r e s i g n i f i c a n t a d v a n c e s a r e p o s s i b l e in the u t i l i z a t i o n o f a g g r e g a t e a g r o f o r e s t r y i n f o r m a t i o n for a v a r i e t y o f r e s e a r c h and a p p l i c a t i o n p u r p o s e s . W h a t is n e e d e d h o w e v e r is a c o m p r e hensive, cohesive, data m a n a g e m e n t a p p r o a c h that can h a n d l e a g g r e g a t e a g r o forestry and a n c i l l a r y i n f o r m a t i o n o f all kinds. S u c h an a p p r o a c h s h o u l d h a v e b o t h the f l e x i b i l i t y to i n c o r p o r a t e n e w i n f o r m a t i o n as it b e c o m e s a v a i l a b l e , and the n e c e s s a r y r i g o r to h a n d l e s e r i o u s s c i e n t i f i c and a p p l i c a t i o n r e s e a r c h .
82 With the accumulation of information over time, such a management structure would ideally allow increasingly comprehensive analyses with greater depth and spatial resolution. In this paper we describe a database approach which utilizes a geographic information system (GIS), satellite imagery, and a large number of case studies of agroforestry and land use for subSaharan Africa. By demonstrating how such a framework can: (1) manage agroforestry information; (2) make it geographically meaningful at scales greater than the local site; (3) how the framework is updatable; and (4) able to incorporate large amounts of ancillary information; we will show how the method can be used to make preliminary estimates for the potential geographic extent of agroforestry types, as well as be relevant to other large scale questions. The central thrust of this paper is a preliminary description of the database approach and its potential utility. Thus we have included only a limited number of case studies of both land use/land cover and agroforestry for the purpose of database construction. As such the analysis here is not intended as a final, comprehensive study on agroforestry application in Africa. Section 5 on 'Potential Improvement' outlines how the approach can be enhanced for increasingly comprehensive studies of agroforestry in Africa. The map presented here is a necessarily generalized example product from the database, and is intended to be illustrative only, and should not itself be used in analysis, which is more appropriately the role of the GIS database. Burrough [1988] provides a good definition of a geographic information system. 'Geographical information systems should be thought of as being very much more than means of coding, storing, and retrieving data about aspects of the earth's surface. In a very real sense the data in a GIS, whether they are coded on the surface of a piece of paper or as invisible marks on the surface of a magnetic tape, should be thought of as representing a model of the real world. Because these data can be accessed, transformed, and manipulated interactively in a geographical information system, they can serve as a test bed for studying environmental processes or for analysing the results of trends, or of anticipating the possible results of planning d e c i s i o n s . . . ' Following a brief description of the overall approach, subsequent sections describe in more detail: 1. the derivation of each layer of the GIS database; 2. how these layers interact to enable the estimation of the ecological and land use boundaries within which certain types of agroforestry could potentially be extended; and 3. the improvability of the database and analysis.
1.2. Ecological zonation, land use, and agroforestry Ecological zonation is undoubtedly the prime factor that determines the possibilities for the spread of agroforestry systems [Nair, 1989b], and it is across this zonation that structural differences exist in agroforestry types. However within this zonation, the practices, patterns, intensity, and spatial extent of land use govern the actual presence of specific agroforestry systems
83 and its potential adoption. But ecological zonation and land use are interconnected, such that the constraints and opportunities of land use are characteristic of specific agro-ecological regions [Nair, 1991]. Large scale consideration of the geography of the interconnections of land use and ecological zone can facilitate investigation into the transfer of present and improved agroforestry systems into new areas within zones [Nair, 1991]. The expansion of agroforestry systems to similar ecosystems and land uses makes particular sense from a development perspective. The failure of many imported development designs which have ignored existing local land uses in the tropics has highlighted the importance of having development efforts utilize in-place land uses as a foundation for development. Tailoring development schemes to existing land uses can facilitate quicker adoption and success while not destroying internal coping and risk reduction strategies inherent to many customary land uses. The spread of an agroforestry arrangement (traditional or modified) from one location to another within the same agro-ecological zone where land uses are similar, is perhaps one of the more rational transfers of an agricultural technology, because such designs have already proven themselves in similar ecological, land use, and in many cases social and cultural contexts. Such an extrapolation in the tropics is today and in the future perhaps more feasible than in the past for a number of reasons. Foremost among these is the increasing homogenization of traditional land uses. This is presently occurring not only within agro-ecological zones, but across ethnic, and cultural lines as traditional groups practicing very different land use ecologies are increasingly incorporated into state monetary exchange systems, and local, regional, national, and international markets.
2. Approach Over 170 site-specific case studies of agroforestry were utilized from ICRAF's database on agroforestry described in Nair [1989c] and from other sources in the published literature. And certainly many more than this number could be used from both sources for more detailed analysis at both continental, country, and within country scales. These descriptions encompass a wide variety of tree density, species, size, temporal and spatial patterning, and management practices. The advantage of using this information is that it contains descriptions of existing, functioning examples of agroforestry as presently practiced. The examples used included the many forms of agroforestry that result from the interplay of ecoclimatic and land-use options in various regions of Africa. As in-place, observed cases, they already embody the many biophysical and human variables involved in functioning agroforestry systems. In this sense, the basis of the work presented here is not hypothetical or theoretical, but instead based on agroforestry as currently practiced. The agroforestry studies were grouped into five broad structural types: silvopastoral, fruit tree, fuelwood, shelterbelts, and timber. Sitvopastoral agro-
84 forestry was further divided into three s u b - t y p e s due to large differences (largely related to precipitation) in tree biomass for fodder-producing trees in pastoral zones. M a n y examples o f the five major types are not practiced alone however, but in combination. Thus, a total o f 21 different c o m b i n e d agroforestry types were considered (Table 1). These 21 types were then distributed o v e r s u b S a h a r a n A f r i c a in the GIS on the basis o f two kinds o f information, each in separate layers within the GIS. First, the geographic regions and precipitation regimes for each type of agroforestry gave an indication of the ecological zone in which the different types are found. Second, a database o f current land use in subSaharan Africa (GIS layer and associated database information) was used to place, more specifically, different types of agroforestry within current, compatible land uses. The type of agroforestry appropriate for a given land use involved the agroforestry case studies themselves, and a set o f rules. For example, silvopastoral agroforestry, because it is practiced by certain variants of pastoralism in the case studies, is considered appropriate for similar pastoralist land uses where the agroforestry system is not presently practiced. Likewise shelterbelt agroforestry is practiced in rainfed crop systems; fruit and fuelwood Table 1. Land areas currently suitable for different types of agroforestry in subSaharan Africa
[Unruh et al., 1993]. Agroforestry type ~ Pastoral 0-400 mm Pastoral 400-800 mm Pastoral > 800 mm Fruit Fuelwood Timber Shelterbelt Pastoral/fruit 0-400 mm Pastoral/fruit 400-800 mm Pastoral/fruit > 800 mm Pastoral/fnelwood 0-400 mm Pastoral/fuelwood 400-800 mm Pastoral/fuelwood > 800 mm Pastoral/shelterbelt 0-400 mm Pastoral/shelterbelt 400-800 mm Pastoral/shelterbelt > 800 mm Fruit/fuelwood Fruit/timber Fruit/shelterbelt Fuelwood/timber Fuelwood/shelterbelt Total:
Area (x 103 ha) 166463 178503 112852 296788 74576 7537 105 0 0 4494 2255 49689 229900 52525 16449 17190 258022 30415 26937 767 23407 1548874
t Agroforestry types involving pastoralism are presented here with precipitation ranges due to large precipitation-related differences in biomass for fodder producing trees.
85 agroforestry are practiced in the fallow period of shifting cultivation systems, alley cropping arrangements and so on. In other words, the connection between land use and the suitable agroforestry system was extracted from the case studies where the connection between land use and agroforestry system has already been made and observed in the field. When more than one type of agroforestry was suitable for a particular land use, two types were assumed to be practiced together for the purposes of this study; although in reality of course many different types of agroforestry can occur together. The number of coexisting types was arbitrarily limited to two for the purposes of this preliminary construction of the database, but certainly more could be included. The connections between land use and appropriate agroforestry system defined a third GIS layer which identified areas for implementation of specific agroforestry types.
3.
Methodology
3.1. Precipitation Precipitation data was obtained from monthly global GIS data layers of corrected total precipitation [Legates and Willmott, 1990]. Individual months were summed to produce a data layer of total annual precipitation in 10 cm increments. This precipitation layer was then further simplified into 4 moisture regimes based on Nair [1989c]: arid (< 500 mm), semi-arid ( 5 0 0 - 1 0 0 0 mm), subhumid ( 1 0 0 0 - 1 3 0 0 mm), and humid (> 1300 mm). Political boundaries were then used to define four geographic regions of subSaharan Africa (west, east, central, and southern) that coincided with regions as defined in the ICRAF agroforestry cases. This 4-region layer was then combined with the layer of four moisture regimes to produce a 16-class data layer of regional precipitation zones. For most of these regions, descriptive data on prevalent agroforestry types within the case studies were of greater detail than the simple 16-class layer, meaning that improvement in this approach is possible. 3.2. Satellite imagery and land use A GIS data layer of present-day land use/land cover with 33 classes was produced from a combination of satellite imagery data and ground studies. This layer is briefly described here and detailed more completely in Houghton et al. [1993], and Unruh et al. [1993]. Land use/land cover are treated as a single category here due to the predominant influence that land use has on land cover. Such that land cover is determined by land use, or lack thereof. The satellite data were weekly, Advanced Very High Resolution Radiometer (AVHRR) 15 km-resolution, global vegetation index (GVI) data obtained from the US National Oceanic and Atmospheric Administration (NOAA). The GVI data are calculated at NOAA from the Normalized Difference Vegetation Index
86 (NDVI), which is the difference between a visible and an infrared hand, normalized by their sum [Holben et al., 1980; Tucker et al., 1980]. Dense vegetation cover results in high values of NDVI, while areas of bare ground, ice and snow, water, or clouds give low values. NDVI is correlated with land cover, phenology, and net primary productivity over broad geographic areas [Goward et al., 1985; Justice et al., 1985; Tucker et al., 1985], but the index has also been shown to have a number of limitations [Goward et al., 1990, 1991]. Three years (1986-88) of recent GVI data were combined to reduce the effects of interannual variability and spatial heterogeneity of climate that are especially problematic in Africa. Weekly GVI data were aggregated into months, taking maximum values from the component weeks, thereby reducing the effects of cloud cover. This monthly data was averaged for the three years to produce average monthly GV! values. An unsupervised statistical classification (the classing together of pixels of similar value) was then performed on the twelve averaged monthly GVI layers to produce a 33-class GIS layer of different classes. Monthly GVI means and standard deviations for each class were then extracted from the GIS and graphed. The resulting 'phenology curves' revealed the type, timing, and intensity of vegetative activity, and were used together with studies from the literature to assign labels of land cover and land use to the classes. Four examples of the 33 phenology curves are presented in Fig. 1. The selection of land use/land cover studies (from the published and grey literature) used to label the 33 classes (several studies used for each class) was based on three criteria. The studies had to have been carried out subsequent to 1980 so as to provide up-to-date ground information. They needed to describe in some detail the vegetation and land use of the area studied, and the study sites themselves needed to cover large enough areas to be descriptive of a significant number of pixels within the class being labelled. These large study areas also enabled the phenology curves to be matched with the ground descriptions of vegetation, seasonality of greenup and land use. Whenever possible, and especially for large classes, several studies were used in labelling. The 33 classes were labeled by using locational information (usually latitude and longitude) contained within the studies, to place them within the classes. The studies thus located within the classes were then used to label each class with respect to land cover and land use. The accuracy of the land cover/land use layer was assessed by comparing it with White's [1983] map of vegetation for Africa. The proportion of the areas (pixels) classified presently as forest and identified as forest by White was examined. A pixel-by-pixel comparison showed 95% agreement. Areas currently classified as woodlands were identified as woodlands with 85% frequency by White. The remaining 31 classes were not readily compatible with White's classification.
87 Desert
Humid Lowland Tropical Forest 140
140
135
135 130
130 _~
125
7
/,t
-
"----
~120 ~ j /
~115
~
/---,"
/
110
125 120
115~ 110-
/ ~ ' J
105
105-
100 -
100 "
95-
g5" lan
feb rna~
apr may
im
iul
aug
sep
oct nov dec
P
Jan
feb
mar
Secondary/Degraded Forest
iun
Id
aug
sep
oct
nov
dec
. od
nov
dec
Miombo Woodland
140
135t 130 ~ 12S'[
apt may
140 L
/t
/
f l
.....
~
f
1301~ -
--
1'51/f
"~.\ \ "
120"
!,,5t " ~
1,ol-f
Z 1051 7" ~00
95 9O ion
feb mar
apr may
ion
j~
aug
sep
oct
nov dec
90
.
jan
.
.
feb mar
.
.
.
opt moy
.
ion
.
i~
.
.
a~g
. ~p
Fig. 1. Four of the 33 examples of phenology curves derived from monthly NDVI means (solid line) and +/- two standard deviations (dotted lines). High values of NDVI indicate that a larger proportion of the land is covered by photosynthetically active vegetation. NDVI is calculated from 15 km resolution AVHR data. 3.3. The intersection of land use and agroforestry The intersection of the 16-class GIS layer of regional precipitation zones with the layer of current land use (33 classes) determined the type of agroforestry suitable at a given location. The intersection was facilitated by the construction of a matrix with the two data layers occupying the axes of the matrix (Table 2). The numbers in the body of the matrix correspond to the type of agroforestry observed (case studies) within the precipitation zones (horizontal axis of the matrix) which matches the land use from the land use data layer (vertical axis). Again the match o f agroforestry type and appropriate land use is a function of the agroforestry case studies themselves; these having already determined what appropriate and successful matches between land use -and agroforestry type can be by virtue of their existence in the field. Zeros in the body of the matrix indicate no existence of the land cover as a broad class in the precipitation zone in question. S o m e areas of land cover, such as dense tropical forest or hyper-arid desert, were less suited for the types of agroforestry considered in this study due primarily to low population densi-
88 II II u'3 c-I
2~
~m ,.-s
II e",
~
r~.S_ .6 ~oax II
~
:.
.o ..~
~-~
c~ ~
~u e~ m
~
2~ g~d ._~
"a .-
o
asN
< ,,
0 ~ 0 0 0 0 ~ 0 0 0 ~ 0 ~ 0
0
~
0
0
0
0
0
0
0
~
0
0
0
e~ ~4
~.~
~ .~
.-
~.~
"~
g
"a,. a p~
H
v~y
~o~
~o
89 0
0
0
0
0
0
0
0
0
0
0
~
0
0
~
0
0
~
0
0
~
0
0
0
0
0
0
0
0
0
~
0
0
~
0
0
0
0
0
0
0
0
0
0
0
~
0
0
~
~
~
0
0
~
0
~
~
,
0
0
~
0
0
0
0
0
0
~
0
0
0
0
0
~
0
0
~
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
4= < 0
0
0
0
~
0
~
0
0
0
0
0
0
0
0
~
0
0
~
0
0
~
0
0
0
0
0
0
0
0
0
0
0
0
0
~
0
~
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
cD
0
~
0
0
~
0
0
0
0
0
0
0
~
0
0
0
0
~
0
0
~
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.~,
0
0
0
o 9 0
E ~'~
E o
~- >
~
~
o
~._
~
~~
o
8
90 ties and the existence of relatively few examples of existing agroforestry. These areas were labeled as having no potential for agroforestry for the purposes of this study. The agroforestry GIS layer resulting from the matrix operation defined the area in subSaharan Africa with potential for agroforestry application at present [Table 1]. Again the agroforestry types included in this study are meant to be broadly representative only, and are not mean to suggest that these are the only types suited to a particular area. There exists considerable potential to improve this matrix arrangement so as to incorporate additional information for more precision in the analysis. The section below on 'Potential Improvement' examines some of these options.
4. Zones of agroforestry potential Table 1 summarizes the land area currently suitable for each agroforestry type, as shown in Fig. 2. A total of 1549 x 106 ha were estimated to be suitable
%
/
.i"
AGROFORESTRY TYPE
9 Pa:.-=4oroI [ ] P(s-;,tora l/Fuel wood m Pastora I/Shelter belt Fruit
i /~
:~iii!FruifjFuel wood , ,q Fuel~voc,. O t h e r Agroforesb_y types
Fig. 2. The location of the different agroforestry types used for this study within the countries of subSaharan Africa.
91 currently for some from of agroforestry in subSaharan Africa. This includes areas presently under agroforestry as well as areas of potential application. Six types of agroforestry accounted for more than 100 x 106 ha each. In descending order these were: fruit tree, fruit tree/fuelwood, pastoral/fuelwood, and each of the three silvopastoral systems. Pastoral systems, either alone or in combination with another type, accounted for a total of 830 x 106 ha. Systems with some component of fuelwood accounted for 639 x 106. And agroforestry where fruit trees were important for at least part of the practice accounted for 617 x 106. The area suitable for agroforestry with timber was approximately 38 x 106 ha. Figure 2 shows the location of the different agroforestry types within the countries of subSaharan Africa. This figure is an example of the type of maps the database can generate, and is much generalized for the purpose of printing. Table 3 shows the areas within each country suitable for the different types of agroforestry considered in this study. Higher resolution examinations of specific regions within a country would allow more detailed analysis of the potential of agroforestry to contribute to agricultural development within specific countries.
5. Potential improvement 5.1. Satellite imagery Utilization of higher resolution imagery (and its classification) would incorporate greater subtleties between land uses and improve the delineation of boundaries between uses; allowing more land use/land cover classes to be utilized. This would improve the precision in the match between land use and agroforestry type. Increasing the number of classes in the imagery would require greater numbers of ground studies to adequately label the larger number of classes, and these could be incorporated into the construction of the land use GIS data layer. The number of ground studies used in this study was not exhaustive of the literature, and the potential exists for much improvement with the addition of more studies, especially from the grey literature. Including new ground studies as they come out would be an important additional facet of the framework presented here, primarily because this would allow the database to be updated as land use changes in type, extent, intensity, etc. This, together with periodic re-classification of the imagery (including the most current imagery) would allow the areal extent and label of the classes to be updated so as to have not only a database that increases in resolution and in the precision of land use labels, but is kept up to date so as to continue to have contemporary utility for agroforestry. In addition, updating the land use layer would eventually create a database of the history of land use, allowing sequences of change (development, degradation, etc.) of particular regions to be observed. Knowing such sequences can be important in designing
92 Table 3. A r e a s ( 1 0 0 0 ha) o f potential a g r o f o r e s t r y a p p l i c a t i o n for s u b S a h a r a n A f r i c a . No
Pastoral
Fruit
Fuelwood
Timber
Shelterbelt
agroforestry
Pastoral/ fruit
option Angola Benin
15610.8 74,3
8466.9 4009.8
40121.5 0,0
0,0 0.0
0.0 0.0
0,0 0,0
1515.4 0.0
1279.9 2192.7 839.0
21431,8 15487.2 20.5
1978.0 0.0 143.2
0.0 0.0 0.0
0.0 0.0 0.0
581.8 0.0 0.0
0.0 0.0 61.4
Cameroon Central African
18652.4 11853.4
4219.0 1651.1
49.3 24.6
0.0 0.0
0.0 0.0
0.0 0.0
0.0 0.0
Chad Congo
69766.5 19825.4
38643.7 0.0
49.5 6416,8
0.0 0.0
0,0 0.0
0.0 0.0
0.0 46.0
Ivory Coast Ethiopia
11885.2 36435.1
432.2 25172.9
216,1 17227.3
0.0 13147.2
168.1 572.7
0.0 0.0
0.0 0.0
Gabon Gambia Ghana
16563.9 94.2 6062.8
233.7 580.7 6671.7
2544.3 0.0 79.4
0.0 0.0 0.0
0.0 0.0 26.5
0.0 0.0 0.0
0.0 0.0 0.0
Botswana Burkina Faso Burundi
Guinea
3586,4
3539.3
0.0
0.0
0.0
0.0
0.0
Kenya Lesotho Liberia
24488.2 0.0 7525.0
12293,8 603,3 60,2
2135.9 2175.7 240.8
2831.3 0.0 0.0
695.4 0.0 0.0
0.0 0.0 0.0
0.0 0,0 0,0
Madagascar Malawi
14479.6 1979.4
651.0 0.0
14322.4 5655.4
4287.7 1131.1
2244.9 0.0
0,0 0.0
0,0 0,0
Mali Mauritania
85885.0 96410.3
25073.8 6659.7
0.0 0.0
0.0 0.0
0.0 0.0
0.0 0.0
0.0 0.0
Mozambique Namibia Niger
5083.9 28512.1 99067.8
905.4 28393.5 26742.5
38559.0 94.9 0,0
10701,8 0,0 0.0
742.9 0.0 0.0
0.0 71.2 0.0
0.0 0,0 0.0
Nigeria Rwanda Senegal Sierra Leone Somalia
3941,5 928.2 2437.7 3337.4 35954.0
35812.7 0,0 11743.3 157.7 22508.6
52.2 376.3 0.0 0.0 49.8
0,0 275.9 0.0 0.0 74.7
0.0 100.3 0.0 0.0 0.0
0,0 0.0 0,0 0.0 0.0
0,0 0,0 0,0 0.0 0.0
South Africa Sudan Swaziland Tanzania
17683.6 121827.4 88.3 13461.8
46573.4 88136.2 0.0 276.3
26962.4 568,9 382.5 36166.1
1546.5 21941,3 29.4 11603.3
44.8 173.2 0.0 1004.6
44.8 0.0 0.0 0.0
0.0 0.0 0.0 0.0
264.6 9509.0
1547.0 362.7
0.0 414.6
0.0 5233.8
0.0 1528.7
0.0 0.0
0.0 0.0
82252.5 4099.9 903.7
12164.3 1436.3 4493.2
60429.9 36037.5 3313.4
0.0 1671.3 100.4
182.7 52.2 0.0
0.0 0.0 0.0
2323.2 548.4 0.0
Togo Uganda Zaire Zambia Zimbabwe Total
874842,4732 457155.237 296787.7055 74575.73272 7536.939481 697.7408596 4494.386704
Percent area for all countries 36.13%
18.88%
12.26%
3.08%
0.31%
0.03%
0,19%
93 Table 3. Continued. Pastoral/
Pastoral/
Fruit/
Fruit/
Fruit/
Fuelwood/
Fuelwood/
Total country
fuelwood
shelterbelt
fuelwood
timber
shelterbelt
timber
shelterbelt
agroforestry potential
12363.6
0.0
46591.9
0.0
0.0
0,0
0.0
6088.9
0.0
470.3
49.5
544.5
24.8
0.0
L24670 11262
2327.0
26388.4
6050.2
0,0
0.0
0.0
0.0
60037
9670.9
0.0
23.1
0.0
23.1
23.1
0.0
27420
204.6
0.0
1493.8
20.5
0.0
0.0
0.0
1783
17665.5
1973.8
98.7
2714.0
2146.5
24.7
0.0
47544
46403.1
0.0
0.0
2045.4
172.5
147.9
0.0
62298
19717,7
0.0
222.7
0.0
0.0
0:0
0.0
128400 34200
2966.9
2920.9
0.0
667.0
46.0
0.0
1311.0
16687,2
1056.5
48.0
912.4
816.4
0.0
24.0
32246
310.2
954.4
10594.1
286.3
10761.1
0.0
6728.7
122190
285.6
5919.4
0.0
830.8
285.6
0.0
103.8
26767
392.4
0.0
0.0
62.8
0.0
0.0
0.0
1130
9054.5
741.3
370.7
264.8
556.0
26.5
0.0
23854
14510.9
0.0
0.0
2689.8
259.5
0,0
0.0
24586
173.9
4942.3
5463.9
769.9
1241.8
0.0
3228.7
58265
54.8
0.0
146.3
0.0
0.0
0.0
54.8
3035
90.3
1444.8
0.0
1083.6
692.3
0.0
0.0
11137
0.0
5208.2
11763.2
2649.0
1055.1
0.0
2042.9
58704
0.0
282.8
2686.3
0.0
0.0
0.0
113.1
11848
12969,2
0.0
0.0
48.0
0.0
24.0
0.0
124000 103070
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
2785.7
17578.0
208.9
2460.7
0.0
835.7
80159
877.7
20209.9
4269.7
0.0
0.0
0.0
0.0
82429
889.7
0.0
0.0
0.0
0.0
0.0
0.0
126700
44374.4
730.9
104.4
1957.7
4907.3
495.9
0.0
92377
0.0
75.3
551.9
0.0
25.1
0.0
301.0
2634
5344.2
0.0
0.0
93.8
0.0
0.0
0.0
19619
2128.5
2t0.2
0.0
1103.7
236,5
0.0
0.0
7174
24.9
1718.0
3386.2
0,0
49.8
0.0
0.0
63766
12147.6
5110.1
8942.6
156.9
0.0
0.0
2891.2
122104
2300.5
0,0
15114,0
0.0
98,9
0.0
420.5
250581
88.3
0.0
1147.5
0.0
0.0
0.0
0.0
1736
0.0
2838.0
26948.8
728.3
25.1
0.0
1456.7
94509
3236.4
0.0
223,9
101,8
305.3
0,0
0.0
5679
466.4
0.0
3679.2
1580.5
0.0
0.0
829.1
23604
35527.0
0,0
32916.7
8744.7
0.0
0,0
0.0
234541
652.9
0.0
28072.7
26.1
0.0
0.0
2663.6
72561
753.0
326.3
28766.4
0.0
0.0
0.0
401.6
39058
280748,7709 11.59%
85837.156ll 3.54%
258022.2038 29796.12197 26709.17588 10.66%
1.23%
1.10%
766.8051094 23406.55087 0.03%
0.97%
94 improved agricultural arrangements, because history of land use has been shown to be very important in the potential development of the landscape [Rindos, 1980; Harlan, 1975; Harris, 1972; Unruh, 1994]. Incorporating a large number of years of imagery into the analysis (beyond the scope of this preliminary work) would allow the calculation of a median idea as to the expected precipitation and its seasonality and the resulting effect on vegetation greenup. This holds the possibility in the future, with more imagery than is presently available, to put extreme years of above and below 'normal' precipitation into perspective; and together with other information allow investigation into the frequency and severity of such extreme years, which can be very important to the long-term success of agricultural technologies involving plants and people. Excluding information on the frequency and severity of extreme years in African rural development efforts has lead to numerous failures in agricultural interventions, often with serious food security repercussions. 5.2. Geographic information system Expanded use of the GIS in the analysis holds considerable potential. Additional information can be included in separate layers of the GIS for more precise and varied types of agroforestry-related investigations. Digitized GIS layers of soils, land degradation, elevation, land tenure types, population (numbers and rates of increase), location and rates of deforestation, infrastructure, watersheds and river courses, development projects, and the areal extent and type of extension services would add greatly to the predictive capability of the relationship between agroforestry and land use. Specialists with their own data layers might then potentially add these to those already in the database for specific investigations and applications. 5.3. Matrix operations Additional agroforestry types can be incorporated into the matrix procedures (greater agroforestry detail). What are now broad categories of land use can be differentiated into their component parts to accommodate the greater agroforestry detail, and allow for more options within the matrix as a result of different variants of land use coming into contact with different types of agroforestry. While matrix construction and operation does entail a good deal of careful work, it can provide great payoff in terms of detail and accuracy. With increasing data types, additional matrices can be constructed to interact with each other, either by constructing matrices with the output from other matrices, or by constructing a 'stack' or 'chain' of matrices which are co-registered, enabling queries in the form of algorithms to be submitted to the 'stack' for different purposes.
95 6. Conclusions What we have attempted here is a preliminary look at the potential for handling site specific agroforestry data in an aggregate manner so as to be able to extrapolate such information to larger areas, and facilitate large scale agroforestry research and application inquiry. The existence of ICRAF's agroforestry database facilitated the operation of the framework described here. And there exists some potential for this approach to further compliment the ICRAF database. For example it might be possible, for each agroforestry case within the ICRAF database, to determine where else in Africa the case might potentially be adopted. Or for any region of the continent it might be possible to determine which agroforestry types could most suitably be implemented. One of the more important larger-scale issues that the field of agroforestry could, and perhaps should be contributing to, is global change. However the agroforestry research effort has been unable to participate in this discussion in a real way because most agroforestry investigations are local and site specific. It is well known that agroforestry addresses problems of agricultural productivity, food security, sustainability, vulnerability, land degradation, fuelwood shortages, and deforestation and reforestation and the associated releases and storage of carbon. These are all global change topics being pursued by various fields of study. However there has not existed a way to approach the management of aggregate agroforestry information which would allow large-scale agroforestry research to contribute to discussions of global change. As a result agroforestry as a mitigation option for biotic emissions of carbon, for developing countries presently under great pressure to pursue activities to sequester and slow the release of carbon into the atmosphere, has not been considered to the extent that it could, in international conventions. This is unfortunate given that agroforestry already exists over large areas in many less developed countries. An analysis of aggregate agroforestry and carbon storage for subSaharan was performed using the framework described here and is detailed in Unruh et al. [1993]. The need for aggregate level agroforestry research is illustrated by the widespread importance of the practice, and the severity and extent of the land use problems that agroforestry is able to address. This importance, and the maturing of agroforestry as a scientific field of study, presents the opportunity for at least a branch of the overall research effort to concern itself with the larger-scale questions that similar fields (e.g. forestry, ecology, geography) are also presently pursuing.
References Burrough PA (1988) Principles of Geographic Information Systems for Land Resources Assessment. ClarendonPress, Oxford
96 Goward SN, Tucker CJ and Dye DG (1985) North American vegetation patterns observed with the NOAA-7 advanced very high resolution radiometer. Vegetatio 64:3-14 Goward SN, Dye DG, Dulaney W and Yang J (1990) Critical assessment of the NOAA Global Vegetation Index data product. Workshop on Use of Satellite-Derived Vegetation Indices in Weather and Climate Prediction Models, pp 34-42. NOAA World Weather Building, Camp Springs, Maryland, US Department of Commerce, National Oceanic and Atmospheric Administration Goward SN, Markham B, Dye DG, Dulaney W and Yang J (1991) Normalized difference vegetation index measurements from advanced very high resolution radiometer. Remote Sensing of Environment 35:259-279 Harlan JR (1972) Crops and Man. American Society of Agronomy, Madison, Wisconsin, USA Harris DR (1972) The origins of agriculture in the tropics. American Scientist 60:183-193 Holben BN, Tucker CJ and Fan C-J (1980) Spectral assessment of soybean leaf area and leaf biomass. Photogrammetric Engineering and Remote Sensing, 46(5): 651-656 Houghton RA, Unruh JD, Lefebvre PA (1993) Current land cover in the tropics and its potential for sequestering carbon. Global Biogeochemical Cycles 7:305-320 Justice CO, Townshend JRG, Holben BN and Tucker CJ (1985) Analysis of the phenology of global vegetation using meteorological satellite data. International Journal of Remote Sensing 6:1271-1318 Legates DR and Willmott CJ (1990) Mean seasonal and spatial variability in gauge-corrected global precipitation. International Journal of Climatology 10:111-127 Nair PKR (ed) (1989a) Agroforestry Systems in the Tropics. Kluwer Academic Publishers, Dordrecht and International Council for Research in Agroforestry, London Nair PKR (1989b) Ecological spread of major agroforestry systems: In: Nair PKR (ed) Agroforestry Systems in the Tropics. Kluwer Academic Publishers, Dordrecht Nair PKR (ed) (1989c) Agroforestry Systems in the Tropics, Appendix I. Kluwer Academic Publishers, Dordrecht and International Council for Research in Agroforestry, London Nair PKR (1991) An approach to developing methodologies for evaluating agroforestry systems in India. Agroforestry Systems 16:55-81 Rindos D (1980) Symbiosis, instability, and the origin and spread of agriculture. Current Anthropology 21:751-772 Tucker CJ, Holben BN, Elgin JN and McMurtrey JE (1980) Relationship of spectral data to grain yield variation. Photogrammetric Engineering and Remote Sensing. 46:657-666 Tucker CJ, Townshend JRG and Goff TE (1985) African land-cover classification using satellite data. Science 227:369-375 Unruh JD (1994) The role of land use pattern and process in the diffusion of valuable tree species. Journal of Biogeography 21:283-295 Unruh JD, Houghton RA and Lefebvre PA (1993) Carbon storage in agroforestry: an estimate for subsaharan Africa. Climate Research 3:39-52 White F (1983) The Vegetation of Africa. UNESCO, Paris
