As discussed in Chapter 1, the process of CS, or flux of C into soils forms part of the global carbon cycle. Movement of C between the soil and the atmosphere is bidirectional. Consequently, carbon storage in soils reflects the balance between the opposing processes of accumulation and loss. This reservoir of soil C is truly dynamic. Not only is C continually entering and leaving the soil, the soil C itself is partitioned between several pools, the residence times of which span several orders of magnitude. Nor is soil C an inert reservoir, the organic matter with which it is associated is vital for maintaining soil fertility and it plays a part in such varied phenomena as nutrient cycling and gaseous emissions. A detailed description and analysis of soil C and organic matter can be found elsewhere (Schnitzer, 1991; FAO, 2001c). Taking into account specific biophysical characteristics of dry areas, this chapter describes different biophysical aspects of CS in dryland soils.
A special feature of many dryland soils is salinity, either through natural occurrence or increasingly as a result of irrigation. Saline soils affect large parts of the drylands (Glenn et al., 1993). Such lands are often abandoned, but halophytic plants are especially adapted to these conditions and offer potential for sequestering C in this inhospitable environment. It has been estimated that 130 million ha are suitable for growing halophytes, which can be used for forage, feed and oilseed. Glenn et al. estimate that 0.6 - 1.2 gigatonnes of C per year could be assimilated annually by halophytes. Evidence from decomposition experiments suggests that 30 - 50 percent of this C might enter long-term storage in soil. Although irrigation would be required to achieve these figures, a complete carbon budget still suggests a carbon actual rate of 22 - 30 percent.
Grasslands are the natural biome in many drylands, partly because rainfall is insufficient to support trees, and partly because of prevailing livestock management. However, grassland productivity and CS have been controversial. The productivity of tropical grasslands is now known to be much higher than was previously thought, and consequently they sequester much more C (Scurlock and Hall, 1998). Estimates for C stored under grassland are about 70 tonnes/ha, which is comparable with values for forest soils. Although many of the grassland areas in drylands are poorly managed and degraded, they offer potential for CS as a consequence.
The average annual input of organic matter into grassland is about double the 1 - 2 tonnes/ha that is contributed to cropped soils (Jenkinson and Rayner, 1977). This fact is borne out by the results of studies from various locations. The data have shown that grassland, even where subject to controlled grazing, generally has higher soil C levels than cropland. Chan and Bowman (1995) found that 50 years of cropping soils in semiarid New South Wales, Australia, had on average reduced soil C by 32 percent relative to pasture. The reduction was linearly related to the number of years of cropping. Similarly, soils of tall-grass pasture under controlled grazing had greater soil C than adjacent cropland subject to conservation tillage (Franzluebbers et al., 2000).
The key factor responsible for enhanced carbon storage in grassland sites is the high carbon input derived from plant roots. It is this high root production that provides the potential to increase SOM in pastures and vegetated fallows compared with cropped systems. Root debris tends to be less decomposable than shoot material because of its higher lignin content (Woomer et al., 1994). Consequently, the key to maintaining and increasing CS in grassland systems is to maximize grass productivity and root inputs (Trumbmore et al., 1995). Grasses have also been shown to sequester more C than leguminous cover crops (Lal, Hassan and Dumanski, 1999). Grasses also have the potential to sequester C on previously degraded land. Garten and Wullschleger (2000) used a modelling approach that estimated a 12-percent increase in soil C could be obtained under switchgrass (Panicum virgatum L.) on degraded land in ten years.
Grazing is a feature of many grasslands, natural or managed. This might be expected to decrease the availability of residues that can be used to sequester C, especially as the quantity of C returned in manure is less than that consumed. However, provided there is careful grazing management, many investigations have found a positive effect of grazing on the stock of soil C. This was found to be the case for a composite pasture (alfalfa and perennial grasses) in the semi-arid pampas (Diaz-Zorita, Duarte and Grove, 2002). Even under harsher conditions in the Syrian Arab Republic, grazing was found to have no detrimental impact on soil C (Jenkinson et al., 1999). Schuman, Janzen and Herrick, (2002) have calculated that with proper grazing management, rangelands in the United States of America can increase soil carbon storage by 0.1 - 0.3 tonnes/ha/ year. For new grasslands, this can rise to 0.6 tonnes/ha/year.
The positive effect of grazing appears to result from the effect that it has on species composition and litter accumulation. Willms et al. (2002) found that when prairie was protected from grazing there was little effect on production but there was an increase in the quantity litter. Reeder and Schuman (2002) also found that there was an accumulation of litter in an ungrazed semi-arid system and that soil carbon levels were higher in the grazed lands. The litter acted as a store of immobilized C. The ungrazed grassland also experienced an increase in species that lacked the fibrous rooting system that is conducive to SOM formation and accumulation.
Therefore, grasslands can play a vital role in sequestering C. However, careful grazing management is essential. The historical record shows how susceptible semiarid grasslands are to overgrazing, soil degradation and carbon loss.
Fires form part of the natural cycle in many biomes and are especially prevalent in grassland ecosystems. However, humans have also used fire to clear areas for agriculture and to clear crop residues. The action of fire would seem to be counter to CS as it returns C fixed by vegetation directly to the atmosphere, thereby preventing its incorporation into the soil. The effect of fire is difficult to generalize because it depends upon the intensity and speed of the fire. These factors are influenced by: the state of the vegetation, i.e. its maturity and woody component; the accumulation of litter; and climate factors such as moisture level. They C in all the aboveground material that is fully combusted will be lost from the system. However, in grassland ecosystems, C lost to fires can be replaced quickly by increased photosynthesis and vegetative growth (Knapp, 1985; Svejcar and Browning, 1988). Even in savannah systems that contain woody species, it has been shown that C lost through combustion can be replaced during the following growing season (Ansley et al., 2002). Regarding the soil, the intensity and speed of the fire will govern the depth to which it is affected. In one study where burning was used to clear forests, 4 tonnes C/ha was lost in the top 3 cm of soil, but this was replaced within one year under a pasture system (Chone et al., 1991).
Not all plant material is fully combusted by fire, and a variable amount of charcoal will be produced. Charcoal is extremely resistant to decomposition; it is not cycled like most organic matter, and has a mean residence time of 10 000 years (Swift, 2001). Consequently, in severely degraded soils that have been affected by several fires, charcoal and charred material can form a substantial proportion of the remaining organic C. However, it is not known whether charcoal and other charred material have any protective effect on the native organic matter. Thus, although fires release CO2 back into the atmosphere, the simultaneous production of charcoal can be regarded as a sequestration process that results in a substantial amount of C being accumulated within the soil for a long period.
Forestry is recognized as major sink for C. However, as well as accumulating C aboveground, it can also make significant contributions to soil C even in drylands. A number of suitable species are available that enable viable afforestation in dryland environments (Srivastava et al., 1993; Silver, Ostertag and Lugo, 2000; Kumar et al., 2001; Niles, et al., 2002). In particular, N-fixing trees generally lead to increased accumulation of soil C. For example, Prosopis and Acacia are adapted to subtropical semi-arid lands and are reported to increase the level of soil C by 2 tonnes/ha (Geesing, Felker and Bingham, 2000).
Some tree varieties are particularly suitable for growing on degraded lands, their deep rooting systems tapping resources unavailable to shallow-rooted crops. Prosopis juliflora has been grown on salt-affected soils in northwest India and increased the SOC pool from 10 tonnes/ha to 45 tonnes/ha in an eight-year period (Garg, 1998). Even where large-scale forestry is not appropriate, there is often scope for introducing trees around farmers fields. Such is the case in semi-arid India where Prosopis cineraria has improved soil fertility as well as sequestering extra C (Nagarajan and Sundaramoorthy, 2000). However, natural systems are complex and trees do not guarantee improved CS. Jackson et al. (2002) found that soil C decreased when woody vegetation invaded grasslands. Although there was an increase in aboveground and belowground biomass, these gains were offset by losses in soil C.
Plant residues provide a renewable resource for incorporation into the SOM. Production of plant residues in an ecosystem at steady-state will be balanced by the return of dead plant material to the soil. In a native prairie, about 40 percent of plant production is accumulated in the SOM (Batjes and Sombroek, 1997). However, in agricultural systems, because plants are harvested, only about 20 percent of production will on average be accumulated into the soil organic fraction. Furthermore, in some farming systems, all aboveground production may be harvested, leaving only the root biomass. Of the plant residue returned to the soil, about 15 percent will be converted to passive SOC (Lal, 1997). Schlesinger (1990) is more pessimistic, suggesting that only 1 percent of plant production will contribute to CS in soil. The actual quantities of residue returned to the soil will depend on the crop, the growing conditions and the agricultural practices. For example, for a soybean - wheat system in subtropical central India, the annual contribution of C from aboveground biomass was about 22 percent for soybean and 32 percent for wheat (Kundu et al., 2001). This resulted in 18 percent of the annual gross carbon input being incorporated into the SOM. In semi-arid Canada, the conversion of residue C to SOC was reported to be 9 percent in frequently fallowed systems, increasing to 29 percent for continuously cropped systems (Campbell et al., 2000).
Unless a root crop is being harvested, all belowground production is available for incorporation into the SOM. Roots are believed to be the major constituent of particulate organic matter although tillage reduces the net accumulation of C from roots substantially (Hussain, Olsson and Ebelhar, 1999). In cool climates, belowground carbon inputs from roots alone can generally maintain soil carbon levels. However, this is not the case in warmer or semi-arid regions where residues are decomposed much more readily, providing sufficient moisture is available (Rasmussen, Albrecht and Smiley, 1998). Consequently, when continuous cropping is practised in drylands, failure to return aboveground plant residue will lead invariably to a reduction in soil C. Many African soils demonstrate this phenomenon. Continuous cropping over a number of years without the recommended inputs has often halved their carbon content (Woomer et al., 1997; Ringius, 2002).
Both the quality and quantity of plant residues are important factors for determining the amount of C stored in soil. The quantity is highly dependent on the environmental conditions and agricultural practices. Differences between crops can be marked. A crop of maize will return nearly twice as much residue to the soil compared with soybean and, consequently, will result in a higher rate of SOM increase (Reicosky, 1997). The advantage that cereals have over legumes for achieving maximum CS rates has also been demonstrated by Curtin et al. (2000). They have shown that while black lentil fallow in semi-arid Canada added between 1.4 and 1.8 tonnes C/ha, a wheat crop would add 2 - 3 times this amount of C annually. Similarly, in Argentina, soybean, which produced 1.2 tonnes/ha of residue, resulted in a net loss of soil C, while maize, with 3.0 tonnes/ ha of residue, lessened the loss of soil C from the system significantly (Studdert and Echeverria, 2000).
Even within one crop group, large differences in organic matter production occur. Abdurahman et al. (1998) compared dry leaf production from pigeon pea and cowpea. While the former yielded 3 tonnes/ha, cowpea produced 0.14 tonnes/ha. These examples illustrate how the choice of crop can have a major influence on how much C an agricultural system can sequester.
The importance of roots relative to shoots for providing soil C is another factor illustrated in an experiment to compare the fate of shoot- and root-derived C (Puget and Drinkwater, 2001). In this study with leguminous green manure (hairy vetch), nearly half of the root-derived C was still present in the soil after one growing season whereas only 13 percent of shoot-derived C remained. This implies that shoot residues are broken down rapidly on account of their higher N content (Woomer et al., 1994) and may serve as a nitrogen source for the following crop.
The chemical composition of plant residues affects their rate of decomposition. On average, crop residues contain about 40 - 50 percent of C but N is a much more variable component. A high concentration of lignin and other structural carbohydrates together with a high C:N ratio will decrease the rate of decomposition. For example, measurement of CO2 evolution from tree leaves of African browse species and goat manure showed a significant correlation with initial nitrogen content and a negative correlation with lignin content (Mafongoya, Barak and Reed, 2000). Legume residues such as soybean are generally of high quality (low C:N ratio) and so decompose rapidly (Woomer et al., 1994). Although the chemical composition of the plant residue affects its rate of decomposition, there is little effect on the resulting SOM (Gregorich et al., 1998).
Where residues accumulate on the soil surface, their physical presence affects the soil. Mulches reduce water loss and soil temperature (Duiker and Lal, 2000), both important factors for drylands, especially where the soil temperature is above the optimum for plant growth. The ability of soil to assimilate organic matter is not clear-cut. A linear relationship between application and accumulation of SOM is often quoted. However, measurement of CO2 flux in central Ohio, the United States of America, showed this flux to increase with added application of wheat straw (0, 8 and 16 tonnes/ha). However, the SOC was 19.6, 25.6 and 26.5 tonnes/ha after four years, suggesting that CS was reaching saturation (Jacinthe, Lal and Kimble, 2002).
Care must be taken when applying residues, as large losses of C can still occur under certain conditions. For example, in western Kenya, 70 - 90 percent of the added C was lost within 40 d when green manure from agroforestry trees was applied during the rainy season (Nyberg et al., 2002). In Niger, the addition of millet residue and fertilizer for five years had no significant effect on the carbon level in the sandy soils (Geiger, Manu and Bationo, 1992). Termite activity may have contributed to the low levels of soil C here because entire surface mulches can be consumed within one year.
The potential amount of residues available for applying to soils can be large. Gaur (1992) estimated that in India about 235 million tonnes of straw/stover are produced annually from five major cereals (not only from drylands). Even if half of this were used for feeding livestock, there would be more than 1 million tonnes available for adding to the soil. However, availability of sufficient plant residues is often a problem, especially where they are required for livestock feed. This conflict of requirement occurs frequently in many dryland farming systems. In West Africa, crop residues are either removed or burnt. Consequently, the amount of soil C has declined steadily. Continuous cultivation and manure application can raise soil carbon levels by 40 percent but this frequently involves mining C from neighbouring areas to support the livestock (Ringius, 2002). Where plant and animal residues are in short supply, possibilities may exist for alternative organic inputs to the soil. For example, in India, waste products from a plant processing coir dust have been incorporated successfully into soil (Selvaraju et al., 1999), and other experiments with industrial glue waste (Dahiya, Malik and Jhorar, 2001) have increased soil carbon levels successfully.
With regard to the complete carbon budget, where plant residues accumulate in situ, there is no extra carbon cost involved. Consequently, the C fixed by plants in photosynthesis is available as a net gain to the soil. The situation becomes more complex where machinery is required to separate the residue from the harvestable components. Where plant residues are transported between fields the energy cost would need to be included.
Perhaps the major issue regarding residue application and carbon accounting is the question of whether C is simply being transferred from one place to another rather than being truly sequestered. If organic material from industrial processes or other sources were to be used for incorporation into the soil, then any C used in transport would have to be accounted for. However, where the material is truly a waste product, there should be no need to consider the C used in its production. In this case, the C available for sequestration should be viewed no differently from the CO2 emitted from a fossil fuel source that is subsequently fixed by plants in photosynthesis and returned to the soil via crop residues.
The application of FYM has long been treated as a valuable source of organic matter to enhance soil fertility. One of the key characteristics of manure application is that it promotes the formation and stabilization of soil macroaggregates (Whalen and Chang, 2002) and particulate organic matter (Kapkiyai et al., 1999). Manure is more resistant to microbial decomposition than plant residues are. Consequently, for the same carbon input, carbon storage is higher with manure application than with plant residues (Jenkinson, 1990; Feng and Li, 2001). Following five years of application, soil receiving manure had 1.18 tonnes/ha more C present than soil receiving plant residues. Even after 15 years, there was a still a difference of 0.37 tonnes C/ha as calculated by the RothC soil carbon model. In the field, Gregorich et al. (1998) found that manured soils had large quantities of soluble C with a slower turnover rate than in control or fertilized plots.
The composition and, therefore, decomposition of manure varies between the species from which it originates and also within species according to their diet (Somda and Powell, 1998). Many field trials have found that manure is the best means for incorporating organic matter into soils and promoting carbon storage. For example, Li et al. (1994) found that manure yielded the largest amount of C sequestered over a range of soils and climate conditions, although soil texture was important, and the greatest rate of sequestration occurred where there was a high clay content. However, many traditional agrosystems add manure in combination with fertilizers (Haynes and Naidu, 1998).
Depending on the system, the application of even relatively high amounts of FYM does not guarantee an increase in soil C. A long-term study in Kenya has shown that SOM declined even when manure was applied and maize residue returned (Kapkiyai et al., 1999). It has been estimated that in order to maintain soil C in this system, 35 tonnes/ha of manure or 17 tonnes/ha manure with 16 tonnes/ha of stover would be required annually (Woomer et al., 1997). Consequently, there is a carbon shortfall in this case, but it is difficult to see how the present system could remedy it. In addition, high application rates of manure can sometimes cause problems in the soil through the accumulation of K+, Na+ and NH4+ and the production of water-repellent substances by decomposer fungi (Haynes and Naidu, 1998). An additional problem in drylands that restricts the quantity of manure that can be applied is burning of the crop when insufficient moisture is available at the time of application. Consequently, farmers often wait until the rains have come before making an application, especially as precipitation is often erratic in arid regions. is often erratic in arid regions.
The production of sufficient manure for application to fields is a real problem for many smallholder farming systems. In Nigeria, direct manure input onto land from dry-season grazing is about 111 kg/ha of dry matter (Powell, 1986). This quantity will have little effect on the soil. More useful is the practice of night-parking cattle as manure production is usually greatest at dawn and dusk. When 50 cattle were penned in an area of 0.04 ha for five nights, they produced the equivalent of 6.875 tonnes/ha of manure (Harris, 2000). Normally, cattle will be penned in fields for 2 - 3 nights in northern Nigeria and this can supply manure at a rate of 5.5 tonnes/ha. Alternatively, in densely populated parts such as the Kano closed-settlement zone, cattle and crop production are fully integrated. Cattle are kept permanently in pens and fed using feed grown on neighbouring fields. Their manure is collected and spread onto the croplands. Although an efficient system, some C will be lost as a consequence of the respiratory and growth requirements of the cattle. An additional problem associated with cattle rearing is that ruminants produce significant quantities of CH4, which is a potent GHG.
There has been some debate as to the usefulness of animal manure in CS. Schlesinger (1999, 2000) has calculated that providing 13.4 tonnes/ha would require 3 ha of cropland in order to produce sufficient cattle feed. This means that manure production requires mining of C from neighbouring lands. Although this is a generalization, it is argued that the 3:1 differential makes it unlikely that manure production per se could be used as a means of providing a net carbon sink in soils. However, in many dryland smallholder cropping systems, manure application rates are much lower, and fodder production is also less efficient. Smith and Powlson (2000) have made the case that keeping cattle is part of many agricultural systems and, hence, that manure should be considered as a byproduct that can be added to arable land without necessarily including the carbon cost of its production. Part of the disagreement with Schlesinger on the usefulness of manure depends on where the system boundary for carbon accounting is drawn. However, when conducting such a carbon audit, it is essential to remember that the purpose of agriculture is to feed people; offsetting GHGs can only be evaluated as a secondary activity.
Fertilization and irrigation are primary means for increasing plant production and crop yield. Any increase in biomass also offers increased scope for CS. Consequently, irrigation and fertilization have been recommended as, and proved to be, successful methods of increasing CS (Lal, Hassan and Dumanski, 1999). Rasmussen and Rohde (1988) have shown a direct linear relationship between long-term nitrogen addition and accumulation of organic C in some semi-arid soils in Oregon, the United States of America. However, these technologies provide no additional organic matter themselves but do carry a carbon cost. Schlesinger (1999, 2000) has pointed out that pumping water requires energy and that the process of fertilizer manufacture, storage and transport is energy intensive. Consequently, Schlesinger (2000) has estimated that the gains in C stored using either fertilization or irrigation are offset by losses elsewhere in the system. Irrigation can also lead to the release of inorganic C from the soil.
Izaurralde, McGill and Rosenberg (2000) have argued that the calculations used by Schlesinger are based on very high rates of fertilizer application. For many dryland agriculture systems in the developing world, farmers do not have sufficient funds to apply large quantities of fertilizer even if they were available. With regard to the energy costs incurred in pumping water, solar-powered systems are being developed (Sinha et al., 2002) and dryland environments with frequently clear skies would be able to make the best use of solar radiation.
These examples serve to illustrate how important it is to consider the whole system when CS is being considered to offset CO2 emissions. The exact carbon cost of irrigation and fertilization would require calculation for each system, but the carbon deficits associated with both technologies make their incorporation into net CS systems difficult to achieve. Water conservation, the growing of legumes and careful nutrient cycling are more likely to yield a positive carbon balance.
Pretty et al. (2002) consider tillage to be one of the major factors responsible for decreasing carbon stocks in agricultural soils. Research and experimentation with reduced tillage practices are most prevalent in the Americas. The mould-board plough and disc harrow are believed to be the causes of the loss of soil C through the destruction of soil aggregates and the acceleration of decomposition by the mixing of plant residues, oxygen and microbial biomass. Soil aggregates are vital for CS (Six, Elliott and Paustian, 2000), a process that is maximal at intermediate aggregate turnover (Plante and McGill, 2002). Of the organic matter fraction, the particulate organic matter is the most tillage sensitive (Hussain, Olsson and Ebelhar, 1999).
It is difficult to quantify the effects of tillage on soil C because the effect are very site dependent, e.g. coarse-textured soils are likely to be more affected by cultivation than are fine ones (Buschiazzo et al., 2001). However, reducing tillage should be most effective in hot, dry environments (Batjes and Sombroek, 1997).
Reicosky (1997) conducted an experiment that used measurements of CO2 efflux to investigate tillage-induced carbon loss from soil. The flux of CO2 was monitored for 19 d following different forms of tillage practice. The mould-board plough buried most of the crop residue and produced the maximum CO2 flux. The C released by the different treatments as a percentage of C in the crop residue was: 134 percent with mould-board plough; 70 percent with mould-board plough and disc harrow; 58 percent with disc harrow; 54 percent with chisel plough; and 27 percent with no-tillage. This demonstrates the correlation between CO2 loss and tillage intensity, and demonstrates why farming systems that use mould-board ploughing inevitably lose soil C. Very large amounts of organic matter would be required to replace the loss incurred by such heavy tillage. Reicosky et al. (1995) estimate that 15 - 25 tonnes/ha manure plus crop residue would be needed annually in North America to offset these losses.
The flux of CO2 from soil generated directly by the tillage process may not always reflect the overall release of CO2 and hence carbon storage of the system. This is illustrated by a comparison of conventional disc tillage and no-tillage in central Texas, the United States of America, by Franzluebbers, Hons and Zuberer (1995). Here, seasonal evolution of CO2 was up to 12 percent greater in the no-tillage system after 10 years. This was despite the fact that more C was sequestered by the no-tillage system. The authors suggest that a change in the dynamics of CS and mineralization have occurred under the no-tillage system. Similarly, Costantini, Cosentino and Segat (1996) found that more CO2 was released from no-tillage or reduced-tillage compared with conventional tillage despite there being increased levels of soil C. They ascribe this difference to an increase in the microbial biomass.
Rates of carbon loss through tillage depend considerably on the site and cropping system. Ellert and Janzen (1999) measured the flux of CO2 following the passage of a heavy cultivator on a semi-arid Chernozem soil in the Canadian prairies. They found that although tillage increased rates of CO2 loss by two to four times, values returned to normal after 24 hours. They calculated that even with ten passes of the cultivator, only 5 percent of crop residue production would be released from this cropping system. In another situation, ploughing of a wheat - fallow cropping system near Sydney, Australia, reduced soil C by 32 percent after 12 years. Elimination of ploughing and adopting a no-tillage approach was unable to prevent a decrease in the carbon stock, although the loss was reduced to just 12 percent (Doran, Elliott and Paustian, 1998). The authors suggest that a fallow period would be required to halt the decline in soil C at this site.
There are many different types of tillage system. Conservation tillage covers a range of practices - no-tillage, ridge-tillage, mulch-tillage (Unger, 1990). Mulch-tillage maintains higher levels of residue cover. With mulches, only a small fraction of the residue is in contact with the soil surface and the microbes it contains. Decomposition is slow, especially as oxygen availability is limited. The physical presence of crop residues on the soil surface also alters the microclimate of the upper soil layer, which tends to be cooler and wetter compared with conventional tillage (Doran, Elliott and Paustian, 1998).
The accumulation of residues also reduces the loss of CO2 from the soil surface. Alvarez et al., (1995) reported an increase in labile forms of organic matter under no-tillage in the Argentine rolling pampa, indicating a decrease in the mineralization of the organic fraction. This study also noted that although organic C increased by 42 - 50 percent under no-tillage compared with ploughing and chisel tillage, there was also a marked stratification in the distribution of C under the no-tillage regime that was not evident in the ploughed system.
Stratification of organic C is common with reduced or no-tillage. Zibilske, Bradford and Smart (2002) in semi-arid Texas, the United States of America, demonstrated that the organic carbon concentration was 50 percent greater in the top 4 cm of soil of a no-tillage experiment compared with ploughing, but the difference dropped to just 15 percent in the 4 - 8-cm depth zone. This is typical of organic carbon gains observed with conservation tillage in hot climates. Bayer et al. (2000), working on a sandy clay loam Acrisol, also found that the increase in total organic C was restricted to the soil surface layers under no-tillage but that the actual quantity depended on the cropping system. An oat/vetch - maize/cowpea no-tillage system produced the largest quantity of crop residues and sequestered the most C: 1.33 tonnes C/ha/year in 9 years.
Reicosky (1997) has compared the results from many no-tillage trials. The data emphasize the effect that crop rotation and quantity of crop residue has on organic matter accumulation. Overall, rates of organic matter accumulation can be expected to be lower in the hotter climates. Nevertheless, even in the very sandy soils of in north of the Syrian Arab Republic, it has been possible to make modest increases in SOM with no-tillage (Ryan, 1997). In western Nigeria, no-tillage combined with mulch application had a dramatic effect, increasing soil C from 15 to 32.3 tonnes/ha in 4 years (Ringius, 2002).
Although no-tillage systems are an excellent tool for combating the carbon losses associated with conventional cultivation, they do have their own special problems. In temperate lands, the reduction in soil temperature commonly associated with plant residue accumulation on the soil surface can retard germination. However, in drylands, where soil temperatures are frequently above the optimum for germination and plant establishment, such cooling is likely to be beneficial (Phillips et al., 1980). No-tillage systems frequently suffer from an increased incidence of pests and diseases; the mouldboard plough and disc harrow are efficient weed controllers (Reicosky et al., 1995). Consequently, no-tillage systems generally rely upon extra herbicides and pesticides. These inputs have an economic price and they also incur a carbon cost. However, in many dryland farming systems of the developing world, purchasing such products is not feasible, and quite often there is plentiful labour available for weeding. Application of N fertilizer can also be problematic when used on undisturbed, no-tillage soils. Where the soil is poorly drained, denitrification can occur and the reduced rate of evaporation increases the risk of nitrate leaching. In addition the native soil N has a lower rate of mineralization in undisturbed soil.
Not all soils are suited to a reduced-tillage approach. Some soils in the Argentine pampa may actually lose more C under no-tillage (0.7 - 1.5 tonnes C/ha/year) compared with conventional ploughing (Alvarez et al., 1995) and periodic ploughing is required to avoid soil compaction (Taboada et al., 1998). Where no-tillage is used on the pampas, the physical status of the soil is a critical factor for the success of the system (Diaz-Zorita, Duarte and Grove, 2002). Similarly, in the West African Sahel, the highest crop yields are obtained with deep ploughing, which is required to prevent crusting and alleviate compaction. In general, the success of reduced-tillage systems is often dependent on soil texture (Needelman et al., 1999).
A particular advantage of the no-tillage system is that it favours multicropping; harvesting can be followed immediately by planting (Phillips et al., 1980). Any cropping system that allows for continuous or near continuous plant growth should yield the maximum capacity to produce plant biomass and, consequently, has the potential to provide the greatest amount of organic mater for inclusion into the soil.
Considering the overall carbon budget, no-tillage systems have a lower energy requirement because tillage is very energy intensive. Phillips et al. (1980) have calculated that no-tillage systems in North America reduce the energy input into maize and soybean production by 7 and 18 percent, respectively. Improved water-use efficiency means that the energy, and hence carbon, cost of irrigation are reduced. However, the impact of energy savings is frequently offset by additional herbicide requirements (Phillips et al., 1980). Kern and Johnson (1993) estimated that the manufacture and application of herbicides to no-tillage systems of the Great Plains is equivalent to 0.02 tonnes C/ha.
Reduced-tillage systems were adopted originally to help combat soil degradation. They were not intended as a means of sequestering C, which is a fortunate side-effect. Although the effectiveness of no-tillage at sequestering C will depend on the specific agricultural system to which it is introduced, there is no doubt that, as the intensity of tillage decreases, the balance between carbon loss and gain swings toward the latter.
The importance of rotation in agricultural systems has long been known and the procedure now forms an intricate part of many conservation tillage practices. The inclusion of rotations has many benefits such as countering the buildup of cropspecific pests and, thereby, lessening the need for carbon costly pesticides and herbicides. Different crop species have a variety of rooting depths and this aids in distributing organic matter throughout the soil profile. In particular, deep-rooting plants are especially useful for increasing carbon storage at depth, where it should be most secure. The inclusion of N-fixing varieties in a rotation increases soil N without the need for energy-intensive production of N fertilizers.
The beneficial effects that rotations have for CS have been proved in many longterm field experiments. For example, Gregorich, Drury and Baldock (2001) made a comparison of continuous maize cultivation with a legume-based rotation. The rotation had a greater effect on soil C than did fertilizer. The difference between monocultured maize and the rotation was 20 tonnes C/ha while the effect of fertilization was 6 tonnes C/ha after 35 years. In addition, the SOM present below the ploughed layer in the legume-based rotation appeared to be more biologically resistant. This demonstrates that soils under legume-based rotations tend to preserve residue C. A positive effect on SOC (an increase of 2-4 tonnes/ha) was also found with legumes and alternate cattle grazing in semi-arid Argentina (Miglierina et al., 2000).
Rotations, especially legume-based ones, are generally regarded as extremely valuable for maintaining soil fertility and have a very good potential for sequestering C in dryland systems. Drinkwater, Wagoner and Sarrantonio (1998) estimate that their use in the maize/soybean-growing region of the United States of America would increase soil CS by 0.01 - 0.03 Pg C/year. The effectiveness of rotations for sequestering C is likely to be greatest where they are combined with conservation tillage practices.
The role that fallows play in CS is varied. Where cropping is not taking place, it is important that vegetation cover is preserved. This is especially so in drylands where exposed soil is most likely to suffer from erosion and degradation. In addition to protecting the soil, cover crops can utilize solar energy that would otherwise be wasted. The CO2 fixed is then available for sequestering into the soil as the plants senesce. The importance of vegetation cover can be illustrated with the results from an experiment conducted at a semi-arid site in Mediterranean Spain (Albaladejo et al., 1998). Four and a half years after the vegetation cover was removed from one site, the SOC had decreased by 35 percent compared with the control plots.
The type of fallow is important. In Nigeria, forest clearance caused a decline in soil C from 25 to 13.5 tonnes/ha in seven years, but 12 - 13 years of bush fallow restored the carbon content (Juo et al., 1995). Conversely, pigeon pea fallow was unable to sequester sufficient C on account of its low biomass production and rapid degradation.
However, fallows can have a negative effect on carbon storage in many situations. The frequency of summer fallows in semi-arid regions has been suggested as one of the major factors influencing the level of soil C in agricultural systems (Rasmussen, Albrecht and Smiley, 1998). Reducing the summer fallow in the semi-arid northwest United States of America is reported to have had a more positive effect on soil carbon retention than that achieved by decreasing tillage intensity. The loss of C in this region is believed to reflect the high rates of biological oxidation that occur here, which can only be offset by very large applications of manure (Rasmussen, Albrecht and Smiley, 1998). Consequently, yearly cropping and the associated organic additions is the recommended practice. Miglierina et al. (2000) also found that reducing summer fallow increased soil C, a consequence of the additional crop residue that was added. Using the CENTURY agro-ecosystem model, Smith et al (2001) predicted that reducing summer fallow in wheat cropping systems (wheat - fallow to wheat - wheat - fallow) in the semiarid Chernozems of western Canada would reduce carbon losses by 0.03 tonnes/ha.
Elimination of fallows can be highly beneficial for soil C simply because most fallows are associated with small inputs of plant residue. The significance of fallows for CS in a given system will depend on whether or not the cropping cycle adds significant quantities of organic matter to the soil. Where it does, then the presence of fallows is unlikely to enhance carbon storage within the system. Conversely, where the cultivation practice is poor and little or no organic matter is added, fallow periods will serve to counter this situation.
Not all soil C is associated with organic material; there is also an inorganic carbon component in soils. This is of particular relevance to drylands because calcification and the formation of secondary carbonates is an important process in the soils of arid and semi-arid regions where, as a result, the largest accumulations of carbonate occur (Batjes and Sombroek, 1997). The dynamics of the inorganic carbon pool are poorly understood although it is normally quite stable. Sequestration of inorganic C occurs via the movement of HCO-3 into groundwater and closed systems. According to Schlesinger (1997) accumulation of calcium carbonate is quite low at 0.0012 - 0.006 tonnes/ha. However, Lal, Hassan and Dumanski (1999) believe that the sequestration of secondary carbonates can contribute 0.0069 - 0.2659 Pg C/year in arid and semi-arid lands.
Although soil inorganic C is relatively stable, it will release CO2 if the carbonates become exposed through erosion (Lal, Hassan and Dumanski, 1999). In addition, irrigation can cause inorganic C to become unstable if acidification takes place through inputs of N and sulphur. The release of CO2 through carbonate precipitation is seen as a major problem if irrigation waters are used in any system that is trying to store C. Furthermore, Schlesinger (2000) has pointed out that the groundwater of arid lands often contains up to 1 percent calcium and CO2. This concentration is much higher than that which occurs in the atmosphere. Consequently, when these waters are applied to arid lands, CO2 is released to the atmosphere and calcium carbonate precipitates. Schlesingers calculations suggest that irrigation of some cropping systems would yield a net transfer of CO2 from the soil to the atmosphere.
An important aspect of agricultural systems in relation to the global carbon balance is the production of trace gases, particularly CH4 and N2O. When CS by soils is being considered as a mechanism for offsetting greenhouse gas emissions, it is necessary to consider all the interacting factors that can influence global warming. Both CH4 and N2O are radiatively active gases and, like CO2, contribute to the greenhouse effect. Although they are present in the atmosphere at much lower concentrations than CO2, they are much more potent. CH4 and N2O are, respectively, 21 times and 300 times more active GHGs than CO2.
Ruminants, composting, biomass burning and waterlogging produce CH4, while N2O is released from soils when N fertilizer or manure is applied (Vanamstel and Swart, 1994). Manure usage is considered to be the major problem with regard to trace-gas emissions in agriculture. This is a potentially serious problem because the application of manure is a major tool for increasing soil C in drylands. Smith et al (2001) calculated that, for European soils, the effect of trace gases is sufficient to reduce the CO2 mitigation potential of some no-tillage and manure-management practices by up to a half. Moreover, climate change is likely to amplify the problem as increased temperature is predicted to promote N2O emissions (Li, Narayanan and Harriss, 1996).
Climate is a major factor involved in soil formation. Consequently, climate change will influence soils. Photosynthesis and decomposition will be affected directly and, hence, have an impact on soils. Whether soil carbon levels increase or decrease will depend on the balance between primary production and decomposition (Kirschbaum, 1995). Overall, productivity is predicted to increase as a consequence of rising CO2 concentration and temperature, and this will lead to increased amounts of residue available for incorporation into the soil. However, higher temperatures can be expected to increase mineralization of SOM because this process is more sensitive to temperature increases than primary production. Kirschbaum (1995) predicts that SOC stocks will decline overall with global warming. However, Goldewijk et al. (1994) suggest that the effects of temperature and water availability on soil respiration will be smaller than those attributable to the CO2 fertilization effect. The direction of change is not certain but the balance of change will most probably operate at the regional level.
It has been argued that agricultural systems are to some extent buffered from environmental effects, while decomposition is not protected (Cole et al., 1993). Hence, increased rates of mineralization might be more significant than any enhancement of production. However, the quality of plant organic matter is expected to decrease under elevated CO2 owing to an increase in the C:N ratio. This would slow the rate of degradation (Batjes and Sombroek, 1997). Globally, the drylands are expected to become moister (Glenn et al., 1993), which should lead to an increase in productivity and decomposition. However, shifts in climate zones are dependent upon a complex array of variables. Predictions based on the CENTURY agro-ecosystem model suggest that, overall, grasslands will lose soil C except in tropical savannahs, which should show a small increase (Parton et al., 1995). Experiments at elevated CO2 have also shown that changes in soil C in agro-ecosystems are particularly dependent upon the crop species grown (Rice et al., 1994).
The full extent of the global rise in temperature associated with climate change may not be felt in many of the drylands because warming is predicted to be greatest at higher latitudes. With regard to the vegetation, some of the best-adapted plants for dryland regions use the C4 photosynthetic pattern. Because these species already have a CO2-concentrating mechanism, they show little or no increase in productivity at elevated CO2. However, they are still likely to receive some benefit from the increased water-use efficiency that accompanies a rising CO2 concentration.