Photo 43. Inventory in Mali: Burkea africana (© Rivière/Cirad)
Once the assessment objectives and framework have been defined, the next question is to implement them within a specific situation. The most suitable tools, arrangements and methods (Annex 2) are determined in accordance with the type of tree and scale of the exercise. The next step is to organize operations.
The choice of tools and methods used to describe or assess Trees outside forests depends on the scale of analysis, kind of data, and degree of exactitude desired (Box 47).
The tools used in the study of Trees outside forests are generally neither specific nor new, what is more original, often enough, is the way they are combined and implemented. Of these tools, aerial photos or satellite images offer the possibility of mapping vast areas of land. The question then arises, whether spatial analysis alone is sufficient, and appropriate, for inventorying trees growing outside the forest.
The chosen procedure depends on just which data managers wish to acquire. Some are easily obtainable whereas others demand more time and money. Categorizing the parameters for inclusion in an inventory of Trees outside forests, we start with the most rapid for large-area coverage (Table 8).
The three levels of assessment
Three complementary levels of assessment, using different tools, can be identified (FAO, 1981).
Usually, most of the data pertaining to the first two levels can be acquired from aerial or satellite images. The eagle's-eye, stereoscopic perspective of the landscape makes aerial photography the ideal tool for counting, demarcating and distinguishing between the various types of Trees outside forests, placing them in their ecological context, situating them with respect to inhabited areas, and perhaps monitoring patterns of change over time.
Large-scale photographs (under 1:10 000) lend themselves most readily to precise interpretation. Some of the earlier mentioned projects such as Kenya and Hong Kong were based on enlargements down to the 1:1 000 scale. With large-scale photographs of excellent quality, crowns a few metres in diameter are easily distinguishable, and a certain amount of data can be obtained directly from the photos, such as density, crown shape and size, stand structure, species identification, and so forth.
Full interpretation of large-scale aerial coverage is a long and costly job for very large area coverage. This is why some projects prefer sampling grids, most often systematic - a method common in national inventories and which can be adjusted for special needs. In the Hong Kong study, for example, the dot grid is used for a specific interpretation (land cover), and for area interpretation (rate of tree cover). For Kenya, the systematic dot grid was used to define stand types. Additionally, all single trees within a central square on the frame were counted and described in terms of crown diameter and shape.
Satellite imagery can be used to support, or even replace, aerial missions. It offers an overview of a territory and is an appropriate tool for agroecological stratification. High-resolution Spot, Landsat and IRS imagery permit an initial demarcation of the major land cover types at reasonable cost, clearly showing forested, farmed and urban areas. Picking out trees in satellite images depends mainly on the pixel size (10 to 30 m) and the type of tree system concerned. A research project estimating oak density in southern Spain's dehesas silvopastoral system used panchromatic Spot images (Joffre and Lacaze, 1991). The study showed that tree cover in these formations could be assessed from the 10 m pixel images. Special effects such as the use of filters to heighten contrast satisfactorily correlated tree densities with the pixel values of the image. Oak has a particularly compact crown, which explains the quality of the results. Another study on Acacia senegal stands in Sudan's Kordofan region (Prince, 1987), led to the conclusion that canopy consistency plays a significant role, and that the phenomenon of spectrum reflectance accounts for the "transparency" of certain trees in tree cover measurements based on satellite imagery. Single trees are harder to pick out than trees growing in rows or clumps. New, one-meter resolution sensors (Ikonos) hold out more promise, but current costs limit their use over large areas.
Table 8. Parameter categories by speed and type of results
Presence, location, spatial organization, area, density
Species composition, structure, crown shape
Height, diameter, volume
Spatial dynamics, volume increase, etc.
Data on uses of wood and non-wood products
Real and potential use of resource
Aside from the use of satellite imagery for stratifying or monitoring the dynamics of a given region, widespread use of satellite data for exact assessment of Trees outside forests probably cannot be recommended at this time. But at the country level, satellite data do allow stratification on the basis of ecological criteria and land cover, providing the basis for a good working document for more detailed later work in representative sectors. In terms of systematic procedures, satellite imagery is a better way of taking regional diversity into account, which is why it was adopted for the woody biomass inventory project in Mali (Nasi and Sabatier, 1988).
The second series of parameters, which can only be acquired by direct measurements in the field, concern dendrometry, species identification, and so forth. There is no specific, all-inclusive national-level inventory for Trees outside forests at the present time (Box 48). Forest inventories seem to have mastered the techniques implemented for this sector, but as Kiyiapi (1999) points out, inventory methods deal with rare objects, such as indicator species, and so lead to overly vague estimates, which really ought to be upgraded.
The literature offers several examples of forestry inventories whose design has been used as a basis for gathering data on Trees outside forests. The U.K. (Smith and Gilbert, 1999) broke down the National Inventory of Woodland and Trees (NWS), launched in 1995, into two separate inventories, The Main Woodland Survey (MWS), and the Survey of Small Woods and Trees (SSWT).
(T.N. I have omitted three footnotes here, as they consist of a translation from the original French into English which would be redundant in the English version).
The Survey of Small Woods and Trees undertook to gather data on groups of trees covering an area of less than 0.1 ha, linear features, and small woods of 0.1 to 2 ha with crown densities exceeding 20 percent. The sampling procedure is based on a systematic, one-kilometer grid covering the country, with a square drawn on 100. The one-km sampling unit was subdivided into 16 squares, 250 m on one side, with the first chosen at random and the second at a distance of six squares. Aerial photographs were used, especially for the preliminary extraction of part of the data such as the number of small woods or the number of line-plantings on the one-km units. The field data collected for clumps, rows and single trees primarily concern the number of living or dead trees.
Ghana (Affum Baffoe, 1995) undertook an inventory of off-reserve trees and forests. The objective was to chart the importance of this resource, tapped, especially by rural communities, for both wood and non-wood forest products. A 0.02 percent sampling rate was deemed adequate to produce reliable estimates at the national and provincial levels. The stratification (or "division") of the country was based on 1:50 000 scale topographical maps, considered as strata, within which sampling plots were selected at random. Slope, type of land use and vegetation were recorded for each plot. For each tree, the species, diameter, status, mortality and causes of mortality were listed.
The Sudan inventory (Glen, 2000 developed a systematic sampling method different from the one described in the above examples, but also slanted towards forestry (including ecological aspects), and ignoring social criteria.
Other countries, however, poorer in forest resources, make more of an effort to link Trees outside forests to the people who actually manage them and upon which they depend for their livelihoods. These more "social" concerns are reflected in the sampling methods, which include villages or households in the sampling plan. In 1991, the Forest Service of India (FSI) did a nationwide assessment of trees in non-forest areas (Pandey, 2000). A pilot inventory was set up in five States to determine the best sampling size. The method adopted was stratified random sampling where a district or group of districts are treated as strata, and villages as sampling units. All standing trees over 10 cm in diameter at breast height (and sometimes 5 cm) in the selected villages were counted and measured. Because the results showed a strong correlation between village area, population size, and the number of standing trees, the FSI decided to modify its methodology. A new stratification was chosen on the basis of agroecological criteria. Fewer villages were sampled, and so the data were slightly less accurate, but field operations were considerably lightened as a result.
From forest inventory to the inventory of trees outside forests
The inventory seems an essential tool for estimating trees outside forests. Forest managers and planners have always considered forest inventories very important. Forest inventory methodologies are constantly updated as and when new questions arise or new techniques and tools are developed. The vast expertise in this domain builds upon many years of experience. Inventories of trees outside forests, instead, are far rarer and more disparate. With few exceptions, the methods employed owe more to experimentation than to careful operational planning. If we want to extend forest resource assessments to Trees outside forests, we will have to answer a number of questions concerning the special role and nature of this resource. We need to ask whether what is feasible for the relatively compact "forest" areas within a given country is equally conceivable for tree systems which are so geometrically varied. Singh, K. (2000) comments that the ideal thing would be a single, integrated inventory. But the current state of knowledge is such that one would tend to recommend not one but several inventories of Trees outside forests, given their highly diverse situations. This idea is supported by a review of the existing literature on the subject.
The proposed methods always attempt to adapt to the local issues, types of Trees outside forests, available resources, etc. What is the common ground between inventories of urban trees, agroforestry parklands, pastoral areas or hedgerows? And yet, it would be a distinct advantage for a country to have sound, reliable data that held for the entire country. What is being accomplished today in some of the temperate countries may appear utopian for others, at least in the short term. The issue merits discussion, and certain methods hammered out in some tropical countries, often under very difficult circumstances, constitute a backlog of experience that cries out for review.
To determine the supply of and demand for wood in the State of Kerala, the Kerala Forest Research Institute (KFRI) inventoried trees on farms (homesteads) in 1988-89). The State was stratified on the basis of area under agricultural use and population density. A three-stage sampling procedure treated villages as first stage sampling units, some households as second stage units, and randomly selected holders of dry lands as the third stage. All trees in the selected homesteads were then counted and measured.
Another, more detailed study, also in the State of Kerala (Mohan Kumar et al., 1994), described these homesteads. The inventory procedure was similar. Administrative units were selected in 14 districts of the State, home gardens stratified into three categories by size, and farmers chosen at random. The study collected data on crops, shrubs and trees (over 15 cm in diameter), using field measurements and on-farm interviews. Very interesting data emerged on the species composition and structure of home gardens, the disparity between provinces, and available wood supplies, especially fuelwood. The team successfully showed, for example, that species diversity lessened as farm size increased, and that property size shrank as population density increased. This type of inventory can also give us a better grasp of management methods and of how these tree systems function.
Field inventories may choose from two types of data as the basis for sampling: ecological units and land cover, or else social or administrative units, the latter being probably the more appropriate choice for densely populated areas. In Kenya (Holmgren et al., 1994), the authors used two stratifications, one agroclimatic and the other social. They did comment, however, that the agroclimatic stratification failed to explain the distribution of off-forest trees. In the inventory done in India (Panday, 2000), agroecological stratification prior to the selection of villages was the preferred method.
As these observations show, the inventory method should not be chosen without good prior social data on the region. The selection of stratification criteria is fundamental for establishing inventory procedures.
To move from an "inventory" of tree densities and species to an "assessment of Trees outside forests" implies looking into the usages and practices of the people using and maintaining these tree resources. Usage and practice vary with cultural, social and economic context. This has already been pointed out, as in the example of coffee-growing in Central America, where people of very different origins and sociocultural histories, though neighbours, opted for very different production styles. The social groups or rural households involved in tree dynamics need to be grasped as a basic decision-making level, especially where decentralization is imperative. A broader approach will not be able to tease out local determinants such as labour supply, social relationships and family structure. And yet social relationships (sex, social status, and the like) are determinants of access to and use of trees and land.
Sociocultural analysis requires an anthropological approach, involving participatory observations, interviews and surveys to discern the various social bonds and rules of access to tree resources. It offers the additional advantage of safeguarding local and traditional lore about the sustainable use of a great many species, which is fundamental to tree management. Ignoring this approach in the case of Trees outside forests (which are so intimately bound up with people's lives) would be tantamount to ignoring their dynamics. The selection of social criteria as prime determinants is clearly relevant for a resource so dependant upon people, especially in heavily populated areas. And indeed the distribution of off-forest tree systems in densely populated areas cannot be explained by climate and ecology alone. Living in a village, observing and conducting surveys, can reveal practices. But it is a more subtle thing to work out how social bonds affect these practices. Someone extraneous to the village and to the society in question will need time to gain the confidence of the local people, and time to grasp the complex sets of social rules underlying the balance of power, eventual conflicts, and tacit understandings.
Despite these difficulties, development agents have come up with a number of tools such as the Rapid Rural Appraisal and the Participatory Rural Appraisal18 to gain a quick grasp of rural concerns. However, for several reasons these standardized survey methods are no substitute for detailed sociological research. The viewpoints recorded tend to reflect the special interests of the more influential members of the community rather than the community's real problems. Underlying or implicit conflicts do not immediately or easily come to the surface.
The great and varied web of social situations demands specific research. Though of only local validity, it will bolster more overarching studies, and is needed for planning management interventions. The aforementioned bibliographical references and the support of anthropologists or sociologists should make it easier to surmount constraints of time and money, so that sociocultural and socioeconomic factors can be incorporated into our assessments.
Photo 44. The sustainable use of tree resources growing outside forests demands a light hand rather than severe pruning for Faidherbia albida, (© Sarlin/Cirad)
Assessments of the environmental services of trees and repercussions of tree management on the environment are still embryonic at this time. Many are still at the study stage and highly demanding in terms of resources and design (Annex 2).
Environmental benefits of trees can be indirectly assessed, however. Easily measurable and environmentally linked variables such as crop yields can be used to judge the impact of trees, e.g., the distance of trees from windbreaks, stream flow, and bird species diversity, but these are still highly localized data. Another possibility is recourse to scientific studies, but these may or may not be applicable to the specific case.
The impact of trees on the environment can only be felt over time because trees grow relatively slowly and ecosystems are resilient. The impact only usually becomes apparent in their absence, in the form of more wind or water erosion, scarcer wildlife or altered wildlife behaviour patterns, lower crop yields, the flight of the local inhabitants, a drop-off in tourism, and so forth.
Measuring the environmental impact of tree management thus remains a problem common to all natural resource planning or management operations. The scientific community should give greater consideration to the development of more easily implemented procedures.
It is important to know the status of Trees outside forests at any given moment in time, but it is even more essential for decision-makers, managers and planners to be able to trace patterns of change over time in the same area. Up to now, there have been two preferred approaches for trees growing in non-forested rural areas.
The first approach is to compare aerial photos taken at sufficiently long intervals in time, as in the case of Nepal and Kenya. Knowing how Trees outside forests have evolved over time will allow us to make a retrospective assessment of the changes involved, such as increase or decline of tree cover or the number of trees, their location, and the like. We can then infer the major trends for an area, and the reasons why the pattern has varied over time, combining the data obtained with historical data of a political, social and economic nature.
The second approach is based on surveys of villager/managers combined with field inventories. These may be retrospective surveys (Njenga et al., 1999; Aalbaek, 1999), showing how Trees outside forests have been managed, and pinpointing the influential factors and constraints in tree management. Retrospective studies may also be prospective studies, indicating possible future trends for Trees outside forests and shedding light on the social demand for this resource.
Other techniques habitually used in silviculture can be harnessed to appraise tree renewal (or its lack), for example, size structures such as diameter curve frequencies, obtainable from inventories.
The first step is to decide whether to establish periodic inventories on permanent plots to measure patterns of change in trees outside forests, as in done in forestry. To the best of our knowledge, some countries (e.g., France and the U.K.) have progressed to this approach, linking such studies to their permanent forest inventories. The challenges inherent in an operation of this sort are obvious, but we can be less sure whether the cost always justifies the effort. And we also need to ask how often assessments are needed. Many forest inventories are made at ten-year intervals. This is justified by the development of forest stands which are basically unaffected by such human interventions as logging, clearing, and the like. But as we have emphasized, the potential growth of Trees outside forests is heavily influenced by a whole spectrum of human activities which vary greatly at different times and places.
Assessing and inventorying trees in urban areas is the responsibility of the city highway or parks departments. Some very sophisticated tools are used for this purpose in the developed countries. Each tree is mapped and identifiable (even electronically monitored, as in the city of Paris), and examined several times a year. But in the large urban agglomerations of the developing countries, only selected green areas such as parks and public gardens are serviced by the city authorities.
A reasonably complete assessment of Trees outside forests requires, inter alia, geographical, ecological, biophysical, social and economic data. The problem lies in the relevant and simplified utilization of this mass of data.
The assessments reviewed here made no mention of how the data were stored and subsequently managed. Even though nowadays computers, software and skilled operators can easily and economically organize structure and store much of the data for later access in a user-friendly fashion, this dimension has to be tackled at the design stage.
Additionally, as we have seen, the demands of the many potential users of assessment data can also vary greatly. If user expectations are to be met, data assembly and data processing will need to consider this diversity of requests, mindful that the assessment data and their presentation will be used for analyses leading to planning and management operations, based on the preliminary inventory. These may be integrated or sectoral, and differentiated by location and size of the territory in question.
These considerations suggest it would be preferable to assemble the data in the form of maps or graphs. These modes of visualizing data offer an expressive and more easily readable way of complementing a series of tables.
Photo 45. Botanical garden in Burundi (Araucaria bidwilli). (© Bellefontaine/Cirad)
At least two types of software should be utilized for these data systems: a relational database management system (RDBMS), and a geographical information system (GIS), which can automatically exchange data, as can most professional software systems. This software system also allow many tasks to be automated (data acquisition and checking, statistical analyses and calculations, data output, requests and so forth), and easily updated or expanded with new data. Data can also be selected, extracted or combined. These options make it easy to decentralize or centralize certain data processing operations.
18 National Inventory of woodland and trees (NWS)
19 Main woodland survey (MWS)
20 Survey of small woods and trees (SSWT)
21 Forest Survey of India FSI.
22 Kerala Forest Research Institute (KFRI).
23 The Rapid Rural Appraisal is a systematic and semi-structured activity carried out in the field by a multidisciplinary team. The objective is to quickly and efficiently gather new data and frame new hypotheses on rural life and resources. The Participatory Rural Appraisal allows rural societies to assume an active role in problem analysis and project design.