Mesolevel

Soil nutrient-balance study Kisii District, Kenya

After the continental nutrient-balance study of land use systems in SSA (Stoorvogel and Smaling, 1990), scale-inherent simplifications were inevitable (Stoorvogel, Smaling and Janssen, 1993). This led to a similar exercise for a well-inventoried smaller area, the Kisii District in southwest Kenya (Smaling, Stoorvogel and Windmeijer, 1993). This area of 2 200 km2 at altitudes of 1 500-2 200 m had about 1.5 million inhabitants in 1990. The district has a high agricultural potential, but the growing population causes overexploitation of the land. Primary data were available on: climate; landforms; soils; land use; use of mineral fertilizer and farmyard manure; crop yields, residues and their nutrient content.

The district was subdivided into two temperature zones and seven LUTs. They included: extensive grazing in bushland, intensive grazing on improved pastures, tea, pyrethrum, coffee, banana, sugar cane, maize and beans (as monocrops or intercropped), sweet potato and fallow. Five rainfall zones were distinguished, with annual precipitation of 1 350-2 050 mm, and 20 soil units were found, mainly formed on volcanic rocks. This resulted in 50 LWCs and, combined with the LUTs, in a total of 107 LUSs.

Methodology

The nutrient balance with five inflows (IN1-IN5) and five outflows (OUT1-OUT5), according to Stoorvogel and Smaling (1990), was used for the calculation.

IN1

Mineral fertilizer input was based on fertilizer use data from 1980. These had to be multiplied by 2.5 for N, 2.0 for P and 3.0 for K, because fertilizer consumption in Kenya had increased considerably. Tea received most N fertilizer, and P fertilizer was applied mainly to maize.

IN2

Survey data for manure application were available. The nutrient contents of the manure were set at 1.3 percent for N, 0.5 percent for P and 1.6 percent for K, based on dry matter. Most manure was applied to coffee and bananas, and came mainly from paddocks and stables.

IN3

Atmospheric deposition was determined using the regression equations from Stoorvogel and Smaling (1990). The nutrient input was linked with the square root of the mean annual precipitation. The regression coefficients for N, P and K were 0.140, 0.023 and 0.092, respectively.

IN4

BNF was the sum of non-symbiotic N fixation and the contribution of beans, the only leguminous species in the study area. Symbiotic N fixation was set at 50 percent of the total N uptake. Non-symbiotic N fixation was determined using the regression equation of Stoorvogel and Smaling (1990):

IN4 = 2 + (P - 1 350) × 0.005

IN5

Sedimentation was not relevant in the study area.

OUT1

Production statistics were available and were multiplied by the nutrient content of the crops. This generated the export of nutrients with harvested products. Insufficient information was available to take differences in nutrient use efficiency into account.

OUT2

The export of nutrient with crop residues was calculated by multiplying the amount of residues by the nutrient contents and a removal factor.

OUT3

Leaching of N and K were determined with a transfer function (based on literature). N leaching was calculated as a percentage of the sum of mineral N in the soil (Nmin) and N applied by mineral and organic fertilizer. The percentages were based on rainfall and clay content (Table 13).

N_min = 20 × N_tot × M

where:

N_tot = total N content of the soil of the upper 20 cm,
M = mineralization rate (2.5 or 3.0 percent).

K leaching was calculated in a similar way with the percentages of Table 13 multiplied by the sum of exchangeable K (in kilograms per hectare) and mineral and organic fertilizer K.

TABLE 13
N and K leaching percentages for different rainfall and clay content

OUT4

For gaseous nitrogen losses, only denitrification was taken into account. A regression function based on an extensive literature research was developed:

OUT4 = (-9.4 + 0.13 × C + 0.01 × P) × (N_min + N_fert)

where:

C = clay content (percent);
P = mean annual precipitation (mm/year);
N_min = mineral soil N (kg/ha);
N_fert = mineral and organic fertilizer N.

OUT5

Erosion was calculated using the USLE. Annual soil loss per hectare was estimated as a function of rainfall erosivity (R), soil erodibility (K), slope gradient (S), slope length (L), land cover (C) and land management (P). The R factor was set at 0.25 for the entire district. The K factor was derived from soil texture, organic matter content and permeability. The S and L factors were determined with:

S = (0.43 + 0.30 × s = 0.043× s²)/6.613
L = (d/22.13)^0.5

where:

s = slope gradient (percent);
d = slope length (m), set at a fixed value of 100 m.

An average C factor was estimated for each LUT.

The P factor was related with the slope (s): P = 0.2 + 0.03 × s

The soil loss (R × K × S × L × C × P) was multiplied by the nutrient content of the soil and an enrichment factor of 1.5 to obtain the export of nutrients by erosion. For P and K, the net loss was multiplied by 0.75 to compensate for soil formation at the root base.

For year-round fallow, equilibrium conditions were assumed, i.e. IN - OUT = 0.

Results

The total nutrient balance, the sum of the four inflows minus the sum of the five outflows, for the entire district was -112 kg/ha for N, -3 kg/ha for P and -70 kg/ha for K. The removal of harvested products (OUT1) and erosion (OUT5) were the strongest negative contributors and, for N, also leaching (OUT3). Table 14 shows the nutrient balance for each LUT component. Losses were highest under pyrethrum and, to a lesser extent, sugar cane and maize.

Discussion

This study was based on the methodology of Stoorvogel and Smaling (1990), but it was possible to calculate several flows in more detail as results of the smaller study area. For IN1, IN2, OUT1 and OUT2, it was possible to use local data instead of estimates or national averages. At this scale, it was also possible to calculate erosion (OUT5) using the USLE instead of estimates. However, the other flows were still based on transfer functions, which are not area specific.

TABLE 14
Nutrient balance of the different land use type components

Soil nutrient-balance study, southern Mali

Nutrient balances were calculated and evaluated economically for southern Mali (Van der Pol, 1992; Van der Pol and Traore, 1993). The study concerned cropping systems in southern Mali, where cotton, sorghum and millet are the main crops. With the withdrawal of fertilizer subsidies, the amount of fertilizer per hectare decreased and production increases were only attributable to expansion of the cotton area. Such a development increases the risk of land degradation as a result of nutrient depletion.

Methodology

The nutrient balance for the region of southern Mali is built up from balances for the various cropping systems. Literature data were combined with locally collected production statistics. The study followed the nutrient-balance calculation approach as described by Pieri (1989) and Frissel (1978).

The nutrient elements in the soil were classified into three pools:

• A: mineral elements available to plants;

• B: elements associated with soil organic matter;

• C: mineral reserves in the soil.

Figure 10 represents the flow of nutrients in and out of the system and between these pools.

Source: after Van der Pol, 1992.

In order to restrict the analysis to long-term dynamics, elements in Pool A and Pool B were considered together as nutrients. The combined size of both pools determines the fertility of a soil to a large extent. Nutrient depletion is associated strongly with a gradual decline of the organic matter content in the soil. The quantities of nutrients being depleted each year were used to estimate the rate of this decline.

On the other hand, elements in Pool C, the mineral reserve, were not considered as nutrients. On a time scale relevant to soil formation processes, an equilibrium may develop between mineral reserves and available and organic nutrients. However, the rates of change of these latter pools are too great to attain such a situation under the influence of human activities. Thus, elements were assumed to become available by transformation and dissolution of soil minerals at a constant rate, i.e. the rate of weathering.

Processes affecting the nutrient pool

In the approach used in this study, the following processes affect the nutrient pool:

· Outputs of nutrients:

- export in cropped products;
- losses by leaching;
- losses by erosion;
- losses by volatilization/denitrification;
- incorporation of P and K in the mineral reserve (‘irreversible fixation’).

· Inputs of nutrients:

- fertilizer application;
- organic manure application;
- restitution of crop residues;
- symbiotic N fixation;
- asymbiotic N fixation;
- recycling of leached nutrients and biological fixation by trees left growing in the fields;
- atmospheric deposition by rain and dust;
- transformation and dissolution of soil minerals;
- import with seed.

The difference between outputs and inputs represents the net nutrient balance.

Acidification

Acidification of soils occurs under increasingly intensive cultivation. As this can be corrected by liming, this study considered Acidification as resulting from a negative ‘lime balance’. Soils acidify by leaching of K, calcium (Ca) and magnesium (Mg). The loss of these elements is part of the nutrient balance but, in addition, Acidification also occurs after application of ammonium fertilizer or urea, as a result of the nitrification reactions. In the calculations on acidification, it was assumed that each kilogram of fertilizer N needed 1.75 kg of lime (CaCO₃) for neutralization.

Reliability margins and factors of uncertainty

The basic data for nutrient inputs and outputs were selected from literature and from production statistics. Data from literature pertained to various sites in Western Africa, but were not necessarily representative for the area of the Compagnie Malienne pour le Développement des Textiles (CMDT) in southern Mali. Taking into account the variation in rainfall, soil properties, etc., some ‘intelligent guessing’ was necessary.

From the data, a ‘most probable value’ for the study region was selected, and a range representing the 95-percent probability level. If the same statistical weight is given to all literature data, the ‘most probable value’ is the arithmetic mean, and the range corresponds to twice the standard deviation of the mean value. However, in reality, the variability in soil properties and rainfall, and the fact that not all literature data were based on the same number of experiments, made a subjective estimate of reliability ranges more appropriate than a purely statistical procedure.

Optimistic and pessimistic views

Three net nutrient-balance values were calculated. The first was based on the use of only ‘most probable’ values. This value best reflected the actual nutrient balance of the region. The second value, labelled ‘optimistic’, was based on the combination of low estimates for the outputs and high estimates for the inputs. The third value, labelled ‘pessimistic’, was based on high estimates for outputs and low estimates for inputs.

Results

Table 15 presents the total nutrient balances for southern Mali. These indicate that N and K are the most deficient elements. Most balances are negative, even with the optimistic view. The balances were also calculated per crop (Figure 11), which enables comparisons to be made between crops and conclusions to be drawn as to whether the balances offered by one crop are more favourable than those of another.

TABLE 15
Calculated nutrient balances for southern Mali

Discussion

The optimistic and pessimistic views are a valuable addition because they provide more insight into the dimensions of the nutrient depletion problem and the variation and uncertainties of the outcomes. Compared to Stoorvogel and Smaling (1990), incorporation of P and K in the mineral reserve as output flow and weathering and import with seed as input flows were included. On the other hand, sedimentation and irrigation, which might be imported especially for rice, were not included.

The pool approach is more realistic than a black-box approach and gives more insight into the soil system. However, in practice it will be difficult to discriminate between the different pools. As in this study, they might have to be combined.

Nutrient-balance studies in India

Soil nutrient-balance studies in various agro-ecological regions of India that are based on broad parameters for input and output flows provide a mesolevel insight into soil fertility aspects. In the two studies reported here, one examined nutrient mining in different agroclimatic zones of Andhra Pradesh State and the other study calculated nutrient removals in Rajasthan State.

Methodology

Andhra Pradesh

Nutrient removal by various crops from soils of different agroclimatic zones of Andhra Pradesh was computed on the basis of nutrient removal per specified economic yield (Singh et al., 2001). In order to proceed with the computations, district-level data on area and production for 1998-99 were used for 15 major crops. District-level fertilizer use data were used in order to calculate nutrient additions through fertilizers. From the district-level data, zone-level nutrient additions were determined by adding the data of the cluster of districts falling in the respective zone.

The share of nine major crops in fertilizer consumption was assumed as 95 percent; the remaining 5 percent was assumed to be consumed by the other crops and vegetables and fruits. Furthermore, while calculating the share of organic sources of nutrients, it was assumed that the total potential of various organic resources and their equivalent plant nutrients were distributed uniformly in the seven agroclimatic zones based on the information of gross cropped area. Finally, 10 percent of the total organic nutrient potential was considered trappable for the purpose of the computations. The nutrient balance was calculated as:

Nutrient balance = [{(A × 0.95) × EF}+ (B × 0.10)] - [TR]

where:

Rajasthan

Nutrient removals were calculated on the basis of published production figures for the major crops and averaged nutrient removal figures from several studies (Gupta, 2001). Figures for mineral fertilizer use were those of fertilizer agencies. The need was perceived for a systematic database with regard to nutrient status of soils, removal of nutrients by different crops/varieties, amount of N fixed by various legumes, and probable contribution of organic manures. The nutrient balance was calculated as:

Nutrient balance = [(A × EF) + D + BNF] - [TR]

where:

Results

For Andhra Pradesh, the overall balance for N was positive (0.207 million tonnes), while both the P and K balances were negative (-0.133 million tonnes and -0.431 million tonnes, respectively). The results varied largely between the different agroclimatic zones.

Discussion

Both studies used a simple nutrient balance with the main inputs and outputs. However, they neglected important outflows such as leaching and erosion. In addition, there is some difficulty in comparing these studies with others as the outcomes are in tonnes rather than kilograms per hectare. However, the results do show the relative differences between the zones, which can form a basis for future strategies, e.g. for increased fertilizer use.

Sub-Saharan Africa soil nutrient-balance study, FAO, 2003

The overall purpose of this study and its macrolevel aspect is reviewed above. The particular hypothesis in this study is that the mesolevel offers a suitable entry point for policy-makers and private sector intervention, where macrolevel and microlevel are not appropriate for policymaking at subnational level. Within such a mesolevel system, the commercial component may function as the engine of the farming system, allowing for intensification and expansion. This cash component can function as a driver for soil fertility management.

The study examined three farming systems with a cash crop or other market-oriented agriculture component in different AEZs: the cocoa-based farming system in Nkawie District and Wassa Amenfi District, Ghana; the tea-coffee-dairy farming system in Embu District, Kenya; and the cotton-based farming system in Koutiala Region, Mali.

Methodology

At the mesolevel, the methodology followed the calculation for the macrolevel. However, owing to the lack of spatial data of sufficiently high resolution, it was not possible to perform calculations on a spatial basis. Therefore, the nutrient balance was calculated on a tabular basis. Relations between land use and soils were established in order to compensate for the lack of spatial data.

At the mesolevel, a 1-km grid is too coarse to represent physiographic differences with sufficient accuracy. Although a land use map can be created on the basis of aerial photographs or satellite images with fast field checks, these were not available for the study areas.

The data availability for the three countries was very different; this prohibited a generic approach at mesolevel. The general procedures for each nutrient flow are described below.

IN1 Mineral fertilizer

The data on mineral fertilizer were derived from farm surveys, recommended fertilizer rates or macrolevel data, depending on the data availability in each study area. Recommended fertilizer rates are in general much higher than the actual application rates. This is because not all farmers can afford or want to apply these quantities. Therefore, the fertilizer rates were multiplied by a factor representing the ratio between the harvested area at mesolevel and the harvested area at national level in order to prevent overestimation.

IN2 Organic fertilizer

The amount of available manure was derived from the number of livestock within the study area, using excretion, nutrient content and loss factors. The application per crop was derived from farm surveys and estimates. Local nutrient content values were used where available.

IN3 Atmospheric deposition

The atmospheric deposition was derived from the macrolevel. Rainfall data from local weather stations were used; where such data were not available, they were derived from the macrolevel.

IN4 N fixation

N fixation was treated in a similar way as for the macrolevel. Where available, specific data related to N on fixation should be included, e.g. agroforestry systems with N-fixing trees.

IN5 Sedimentation

Irrigation was not relevant for the three case-study areas. Sedimentation was estimated for crops grown in river valleys, e.g. rice. The LAPSUS model was not used at the mesolevel because of the lack of spatial data.

OUT1 Crop products

Crop production data were multiplied by nutrient contents of the crops. Where available, local nutrient content factors were used as these can be significantly different from the average continental values used at the macrolevel.

OUT2 Crop residues

Crop residue removal factors were derived from farm surveys or estimated by local experts. These factors were multiplied by the production and the nutrient content factors of the crop residues.

OUT3 Leaching

The regression models used for the macrolevel were used to calculate N and K leaching.

OUT4 Gaseous losses

The regression model used for the macrolevel was used to estimate the gaseous nitrogen losses.

OUT5 Erosion

Estimates of erosion were made for each crop. These estimates took regional differences in topography and soils into account, and were based on literature and expert knowledge. Although suitable for mesolevel, the LAPSUS model was not used because of the lack of spatial data.

Results

The study compared two districts in Ghana. Nkawie District is more densely populated and has a long land use history under cocoa. Wassa Amenfi District has experienced a large increase in the cocoa area in recent decades event though the area is less suitable for cocoa.

TABLE 16
Nutrient balance for two cocoa districts, Ghana

Table 16 shows the resulting nutrient balances per crop for both districts. The balances for Nkawie District are more negative than those for Wassa Amenfi District. The main reason for this difference relates to the area under cocoa, which is 58 percent for Nkawie District and 90 percent for Wassa Amenfi District. The nutrient balance for cocoa is only slightly negative, unlike most crops. In particular, cassava, yam and cocoyam have strongly negative nutrient balances.

Discussion

The study shows that a mesolevel nutrient balance can be assembled properly, provided that sufficient data are available. The mesolevel results provide information that cannot be deduced from macrolevel and microlevel studies. Mesolevel nutrient balances will have greater potential when they are made spatially explicit. In this study, not enough spatial data were available. In particular, the erosion estimates can be improved significantly using the LAPSUS model. Without the spatial component, the mesolevel approach is not very different from the previous mesolevel cases.