Nutrient flows and balances are not very meaningful without knowledge of nutrient stocks. This is because the rate of soil fertility decline is not simply a per hectare per year unit, but also a ratio indicating the percentage change of the total nutrient supplies. Chapters 4 and 5 consider this aspect in more detail. Furthermore, nutrient stocks play a role as input data for the calculation of the nutrient flows, sedimentation, leaching and erosion, and to a lesser extent gaseous losses. The WISE database, developed by the International Soil Reference and Information Centre (ISRIC) (Batjes, 2002) was the source of all soil data for the macrolevel and part of the data for the mesolevel. The WISE database consists of a set of homogenized worldwide data of 4 382 geo-referenced soil profiles, classified according to the FAO-UNESCO original legend (1974) and the revised legend (1988). This database yielded the soil profiles for Africa: 1 799 different soil profiles, describing 81 different soil units. The FAO soil map of the world distinguishes 95 different soil units for Africa, apart from the classes: water, salt, desert, rock and no data. Similar soil units provided the basis for estimates of the properties for the remaining 14 soil units. Most missing soil units in the WISE database were the overall major soil groupings, e.g. Luvisols and Xerosols, because all profiles were classified at soil unit level, e.g. gleyic Luvisol or luvic Xerosol.
TABLE 25
Comparison of data and procedures for
calculating nutrient flows for each scale level
Flow |
Macrolevel |
Mesolevel |
Microlevel |
IN1 |
Fertilizer use data per crop (IFA/IFDC/FAO, 2000) and total consumption from FAOSTAT |
Regional fertilizer use data and factors per crop (same as macrolevel, if no other values are available) |
Local fertilizer use data per crop |
IN2 |
Conversion of continental livestock density maps (FAO, 2000) in nutrient input by multiplying with the nutrient content, manure production and loss factor |
Number of livestock (regional data) with conversion factors for nutrient input and manure management |
Local livestock data, conversion factors for nutrient input (where not determined) and manure management data (or amount of manure application where available) |
IN3 |
Related to rainfall (Leemans and Cramer, 1991) and dry deposition derived from created Harmattan dust map |
Fixed value derived from macrolevel (or directly where literature available) |
Fixed value derived from macrolevel (or directly where literature available) |
IN4 |
Percentage of leguminous crop production and related to rainfall (Leemans and Cramer, 1991) |
Percentage of leguminous crop production and fixed value |
Percentage of leguminous crop production and fixed value |
IN5 |
Irrigation areas receive 300 mm irrigation. Sedimentation calculated using the LAPSUS model (Schoorl et al., 2002) |
Estimated amount of irrigated areas and sedimentation areas |
Amount of irrigation water and measured or estimated sedimentation |
OUT1 |
Harvested areas and yields per country from FAOSTAT multiplied with nutrient content |
Regional data on harvested area and yield (or production) multiplied with nutrient content |
Local production data multiplied with nutrient content |
OUT2 |
Factor nutrient content and crop residue removal factor per crop and country |
Factor nutrient content and crop residue removal factor per crop |
Factor nutrient content and crop residue removal factor (management) |
OUT3 |
Regression models for N developed by De Willigen (2000) and for K in this study |
Regression models for N developed by De Willigen (2000) and for K in this study |
Regression models for N developed by De Willigen (2000) and for K in this study |
OUT4 |
Regression model based on data from a study on global N emissions from agricultural land (IFA/FAO, 2001) |
Regression model based on data from a study on global N emissions from agricultural land (IFA/FAO, 2001) |
Regression model based on data from a study on global N emissions from agricultural land(IFA/FAO, 2001) |
OUT5 |
Quantitative erosion- sedimentation map, computed using the LAPSUS model (Schoorl et al., 2002) |
Estimated values per crop, based on literature and macrolevel |
Measured or estimated erosion and soil nutrient content data |
OUT6 |
Not relevant |
Not relevant |
Estimated, based on number of consumers and crop products used for own consumption |
This study calculated the following soil properties for each soil unit: clay, pH, organic carbon, total N, exchangeable K, CEC, available P and bulk density. Soil depth and erodibility are not parameters in the WISE database. These were estimated for each soil unit because they are necessary for the erosion-sedimentation model. The WISE database describes soil properties per horizon, but this study used only one value per soil unit. The horizon data were converted to one value per soil profile. This study considered only the upper part of the profile because this is the most important part for agriculture and because most nutrients are stored in the topsoil. The upper horizons should include at least the upper 30 cm of the profile and the lower boundary should not be deeper than 60 cm. This was done to prevent only a thin top horizon or a deep subhorizon being taken into account. These selected horizons provided the basis for calculating an average value. Annex 6 lists the resulting values.
In order to calculate the loss of P and K by erosion, it was necessary to recalculate the values to obtain values as percentages of the total soil mass. For exchangeable K, this is relatively easy, using the bulk density and the atomic mass of 39.1. For P, no direct relation exists between the total amount of P and the amount of available P, as derived from the WISE database. Different analytical methods exist to determine the amount of available P and each has a different relation with the total amount of P in the soil. According to the WISE database, 1 135 of the 1 799 profiles were analysed for available P. The Olsen method was used for 943 profiles, or 83 percent of all analyses. The Bray method was used for 6 percent and the Truong method for 3 percent of all analyses. According to Landon (1991), the values of P-Olsen correspond with total P as follows: > 15 is high, 5 - 15 is medium and < 5 is low for P-Olsen, whereas for total P, > 1 000 mg/kg is high, 200 - 1 000 mg/kg is medium and < 200 mg/kg is low. This rough classification is currently the best available. A regression equation based on the Olsen method was developed to relate available P to total P. It was modified slightly for the influence of the Bray and Truong methods:
Ptotal = 13 × Pavailable1.5