Trees reduce the amount of sunlight reaching soils and crops through shading. The extent of reduction varies according to crown dimensions, tree phenology and leaf density (Breman and Kessler, 1995). Available evidence on reduction of solar irradiance by tree canopies is reported in Box 3.4.
Several studies in savannas suggest that tree shade increases understorey herbaceous productivity because of the reduction of temperature and evapotranspiration (Tiedemann and Klemmedson, 1977; Bernhard-Reversat, 1982). In Kenyan savannas, however, artificial shade increased herbaceous productivity in only one open site (Belsky, 1994). This was attributed to the physiological adaptation to shade of particular plant species, consisting in their capacity to reduce their stomatal apertures and conserve moisture at low light levels (Amundson et al., 1995). Belsky (1994) could not establish clearly whether microclimate or nutrient enrichment was responsible for increased productivity under trees. But if stomatal conductance is primarily responsible for the increase in productivity in open sites, one would assume that agricultural crops, especially C4 crops which are generally adapted to full light conditions and sown uniformly under and outside parkland tree cover, could not benefit significantly from conditions associated with reduced light intensity. In turn, higher soil nutrient levels may be the dominant advantage of parkland trees for crop production.
How much do parkland tree canopies reduce sunlight intensity?
|Solar irradiance was reduced by 45 to 65 percent under Acacia tortilis and Adansonia digitata (Belsky et al., 1989). Only about 20 percent of total radiation reached the understorey of A. tortilis and Balanites aegyptiaca at midday in a Sahelian savanna (Akpo and Grouzis, 1996). Based on measurements of shade contours and corrections with photometer data, sunlight intensity was reduced to 45 percent under 10 to 13 m high Vitellaria trees and 20 percent under 14 m high Parkia trees, with successively less shading further away from the bole (Kessler, 1992). Vitellaria trees of 7 m height and 4.7 m crown diameter also decreased photosynthetically active radiation (PAR) directly under and outside crowns by 40 and 20 percent respectively (Boffa et al., 1999). Under larger trees (average crown diameters of 8.4 to 11.2 m for Vitellaria and 9.5 to 17.1 m for Parkia trees), PAR was reduced by 75 percent on average (Jonsson, 1995).|
|In contrast to other parkland trees, Faidherbia albida generally loses its leaves shortly before the rains and remains leafless during the growing season. Light interception by this species is therefore generally considered to be small and does not affect crop production significantly. However, the extended leafing resulting from pruning (Box 3.2) suggests that crops are at least partially shaded. Shading by unpruned trees has not been quantified, but it appears to be slight because of the reduced crown and its partial recovery and densification during the initial three to four months of cropping (Depommier, 1998). It may have a positive effect on crop establishment, but a negative one at the grain filling stage when high insolation is required for cereals.|
|When trees are defoliated, the reduction of solar irradiance by branches, especially low ones (Poschen, 1986), may be important. Maximum radiation around midday in May under a nearly leafless F. albida was about half that in the open and shade had a substantial impact on soil temperature (van den Beldt and Williams, 1992). Radiation was also reduced by 30 to 40 percent under leafless Acacia tortilis and Adansonia digitata during the dry season (Belsky et al., 1989).|
In contrast to herbaceous productivity in savannas, substantial evidence shows that light interception by tree canopies is associated with decline in C4 crop productivity. Reduction of photosynthetically active radiation (PAR) generally results in a decline of C4 grain productivity, as shown in a standard light curve for maize (ICRAF, 1993). Sorghum yield decline from a distance of two crown radii towards the bole was strongly correlated with a consistent decrease in average light intensity estimates (Kessler, 1992). Boffa et al. (1999) found a significant decrease in sorghum yields with increasing canopy size in plots of fixed location bordering the canopy edge. Light was a determining factor of grain yield variation in plots located under tree crowns. Shade of F. albida does not appear to have a significant impact on cereal yields, but this would need to be precisely tested in the case of intensely pruned trees. However, Poschen (1986) found the shade from branches significant and recommended that F. albida branches up to 4 m be pruned in order to minimize light competition with crops.
Smaller V. paradoxa trees were also associated with less yield depression (Kessler, 1992) or higher production (Diakité, 1995) than larger ones. Kessler recorded a greater depressive effect for both Vitellaria and Parkia trees than Maïga (1996) or Kapp (1987), and attributed this to the larger tree size (10 to 13 m) in Oula. Likewise, a smaller yield decrease of 10 percent was found in Vitellaria parklands with average tree height of 7 m (Boffa et al., 1999). This was not the case in Saponé, Burkina Faso, where neither the percentage of light reduction nor crop yields changed markedly according to tree size under small and large Vitellaria and Parkia trees (Jonsson, 1995). Indeed, millet biomass production was found to be higher under a large Parkia tree than under a small one or under Vitellaria trees or in controls, maybe due to a relatively higher nitrogen content. Despite this isolated case (presented in more detail below), one may conclude that reduction in light intensity varies according to tree shape and size and generally contributes to yield depression of C4 grain crops in the proximity of these parkland species. In contrast, C3 crops are adapted to lower light intensities and their yield may be less depressed by partial shade.
Although crop establishment and growth are impeded by extreme heat, there is no conclusive evidence of shade being responsible for better yields under Faidherbia trees, than in the open.
Temperatures are lower under tree canopies due to shading. In semi-arid Kenya, soil temperatures 5 cm below the surface were at least 5–9°C lower under trees than in the open grassland, both at the beginning of the growing season and when grass cover was at a maximum. The difference between locations decreased with soil depth (Belsky et al., 1989). Soil temperature was also substantially lower under Vitellaria and Parkia trees than in the open (Jonsson, 1995). An almost leafless crown of F. albida resulted in a soil temperature decrease of up to 10°C at 2 cm depth (van den Beldt and Williams, 1992). In northern Senegal, air temperatures under and outside tree canopies differed by 6°C at maximum temperatures (Akpo and Grouzis, 1996). In addition, the variation of soil temperatures at 10 cm depth during the day was lower under canopies (3°C) than in full sunlight (9°C). Maximum and minimum air temperatures are moderated by tree crowns because of reduced solar radiation during daytime and reduced reflection of infrared radiation at night (Dancette and Poulain, 1969).
Temperature reduction has been held responsible for enhanced yields under Faidherbia crowns. There is some evidence that extreme heat negatively affects crop establishment and subsequent growth (Ong and Monteith, 1985; Peacock et al., 1990; Mclntyre et al., 1993). Using vertical artificial screens, van den Beldt and Williams (1992) at the International Crops Research Institute for the Semi-Arid Tropics (ICRISAT) showed that the effect of shade on soil temperatures contributed to better millet growth during seedling establishment. They argue that root damage due to high temperature rather than water deficits caused differences in millet performance, and that crops would not be able to take advantage of the greater soil fertility around F. albida trees without the moderated temperature associated with them. However, this study was conducted on-station where soils are possibly more fertile than at onfarm sites. Furthermore, it was done in May when temperatures are very high, whereas millet is normally sown in late June or early July in this area, and it was conducted for only six weeks, thus not encountering nutrient constraints.
A subsequent on-farm experiment in N'dounga, Niger, attempted to quantify the effect of soil fertility independent of other F. albida microclimate-related effects, with treatments including single and combined doses of nitrogen and phosphorus fertilizers (ICRAF, 1996). No effect of tree canopy or fertilizer treatments on seedling survival was visible 25 days after sowing, and the difference between F. albida and open sites in millet canopy temperature was only 1.8°C. The F. albida effect was made up of the effects of nitrogen and phosphorus which were the most limiting resources for millet growth in this study, while shade had no direct influence. (Nitrogen fertilization cancelled the difference between yields in the crown zone and the open.) Thus, temperature moderation by F. albida may not influence cereal stand establishment as much as was previously believed.
The potential positive effect of temperature reduction in species such as V. paradoxa and P. biglobosa is usually not visible, due to reduced yields blamed on PAR reduction and on increased humidity resulting from shade. Surprisingly, however, in Saponé, Burkina Faso, millet productivity under both canopies was relatively high (highly variable and not significantly different) compared with the open in a year of far higher rainfall than average, in spite of a 65–75 percent reduction of PAR (Jonsson, 1995). Jonsson thus assumed that the positive effects of soil temperature reduction due to tree cover offset the negative impact of shade on photosynthesis. She supported her findings by suggesting, first, that supra-optimal soil temperatures impede seed germination and seedling establishment (Peacock et al., 1993). There is also evidence that they significantly constrain subsequent growth and development (Mclntyre et al., 1993). Secondly, lower subcanopy temperatures caused a 15-day extension in crop duration, which is advantageous in years of extended rainfall. In turn, a 15-day reduction in crop duration would generate a 15 to 20 percent production decline (Squire, 1990, cited in Jonsson, 1995). The effect of high humidity on crop failure due to pest attacks, which was found in other studies, was not reported. Given the opposite results found in other studies on the same tree species, Jonsson suggested that the overall impact of microclimate moderation on crop production might depend on annual rainfall conditions for reasons not yet fully understood.
Diakité's (1995) results in the western part of the Middle Bani area of Mali also deviate from the yield depression generally observed under Vitellaria crowns. However, the overall average yield increase of 26 percent (Table 3.3) under and away from tree canopies hides substantial differences related to tree size, with sub-canopy yields in the larger tree size group (15.4 m diameter) appearing to be lower than yields in the open (despite the opposite pattern for the whole sample). This suggests that the reduction in solar radiation under larger canopies does indeed have a negative effect on sorghum production. Nevertheless, tree size in the smaller group is similar to that in other studies (Kessler, 1992; Kater et al., 1992) but gives opposite results. The dramatic increase in sorghum grain (80 percent) and straw yields (70 percent) in the vicinity of smaller trees (9.9 m in crown diameter) compared with plots around trees of larger crown diameter (15.4 m) in a year of low rainfall suggests a significant influence of the microclimate. Diakité attributed it to the higher interception and evaporative demand of the larger canopy size group.
A higher topsoil moisture under woody canopies than in treeless sites appears to be a common pattern during the rainy season and some time afterwards. For instance, soil moisture at 0–10 cm depth during the end of the rainy season was twice as high under F. albida canopies as in the open (1.4 versus 0.7 percent), probably due to lower evapotranspiration (Charreau and Vidal, 1965). A similar pattern was observed under pruned Faidherbia trees in Burkina Faso in the early and late season (Depommier, 1996a). Dancette and Poulain (1969) also found that soil moisture was higher under F. albida trees than in open controls in the top 120 cm, but was lower in deeper horizons (down to 4 m) due to water absorption by deep tree roots. (During the rainy season, water reaches deeper horizons under trees than in the open because of better infiltration (stemflow, soil texture) and reduced evapotranspiration.) Rhoades (1995) recorded increased soil water in the top 15 cm of soil beneath F. albida canopies in Malawi. The increase over open sites rose from 4 percent at the end of the dry season to 53 percent at the end of the cropping season.
Soil moisture in unspecified soil depth and period was significantly lower (4.7 versus 9.3 percent) in the open than under both Hyphaene thebaica and F. albida in Kareygorou and Say, Niger (Moussa, 1997). There was also a decrease in topsoil moisture with increasing distance from V. paradoxa trees at three dates toward the latter part of the rainy season in southern Burkina Faso (Boffa et al., 1999). In northern Cameroon, soil water available to cotton, as calculated from soil humidity measured by probes, seemed little affected by F. albida trees in the 0–80 cm soil horizon in which cotton roots were concentrated and where hardly any tree roots were detected. Moisture available to plants appeared higher under trees at the beginning and end of the season than in open field controls, but this difference was significant only for one site (Libert and Eyog-Matig, 1996).
The higher soil moisture in tree sites described above is generally assumed to be due to reduced soil evaporation and plant transpiration caused by shading and the resulting lower temperatures (Belsky et al., 1989). To illustrate, in a F. albida stand of 25–30 trees/ha, potential evapotranspiration was 50 percent lower during the dry season when trees were in full foliage and 10 percent lower during the rainy season than in a bare field (Schoch, 1966). In Bambey, Senegal, evapotranspiration was reduced from 2 200 mm in an open degraded field to 1 850 mm in F. albida windbreaks and 1 520 mm in a location sheltered with constructions and neem or Prosopis hedges. Over a period of two years, the lowest reduction in a field with scattered Faidherbia trees was 5 percent compared with the open field (Dancette and Niang, 1979). During two short periods in July in a season of exceptionally high rainfall in Burkina Faso, evaporation under Vitellaria and Parkia trees was 22 percent lower than in the open (Jonsson, 1995). Traditional rice fields located in bottomlands or on slopes, and surrounded by trees, showed a 40 percent reduction in evaporation compared with a modern treeless rice-growing area in Djibélor, Senegal (Niang, 1998).
Wind speed reduction in agroforestry parklands
|The role of trees in wind protection has been studied mainly in relation to linear shelterbelts. In a state-of-the-art paper on the effects of parklands on microclimate, Stigter et al. (1996) presented recent insights on air movement around shelterbelts and in canopies, and noted the lack of understanding of air movement and the lack of research on scattered trees in typical parkland situations.|
|Agroforestry parklands appear to reduce wind velocity in certain conditions. In Burkina Faso, Jonsson (1995) measured a significant difference in wind speed under small and large Vitellaria and Parkia trees compared with the open. Wind velocity reduction was held responsible for a 15–20 percent decrease in overall water consumption in Faidherbia albida stands as compared with bare land (Dancette, 1966). In Tanzania, wind speed patterns were investigated along lines parallel to wind flow into a woodland made up of scattered trees, mostly Acacia tortilis and a few other species such as Acacia mbuluensis, Commiphora schimpere and Balanites glabra of 5.5 m average height (Kainkwa and Stigter, 1994). Wind speeds decreased approximately linearly into the woodland until they became almost constant (saturation reduction). The minimum length of woodland (estimated density of 120 trees/ha) for a maximum of 50 percent wind reduction was about 110 m at 1.0 m height and about 80 m at 2.5 m height. Tunnelling effects under tree canopies seem to cause lower wind speed reductions at 1.0 m than at 2.5 m. With a lower tree density of about 60 trees/ha, significant wind reduction was found only at 2.5 m. Therefore, scattered trees in parklands, much like shelterbelts, can reduce wind speed significantly at given densities (especially 120–150 trees/ha). Furthermore, once the saturation point in wind reduction is reached, small gaps in tree cover are likely to have little influence because air affecting the woodland decouples from the main winds above the system. Likewise, wind reduction will be more pronounced in large expanses than in isolated patches of parklands. Vertical and horizontal tree biomass distribution (including stem height) and species composition influence wind reduction patterns, as demonstrated around single trees (Gross, 1987; Stigter et al., 1996).|
|Taking into account the paucity of data available, it nevertheless seems that the minimum densities at which sufficient wind reduction occurs are higher than most parkland densities observed in the region. Lower densities of higher and larger trees, as often found in Sahelian parklands, may be similarly effective, however. Therefore, further studies focusing on parkland examples, and aimed at identifying optimum characteristics (canopy management, tree density, size and species combinations), are desirable.|
Mechanisms involved in evaporation are similar on plant leaves and on a water surface. Evaporation is proportional to wind speed and the vapour pressure gradient between the evaporating surface and air. Wind reduction by parkland trees (Box 3.5), therefore also contributes to decreased evapotranspiration, even though few data are available to prove this relationship. The additional link between wind reduction at the scale of a spatially defined population of parkland trees, rather than at the scale of an individual tree, and crop productivity, could be the primary cause of the ‘parkland effect’ which was discussed earlier.
Differences in soil humidity recorded under trees and in the open vary according to season. At the onset of rains, especially light ones, sub-canopy soil moisture is usually lower under savanna trees than in the open because of at least partial canopy interception and subsequent evaporation (Box 3.6). A lower sub-canopy moisture may continue during the first part of the rainy season in northern and drier latitudes, as the tree and herb layers together may utilize more water than a sole grass or crop component. Effective evapotranspiration of an Acacia tortilis and Balanites aegyptiaca grassland community measured over a soil depth of 475 cm was higher than for a solely grassland area monitored in the top 60 cm (Nizinski and Grouzis, 1991).
Rainfall interception by parkland trees
|The amount of rainfall intercepted by trees varies according to rainfall intensity. In Senegal, a lower rainfall was recorded under Faidherbia albida trees during light rains because of canopy interception and subsequent evaporation, while there was a sub-canopy rainfall surplus during heavy rain storms to the detriment of the area leeward of the trees. Over a month and a half of rainfall measurements, the balance was a 9 percent surplus under F. albida trees compared with the open, but this did not include stemflow (Dancette and Poulain, 1969). In Cameroon, there was 9 and 6 percent less rainfall in pluviometers located under and at the edge of Faidherbia canopies, respectively, than in open controls over the whole season (Libert and Eyog-Matig, 1996). The fact that only one (unspecified) transect orientation per sample tree was monitored may explain the difference between these and the Senegal findings. Cordyla pinnata canopies in Senegal intercepted 22 percent of gross rainfall, and interception declined with increasing gross precipitation. The storage capacity of C. pinnata canopies was 5 mm (Samba, 1997).|
Later during the rainy season, tree sites tend to be more humid than open locations, because of reduced soil evaporation and plant transpiration due to lower temperatures and reduced radiant energy (Akpo, 1993). The pattern of higher moisture under trees during the middle of the season does not always apply. The distance from Vitellaria trees did not affect soil water content in Mounzou, Mali (Diakité, 1995). Under F. albida in Burkina Faso, differences in soil moisture content subsided in the mid-season as rainfall was abundant and leafless trees caused little shade and rainfall interception. Topography may also influence the tree-related variability of moisture content. Thus no difference was noted in lower toposequence positions, which are rapidly saturated with rain water, during the agricultural season (Depommier, 1996a). The higher plant productivity, due to better soil fertility under tree crowns in savannas, and the associated rise in evapotranspiration may reduce the sub-canopy moisture surplus and lead to similar moisture content in surface soils (Belsky et al., 1989). As the dry season proceeds the topsoil gradually dries out uniformly.
One would expect greater declines in temperature under the total area of large, dense and spreading canopies than under small, light and erect tree crowns. These would also imply lower soil evaporation and crop transpiration as well as increased soil humidity. In Malawi, soil moisture was not higher under small F. albida trees than in the open, as their leafless canopies did not create enough shade to reduce evapotranspiration. However, surface soil moisture was consistently higher under large crowns throughout the growing season (Rhoades, 1995). Similar findings result from comparing trees with different crown sizes due to pruning. In the Bulkiemdé parklands in Burkina Faso, for example, Zoungrana et al. (1993) measured a more rapid decrease (5 versus 8.6 percent) and lower content of soil moisture (10 versus 12 percent) under Azadirachta indica trees (crown radius £3 m) pruned two years earlier than under unpruned ones (radius ≥ 3 m) over a 12-day period during which the understorey crop was in the flowering stage (statistical significance was not tested). It should also be noted that differing tree shape affects the way species influence microclimate and soil water. In northern Senegal, the higher evapotranspiration in A. tortilis than B. aegyptiaca stands was attributed either to the higher interception of the former's spreading crown or to its higher absorption/transpiration (Nizinski and Grouzis, 1991). Additional research into water use of parkland trees in managed conditions is desirable.
In contrast to the above, larger canopies may also reduce soil moisture by generating more evapotranspiration due to their more extensive exchange surface between foliage and air. This is especially true towards the late part of the rainy season when rainfall events become less intense and frequent and temperature rises. In Mounzou, Mali, for example, Diakité (1995) reported that soil moisture under small V. paradoxa canopies (9.9 m diameter) was significantly higher than under larger crowns (15.4 m diameter) in September (0.144 versus 0.131 g water/g of dry soil) and October (0.0824 versus 0.0644 g).
The apparently conflicting results presented here may be reconciled by acknowledging that the process of reduced crop transpiration and soil evaporation under larger crowns may be generally valid, while the mechanism of greater evapotranspiration in canopies may apply primarily in the latter part of the rainy season and be felt more intensely under large tree crowns. As microenvironmental benefits of trees are conferred to the system in the long term, a finer analysis along a gradient of tree size would be useful to indicate when these benefits may arise.
Air humidity, which results from lower air temperatures and higher top soil moisture under trees, is usually higher under tree canopies than in the open (Breman and Kessler, 1995). Relative humidity in July was 59 percent under F. albida trees and 51 percent away from them in Senegal (Dancette and Poulain, 1969).
Apart from intercepting solar radiation, canopies may also reduce crop viability through increased humidity. Kater et al. (1992) observed that cereal crop survival was lower in the proximity of trees than in the open and attributed this to higher air and topsoil humidity favouring fungal infection. Crop establishment failure made a relatively greater contribution to the decline in overall grain productivity for sorghum and millet, under both Vitellaria and Parkia trees, than for cotton under P. biglobosa. Rather than pruning treatments, Kater and colleagues recommended crop mixtures adapted to higher moisture conditions under these trees. Only 8 percent of plants sowed under V. paradoxa failed to reach maturity in southern Burkina Faso (Boffa et al., 1999). Kessler (1992) reported that crop maturity under trees was retarded and that about 5 percent of sorghum panicles under trees were infected with a virus, Spacelotheca sorghi. Accumulated water was observed in flag leaves of sorghum long after rainy events under Vitellaria canopies, suggesting high air humidity, and possibly increased susceptibility to fungal attack (Boffa et al., 1999).
There are few references to understorey species composition in agroforestry parklands because wild herbaceous species are replaced by agricultural species. Nevertheless, research in Watinoma and Dossi, Burkina Faso, has shown that F. albida significantly reduces the density of the plant parasite, Striga hermontica, which causes considerable yield loss in degraded soils. The extension of shade under pruned trees during the agricultural season seems to be detrimental to the species which requires high solar intensity for its development. Besides higher weed growth under trees, the floristic diversity in weeded plants growing under F. albida crowns tended to be lower than in the open in Cameroon (Libert and Eyog-Matig, 1996). More palatable and shade-loving grasses (Panicum maximum and Brachiaria spp.) were dominant under P. biglobosa canopies, while less palatable Andropogon and Imperata spp. prevailed outside canopies (Sabiiti and Cobbina, 1992).
Vegetative development of herbaceous plants is generally longer in plants growing under woody canopies (Weltzin and Coughenour, 1990). Weed growth tended to start earlier under F. albida canopies than in open controls (Libert and Eyog-Matig, 1996). Duration of the vegetative plant development was longer in cotton in the proximity of F. albida (Libert and Eyog-Matig, 1996), as well as in sorghum under Vitellaria and Parkia canopies (Kessler, 1992).
In the Acacia caven-based silvipastoral system, the vegetative development cycle of plants growing in the shade of these trees was 25 to 35 days longer than in the open (Ovalle and Avendano, 1987; Akpo and Grouzis, 1996). The proportion of species displaying a short vegetative cycle was also higher in stands where tree cover had been eliminated than in stands with an A. caven cover of 80 percent. Plant emergence took place two weeks earlier in tree-adapted species than in plants found outside, while flowering stages started earlier, lasted a shorter time and had a lower success rate in the open than under tree canopies (Akpo and Grouzis, 1993).
The interactions between different biophysical factors in determining the productivity and sustainability of parkland systems are very complex. An overall understanding is still lacking because most studies to date have focused on the various interactions in isolation and have also extrapolated from experiments on individual trees to the parkland level.
The effect of parklands on soil fertility is of particular interest. Soils under tree canopies, in both savannas and parklands, generally display a higher fertility than those in the open. This pattern of higher fertility under tree crowns is usually characterized by a gradual decline with distance away from the trunk as well as with soil depth. Nutrient enrichment by trees also appears to increase with tree size. Studies of the relationship between soil fertility and tree density have so far been inconclusive. Comparative increases in nutrient content are highly significant under Faidherbia albida trees, and less remarkable though still common in other parkland species. Cardinal orientation may also affect nutrient concentrations in the subcanopy zone, with differential organic matter accumulation due to wind or asymmetrical crowns.
Several mechanisms may account for the increased fertility under trees. Nutrients are returned to the soil through deposition of litter, root decay and exudation, as well as the leaching of tree nutrients in rainfall. Biological processes appear to be particularly important with tree sites being characterized by higher macrofaunal and microbial activity, as well as higher mineralization rates, lower bulk density and better water infiltration rates than treeless locations. A few studies report that subcanopy soils are associated with a larger proportion of fine-textured elements than in the open, which may be due to limited soil erosion, increased termite activity or pre-existing soil differences. Interception of wind-blown material may be another important yet seldom measured factor.
Nitrogen fixation increases soil nitrogen content but appears relatively limited in adult tree populations and constrained by availability of phosphorus in the Sahel. Nutrients are concentrated under trees by livestock looking for shade and fodder, as well as by wild fauna. The few quantified measures indicate that livestock droppings probably have a significant responsibility in enhanced crop performance under trees in F. albida parklands. Fertility distinctions between tree sites and interspaces may also be related to pre-existing fertility or different soil management practices.
Where trees are associated with increased soil fertility, the question remains as to whether this is due simply to a redistribution of nutrients or whether trees engender a process of nutrient enrichment. Trees can contribute to nutrient redistribution by lateral root uptake and animal and atmospheric deposition. Alternatively, they may increase nutrient availability in the system by reducing losses through leaching and soil erosion and by adding nutrients through N-fixation, with nutrients absorbed by deep roots, and mycorrhizal infection. Although these enrichment processes have been documented they may not be widespread. Thus deep rooting is frequently limited by root barriers in semi-arid West Africa and its nutrient contribution is low, due to the low nutrient content of deep soil layers.
The effect of parkland trees on soil fertility is just one component of their influence on crop productivity. The vertical distribution of tree and crop roots suggests that competition and facilitation both occur in agroforestry parkland systems. Competition patterns are linked to tree root distribution, which often extends far beyond the crown area in arid conditions, while it is confined within or close to the crown area in subhumid zones. Available data suggest that F. albida's underground competition with crops is small because of its deep taproot and reverse foliation (although pruning may cause it to refoliate during the rainy season). In contrast, competition for nutrients may be higher for parkland tree species with a similar phenology to crops. Even so, its depressive influence may be less significant than that of shade in these species.
In cereals, F. albida is responsible for a substantial increase of grain yields under its canopy, often equivalent to the yield improvements expected from fertilizer additions. The F. albida effect is more pronounced on soils of low fertility, in years of below-average rainfall, and in the absence of fertilizers. It is also positively related to tree diameter, but can be obscured by pruning. The distribution of yields within the canopy zone usually decreases regularly from the bole to the edge of the crown or peaks within the second half of this zone. In the case of Vitellaria paradoxa and Parkia biglobosa, crop performance may be significantly reduced, but there is also some evidence to the contrary. Neem (Azadirachta indica) depresses crop yields to a lesser extent while Hyphaene thebaica stimulates cereal crops, probably because of its high, small crown and limited horizontal root extension.
With the possible exception of Prosopis africana, yield depression is more pronounced on soils of low fertility. C3 crops are generally less affected by trees than C4 plants. Tree species with contrasting branch architecture and rooting patterns, as illustrated by P. biglobosa and V. paradoxa, differ in underground and above-ground competition patterns with crops, making a reduction of crown area and/or limiting root competition desirable. Tree size is generally positively correlated with production gains under canopies of F. albida but there is no consensus on its effect in species with typical phenology.
The effect of trees on crop production may be accounted for in part by changes in microclimate. Trees generally intercept 20 to 80 percent of incident solar radiation depending on size and species. Even when leafless, savanna and parkland canopies are associated with lower temperatures (and evapotranspiration) than in the open because of reduced solar radiation during the day and reduced reflection of infrared rediation at night. Studies in savannas suggest that tree shade thus induces higher understorey productivity. However, although crop establishment and growth are impeded by extreme heat, there is no conclusive evidence of shade being responsible for better yields under Faidherbia trees, than in the open. The positive temperature effect of shade in species such as V. paradoxa and P. biglobosa is usually not apparent in the overall reduced yields, but has been said to offset the decline in photosynthesis in a year of high rainfall in Burkina Faso. The magnitude of the negative shade effect rises with increasing canopy size and is more prevalent with C4 than C3 crops. The specific contribution of nutrient enrichment and microclimate will need to be clarified through experiments with artificial shade, fertilizers and their combination under and away from parkland trees and over several years.
Moisture in the topsoil is generally higher under woody canopies during the rainy season because of reduced soil evaporation and plant transpiration as a result of shading and lower temperatures. This pattern may not apply at the beginning of the season, because of interception of rainwater, as well as later due to higher evapotranspiration of increased understorey plant biomass, and/or absence of leaves in pruned Faidherbia trees. Wind speed reduction probably contributes to reduced evapotranspiration and crop productivity, but this link has not yet been established. The higher air humidity under Vitellaria and Parkia canopies appears to be conducive to pest attacks and crop failure, while F. albida shade is associated with higher weed growth, lower weed diversity and a significantly lower density of the plant parasite Striga hermontica than in the open. The vegetative development of crops generally takes longer under than outside of woody canopies.
Relationships between parkland density and crop production are complex and available information is limited because of logistical difficulties. A concept of critical canopy cover has been proposed. Below this critical level, herbage production is no longer increased with further tree cover reductions while, above this level, herbage production decreases due to tree-herb competition. There may also be a positive ‘parkland effect’ on crop production linked to the spatial arrangement of scattered parkland trees, which would not exist in the presence of isolated individual trees, consisting of a synergetic increase in soil and air moisture, as well as less air circulation. However, no scientific evidence of this effect has yet been provided.