|Environment, Biodiversity and Agricultural Change in West Africa (UNU, 1997, 141 pages)|
|Pilot study of production pressure and environmental change in the forest-savanna zone of southern Ghana|
|6: A multidisciplinary integrated methodology|
In-depth field study concentrated at the three sites, Yensiso, Amanase and Sekesua, located between 60 and 120 km inland by road from Accra. They were inhabited before or from the year 1910, when the cocoa boom was well under way in Ghana.
Yensiso site in the Akuapem district, cradle of Ghana's crucially important cocoa industry, covers the Yensiso, Adenya, Gyamfiase and Kokormu cluster of villages occupied predominantly by the native Akuapem and Ayigbe and Ewe migrant tenants. Amanase site in the historically important cocoa district in southern Akyem covers the juxtaposed villages Amanase and Whanabenya, settled by migrant Akuapem and Siade/Shai landowners, and by a mixture of other migrants, including Ayigbe and Ewe tenants. Sekesua site in a migrant Krobo district covers the adjacent villages Sekesua and Osonson. These areas had attracted a certain amount of research that could form a basis for further work to inform land use and environmental policy (Field 1943; White 1956; Hill 1963; Huber 1963; Hunter 1963; Gyasi 1976, 1979; Gyasi et al. 1990; Gyasi 1991; Amanor 1994; Gyasi 1994b, 1996).
The field study to generate firsthand information was carried out on a team basis among the interdisciplinary research group by rapid rural appraisal and more deliberate methods in October 1993. Work included group discussions; the popular questionnaire and non-questionnaire interviewing by sample; ground transect involving quadrat sampling and visual observations; and market survey (Gyasi 1994a). Other sources of information included historical records and aerial photographs. These variously assembled data were analysed by laboratory, statistical, cartographic and other techniques. In so doing, we drew on various ideas concerning the evaluation of sustainability and unsustainability.
The group discussion and transect methods are singled out for discussion because of space limitations, and also because they appear to be particularly efficient for studying agroenvironmental trends and would, therefore, appear to warrant more publicity than the other, better known methods.
The small farmers who are the major users and, therefore, the principal modifiers of the biophysical environment were the primary target of the survey by group discussion and the other methods.
Table 6.1 Distribution of Farmers with Whom Discussions Were Held on a Group Basis
|Yensiso: Yensiso village, Kokormu and Gyamfiase||50||35||85|
|Amanase: Amanase village and Whanabenya||35||15||50|
|Sekesua: Sekesua village and Osonson||40||20||60|
1. Figures are approximate.
2. In two cases, there was a sprinkling of women among the predominantly male groups.
For the purpose of the group discussion, our original plan was to segregate the farmers into the following four groups: elderly males including the chiefs; young males; elderly females; and young females. However, because of time constraints and difficulties in mobilizing the farmers, it was found necessary to combine these four groups into two principal ones, males and females. In two cases, the men and the few women had, unavoidably, to be combined into one group because of a severe time limitation.
We assumed, a priori, that the farmers' perception of their agroenvironmental situation and changes therein would show significant variations between the sexes, and between the different age categories, because of differences in their socioeconomic role, experiences and aspirations, hence the segregation.
Using a check-list of relevant agroenvironmental parameters and a cassette recorder, we held separate discussions with each of the various groups of men and women farmers, totalling approximately 200 in the three study sites, with one or more of us leading or moderating the discussion (table 6.1). These discussions were held in the mornings on non-farming days, or on the market day, when the farmers could easily be rallied by public notification through the chief, chief farmer and other community leaders.
On the whole, the information obtained through the group discussions was similar to that obtained through the individual interviewing by the structured questionnaire method. Both methods revealed a popular awareness of major changes, generally of a negative character, in the biophysical environment. They also revealed a popular awareness of some possible solutions to the problems, and a general willingness to participate in corrective programmes. However, evidently because of the debate that it engendered among the broad spectrum of farmers, the group discussion approach generated more information, apparently of greater reliability, about changes in biodiversity, including reduction in the quantity and variety of wild mushrooms, plants and animals, and in traditional crops and farming systems, as reported in ensuing chapters. Moreover, the group discussion was faster because it did not entail disparate individual discussions with farmers in their scattered households, as was the case with the structured questionnaire method.
An important finding from the group discussion was that the women, when mixed with the men, hardly spoke. However, they opened up when discussions with them excluded the men, especially their husbands. An illustrative case is recalled. It emerged, during discussion with an exclusively female group, that one member, a young woman, was expecting her fifth child, despite her complaints of overpopulation, poverty and difficulty in caring for her four children. With a dare, a male moderator attributed her obviously high birth rate to overindulgence in nocturnal matrimonial entertainment occasioned by the boredom associated with the lack of electricity and alternative entertainment in the village; whereupon the young woman promptly responded that the boredom and, hence, the overindulgence the moderator had alluded to, were not limited to the night, but extended to the day as well! Such is the kind of frank free-flowing information that group discussions could elicit.
The group approach confirmed our original hunch of important differences in the environmental perceptions between the sexes. For example, when talking about the effects or symptoms of environmental degradation, the women emphasised a shortage of firewood, a household item which, traditionally, the women are required to furnish; while the men emphasised the proliferation of weeds, e.g. C. odorata, whose clearance is, by custom, the responsibility of the men. Furthermore, because of its participatory character, this approach appeared to enhance researcher-target group interaction. This underscores the great potential of group discussion as a tool for sustainable community-based environmental management programmes.
A complementary method of surveying vegetation and land use in different ecosystems is ground transect or traverse across the agroecological landscape. This method is greatly facilitated by the use of quadrats along the transect line, although, it should be noted, the quadrat method could be used independently of the transect method.
The multidisciplinary team (composed of a botanist, soil scientist, and geographers specializing in land use and agricultural systems, all backed by technicians) employed a combination of transect and quadrat methods for the survey of the agroenvironmental conditions along footpaths radiating up to approximately 2 km each from the following villages:
Overall, a total of about 5 km was traversed.
Ideally, as briefly argued below, the transect should have followed a straight linear distance. However, as this approach would have entailed cutting through the bush at considerable cost in time, which we could ill-afford in view of the severe time limitation for our study report, we used the pre-existing footpaths or roads devoid of major obstacles.
Along the traversed paths, distances were measured by the chain and tape, while observations on soils, plants, farms, fallow lands and other agroecological conditions were made by the eye, by instrument, or by counting, often within rough-and-ready quadrats measuring approximately 30 x 30 m2.
Although the transect method employed generated considerable useful information at little cost in time and money, it nevertheless raised several issues, including the following:
Typically, footpaths are crooked, reflecting, as they generally do, lines of least resistance to human movement, and not necessarily the typical biophysical and agricultural conditions of the study site. Therefore, transects along footpaths might give a biased or distorted picture of the general agroenvironmental situation. A more objective alternative would seem to lie in a linear transect moving in a direction that, on the basis of aerial photographs, satellite images and reconnaissance field surveys, including discussions with the local inhabitants, promises a balanced representation of the general agroecological (including land use and cover) conditions or contrasts.
The optimal transect length might be 4-5 km, which is considered to be the maximum distance that could be conveniently walked in a day for farming purposes.
As for interesting peculiar features far removed from the traverse path, these might be singled out for survey by offsets and other related techniques despite their great deviation from the path.
A 1 x 1 m2 quadrat would seem to be appropriate for a sample survey of species characteristics in terms of type, frequency, dominance etc., while a 25 x 25 m2 quadrat appears adequate for the survey of trees irrespective of height.
There is no easy answer to the question of quadrat sample size or number. However, a general statistical rule is to avoid the following:
In sum, the optimal sample size is the one that minimizes, at the least conceivable cost, the margin of error to a reasonable or tolerable level, which can be expressed in terms of confidence limits.
Assuming that L is the tolerable error in the sample mean (X) and there is a 5 per cent chance of the actual error not exceeding L, then the required sample size n could be represented thus:
where n = the required sample size; a = the population standard deviation; L = the allowable error or given or acceptable limit of error.
A requirement for the operationalization of this equation is an estimate of the population standard deviation (s), which could be guessed, with a fair degree of accuracy, from the results of previous samples of a similar character. In order to expedite our work in the wake of the severe time constraint, the quadrats were limited to one side of the transect path. However, a more thorough or broadly based quadrat survey might better be earned out along both sides of the transect, either continuously or, more realistically, in an alternative or staggered formation. However, this and the other issues raised above require further investigation.