Management involves a series of mechanisms put into practice by rural people, who, in many cases, are coordinating their actions with others, ideally under orders from some local authority they regard as legitimate. In many cases, management is conducted almost entirely by people with rules in their heads and without coercion. In other cases, it is as important to know why people obey rules as it is to know what the rules are.
In reviewing the African literature, to my surprise it became clear that we allow ourselves to fall into a trap if we talk about management of woodland as distinctly separate from management of trees on farmland. There are three reasons for this:
Thus, it seems to us a pity not to give some of the flavour of this process, since much of the information gleaned has useful implications for any proposed woodland management.
It is now standard practice to deplore the inability of Third World governments, and colonial regimes before them, to distinguish between common property resources (CPRs) and open access land. We would be unwise to make the same mistake by lifting the woodland managed by local people from the matrix in which it exists, the primary economic activities of farming or herding, and treating it like a forest reserve. By and large, the tenure and authority regimes which once governed the successful use of forest, are also those that govern the use of farmland and all other local resources. It is divorced management of the two resources, by different ministries, by local and non-local people, which has led to many present day problems.
The starting point for this study, then, is that ownership and management go together and that, in many cases, farmland or rangeland and woodland cannot be separated. Only by understanding tenure fully will we understand the conditions for successful management.
1. The hybrid term "swidden-fallow" has been used for two reasons. First, it links two terms which, rather arbitrarily, are normally used respectively for non-African and African situations.
Second, it stresses (in a way that terms such as "shifting cultivation" and "slash-and-burn agriculture" do not) the fact that farmers do not abandon cleared land when they leave it to restore its fertility and plant crops on new land. They obtain other products from it, may retain continuing control over trees growing upon it and will return to it when their own particular cycle is complete.
Management of natural woodland is practised by those to whom it belongs, and as has been seen so clearly in tree planting, no serious investment of time and effort will be made unless the resource is owned. Thus, the mechanism for ownership must be our initial focus.
The herding lineage
In many of Africa's herding groups, the genealogy of the lineage is the charter for access to land. All the male descendants of one ancestor share one large area but subgroups of more closely related kin intensively share smaller portions for certain purposes. Thus, any one piece of land is shared by small numbers of owners with strong claims and large numbers of owners with weak claims. Cases such as the Libyan Bedouin , the Baggara of the Sudan  and the Turkana of northern Kenya [7, 8, 9] demonstrate the flexibility of this system. The Libyan and Turkana cases, in particular, show how tenure is related to predictability of rainfall or access to perennial water. Access narrows, however, when a perennial asset is at stake and broadens where low and erratic rainfall renders uncertain the location of the best grazing from year to year.
In addition, of course, agreements have to be made about the boundaries between adjacent tribes or clans. These were created in the past by political and military means and have always fluctuated to a degree . In Somalia's Bay Region, elders explicitly linked the right of access to grazing land to the duty to fight for it when need arose .
This "nesting" model for lineage territory, with land belonging to tribe/lineage/lineage-section/immediate kin group but with the focus dependent upon context, is the most flexible when it applies to fully mobile pastoralists. Although, as an ideology, it is also found in modified form among semi-sedentary groups, for example, northern Kenyan Samburu on group ranches  and, also among agriculturalists such as the matrilineal Aouan of Cote d'Ivoire  or the gum farmers of Kordofan, Sudan .
In the case of mobile pastoralists, the actual extent of the range owned by the lineage will vary over time, and individuals with their herds will find themselves in different parts of the current range at different times of the year, or in different years. There is no necessary relationship between individuals and specific pieces of land as in the case of farmers. The membership of the lineage is enough, and after that flexibility is considerable.
Sedentary kinship groups
For sedentary farmers, while a knowledge of genealogical links is important, the "moral community" which holds land is defined by both descent and residence, rather than descent alone. Very often this is expressed as the chief holding the land on behalf of the collectivity and the lineage members who live in his area drawing usufruct rights through him. This is the case in the current collection of summaries for the Unyamwezi of Tanzania [I], areas of southern Ghana [65,70], the Kakamega district in western Kenya  and the Peul of the inner Niger delta in Mali .
Alternatively, we are told that the clan is the descent and residence group, and often this means that no single chief allocates rights but that he does so with groups of elders. This seems to be the case for the Gush in Western Kenya , the Somali Bay Region , in eastern Botswana [57, 88, 96] and in various parts of Niger .
The village itself is at times the focus of descent and residence rights, with the village headman (or sheikh) making land allocation decisions over which the clan chief, a superior in descent terms but a non-resident, has no control. We find this among the Nyakyusa , the Sukuma [64, 94] of Tanzania and in the northern Sudan [45, 48].
Finally, there is mention of village management of resources without clear specification regarding if and how descent and residence rules work. The Dogon of Mali operate a management system run by representatives of village quarters  and in Okwe, southern Nigeria, adjacent villages manage aspects of their swidden-fallows as a collectivity .
The household head and the household
Just as the lineage fits within the tribe, and the village is often a subset of the lineage defined by residence, so the household is often seen merely as the lineage writ small. This is clear from the fact that land would revert to the next collectivity upon the owner's death, rather than to his children. There are examples here from the Sudan , Ghana  and among the matrilineal Luguru of Tanzania .
In some systems in the past, the household head had to accept that co-villagers had some rights to self-seeded trees on his land but, like the larger groups of which he was also a part, he could limit the rights of outsiders and control the access of different categories of insiders .
However, the household head's position is continuously gaining in importance as other levels of the kinship system cease to have political meaning and as land registration becomes the norm [34, 35, 73, 109]. In the future, user groups for particular natural resource assets are more likely to be aggregated upwards from the household, rather than from a sub-section of a larger collectivity.
The right to manage: actions undertaken by rightholders
For chiefs and lineage elders of herding lineages, the most common management actions undertaken are likely to be the exclusion of outsiders [95, 93], adjudication between insiders and the promulgation of new rules. Outsiders need to ask permission before bringing animals to graze within the lineage area , or they should stay away if, as do charcoal burners, they have competing designs on the trees . Samburu elders , for example, decided to allow poor local men temporary permits to make charcoal but banned charcoal making by outsiders. Turkana elders, in consultation with project staff, issued new instructions about the lopping of Acacia tortilis, so that only side branches were taken and the leader shoots had a chance to grow rapidly above goat-browsing height .
Chiefs, elders and village leaders seem to have attempted a wide variety of management initiatives. Many areas with strong chiefship or even kingship institutions have attempted to preserve the most valuable tree species in the area and have linked them to the very office they held. In some cases, the effort was linked to an altruistic attempt to encourage farmers to keep alive the species they most needed on their swiddenfallows, in others a personal perk seems to have been the driving force behind the edict. Examples of the former include the protection of Faidherbia albida [25,43] in Niger and in the Sahel, in general , and the protection of Acacia senegal for gum arabic in the Sudan . An example of rulers seeking to extract personal benefit from control of species associated with the chiefship occurs in northern Ghana, where both the Parkia clappertonia and Butyrospermum parkii produce a commercially valuable product [71 ].
In southeastern Botswana, village chiefs banned the felling of village amenity trees (for shade, and lavatory shelter for those too elderly to go out to the bush [86, 88]) and arranged elaborate zoning for different categories of fuelwood collectors . Around Mount Kenya, and widely elsewhere in Africa, sacred groves were used as the meeting places or burial grounds of chiefs or the members of senior age sets, and, to the extent that these groves symbolized the patrilineage, women were banned from entering them. In the Cote d'Ivoire , elaborate arrangements for protecting and resting the forest were originally in force. Punishments for forest misuse, too, are often graded so that villagers are more likely to pay lower fines for local crimes than are outsiders [ 102].
At the individual household level, the lineal household head controls access to the trees on his land. He makes rules for his wife and neighbours, who are both categories of insider-outsiders, as to the species that may be cut such as for the Kamba in Kenya . Just as lineage heads do for would-be immigrants and clients, he also makes the rules for the thickness of branches that may be cut (e.g. for a Wolof village in Senegal [801), for the volume that may be cut (e.g. for the Kenyan Maasai ) or, indeed, whether any cutting will be allowed at all .
Rights of secondary users
What we see in all these cases is a system perhaps best described as a series of background, and foreground rights. In the background, are broad rights in forest and farmland with the potential for crisper definition, while in the foreground are firm rights, narrowed down to a smaller number of right-holders, which have been won by deliberate action. Clan and lineage mechanisms manage large areas lightly [3, 5], while smaller groups consolidate their rights by specific action.
At the same time, secondary rights remain very important. Among pastoralists, for instance, one of the risk-spreading actions has always been to gain secondary in-law rights to other areas by marrying daughters to remote lineage sections or to a man from another lineage entirely. Limited rights can be a great deal better than no rights at all, as is clear from the example from Burkina Faso  in which herders are allowed to graze on farmers' fields after harvest but not, even though the farmers themselves do, to lop browse from farm trees for their own animals.
In Senegal , all co-fallowing villagers may take firewood from one another's land in the dry season, only wives may collect from their husbands' fields in the wet season, and only the farmer himself can cut larger branches, no matter what the season.
Several of the summaries make it clear that it is the investment of labour which creates ownership. In the case of land, for the agriculturalist this means being the first to clear and plant land once under forest or woodland [ 10, 14, 37, 65]. In all cases, until present-day land registration procedures, if it was abandoned, such cultivated land reverted not to the wild but to the group to which the clearer belonged . In this way, the individual had created rights for others as well as himself. The creation of pastoralist tenure, implicit in the Turkana examples [7, 8, 9], takes place in the dry season and is manifest in the successful defence of the key dry season assets of grazing browse and water. Here, labour investment may mean simply defence, although in some areas the annual restoration of wells or clay lined reservoirs  is a key tenure-defining piece of labour.
Labour creates ownership in various ways. Among the Tanzanian Sukuma , banging pegs into a tree suitable for hanging honey barrels ensures exclusive access to the tree. In the Samburu forests in Northern Kenya, cut and stacked piles of firewood will not be touched by other women .
It is clear that tenure must be worked at in some of these cases and that, as a general rule, individual tenure comes from the greatest, and the most constant, labour investment. In the Senegal case , for example, land was lost to the owner if it remained uncultivated for more than ten years. Similar potential loss of tenure of dry season grazing is reported from the Turkana . In the case of the Cyrenaican Bedouin [ 12], investment in water leads to greatly strengthened tenure.
Tree planting, because it is more work than tree use, also creates tenure. Planters of valuable or exotic trees  strengthen their rights to the land on which the trees are planted mainly because they are replacing seasonal with perennial crops.
It is important to note, however, that the investment of labour must come from its owner's own free will. Neither slaves [ 101 ] nor tenants  can easily create ownership for themselves through their labour, since it has already been made clear in their case that their labour belongs to another within the terms of their status or contract with the landowner.
Management, in the sense in which it is used here, means some individual or group activity which organizes the utilization of tree resources so that the resource is more equitably shared, more likely to remain into the future (more sustainably used) or so that it will grow better than if the management practice was not taking place.
Classical planned natural woodland management, of the kind practised in European forests in the past, is somewhat different, for two reasons:
Systems described here will show that planned actions which encourage some species and eliminate others, or which encourage trees to produce different end products - e.g. thin poles or thick timber - are by no means uncommon. However, many of them take place in the context of farming, rather than dedicated forest.
What they are doing is devising rules for sparing certain species or certain size classes, or simply saying who may and who may not use certain tracts of woodland. They may mean management of the individual tree but more often they mean management of the space which the trees occupy, along with the grass and water there.
Here, all and any planned and deliberate activity which enhances the quantity and quality of woodland or makes its use more sustainable is defined as management.
Swidden-fallowing has been the most important method of woodland management in Africa for hundreds of years. The selection of summaries here, while giving some flavour of the variety of patterns and types of fallow to be found on the continent, is only a minute fraction of a vast literature on the subject. Some order can be made, partly by focusing on the types of swidden-fallow associated with particular tree species, and by attempting to identify the management procedures which governed the progression of the fallows.
One of the most useful documents in this regard is by J.P. Raison . Among others, it deals with the Sahelian and Sudanian zones of Africa, where the inhabitants practise both agriculture and livestock raising, and where two particular swidden-fallowing systems predominate. In these zones, settlements create a variety of "parklands" placed in concentric rings around the village, which are constantly evolving. Leaving aside the permanently cultivated village home gardens, we find two zones before the unadapted bush is reached:
the zone of permanent fields set with Faidherbia albida, and
the zone of fields cleared in the bush, and of long swidden-fallows, where Butyrospermum parkii (Vitellaria paradoxa or karit), Parkia biglobosa and Ficus platyphylla are found.
The following three descriptions of these zones summarize Raison's insights.
Bush Swidden-Fallow (Long Fallow) Butyrospermum parkii Parkland [14, 28, 29, 30, 31, 38, 49, 54, 68, 71, 74, 82, 84]. This formation is found throughout the Sudanian zone, apart from Senegal, and is associated with long swidden-fallows. In the untouched wooded savannah, with initially perhaps 1,000 trees per ha (diameter range 5-60 cm), all but 100 or so are burned so that they die. Butyrospermum parkii is the main species preserved. Over the next four years, much of the dead wood is used up as fuelwood, as the area is cultivated, then the whole patch is left to rest for 20 years while the farmer moves on outwards from his village. If one area is re-used by one village for 2-3 successive cycles, i.e. 40-60 years, with the B. parkii protected all that time, after 60 years the shade will be so great that agriculture will suffer. So the whole village will move, using the earlier area for fruit gathering, and the process begins again. After two or three moves - 150 to 200 years - the B. parkii will be very old and it will be time to start the whole process of major land clearance again, preserving younger trees.
In the documents identified in this survey, references to the manner of fallow management at village or group level are scanty. It is implied, however, that in Niger, land management systems were organized by the village chief, elders and household heads [31 ]. What is impressive in the Butyrospermum parkii cycle is the lengths of time involved, the coordinated nature of village planting and protection and the extent to which the landscape in these regions is manmade.
Savannah Swidden -Fallow (Short Fallow) Faidherbia albida Parkland [2, 9, 10, 14, 25, 29, 30, 38, 43, 49, 51, 54, 61, 68, 74, 75, 80, 82, 84, 87, 89, 91, 99, 104, 105]. This is the most interesting and developed of the types of parkland created by Sahelian farmers. Because the trees are in leaf in the dry season and leafless in the rains, permanent cultivation under them is possible. At a density of 10-30 per ha, they fertilize up to 50 percent of the area. Crop rotation is practised under the Faidherbia albida trees.
However, it is clear that F. albida alone cannot restore soil fertility - animal dung is also needed. A hectare of F. albida parkland with 20 trees on it supports six cattle in the dry season which is enough to keep the area fertile. In the rainy season, however, the trees are leafless, hence there are no agricultural residues available for fodder, and the bushland is then an essential complement. If bushland is nearby, livestock can browse there and be tethered on currently exploited fields at night to deposit their manure. If there is no bush nearby, more labour intensive solutions must be adopted, such as sending animals north for part of the year or growing forage for them. But more recently, with the departure of the young people as labour migrants, animal herding has become too labour intensive an activity. The result is often that F. albida parkland fails and reverts to less intensively managed Butyrospermum parkii parkland.
There are equally few clues to the style of management of these parklands, although several writers have noted the symbiosis between cattle and trees - F. albida - making it possible to keep stock on the land in the dry season, resulting in the additional fertility of the field from manure [61, 75].
Gum Gardens Acacia senegal [31, 45, 46, 82, 84, 87, 89, 90, 99, 1091. In the case of gum arabic, we lack Raison's clear descriptions, and, therefore, our understanding of the swidden-fallowing system is patchier. It seems that in the Sudan, each household was allocated by the village sheikh three or four plots of land, each of which was farmed for three to six years on a rotational basis. Acacia senegal naturally regenerated in cleared plots and was then protected as it grew up. Very occasionally, Acacia senegal seeds were sown. As the trees came to maturity on the fallows they were tapped for gum by the respective fallow-owners [45, 46]. Finally, when the trees were old, they were felled and sold for charcoal. It is not clear from descriptions whether the circular swidden-fallowing pattern around settlements was the norm or whether farmers' fallows and cultivation were arbitrarily placed. Centrally allocated gum plots would suggest the former. The system is now collapsing partly because population densities are now too high to sustain such large amounts of land per household.
Private fallow systems
Much fallowing in semi-arid Africa would seem to have involved village-wide agreements about how the surrounding land was to be used, not least so that the whole settlement could ultimately decamp together. A few documents stress the private nature of particular swidden-fallowing systems. In the Bougage strip rotation system in Niger , where the household controlled 17 strips, at any time, 12 were being fallowed, four in their first, second, third or fourth year of cultivation and one as a house site and cattle pen. While it is possible that some groups practised long swidden-fallowing in a more "private" way than others, it is much more likely that different writers have focused on different things. While fallowed plots were most likely to be returned to by the household who had cleared them in the first place, the pattern of each household's fallows would also have to fit with village-wide patterns.
The conservative use of resources sounds like an activity any individual would undertake without even thinking about it. However, when we find such attempts in large areas of woodland, they not only let us know that we are in the presence of a common property resource but also that some kind of authority in the background is passing on the rules of good resource use and checking to make sure that they are obeyed.
In the literature, we find four instances noted:
Most settlement areas in semi-arid Africa began with a surplus of trees, so that the people who lived there had the luxury of taking out unwanted species and concentrating on the preservation and enhancement of others [ 109]. In the past, the result has been a woodland composition strongly affected by human selection . We have already seen this process at work in the section on fallowing [Sect. 3.1]. Other species which have clearly been promoted in this way deserve mention, as well.
The literature in which the fruit Parkia biglobosa is mentioned [28, 29, 30, 31, 38, 49, 54, 68, 74, 84, 89,104) indicates that it was enthusiastically preserved in the same swidden-fallowing systems as Butyrospermum parkii. We also find that Adansonia digitata [ 14, 29, 30, 31, 38, 49, 54, 74, 75, 80, 84, 89, 91, 99, 104) is frequently promoted, planted and saved, above all around villages, where it is valued for its edible spinach-like leaves, its ascorbic fruit, a rope which can be made from its bark and, as in the Sudan, the water storage potential of its hollowed-out trunk. It is also a good honey-barrel tree because it is not spiny to climb. For archaeologists the clumps of this species frequently found standing alone in the savannah are memorial indicators of long abandoned villages .
Among many other valued trees, Borassus aethiopum [39, 74, 75, 89, 91, 92] is valued for basket making. And Balanites aegyptiaca [8, 10, 38, 46, 68, 73, 74, 80, 82, 84, 89] is important for fodder, dry season fruit, as a leafy vegetable and for construction, tools and utensils. It is occasionally even regarded as sacred .
Valued trees are cut and pollarded by pastoralists so that they live to be cut again [6, 8]. Species that tolerate and even thrive on lopping are encouraged, such as Balanites aegyptiaca among the Pokot in Kenya , Combretum nigricans in Guesselbodi, Niger  and many more.
There is much evidence of the attraction of indigenous fruit trees in the literature, from those preserved in Zimbabwe  to the Sahelian-Sudanian species, such as Tamarindus indica and Sclerocarya birrea, as well as the others mentioned here. From Senegal to the Ethiopian Rift, and extending southwards as well, we also find similar tree species preserved among crops and similar species left and planted to enrich the woodlands [50, 74].
As usual, there is far more evidence in the literature of what has been preserved, than there is of how it was preserved. As far as group management is concerned, we have to reconstruct a picture from the few good accounts we have. Brokensha and Castro  have an excellent description of an island of forest managed by the elders in Mwea at Mount Kenya on behalf of rightholders who had to get permission to fell timber trees in it. We also read of effective village management of nearby woodland in the northern Sudan  and in northern Ghana [71 ], and of the protection of trees on farmland, even in the dry season, in Burkina Faso . There is more data on individual management of trees on farmland or fallows but we know that such action is easier to organize [38, 73, 91].
Considering the thesis of this study, that trees are not set aside in indigenous management systems but are lived among and used so much that it is hard to say where woodland ends and farmland begins, it is surprising that reservation exists at all. Yet it does. From many accounts it is difficult to determine whether trees are actually being reserved for soil conservation or watershed management reasons and whether the ban is merely wrapped in religious language to give it force, or whether the preservation of the trees has nothing to do with "rationality" of this kind. It is certainly worth noting that in all Bantu African languages there is a unique noun prefix for all trees, gods and spirits and that many tree species are thought to house spirits. The summaries in  and  make it clear, too, that such veneration goes beyond Bantu boundaries.
Some sacred groves acquire their importance from the fact that ancestors' graves are clustered in them, and economically or socially important species are clustered there as grave markers. It is customary to put such clusters of graves on a hilltop or ridge in some cultures - hence, the trees may have an inadvertent conservation effect, as well. Among the Kikuyu, such sites were commonly adorned with Ficus natalensis trees [22, 58]. Among the Lozi of Malawi, a small grove was planted at a king's grave with seedlings transplanted from the bush [41 ]. Some sacred groves clearly symbolize the patrilineage, and women may not enter those groves . Some huge trees with their many large and smaller branches may also symbolize patrilineage .
At the practical rather than the sacred level, there are a few instances of reservation for patently economic reasons. The grazing and browse reserves of the Pokot and Turkana are one such example from primarily pastoral areas [8, 9, 60], while the hilltop grazing and woodfuel reserves of the Sukuma in Tanzania provide a more agricultural example .
Finally, in the past, some groups planted and guarded particular tree species as a reserve against famine [54, 84].
In both agricultural and pastoral areas, it is commonplace to say that there are great differences between the dry season and the rainy season. From the point of view of tree management, trees in swidden-fallows and on fields are protected while there is a growing crop but lie open, as does the land itself, to all comers in the dry season. For pastoralists, there is a plethora of browse in the rainy season but the retention of browse sources for the dry season is of paramount importance.
The agricultural point is made by the description, for Senegal , of women's exclusive access to minor forest produce from their husbands' fields in the rainy season and their competitive access, with all other villagers, in the dry season. In Botswana, "to prevent hail", which might damage the crops, the chiefs used to put a ban on the cutting of certain unidentified tree species from land preparation to harvest [86, 88]. When the harvest was in, and the slack, house-repair season about to begin, the chief would "open" the tree cutting season again by making the first cut and insisting on repair to the kgotla, the village meeting place, before private repairs were begun. We also read of open and closed cutting seasons in other areas, for example, Marsabit in north east Kenya  and the management of common property resource grass destined for thatching in Mwanza, Tanzania  by the setting of a cutting date.
For pastoralists, the earmarking or enclosure of dry season fodder is the key response to seasonal vulnerability such as for the Turkana [7, 8, 9, 60] and for the Baggara of the Sudan . Many pastoralists also leave certain categories of fodder until last, usually that near their home wells where they will return to wait out the last of the dry season, or that high up in trees, inaccessible to animals but which can be lopped for them as a fall-back resource .
Other seasonal uses recorded are the utilization of wild plants for human food during the agricultural growing season , and the prescribing of rest days for gathering in the forest so that the forest may have time to recuperate .
Much woodland management is, of course, management by area, and the rules which define what the area is, and who the legitimate insiders and outsiders are, have been analysed in some detail [Sect. 3.2]. However, since there the emphasis was on ownership, and here we focus on management tools, the defined area is mentioned again here as one of those tools.
Religious sanctions in the African context are a way of connecting elders or rulers to God and the ancestors and using their authority to strengthen the authority of the living individual: of wrapping up management rules, in short. There are examples in this collection of the following sorts of sanctions:
While we cannot always understand these taboos, partly because of the often incredulous or superior way in which they are reported, and partly because of the way in which the original informant explained the information, perhaps, it is clear that they were usually obeyed and that they had meaning for those who obeyed them. We should assume, therefore, that such sanctions are rooted in historical observation and in a good understanding of the local ecosystem.
For instance, much to the irritation of community forestry projects trying to get started, it is not infrequently reported in some parts of Africa that villagers will say (as they did around Tamale in northern Ghana [71 ]) that anyone who plants trees will die. Or (as they did in southeast Botswana ) that only God may plant trees and that it is impious for humans to do so. As bald statements, these can be dismissed as mere conservatism or as proof that more publicized information on the value of trees is called for. Closer investigation reveals in each case that the proposed tree planting was on village CPR land, that the trees would diminish the CPR rights of others, and that it is other humans, not God, who may be wronged in due course. The speaker is simply telling the project, in the strongest language he can use, that elaborate arrangements are under threat.
Fire is a very important management tool [99, 111] one on which there has been much conflict in the past between European foresters working in Africa and local people. Unfortunately, the set of references gathered together here does not do justice to its importance, and it is probable that there is a need for further new field research on indigenous knowledge of management by fire. Two technical articles have been included to indicate what some of the conflict areas have been.
Fire is plainly both a cleanser and enricher of the earth and a potent danger to savannah woodland. It is an essential part of agriculture, as well. The evidence would seem to be, however, that the Sahelian (and miombo) farmer has had a sophisticated understanding of the effect of fire on the tree and grass species he needed both for the presence of animals, and for bush fallow cultivation.
The negative tone of much professional comment on Sahelian herders' fire management practices can be read in  where fire-risky lopping is reported. We do not know whether the observer misunderstood what he saw, or whether indeed this particular herdsman was off his home range and treating the property of others with scant regard, as is certainly common . Yet, elsewhere, we hear of farmers planting firebreaks of bamboo in their agricultural system  and of a most efficient system for the policing of careless fire users . Here, on the left bank of the Niger river, the Fulani would put out bush fires they came across and, once the fire was under control, identify the culprit(s) and impose a fine: the provision of a feast for all the men who had put out the fire.
Two fire control experiments are reported on. One was conducted in savannah woodland close to, and derived from, closed forest in Nigeria, over a 28-year period from the 1930s to the 1960s, and with plots which had been treated with different management methods . After 28 years:
In the second, experiments were conducted in Anogeissus-Combretum forest in Nigeria (600-900 mm of rainfall), on a variety of combinations of protection and burning regimes, to discover which gave the greatest mean annual increment and which the greatest number of large stems per hectare : early burning every fourth year and fire protection the rest of the time, early burning and fire protection every alternating year, fire protection alone and or early or late burning annually. In order to obtain a large number of large stems per hectare (50 cm and over), protection without burning was the best regime, even though some mean annual increment (MAI) was sacrificed.
These experiments suggest two things:
The very range of vegetation to be found in the semi-arid areas of Africa is, to an unknown extent, the result of the planned and unplanned impact of grazing and browsing animals. The mean annual increment of woody vegetation can be increased by the reduction of herbaceous competition offered by grazing animals, so long as there is enough rain for it to grow at all . Grazing also keeps the balance between annual and perennial grasses, and between grassland and bush, while reducing the incidence of inopportune fires .
Africa's pastoralists are skilled at managing a broadly spread, light exploitation of plant cover by keeping a wide range of different stock with different grazing and browsing needs .
As is evident from the discussion in the introduction, short fallow systems are fundamentally dependent on animals and their manure, while Faidherbia albida enables the farmer to keep at least some of his herd on the land through the dry season [75, 80].
Mariam Niamir, in her survey , noted few published details of management of individual trees. J.P. Raison, while describing the pollarding and removal of side branches practised on trees on terraces among the Mafa of Mandara, Cameroon , also comments on the lack of single tree management examples in the literature survey he conducted. Our survey, however, has turned up a small group of examples:
Lopping and pruning
The Kamba in Kenya, lop side branches from Grevillea robusta and Eucalyptus spp. in order to produce a central pole for construction timber in the long run and some fuelwood in the short run . Many Africans are aware of the species that respond best to lopping and will readily sprout again, and plant and preserve them where they can .
A case in Sierra Leone is cited by Hoskins  of trees being cut to different heights when a fallowed field was being cleared in order to favour selective re-growth. In Faidherbia albida fallows, all of the F. albida trees are lopped on a rotational basis every 3-5 years and fed to livestock, while the wood is used for fencing and fuel. The trees are thus kept to a manageable size in the field . Ostberg presents an interesting, if puzzling, management example . The author notes thorn-fenced fields containing heavily lopped trees, and fields surrounded by living hedges, which are thus more permanent, in which the trees are carefully and lightly lopped. Tenure difference would seem to be at work.
In Ghana, trees on swidden-fallows are pollarded so that, when they sprout again, the shoots will be above grazing height . Pollarding is noted among the Mbeere in Kenya by Castro and Brokensha . Poulsen  writes similarly of the unusual variant on pollarding practised on Mount Kenya where Grevillea robusta is heavily pruned every two or three years for fuelwood and small timber, while a larger pole can be grown centrally. The trees can be pruned for 30-50 years.
The Turkana climb high into fodder trees to pollard them when selecting dry season fodder . And this practice is widely reported by ILCA  for the sub-humid zone just before the rains when grass is no longer available.
In some silvopastoral systems, trees are cut to stump level for purposes such as timber, fuel, browse, and then left untouched for several years in a swidden-fallow period .
The tools for the job
The usual tool for all of these actions is a bush knife (panga) or, in some cases, a small axe. Only commercial fuelwood traders use heavier tools, such as long-handled axes or chainsaws .
Many of these examples seem to show management of the individual tree to diversify its products and its usefulness and to make it work harder for its keep, as it were. More intensive tree management is found mainly on farms, presumably because while an ample bush supply still exists, diverse products will be met from diverse trees rather than by management of one species alone.
One of the facts that emerges from the literature is the tremendous paucity of formal forester knowledge about the management of tropical dry forest, although enormous changes were wrought upon them in the name of better management and state control.
Recent survey articles hammer the first point home. According to Bonkoungou and Catinot , there is only very slight experience of silvopastoral management based on natural regeneration techniques for mixed forest and grassland. Much indigenous knowledge waits to be collected. According to Jackson, who has done the most conclusive survey of the Sahelian literature on woodland management [54, 55], there were only two examples, up to 1983, of formal forest management in the Sahel: that of Bandin forest, Senegal, which failed in its aims, and the relatively successful management of Acacia nilotica on the Blue Nile in the Sudan. Other projects failed because they neglected to build upon the much longer technical experience of local people (e.g. a windbreak project in Cameroon ).
Set against this thin knowledge, we find the imposition of European concepts of property and land tenure, with disastrous effect. The most important gap was the failure to understand swidden-fallowing systems, under which the landscape was sustainably used for hundreds of years. The systems were simply invisible to outsiders , and land being fallowed often looked abandoned and ownerless to the northern forester's eye. As a result, fallows, rested for years, well-wooded and almost ready to be felled again, were gazetted by the state and converted into forest reserves (forets classees); thus, at a stroke, intensively managed CPRs were turned into open access land, as far as villagers were concerned, to be poached from rather than managed [49,110]. People became wary of fallowing their land [ 103] and instead overworked it. In the literature under review, instances of state appropriation of CPRs and fallows in this way occur in no less than eighteen cases [5, 12, 16, 27, 31, 32, 40, 49, 62, 80, 90, 93, 95, 97, 98, 103, 105, 110].
There is a further confusion. On the one hand, in the literature which deals with francophone west Africa [e.g. 99], land tenure and tree tenure are often described as if they are two distinct entities and have been so since pre-colonial times. On the other hand, writers in the anglophone literature appear to see the two as always going together [e.g. 11, 12, 35, 86], even while it is made clear that people do have limited usufruct rights in trees on each others' land in the dry season or while it is being fallowed. The apparent difference is resolved if we remember that land farmed by individual farmers was itself only "borrowed" from the lineage to which, when not under annual crops, it returned, along with any trees growing on it. Even the association of individual valuable tree species with the local ruler was [discussed in "Owners of natural woodland and other land with trees on it"] no more than an expression of this communal tenure.
In our opinion, and that of Norton , however, the separation of tree and land tenure became a reality in some countries only in the colonial period - as the result of poor understanding of fallowing rules. Indirect support is lent for this view by Thomson's work on Niger , and by Seif el Din's work on the Western Sudan . The assumption of a historical split between land and tree tenure has been used to reserve certain tree species for the state to manage, to the ridiculous point where it is as impossible to fell a Faidherbia albida tree today in Senegal  as it was under the ferocious Sultan Tanimoun in 19th-century Niger, who also seemed to misunderstand the nature of fallowing systems . Thus, State management has not only lacked insight, it has eroded working systems and concepts to a serious degree.
Tenure and land use changes
Villagers have been turned from landowners into leaseholders in many countries (e.g. Senegal  and Sudan ), and from landowners to landless wanderers in the case of pastoralists. Barrow  points out that customary tenure was taken into account in Kenya when agricultural land was demarcated but it was neither investigated nor recorded in the dry areas. The best solution offered to Kenyan pastoralists has been group ranches, which lack the area flexibility of lineage range management  and begin to look like some of Africa's other tenure ghettos, such as the communal lands in Zimbabwe .
Political changes affecting land tenure usually destroy CPRs. For instance, during the Ujamaa period in Tanzania, the Sukuma were asked to leave their dispersed granite outcrops and to cluster in villages. Unoccupied, these areas fell prey to urban charcoal burners .
Land use changes relating to economic change
The introduction of the groundnut has had a negative effect on the value of oil-bearing tree species, such as Butyrospermum parkii, and the systems into which they used to be set . It has also led to excessive clearing of forests in many areas . On the other hand, it is a crop which fits well into curtailed swidden-fallow cycles.
The loss of authority of the elders
In Turkana, the authority of the clan elders, who were originally solely responsible for livestock and natural resource management, is being eroded by modern education for the young and by the presence of government officials in the area. Among the Kikuyu, as long ago as the 1930s, the weakened position of the elders made the institution of sacred groves no longer tenable .
Better health conditions and food security for humans and the increased use of veterinary drugs has increased both human and animal pressure on rangeland and farmland. More people are living in fixed, rather than semi-mobile, conditions, with a corresponding heavy intensification of land use and biomass off-take. These increases have not been incorporated into land management strategies even though the range in dry areas was already used at close to its subsistence potential [6, 103].
Many parts of Africa have experienced bigger influxes of people than they can easily absorb because of those moving southwards away from drought areas , because of refugee problems  and because of the artificial withdrawal of large tracts of land, as in the mechanized farming schemes of the Central Sudan [ 111 ].
As populations increase, there is also an increased sedentarization , the household grows in importance, while the clan and lineage structures shrivel , and grazing is privatized [ 11 ] or shared by a much smaller subsection of a larger entity . All such actions sound the death knell of successful woodland management.
It is true that rising population pressure on forests will cause deforestation. For areas with sufficient rainfall, the inevitable long-term result of population pressure is, in fact, a move to tree planting. In Rwanda , a survey revealed that people plant trees most where there is the most pressure on land and much cash-cropping. Male farmers who got fuelwood from their own land had planted more than those who collected dead wood or whose wives got the fuelwood.
The shortening of fallows
Population pressure seems to have led to the shortening of fallow periods throughout semi-arid Africa [2, 16, 28, 89, 90, 109]. In the Western Sudan the whole cycle has more than halved in length, and the fallow period is less than a third of what it was . The tendency is for the fallowing system to shrink in the end to the point where it is replaced by a crop rotation of alternating millet and groundnuts .
The concentric farmed parklands are under other kinds of pressure too. If they cannot be abandoned at the end of their cycle because of overall population densities, they degrade, and tree stocks fall to perhaps only five trees per ha [80, 84]. Labour migration from the Sahel has encouraged extensification of agricultural techniques, for, although the population overall has grown, more men and more young people have left the area forever. Instead, people are trying to cultivate the most land they can in the least time, at the very beginning of the rainy season. Each household head now tries to spread his bets by sowing over as wide and varied an area as possible. The orderly management of land practised in the past has tended to fall apart, and more and more marginal lands are farmed, with accompanying destruction of bush areas. Older practices, such as manuring, intensive sowing and weeding, planned swidden-fallowing and water conservation, have all been replaced by quick easy farming .
The growth of towns
The growth of towns is one of the largest threats to the dry tropical forests, because the clustering of large numbers of people who will need rural biomass for fuel and the low MAI of that rural biomass mean that the fuelwood shadow from a large town or city spreads hundreds of km out into already hard-pressed rural areas. Yet population growth in the Sahel is of the order of 2 percent or less in rural areas and 6-8-10 percent in urban areas. In addition, urban entrepreneurs are keeping cattle for cash near towns, where before they would have travelled widely with them [15, 16, 29, 30, 46, 59].
Roads are built from towns to rural areas, and these then facilitate more commercial transactions, such as the sale of charcoal or wood-cured tobacco to towns . Labour migration to towns leaves rural economies unable to farm properly and leads to the extensification noted above. Furthermore, the World Bank now believes that urban-biased economic policies bear responsibility for further extensive agricultural production and the shortening of fallows, because of low producer prices .
Thus towns grow at the expense of rural people, shrinking their resource base and extracting huge quantities of precious biomass from them . Urban people can make rural resources come to them, in effect, and can even be absentee herders, enjoying the wealth animals bring without having to take care of the resource the animals live from in the way true pastoralists do .
The state itself, as an urban-based institution, has exploited but failed to understand, the dynamics of rural systems, has seen its own urban needs as paramount, and, in some cases, has made it harder for rural people to build a surplus of animals, grain, because of labour shortages, or fallowing land. The result has been much increased vulnerability to cyclical droughts, let alone the more substantial dry phase we are witnessing at the moment.
Finally, and beyond human control, it is clear that the last twenty years have been exceptionally dry in semi-arid Africa, in contrast to the period which immediately preceded it. Up to 1968, livestock raising increased steadily and natural bush regeneration was abundant. Since 1968, however, natural regeneration has not been favourable and woodlands are changing their species composition, and ageing. In some places, as rainfall decreases, we see species changes, such as Acacia tortilis taking over from Acacia laeta and Acacia senegal ; in others, Acacia tortilis and Commiphora africana have been dying, woodlands have opened up and erosion has set in. Also, by the 1970s, gum stands from the 1940s and 1950s had almost disappeared and no regeneration was present .
Ground water deficits have stopped tree species like the Butyrospermum parkii and Parkia biglobosa from regenerating [ 15] and have caused the Niger River to drop by 50-60 cm, with immense recharge deficits as a result .
Consequently, a forest like that of Rawashda in the Eastern Sudan , originally reserved for urban fuelwood needs, now would actually need to be much larger to satisfy the same needs as before, because of reduced MAI. However, the whole area is already under much more pressure than it was when it was reserved.