Six rules for the rapid restoration of degraded lands: Synthesis of 17 case studies in tropical and Mediterranean climates

ARTICLE

sec.2011.0300

Auteur(s) : Eric Roose¹ eric.roose@ird.fr, Ronald Bellefontaine² ronald.bellefontaine@cirad.fr, Marjolein Visser³ marjolein.visser@ulb.ac.be

¹ IRD UMR 210: ECO & SOL BP 64501 34394 Montpellier cedex 5 France

² Cirad UPR Génétique forestière 34398 Montpellier France

³ Université libre de Bruxelles École interfacultaire des bioingénieurs CP 169 50, avenue F. Roosevelt 1050 Bruxelles Belgique

Tirés à part : E. Roose

In the tropics, only forests and savannahs protected from adverse effects of livestock and fire can maintain or improve soil fertility. Tillage inevitably reduces soil fertility in the short or the long run (Roose, 1996; Conedera et al., 2010). As soon as the permanent plant cover is taken away, plant litter stops protecting the soil and soil organic matter (SOM) content plunges. With the decrease of SOM content, the chemical, physical and biological qualities of the upper horizons of the soil decrease as well. Fire is often used to clear the land, but emits NO+CO₂ and brutally mineralises biomass. The resulting ashes temporarily help increase soil pH and fertility but are also very sensitive to wind and water erosion. Repeated burning in a short time span decreases soil quality. Tillage on the other hand brings oxygen into the deeper soil layers. This extra oxygen accelerates the mineralisation of SOM and mixes humic horizons with mineral horizons. As with fire, even when this means a temporary boost of chemical fertility, repeated tillage reduces the biological activity of the soil, hence productivity, and increases soil sensitivity to the impact of raindrops and runoff (Roose and Barthès, 2006). Tillage brings about a 50% loss of SOM in 4 years for sandy soils and 12-14 years for clay soils (Roose and Barthès, 2001). This is the main reasoning behind efforts to conceive systems of direct seeding in mulched soils (Seguy et al., 2004; Vander Mijnsbrugge et al., 2009). The insight that soil needs to be covered with plant litter to keep its productive potential helps explain why soils in forest are never cultivated – they actually do not need it. They are fertile and drain well thanks to their optimal soil life in the upper horizon (Roose, 1996).

Soil scientists usually teach that the soil is a non-renewable natural resource at the scale of a human generation. That holds true in the case of the erosion of a shallow humic horizon overlying a resistant mother rock, such as granite or calcareous deposits. In tropical conditions, it indeed takes 200,000 to 300,000 years to weather 1 m of granite (Leneuf, 1956). The statement is much less true in the case of certain soft and brittle rocks, such as claystone, marl, sandstone, and soft schist, and even for basalt, on the one hand because weathering of such rocks can produce 0.5 to 1m of soil in less than a century (Leneuf, 1956; Roose, 1996). On the other hand smart farmers have developed a variety of systems to create new soils with weathered material or volcanic ashes (Roose, 1996). Apart from traditional fallow systems (Floret and Serpantié, 1991), farmers have also developed techniques to restore eroded soils that rely on straw mulching, organic and mineral amendments and agroforestry (Aronson et al., 1993; Roose et al., 1993; Ouedraogo et al., 1996; Roose et al., 1996; Sawadogo et al., 2008). In the African Sahel region, to recuperate exhausted croplands, traditional zaï techniques have been developed that combine water harvesting with micro-catchments, manuring, and fostering termite activity (Roose et al., 1993; Ouedraogo et al., 1996; Roose, 1996; Roose et al., 1996; Sawadogo et al., 2008).

In ecology, the restoration of an ecosystem in the strict sense consists in taking away the factors causing degradation. If the degradation is rather mild, the ecosystem will recover spontaneously in a short time span. If, however, the degradation is very advanced, the recovery needs to be assisted. Recovery means the ecosystem is allowed to evolve toward a pre-defined reference ecosystem, with its characteristic fauna and flora and its associated physical, chemical and biological soil properties (Aronson et al., 1993). In this article, the focus is on restoring agro-ecosystems in a short time span. We bring together a number of case studies where labour-intensive but simple operations have been shown to rapidly and sustainably restore the productivity of degraded agro-ecosystems, hence their prime role in providing food, feed, fuel, and fibre for rural land users (Floret, 1991; Ouedraogo et al., 1996). Table 1 compares some environmental parameters of the experimental sites of these case studies. From these case studies six rules for rapid restoration of cultivated soils are derived.

Table 1 A synopsis of climate and soil data of the case studies and their references.

Case studies

Creation of new productive soils from weathered rock

A first series of case studies brings together examples where new soils are created in situations where almost none is available (steep slopes) or else was left after intense erosion.

In Haiti, the continuous cultivation of brown clay soils overlaying basalt on steep slopes has led to copious runoff. The humic horizon disappears in a few years time, and when basalt stones appear at the soil surface the land has to be abandoned (Smolikowski, 1993). These “dead lands” can, however, be brought back to life. Some courageous and ingenious farmers buy them up. With a crowbar, they dig holes each 2 to 5m in staggered rows. The 0.2m³ volume of each hole is filled with a mixture of fine soil (extracted from other places like constructed areas or from the river sediments) and manure, in which a fruit tree and creeping legumes are planted. Little ridges concentrate runoff water. The manure comes from a goat or a pig that is penned and fed in a nearby hole, to produce the necessary manure to vivify the weathered rock. After about ten years, the plot that was treated as a “collection of plant pots” concentrating runoff water and sediments, animal manure and plant litter, becomes fully productive again, even if more slowly in between the pots.

In Mali (Kassogué et al., 1996), centuries ago the Dogon people were forced to make a living in rocky hills between sandstone boulders to escape muslim invaders that settled in the plains. They built new soils on subhorizontal sandstone banks by arranging stone bunds in a dense beehive pattern. The 1m² basins are filled with a mixture of sandy soil and manure transported from the valleys. Each basin is planted with sweet onion. In the dry season, the crop is watered with from a well or from a nearby micro dam. To conserve as much water and soil as possible, a variety of techniques are applied, ranging from simple alignments of stones or of faggots of sorghum straw and stone walls on steep slopes to cultivation techniques that concentrate resources rather than spreading them equally over the soil surface. The Dogon people can survive with a density of over 80 inhabitants per km² in a climate with less than 450mm of summer rain per year, thanks to the onions that are highly prized all over the region.

In the granitic Mandara Mountains of North Cameroun (Seignobos, 1998), the Mofu accelerate the soil formation of narrow and elongated contour terraces that are built with stonewalls. This physical arrangement is completed with inputs of manure and sand and of plants chosen for root systems that can explore the rock fissures (sorghum, millet, Cynodon dactylon), different fern species and forage plants, Ficus and Acacia polyacantha. Manure is concentrated by raising young cattle in pens. To get new substrate, they light fires at the base of a rocky slope. Upon heating, rocks explode in smaller pieces and sand particles. In this way it has been possible to reintroduce dozens of local woody species.

In Mexico and Ecuador (Quantin et al., 1993; De Noni et al., 1994), experience has shown it is possible to restore indurated volcanic ash soils, locally named tepetate, by the mechanical fragmentation with powerful machines, terracing and fertilising a crop rotation of wheat, maize and a legume (vetch, faba beans or Phaseolus beans). From year 3 on, yields have recovered to the regional average and the operation can start paying back.

In El Salvador (Collinet and Mazariego, 1993), the introduction of chicken manure and leaves of Gliricidia sepium grown in hedges hastened the restoration of the structure and the productivity of an andosol on young volcanic ashes.

In Martinique (Roose and Barreteau, 2003; Roose et al., 2005), the erosion of a denudated steep slope (45-60%) threatened to suffocate the coral reefs of a nearby sea inlet (the nature reserve of Caravel). With the Universal Soil Loss Equation (USLE) model (Wischmeier and Smith, 1978), the potential erosion was estimated at 120 metric t/ha/year (Roose et al., 2005). Somehow, the spontaneous vegetation cover of that slope had to be restored. A first attempt to plant local woody species upslope a series of planks maintained by two stakes (pseudo-terraces) failed because the eroded sediments of the slope to be stabilised were very acid, low in SOM and nutrient poor. The saplings just did not take off and erosion continued to be excessive (> 30 t/ha/year).

A second attempt worked with five micro-catchments of 67 to 130m² surrounded by stone bunds that were cemented onto the base rock. This was combined with a tank (1m³) dug at the lower end of each micro-catchment. These tanks were filled with sediments that came down the slope (runoff + suspension + coarse sediments). The USLE model makes it possible to estimate an erosion potential of 120 metric t/ha/year(Roose et al., 2005). The slope was then covered with a 2-5cm mulch of sugar cane residues of the nearest sugar refinery and three treatments were compared for the trees planted in new cubic 40cm-wide basins: (1) only crushed rock with sugar cane mulch; (2) crushed rock, sugar cane mulch and 10 litres of compost; and (3) crushed rock, sugar cane mulch, 10 litres of compost and 50g of NPK fertiliser.

The survival and growth rates of the trees under treatment (1) were near zero, of those under treatment (2) better, and best for those under treatment (3) (Roose and Barreteau, 2003). All the pockets, including the 50% with dead trees, developed a spontaneous cover of grasses and legumes and increased faunal activity (especially ants) was recorded. During the first three years and before its complete mineralisation, the sugar cane mulch was able to stop the erosion, which gave the necessary protection to the saplings to start growing and take over the role of the disappearing mulch. However, the transport and the manual spreading of this mulch (50 t/ha) proved too costly to be feasible (7,000 euros/ha). For the future, solutions should be found to concentrate the mulch under the tree planting lines and to mechanise the mulch spreading work.

In the High Atlas of Morocco (Roose, et al., 2010), two techniques of soil creation coexist, depending on the slope and correlated difficulty of the work. In both cases there is very little prime cropland available (in the case of the Rheraya valley, less then 10% of the mountainous landscape consists of cropland). The first technique is applied in floodplains of mountain wadis. Pebbles are used to enclose rectangular fields with permeable walls. Inside, floodwater is allowed to stagnate and sediments build up. A grassy sward develops gradually on the sediment layer, which accelerates the process of sediment build up. As soon as there is 10cm of sediment, farmers can manure and till it and get a first crop (usually a cereal-legume mixture such as vetch-oats). After a few more floods, this new soil can reach 40cm of depth and fruit (walnut, cherry, apple) or forage (willow, ash, poplar) trees can be planted. Trees thrive and can survive the summer drought without irrigation because their roots can reach the underflow of the riverbed that is fed by summer snowmelt from the highest peaks (>3,500m). Tree growth has to be fast because the trees need to survive the first new floods. However, flooding is variable from year to year. Each 20 to 50 years there will be an extreme flood that destroys these fields, after which farmers will spend about 10 years rebuilding them.

The second technique of the Moroccan High Atlas (Roose et al., 2008) is the terracing of steep slopes (20-40%) with shallow stony soils. Stonewalls are equipped with drains and the area immediately behind each stonewall is filled with humic soil that is found in pockets lower down the slope. The work needed to sort suitable rectangular stones, clear the foundation space, build tilted stonewalls, fill the terrace with humic soil and incorporate manure can be estimated as two hours per square meter of stonewall, or 700 to 1,400 workdays/ha for slopes of 20 to 40%, respectively. This is without counting the transport of manure (10 t/ha) and the construction of an open irrigation canal to divert water from an uphill source or stream. Farmers spend decades perfecting these terraces but their human toil is rewarded with land property rights.

Peasant methods to restore degraded soils

A second series of examples brings together cases where degraded soils are still partially in place despite erosive practices, and where the focus is on restoring functioning and fertility of the remaining soil.

Fallowing

The most widely used (and easiest) technique to restore abandoned cropland is the long fallow (10-50 years depending on the pedoclimate). As long as the prevailing climate allows fair plant cover of the soil, spontaneous ecological succession can take place. As soon as a few months later the soil can be well covered (and thus protected against the impact of the raindrops) by a grassy sward. Grasses are known for their structure-restoring properties. They restore soil structure because their dense fine root systems hold together the soil aggregates. Sooner or later woody species get established in the grass sward and the ecosystem evolves to scrub and eventually to woodland and forest. Woody species root deeper, bring cations deep into the soil and act as nutrient pumps in the reverse direction. Leaf litter enriches the superficial soil layers with carbon and nutrients (Floret and Serpantié, 1991; Roose and Barthès, 2001; Seguy et al., 2004). However, under population pressure, long fallow declines to short fallow, and the natural fertility-restoring processes have to be accelerated artificially. This is possible by sowing forage species to create “forage fallows” of selected species, usually again a mixture of grasses and legumes (Roose, 1996).

As an example of intensified short fallow we can cite a simple crop rotation of South Benin (Azontonde, 1993). A fallow of only 7 months sown with Mucuna pruriens (an annual legume pulse of tropical origin and widely used) in between two maize crops can reduce runoff and erosion, increase SOM content and improve maize grain yield from 0.2 to 2.8 t/ha/year. This rotation has been introduced on “terres de barre” (desaturated and indurated lateritic soils developed on tertiary sandy-clay sediments) under conditions of high population density.

Restoration of acid lateritic soils through mulching and organic and mineral amendments

In Central Burundi (Rishirumuhirwa and Roose, 2004) erosion rates were measured over three consecutive years in the experimental site of Mashitshi (slopes <25%), near Gitega under three 300m² main plot treatments: (1) fully mulched with local grass; (2) banana plantations of varying densities with mulch of banana leaves; and (3) bare soil (tilled but without mulch, nor crop). The cumulated erosion after three years was 0.15 t/ha on the fully mulched soils, 17-54 t/ha under banana plantation and 154 t/ha on bare soil. In the fourth year each of the three main plot types was split into four subplots and sown with maize under four types of soil amendments: (1) no amendment, (2) peasant practice: manure at 20 t/ha; (3) manure at 10 t/ha + NPK-fertiliser; and (4) manure at 10 t/ha + NPK-fertiliser + lime at 200 kg/ha to reduce pH and Al-toxicity. The response of the maize crop to the amendments was inversely related to preceding erosion intensity. The bare soil with highest erosion rates yielded zero kg maize grain per hectare with no amendment and maximally 500 kg maize grain per hectare with manure and NPK and the fully mulched soil yielded 1,700 kg maize grain with no amendments (similar to the productivity after burning secondary forest) and maximally 3,000 (manure only) to 4,000 (manure + NPK) kg maize grain per hectare. Interestingly, liming did not improve those yields (figure 1).

In the experimental sites of Rubona and Butare in South West Rwanda (Roose and Ndayizigiye, 1996; Koenig, 2005), with the same soil types (table 1) but with steeper slopes (>25%), hedges of leguminous shrubs (Calliandra calothyrsus and Leucaena leucocephala) reduced runoff (yearly runoff coefficient <2%) and erosion (<2 t/ha/year) rates while adding significant amounts of nutrients to the soil (100kg N/ha/year; 10kg P/ ha/year, and 20-40kg Ca, Mg and K per hectare and per year). However, the yield of the crops between those hedges did not improve unless farmyard manure and/or NPK fertiliser was added. This illustrates well our idea that soil and water conservation measures on their own are not enough to improve crop yields. They have to be complemented with fertility restoring measures, especially those that increase the amount of nitrogen and plant-available phosphorus (Roose, 1996; Roose and Ndayizigiye, 1996; Koenig, 2005).

Restoring ferruginous sandy soils in the Sudanese climate zone of Africa

In North Cameroun (Boli and Roose, 1998; Boli and Roose, 2004) and on ferruginous sandy soils that were degraded by thirty years of continuous cultivation, ploughing in crop residues did not improve water infiltration properties and did not reduce erosion rates, in stark contrast with lateritic clay soils, because sandy aggregates are much more fragile than clay aggregates. On a 2% slope, the erosion rates cumulated over four years reached 160 t/ha on bare soil (equivalent to 10mm of top soil removed), 90 t/ha (6mm of top soil) under a maize-cotton rotation with yearly tillage and 30 t/ha (2mm of top soil) with the same crops with reduced tillage. In the fifth year, maize was sown uniformly in all plots and as in Burundi, also here the after-effects of the preceding erosion rates were visible. Figure 2 shows the maize yields in year 5 in relation to the way the plots were treated the previous four years. After a plot history of tillage and correlated high erosion rates (several treatments on the left of figure 2) yields were 40% lower than after a plot history of reduced tillage and correlated low erosion rates (several treatments on the right of figure 2). The yield reduction caused by selective sheet erosion is of the same order of magnitude as the yield reduction after experimental non-selective removal of a tenfold quantity of soil. In other words, selective sheet erosion is ten times more degrading than non-selective soil removal. Overall, the experimentation demonstrated that the productivity of even badly degraded soils could be restored in a very short time-span with very simple techniques. Their physico-chemical behaviour toward the aggressive rains had changed completely.

Restoring soils with zaï technique in the Sudanese-Sahelian climate zone of Africa

In Burkina Faso (Marchal, 1986; Zougmore et al., 2004), many fields are abandoned after a history of 10 to 15 successive years of extensive cropping with tillage. These fields are characterized by a thick and impermeable erosion crust, which hampers spontaneous secondary succession during the fallow. Around 1986, over 20% of the croplands of the Sudanese-Sahelian zone (400-800mm of rain in four to five months) were desertified in this way (Marchal, 1986). Nothing grew anymore on these “zipellés”, unless they were recuperated at great expense by the landless under population pressure, applying a variety of zaï techniques. With zaï, most of the work has to be done during the dry season. The landless dig 8,000 to 12,000 shallow holes per hectare, roughly 1 hole/m². The holes are 20cm deep and 40cm in diameter and the dug-out earth is arranged as crescents downward the slope, in order to create micro-catchments. The soil in the hole is then mixed with manure (usually dried powdery goat dung) at rates of 1-3 t/ha and/or plant litter. Finally, sorghum or millet is sown in clusters of 10-12 seeds per hole, so that the collective power of the germinating seeds can lift the sedimentation crust that will have been produced by the first rains. These first rains produce runoff from between the holes into the holes and the drainage water produces resource pockets that can feed the sorghum and millet plantlets for over three weeks. Right from the first year on, these recuperated zippelés produce as much grain as the regional average of 600kg/ha but with a complement of 60kg N and 30kg of P per hectare, it is frequent to reach 1,500kg/ha, eight times the productivity without zaï ([Roose, 1986], figure 3). The added goat dung has been observed to contain viable seeds of many different plant species: 15 different leguminous shrubs and 26 herbaceous species (Hudson, 1987). The dung thus reintroduces many herbaceous and woody species that covered these fields before their desertification (figures 4 and 5).

These techniques are tremendously labour-intensive, which explains the reticence to do the job in the hottest period of the year. About 350 man hours/ha with a pick are necessary, to which has to be added the production and transport of 3 tons of manure and the collection of 10 tons of stones to add anti-erosive stone bunds around the fields (Roose, 1986; Zougmore et al., 2004).

Restoring rendzinas on calcareous rock and brown loam-clay earths on marls in North Africa

In North Africa (Bellefontaine, 2010; Bellefontaine et al., 2010), the soils on marls are easily degraded and crust-covered by the impact of the rain drops on bare soil after many extensive cereal cropping cycles or because of overgrazing. Soils on calcareous rock, on the other hand, are rarely deep, because the weathering of calcareous rock leaves very little solid residue (sands, Fe/Al hydroxides and some fossils). The carbonates are dissolved by the rainwater and humic acids, but do not contribute to soil formation. Foresters have adapted the previously decribed zaï techniques to these conditions to plant Mediterranean fruit (almond, olive) and timber (eucalyptus, pine) trees. The micro-catchments are bigger and wider spaced and the area of the catchment should be 4 to 20 times the surface of the cultivated area, depending on the edaphic aridity (Hudson, 1987). Also here, well-decomposed manure should be used to foster tree growth. A recent example of adapted crescent-shaped bunds around plant pits in Morocco is the regeneration of the argan tree forests (figure 6). The argan (Argania spinosa, syn. A. sideroxylon Roem. & Schult.) is a tree species endemic to the calcareous semi-desert Souss valley of southwestern Morocco and to the Algerian region of Tindouf in the western Mediterranean region. In Morocco, argan tree forests now cover some 8,280 km² and are designated as a UNESCO Biosphere reserve (Bellefontaine et al., 2010). Their area has shrunk by about 50% over the last 100 years, due to charcoal-making, grazing, and increasingly intensive cultivation. The best hope for the conservation of the trees may lie in the recent development of a thriving export market for argan oil as a high-value product.

Local populations noticed the complete lack of spontaneous regeneration of the argan trees (Bellefontaine et al., 2010). In 2000, foresters were contacted with the view of restoring the argan tree forests. This led to a plantation programme with bulk seed in exchange for protection measures against goat browsing of young saplings. Planting holes were deeper (70-80 cm) than the usual 30-50cm generally practiced in the region, which helped roots to access deeper soil layers. Holes were filled with amended soil only up to 80% of their total depth. The remainder of the soil was arranged as a bund around the hole. The preparation of planting holes in staggered rows improved the water reserves available for each sapling individually and encouraged the return of weeds and local shrubs of medicinal use in between. What remains to be done is the genetic improvement of the plant material. Bulk seed should be replaced by cuttings of selected individual trees raised in a nursery. It has been shown that rigid deep containers give much higher quality roots (no rolling up of roots during extended waiting period in nursery) hence stronger saplings than the conventional polyethylene bags (Bellefontaine, 2010). The plantation cost is about 3.6 euros per plant, which, at planting densities of about 150-200 trees per hectare comes to 636 euros, including digging, refilling, catchment building, plant transport, plantation, irrigation, replacing dead saplings, maintenance and two waterings per month during the first months following plantation. This price does not include the seed and nursery costs, nor the training and transportation costs of the foresters. After four successive plantation seasons, a survey in 2008 checked mortality and the saplings showed no signs of goat browsing. It concluded with a general mortality rate of only 5-8% if no account is made of holes left unplanted, which would be exceptionally low compared to mortality rates of other forestry plantations in the region (Bellefontaine et al., 2010).

Restoring steppic ecosystems in arid Tunisia

In arid Tunisia (100-400mm of annual rainfall) (Le Floc’h et al., 1999; Visser, 2001; Visser et al., 2008; Visser et al., 2010), desertified fallows after cereals with near zero perennial plant cover and productivity are a prominent feature of the landscape. They have high restoration potential but there are serious obstacles. In arid Tunisia, a first category of obstacles centered around the question: where should the seed come from? That question can now be considered as resolved (Visser et al., 2008). Native seed production has focused on two highly palatable and productive perennial grasses (Stipa lagascae R. & Sch., Cenchrus ciliaris L.) and one herbaceous pluriannual legume (Argyrolobium uniflorum (Decne.) Jaub. and Spach) (Visser, 2001). A second category of obstacles is related to the technicalities of reseeding in a very agriculture unfriendly environment. Reseeding is a high-risk endeavour even if quality seed of native species is available, and this risk increases with climate change in the Mediterranean area. Very few reseedings have been reported on in the Maghreb. We are aware of just two publications (Le Floc’h et al., 1999; Visser et al., 2010) and many (not always failed) reseedings remain unreported. Still, in fact, the technicalities of reseeding cannot be addressed without tackling a third and the most important category of obstacles: for whom and for what immediate use will we reseed desertified cereal fallows? Restoring desertified drylands with native species is as much an agro-ecological as a societal challenge.

In a context of extensive rangelands bordering the Sahara desert that were traditionally grazed by tribally-owned small ruminant herds, it is important to appreciate that cereal fallows have become de facto privatized land under “sharia” law. Private land users have to be convinced of the advantages of reseeding fallows with locally adapted grasses and legumes rather than with barley, but once they are convinced, how can we technically ensure reseeding success with the best available native seed? In these desertified landscapes, the above described techniques create micro-catchments that spatially concentrate water and nutrients to jumpstart growth of fertility restoring plant species, which in turn enhances overall agro-ecosystem performance. However, since we work in an extremely depleted soil, nutrients have to be added to jumpstart growth of reintroduced seed effectively.

A key issue in the ecological restoration of the Maghreb drylands is sourcing adequate amounts of organic amendments. In particular, the rangelands of Northern Africa are severely phosphorus-depleted (Visser, 2001). Phosphorus deficiency is the main factor limiting the full expression of the growth potential of resource-responsive legumes, which in turn need to fix the necessary nitrogen for the full expression of the growth potential of resource-responsive grasses. Even though Tunisia exports high-quality rock phosphate and superphosphates, these fertilizers prove very expensive for local users and quickly become plant-unavailable in calcareous soils. The alternative is manure from stable-fed small livestock, which is more balanced with regards to other nutrients, plentiful, cheap and easily accessible. Along with the other nutrients, the phosphorus of the imported feeds can become of benefit locally, by replenishing soil fertility of an agro-ecosystem otherwise left bare for years to come. However, total P-content of this manure will rarely exceed 0.1% and will be released slowly. As above in the Burundi example, a compromise between manure for overall soil fertility and small amounts of concentrated P to jumpstart growth just after seeding might be the best compromise between expenses and results.

Discussion

Mechanical soil and water conservation techniques rarely enhance growth and yield on farmers’ fields. Even biological techniques that use trees or shrubs to conserve soil and water rarely do so. To enhance growth and yield on farmer's fields, we should not just concentrate rainwater and conserve soil in order to stock this water, we should also provide for plant nutrients (Roose et al., 1993; Ouedraogo et al., 1996; Roose et al., 1996; Zougmore et al., 2004; Roose et al., 2008; Visser, 2001) in order to make use of this newly acquired water and soil. Moreover, these nutrients should be provided essentially in an organic form, so that organic matter and soil biological activity can build up again and thereby increase the water holding capacity and the structural stability of the soil. Mineral nutrients, if available and affordable, could complement organic nutrients. However, the exclusive use of mineral fertiliser brings new problems such as soil acidification, which in turn renders unavailable certain micro-nutrients.

However indispensable to jumpstart the soil recovery process, providing adequate amounts of nutrients is often the bottleneck of the endeavour. In order to avoid unnecessary losses, neither organic nor mineral nutrients should be spread over the total area to be restored, but rather concentrated in the micro-catchments, near the seeds or the plants. If possible and feasible, mineral nitrogen fertiliser should be added in successive small doses. For example, in cereal cropping, key moments for extra nitrogen addition are during tillering, heading and flowering. This bottleneck means that all available biomass should be put to use to yield adequate amounts of organic amendments, such as dung, human faeces, weeds, crop residues, kitchen scraps, fire ashes.

In contrast with impermeable earthen barriers, permeable barriers with a biological component (hedges, grassy bands, micro-catchments with local woody or grassy species) are better suited to reduce the speed and the volume of surface runoff (both water and suspended sediment load). Living barriers allow for the superficial drainage of rainwater in case of heavy and concentrated showers, which is the way rain often falls in drylands. If the field in between barriers is covered with a mulch (plants or stones), solid transport can be reduced to an acceptable minimum (Roose and Barthès, 2001; Seguy et al., 2004; Roose et al., 2010; Roose et al., 2008). The other advantage of living barriers is that they almost always increase biodiversity, either intentionally through the choice of local plant material, or as a side-effect of the creation of resource islands that become attractive for local wildlife.

However, many big multilaterally financed projects use exotic species to create living barriers, even if these species are classified as invasive (Tassin et al., 2009), or already known to fail if the environment is too harsh, as in the case of many exotic fodder species planted in North-African drylands, in particular Acacia saligna. Many failed experiments have shown that, as a precautionary measure, only plant material that is already of local use (Bellefontaine, 2005; Tassin et al., 2009) should be considered, which also enhances social acceptance of such restoration projects. For most grass and tree species, a simple clonal selection program, followed by multiplication of the best clones, could quickly yield the necessary amounts of local plant material of sufficient quality and genetic diversity (Visser et al., 2008; Tassin et al., 2009; Bellefontaine, 2005; Bellefontaine, 2010). Ample genetic diversity of local origin is preferable to poor genetic diversity often associated with exotic plants. If it is difficult to obtain seed, low-cost vegetative propagation techniques (Bellefontaine, 2005; Meunier et al., 2006; Tassin et al., 2009) can be envisaged on the condition that a broad pool of genetically different clonal lines is selected.

A final issue that merits discussion is that despite the seeming similarity of techniques applied in a wide range of environments that cope with drought and nutrient stress, the need to adapt them to local constraints should not be overlooked. Between the Sudanese-Sahelian and Mediterranean drylands south and north of the African Sahara in particular, the ecological differences are startling. The Sahelian landscapes are often long glacis with slopes <2%. The rain regime is one of summer rainfall, concentrated in three to five months, leading to very high Potential Evapotranspiration (PET) and rapid plant growth. After the rainy season, the final growth stages of the cereal crop depend on remaining soil water reserves, which in turn depend on how much water was stored during the preceding months (to be influenced with zaï techniques) and on the physical properties of the soil surface when the soil is drying out. Mediterranean rains on the other hand are concentrated in the winter months, which makes for much lower PET and slower plant growth. Hence, in Mediterranean regions the area of micro-catchments for tree plantations must be much larger than in Sahelian areas. In addition, because there is much more relief in the wider landscape, the small-scale creation of extra soil roughness through barriers and micro-catchments is less efficient to concentrate water and nutrients on steeper slopes (Roose, 1996). As a consequence, the area of the catchment to be provided per tree has to be larger, between 4 and 20 times the surface of the cultivated area (Hudson, 1987).

Conclusion: Six rules for the rapid restoration of degraded land and agro-ecosystems

We brought together results from 17 in situ experiments, observations of farmers’ practices and summaries of individual experiments to demonstrate the feasibility of low-tech solutions with mainly local resources to solve the complex problem of restoring degraded tropical soils worldwide. It is possible to speed up rock weathering and to restore degraded soils if water and soil conservation measures are combined with nutrient addition. We can summarize these case studies with a set of basic rules to foster restoration success in a few years’ time.

• 1) Management of runoff water with a combination of adapted techniques and design: stone bunds, mulching, micro-watersheds, terracing, hedges, etc;
• 2) Locally work the soil to recreate soil macroporosity for enhanced infiltration of captured runoff water (rule 1) and stabilise soil structure adding structural organic matter (with high C:N ratio);
• 3) Stimulate soil life locally by adding compost, manure, easily degradable plant litter (legume biomass for example) with lower C:N ratios;
• 4) In the vicinity of plants, correct soil pH if necessary to alleviate Al-toxicity of very acid soils and increase plant phosphorus availability (irreversibly fixed outside the 5<pH<8 range);
• 5) Use the newly created resource islands (rules 1-4) to nurture plant growth. Especially during the delicate establishment phase, add enough directly plant-available macro-nutrients (especially N and P) through NP-rich organic amendments (low C:N:P ratio) and, if feasible, carefully dosed mineral fertilizer complements. Always add these close to the developing plant roots;
• 6) Choose genetically diverse plant material and prefer the species already in demand by local land users.

These rules emphasize organic matter and its synergetic roles to restore biological, chemical and physical soil functioning. Since plant growth is the ultimate source of soil fertility, and since degraded soils have been typically depleted of this soil fertility, any restoration technique aimed at promoting plant growth should start by adding organic matter and possibly mineral nutrients. As a consequence, the lack of availability of the right organic amendments in adequate quantities will often be a bottleneck of any large-scale soil restoration project, as these materials might have to be sourced elsewhere, at the expense of other agro-ecosystems.

In addition to the rapid restoration of agricultural productivity, the application of these rules also enhances local plant and animal biodiversity. If the choice of threatened local plant species is in tune with the demands of the local populations (in particular wild fruit trees), these species will get extra protection.

Notwithstanding the apparent simplicity of these rules, they require a lot of work and commitment in a short time span. It should be crystal clear that not one peasant will be ready to make a considerable initial investment (human labour and/or machinery with energy expenditure, concentrating adequate quantities of organic amendments, nursery requirements, initial follow-up after planting and seeding, etc.) if the farmer cannot be assured of a rapid and considerable return on this investment (ideally from year 1 on). If there are problems with land ownership or with benefit sharing, peasants will stick to labour-extensive cultivation methods and continue degrading the soil. It is up to national, regional and local policies to foster, not discourage, large-scale soil restoration.

Many Mediterranean and tropical lands have been degraded by traditional cropping methods that are no longer adapted to growing population densities. It is time to reorient research towards agro-ecological techniques that reverse and prevent land degradation with local resources and based on local knowledge. We invite our colleague restorationists of the world to revisit traditional soil creation and restoration techniques and test improvements of these techniques with the help of new insights into the management of biodiversity and genetic resources, the importance of soil life and landscape configurations, and sound agronomy (using reasonable irrigation and fertilization techniques) to combat soil erosion.

References

J Aronson, C Floret, E Le Floc’h, C Ovalle, R. Pontanier Restoration and rehabilitation of degraded ecosystems in arid and semi-arid lands. I. A view from the South Restoration Ecology 1993; 1: 8-17.

Azontonde A, 1993. Dégradation et restauration des terres de barre (sols ferrallitiques faiblement désaturés argilo-sableux) au Bénin. Cah Orstom, Ser Pedol 28: 217-26.

Bellefontaine R, 2005. Pour de nombreux ligneux, la reproduction sexuée n’est pas la seule voie: analyse de 875 cas. Texte introductif, tableau et bibliographie. Sécheresse 16: 315-7 et Sécheresse 3E, 2005. http://www.secheresse.info/article.php3?id_article=2344.

Bellefontaine R, 2010. De la domestication à l’amélioration variétale de l’arganier (Argania spinosa L. Skeels). Sécheresse; 21: 42-53. doi : 10.1684/sec.2010.0226.

Bellefontaine R, Ferradous A, Alifriqui M, Monteuuis O, 2010. Multiplication végétative de l’arganier (Argania spinosa) au Maroc: le projet John Goelet. Bois For Trop; 304: 47-59.

Boli Z, Roose E, 1998. Degradation of a sandy Alfisol and restoration of its productivity under cotton/maize intensive cropping rotation in the wet savannah of Northern Cameroun. Advances in Geoecology 31: 395-401.

Boli Z, Roose E, 2004. Effet du labour classique et du non-labour sous litière sur le fonctionnement de deux sols ferrugineux tropicaux sableux à Mbissiri, Nord Cameroun. Bull Réseau Erosion 23: 431-7.

Collinet J, Mazariego M, 1993. Réhabilitation de sols volcaniques dégradés à l’aide d’amendements organiques au Salvador (Amérique Centrale). Cah Orstom, Ser Pedol 28: 187-202.

Conedera M, Bomio-Pacciorini N, Bomio-Pacciorini P, Sciacca S, Grandi L, Boureima A, Vettraino AM, 2010. Reconstitution des écosystèmes dégradés sahéliens. Bois For Trop; 304: 61-71.

De Noni G, Janeau JL, Prat C, Trujillo G, Viennot M, 1994. Hydrodynamique, érodibilité et conservation des sols volcaniques indurés d’Amérique latine (Equateur, Mexique et Nicaragua): impact du matériau originel et effet de la réhabilitation agricole. ISSS-MSSS, Acapulco (Mexico), 15^e Congrès de la Société des Science du Sol; 6: 554-70.

Floret C, Serpantié G, 1991.La jachère en Afrique de l’Ouest. Série Colloques et Seminaires. Montpellier : Orstom éditions.

Hudson NW, 1987. Soil and water conservation in semi-arid areas. Soils Bulletin, n^o 57. Rome : FAO.

Kassogué A, Komota M, Sagara J, Schutgens F, 1996. Développement des techniques de CES en milieu Dogon (Mali). In: Reij C, Scoones I, Toulmin C, eds. Techniques traditionnelles de conservation de l’eau et des sols en Afrique. Wageningen (The Netherlands) ; Paris : CTA ; CDCS ; Kharthala.

Koenig D, 2005. Agroforesterie au Rwanda: efficacité et limites. In: Ratsivalaka S, Serpantié G, De Noni G, Roose E, eds. « Erosion et GCES. » Journées scientifiques AUF, Antananarivo.

Le Floc’h E, Neffati M, Chaïeb M, Floret C, Pontanier R, 1999. A rehabilitation experiment in the arid zone. The case study of Menzel Habib (Tunisia). Arid Soil Res Rehab 13: 357-68.

Leneuf N, 1956. L’altération des granites calco-alcalins et des granodiorites en Côte d’Ivoire forestière. Thèse, faculté des sciences Paris.

Marchal JY, 1986. Vingt années de lutte antiérosive au Burkina Faso. Cah Orstom, Ser Pedol 22: 173-80.

Meunier Q, Bellefontaine R, Boffa JM, Bitahwa N, 2006. Low-cost vegetative propagation of trees and shrubs. Technical Handbook for Ugandan rural communities. Montpellier: Cirad éditions.

Ouedraogo M, Kaboré V. Le zai, 1996. Technique traditionnelle de réhabilitation des terres dégradées au Yatenga (Burkina Faso). In: Reij C, Scoones I, Toulmin C, eds. Techniques traditionnelles de CES en Afrique. Wageningen (The Netherlands) ; Paris : CTA ; CDCS ; Karthala.

Quantin P, Prat C, Zebrowski C, 1993.Erosion et restauration des sols volcaniques indurés « Tepetates » de la région de Mexico et de Tlaxcala, Mexique. Cah Orstom, Ser Pedol 28: 175-86.

Rishirumuhirwa T, Roose E, 2004. Influence de la gestion de la biomasse sous bananeraie sur l’érosion, le carbone et les propriétés d’un sol ferrallitique acide des hauts plateaux du Burundi. Bull Réseau Erosion 23: 250-61.

Roose E, 1986. Terrasses de diversion ou microbarrages perméables ? Analyse de leur efficacité en milieu paysan ouest africain. Cah Orstom, Ser Pedol 22: 197-208.

Roose E, 1993. Innovations dans la conservation et la restauration des sols. Cah Orstom, Ser Pedol 28: 147-56.

Roose E, 1996. Land husbandry: component and strategy. Soils Bulletin, n^o 70. Rome: FAO.

Roose E, Barreteau D, 2003.Erosion et environnement à la réserve naturelle de la Caravelle, Martinique. Antilla 1030: 18-19.

Roose E, Barthès B, 2001. Organic matter management for soil conservation and productivity restoration in Africa: a contribution from francophone research. Nutrient Cycling in Agroecosystems 61: 159-70.

Roose E, Barthès B, 2006. Soil carbon erosion and its selectivity at the plot scale in tropical and Mediterranean regions. In: Roose E, Lal R, Feller C, Barthès B, Stewart BA, eds. Soil erosion and carbon dynamics. Advances in Soil Sciences. Boca Raton (Florida): CRC Press.

Roose E, Ndayizigiye F, 1996. Agroforestry and GCES in Rwanda. Soil Technology 11: 109-19.

Roose E, Kabore V, Guénat C, 1993. Le zaï: fonctionnement, limites et améliorations d’une pratique africaine de réhabilitation de la végétation et de la productivité des terres dégradées en région soudano-sahélienne (Burkina Faso). Cah Orstom, Ser Pedol 28: 159-73.

Roose E, Kabore V, Guenat C, 1996. The zaï practice: a West African traditional rehabilitation system for semi-arid degraded land (Burkina Faso). Arid Soil Res Rehab; 13: 343-55.

Roose E, Vernière N, Laune P, 2005. Lutte antiérosive et revégétalisation assistée d’un versant décapé de la Réserve Naturelle de la Caravelle (Martinique). Compte rendu final, Montpellier (France) conv. PNRM-IRD.

Roose E, Albergel J, De Noni G, Sabir M, Laouina A, eds, 2008. Efficacité de la GCES en milieux semi-arides. Montpellier ; Rabat ; Paris : IRD ; ENFI ; AUF ; EAC.

Roose E, Sabir M, Laouina A, 2010. Gestion durable de l’eau et des sols au Maroc. Valorisation des techniques traditionnelles méditerranéennes. Montpellier ; Rabat ; Paris: IRD ; ENFI ; AUF.

Sawadogo H, Bock L, Lacroix D, Zombré NP, 2008. Restauration des potentialités de sols dégradés à l’aide du zaï et du compost dans le Yatenga (Burkina Faso). Biotechnol Agron Soc Environ 12: 279-90.

Seguy L, Bouzinac S, Maronezzi A, Scopel E, Belot JL, Marti J, 2004. From destructive agriculture with soil tillage to sustainable agriculture with direct seeding mulch based systems: 20 years of research by CIRAD and its Brazilian partners in the Cerrados, Brazil. Bull Réseau Erosion 23: 352-61.

Seignobos C, 1998.Pratiques antiérosives traditionnelles: l’élaboration des terrasses des monts Mandara et la récupération des terres hardé (Nord Cameroun). Bull Réseau Erosion 18: 300-5.

Smolikowski B, 1993. La GCES, une nouvelle stratégie de lutte antiérosive en Haïti. Cah Orstom, Ser Pedol 28: 229-53.

Tassin J, Bellefontaine R, Roger E, Kull C, 2009. Evaluation des risques d’invasion par les essences forestières introduites à Madagascar. Bois For Trop 299: 27-36.

Vander Mijnsbrugge K, Bischoff A, Smith B, 2009. A question of origin: where and how to collect seed for ecological restoration. Basic and Applied Ecology 11: 300-11.

Visser M, 2001. Produire des semences autochtones pour réhabiliter des terres dégradées. Le cas de Stipa lagascae R. & Sch. en Tunisie présaharienne. University of Ghent, Belgium.

Visser M, M'Seddi K, Chaïeb M, Neffati M, 2008. Assessing yield and yield stability of remnant populations of Cenchrus ciliaris L. in arid Tunisia: developing a blueprint for native seed production. Grass For Sci 63: 301-13.

Visser M, Ouled Belgacem A, Neffati M, 2010. Restoring Mediterranean cereal fallow drylands using Stipa lagascae R. & Sch.: influence of defoliation during the establishment phase. Grass For Sci 65: 23-7.

Wischmeier WH, Smith DD, 1978. Predicting rainfall erosion losses. A guide to conservation planning. USDA Serie Agricultural Handbook, n^o 537. Washington : USDA.

Zougmore R, Mando A, Ringersma J, Stroosnijder L, 2004. Effet synergique des techniques de gestion de l’eau et des nutriments sur le ruissellement et l’érosion en zone semi-aride du Burkina Faso. Bull Réseau Erosion 23: 540-53.