Previous Page Table of Contents Next Page


Chapter 2

Methodologies for assessing soil nutrient balances


Macrolevel

Sub-Saharan Africa soil nutrient-balance study, FAO, 1983-2000

The study assesses the state of soil nutrient depletion in sub-Saharan Africa (SSA) for 1983 and 2000 (Stoorvogel and Smaling, 1990). It provides data on the net removal of the macronutrients N, phosphorus (P) and K from the rootable soil layer on a country-by-country basis.

The development of a method to make this assessment was the focal point of the exercise. FAO provided production figures (1983) and projections (2000) per crop and per country. These were further specified for six mainly climate-based ‘land/water classes’ (LWCs): low, uncertain and good rainfall areas (rainfed agriculture), problem areas, and naturally-flooded and irrigated areas. Data on fertilizer use (1983) and projections for (2000) were given per country and per crop. The next step was to define and quantify factors determining the flow of N, P and K into and out of the soil for the smallest constituents of each LWC: the land use systems (LUSs). Soil fertility dynamics in an LUS are governed by five input (IN) and five output (OUT) factors.

Methodology

TABLE 1
Attributes of land use systems and their specification

Attribute

Specification

Rainfall (R)

Average for LWC (mm/year)

Soil fertility (F)

Classes: 1 - low; 2 - moderate; 3 - high

Management level

Differentiated in low (L) and high (H)

Fertilizer use

Weighting factor 0.0-3.0, related to regional distribution of total national consumption

Manure application

0, 500, 1 000, 1 500 kg/ha/year or ‘during grazing’

Residue removal

% of crop residues removed from the field or burnt

Erosion

Soil loss (tonnes/ha/year)

Crops

FAO database

Assumptions had to be made for describing and quantifying the mechanisms that contribute to the flow of N, P and K into and out of the soil. This was the pivotal stage of the exercise. An important decision in this respect was the further subdivision of LWCs into LUSs. An LUS is defined as a well-defined tract of land with its pertinent land use type (LUT) (FAO, 1976). This study included the further assumption that an LUS is a homogeneous entity. This formed the basis for calculating the nutrient balance.

Table 1 lists the attributes of an LUS. Each LWC comprises one or more LUSs. The description of an LUS is based on relevant, country-specific literature.

At any one time, a certain amount of organic and inorganic N, P and K is present in the soil in stable or labile plant-available forms. When measured one year later, these amounts are not necessarily the same. This is because various processes cause nutrients to flow into and out of the rootable soil layers. In spite of the uncertain nature of the many factors affecting soil fertility, a relatively simple model should serve the purpose of simulating the processes. The five input and five output factors considered in this study are listed in Table 2 and presented in Figure 1.

FIGURE 1
Nutrient flows in the soil

TABLE 2
Input and output factors governing nutrient flows in the soil

Input


Output


IN1

Mineral fertilizers

OUT1

Harvested product

IN2

Manure

OUT2

Crop residues

IN3

Deposition

OUT3

Leaching

IN4

Biological N fixation

OUT4

Gaseous losses

IN5

Sedimentation

OUT5

Erosion

Eight factors have a clear role in enriching (IN) or depleting (OUT) the soil. As livestock in Africa feed largely on crop residues, the two factors IN2 and OUT2 can interact. Consequently, part of the crop residues is removed temporarily, to be returned later as manure.

Mineral fertilizers (IN1)

The FAO database contained information on actual total fertilizer consumption per crop per country for 1983 and projections for 2000. However, these data were not specified per LWC. Hence, the total amount needed to be distributed over the LWCs. Two situations arose:

  1. The literature provided raw data on the regional distribution of fertilizers within a country; where so, these data were used.
  2. Where such information was not available, the assumption was that the use of fertilizers was not distributed evenly within a country, and each LUS received a weighting factor as indicated in Table 3.

TABLE 3
Weighting factors for calculating mineral fertilizers (IN1) per land use system

Land/water class

Weighting factor

Low management

High management

Low rainfall (LR)

0.2

0.4

Uncertain rainfall (UR)

0.6

1.2

Good rainfall (GR)

1.0

2.0

Problem area (PR)

1.0

2.0

Naturally flooded (NF)

0.6

1.2

Irrigated area (IR)

1.5

3.0

Manure (IN2)

Although livestock is an essential element of African farming, the study did not consider extensive grazing; it considered only arable land. However, two forms of manuring occur in the LUS description:

A) Manure collection from bomas, kraals and other storage places, and application to arable fields prior to planting (LUS 0, 500, 1 000 or 1 500 kg/ha).

B) On-the-spot manuring by livestock feeding on crop residues (LUS ‘during grazing’; interaction with OUT2).

In order to calculate A), each LUS with a manure input but without grazing on the arable field was characterized by one of four classes indicating the amount applied to the fields. These amounts were set at 0, 500, 1 000 and 1 500 kg fresh weight/ha/year.

Although the chemical composition of fresh manure varies widely according to its nature and moisture content, for calculation purposes it must be set at constant values for groups of LWCs. Table 4 was constructed on the scarce information available in the literature.

For B), where livestock feed on crop residues left on the field, some of the manure input is realized ‘during grazing’. Three questions arose:

  1. What is the fraction of the crop residues that is grazed?
  2. How many hours a day do the animals spend on the grazed field?
  3. What is the fraction of the nutrients that remains inside the animals?

The answers were stipulated as follows:

  1. This differs for each LUS and is indicated as such in its description.
  2. 12 hours (fixed value for all LUS).
  3. 10 percent (fixed value for all LUS).

TABLE 4
Chemical composition of manure land/water classes


N

P2O5

K2O

Land/water class

(% of fresh weight)

Low rainfall, uncertain rainfall, irrigated area

0.48

0.40

0.65

Problem area (< 1 200 mm/year rain)




Good rainfall, naturally flooded

0.42

0.35

0.55

Problem area (> 1 200 mm/year rain)




Deposition (IN3)

The processes of wet and dry deposition supply considerable amounts of nutrients to soils. Because of an uneven distribution of data over the continent, the calculation procedure was split into two, relating to:

  1. Areas within Harmattan influence (West Africa); the literature provided sufficient point data to allow interpolation.
  2. Areas outside Harmattan influence: data on the factors were scarce, but there was a correlation with rainfall; regression analysis for the different nutrients resulted in the equations listed below. They were used to calculate the contribution to soil fertility by IN3 for areas outside Harmattan influence.

The calculations were:

IN3 (N)

= 0.14 × (rainfall)1/2

IN3 (P2O5)

= 0.053 × (rainfall)1/2

IN3 (K2O)

= 0.11 × (rainfall)1/2

where: IN3 is expressed in kilograms per hectare per year; and rainfall in millimetres per year.

Biological N fixation (IN4)

An important source of N in several agricultural systems is N2 from the atmosphere. Leguminous species and wetland rice draw considerably from this source. Based on information from the literature, three stipulations could be presented, depending on total N demand by crops:

  1. Of the total N demand of leguminous crops (soybean, groundnuts and pulses), 60 percent is supplied through symbiotic N fixation (Rhizobia).

  2. Of the total N demand of wetland rice (LWC, naturally flooded and irrigated area) 80 percent is supplied through chemo-autotrophic N fixation (Azolla, other algae), up to a maximum of 30 kg/ha/year. Higher uptakes are drawn from soil N.

  3. All crops benefit from N that is fixed non-symbiotically (Azotobacter, Beijerinckia and Clostridium) or by N-fixing trees that are left on the field (Rhizobia and Actinomycetes spp.). This contribution is partitioned in Table 5.

TABLE 5
Contribution of scattered trees and of non-symbiotic fixation to biological N fixation

Land/water class

Input
(kg/ha/year)

Low rainfall

3

Uncertain rainfall

4

Good rainfall

5

Problem area

> 1 200 mm rain/year

5


< 1 200 mm rain/year

2

Naturally flooded

2

Irrigated area

2

Sedimentation (IN5)

In parts of the LWC ‘naturally flooded’, sedimentation takes place. Hardly any information on the nutrient content of this sediment could be traced. However, it was necessary to make an assumption on the importance of this input factor. The group of experts reached consensus on a nutrient balance being in equilibrium in this LWC. Input and output factors were calculated, but the deficit (IN5) was assumed to be supplied by the floodwater and its sediment.

In LWC ‘irrigated area’, the nutrient content of the irrigation water was also considered as an input factor. Literature and consultations led to the assumption that, on average, 300 mm/ha/year of irrigation water is supplied to irrigated land. The calculation of IN5 was now governed by the concentration of the three macronutrients in this amount of water. Limited information on this aspect indicated that the following values could be used:

N:

10 kg/ha/year,

P2O5:

3 kg/ha/year,

K2O:

5 kg/ha/year.

Harvested product (OUT1)

Different crops withdraw different amounts of the various nutrients from the soil. A considerable amount of literature is available on this subject. Average values for each crop (excluding outliers) were compiled. In order to obtain an estimate of OUT1, these data needed to be combined with the production figures provided by FAO.

Crop residues (OUT2)

An estimate of the amount of crop residues removed from the arable field was obtained from the literature. It was found that farmers’ attitudes towards utilizing crop residues differed considerably among and within the countries studied. The actual removal is given in the LUS description. The removal can be complete (e.g. used for fuel, roofing or manufacturing) or incomplete (e.g. grazed or burnt). Where there was grazing, this was mentioned in the LUS description. IN2 outlines the effect of grazing on soil fertility. Burning practices are difficult to portray on a continental scale. In the study, only the residues of cotton were assumed to be burnt completely for reasons of field hygiene. Removal of N and K through burning was calculated in OUT3 and OUT4.

A complete raw data set on the uptake of nutrients by above ground crop residues was compiled. Average values of the amount of nutrients in crop residues per tonne harvested were also compiled as were ranges for several crops. Where ranges were given, the general level of management and thus the LUS description were used. The lower value of the range (few nutrients in residue per tonne of harvested product) represents a high level of management, whereas the higher value represents a low level of management. More favourable grain-straw ratios related to genetic improvements are the explanation for these differences. To calculate OUT2, the formula was:

Leaching (OUT3)

Leaching is a significant loss mechanism for some nutrients. In tropical soils, P is often bound tightly by soil particles. Therefore, this study assumed that leaching only played a part with respect to N and K. Research on leaching is confined mainly to point observations, which have an uneven distribution over the continent. These few data are not enough to support a model that should have a spatial significance. Therefore, the literature was reviewed extensively. Together with expert consultations, this review provided clues for correlation. Multiple regression showed leaching to correlate positively with:

The following regression equations were found (in kilograms per hectare per year):

OUT3

(N) = 2.3 + (0.0021 + 0.0007 × F) × R + 0.3 × (IN1 + IN2) - 0.1 × UN

OUT3

(K2O) = 0.6 + (0.0011 + 0.002 × F) × R + 0.5 × (IN1 + IN2) - 0.1 × UK

Gaseous losses (OUT4)

N is lost to the atmosphere by two processes: denitrification and volatilization. Denitrification takes place under anaerobic conditions. A soil does not have to be entirely saturated for denitrification to occur. A moist soil already loses nitrate through microbial processes in wet films and pockets. The loss through denitrification is expected to be greatest in wet climates, on highly fertilized and clayey soils, and for crops that withdraw relatively small amounts of N. Ammonia volatilization plays a role mainly in alkaline environments. Because such soils are not common in SSA, volatilization and denitrification were not treated separately.

TABLE 6
‘Base’ denitrification per land/water class

Land/water class

Denitrification
(kg/ha/year)

Low

3

Uncertain

5

Good

8

Problem

> 1 200 mm rainfall

12


< 1 200 mm rainfall

5

Naturally flooded

12

Irrigated

11

In general, information on both factors was scarce and unevenly distributed. Therefore, correlations were again sought. Multiple regression analysis provided the following equation for the output factor (in kilograms per hectare per year):

OUT4 (N) = ‘Base’ + 2.5 × F + 0.3 × (IN1 + IN2) - 0.1 × UN

where:

‘Base’:

a constant value, covering relative wetness of the soils specific for LWCs (Table 6).

F:

soil fertility class (1 - low; 2 - moderate; 3 - high),

IN1 + IN2:

total application of fertilizer and manure (LUS-specific; in kg/ha/year),

UN:

total uptake of N (crop and yield specific; in kg/ha/year).

Erosion (OUT5)

Research findings on soil loss through erosion were reasonably well documented for most countries. An estimate of soil loss based on this information was given in the description of each LUS. A soil with a high fertility has more to lose than a poor soil. Table 7 lists the assumed nutrient contents of eroded soil material of the three fertility classes. These classes were indicated in the LUS description. They were also used to assess OUT3 (leaching) and OUT4 (gaseous losses).

TABLE 7
Nutrient contents of eroded soil at three levels of soil fertility

Soil fertility class

N

P2O5

K2O


(%)

1

0.05

0.02

0.05

2

0.1

0.05

0.1

3

0.2

0.1

0.2

The difficult part was the assessment of the nutrient content in the eroded soil material. Based on the limited literature available, a so-called ‘enrichment factor’ was established. As the fi nest soil particles are the first to be dislodged during erosion, eroded soil material tends to contain more nutrients than the original soil. In the study, the enrichment factor was set at 2.0 for N, P and K, implying that a ratio of two between the nutrient content of the eroding soil material and the nutrient content of the original soil material.

As topsoil erodes, the roots of crops start to enter layers that were previously beyond the rootzone. Hence, part of what is lost on top is gained at the bottom of the described system. The implication is that the calculated P and K losses through erosion are offset partially by the downward extension of the rootzone. The contributions were set at 25 percent of the calculated losses for the two elements.

Cropping intensity

The FAO database provided the area values of both harvested land and total arable land for each LWC. The ratio between the two, expressed as a percentage, is called the ‘cropping intensity’ (CI). Where this ratio was less than 100 percent, part of the arable land was considered fallow. Its area was calculated as follows:

Fallow area = ((100/CI) - 1) × harvested area (ha)

During a fallow period, a gradual buildup of nutrients takes place. IN3, IN4 and IN5 provide external contributions to soil fertility. In addition, part of the plant-available nutrients is retained in the fallow biomass instead of being leached or eroded. During years of fallow, while the ongoing processes of weathering and mineralization do not increase the total amount of nutrients in the soil, they do replenish the labile pools of the nutrients.

On the other hand, woody species from a fallow are often used as a source of fuel or sold along the roadside (OUT1), a fallow is partly depleted by grazing animals that do not return all they have taken (OUT2 - IN2), and the slash and burn practices prior to cultivation enhance the loss processes OUT3-OUT5 strongly. In addition, extra input in West Africa through deposition of dust (IN3) is offset by extra output owing to the scarcity of fuelwood (OUT1).

These considerations, combined with findings in literature and expert consultations, led to the decision to set the nutrient input by fallow at fixed values of 2 kg/ha/year for N, 2 kg/ha/year for P2O5, and 1 kg/ha/year for K2O irrespective of the LUS. Where cropping intensity equalled 100 percent, the fallow acreage was set at 0 ha. Where the cropping intensity exceeded 100 percent, multiple cropping took place and it was assumed that there was no fallow. In this case, the harvested areas and yields of the annual crops were adapted so that the total area equalled the arable area, and the total production remained the same.

FIGURE 2
Nutrient depletion rate in sub-Saharan Africa, 1983

Results

The results of the study showed N, P and K balances by land use system and by country. They revealed a generally downward trend in soil fertility in Africa. Densely populated and hilly countries in the Rift Valley area (Kenya, Ethiopia, Rwanda and Malawi) had the most negative values (Figure 2), owing to high ratios of ‘cultivated land’ to ‘total arable land’, relatively high crop yields and erosion. For SSA as a whole, the nutrient balances were: -22 kg/ha in 1983 and -26 kg/ha in 2000 for N; -2.5 kg/ha in 1983 and -3.0 in 2000 for P; and -15 kg/ha in 1983 and -19 kg/ha in 2000 for K. Table 8 lists nutrient balances for several SSA countries. The prediction for 2000 was for a more negative nutrient balance for almost all countries. This was influenced by the optimistic FAO estimates for crop production in 2000 (high OUT1) and the expected decrease in fallow areas in 2000.

Discussion

Following earlier work by Pieri (1985), this study was the first with a clearly defined nutrient balance and quantified nutrient flows. It has formed the basis for most subsequent nutrient-balance studies. The basis of the nutrient balance with five inflows and five outflows has been used widely. However, other studies with different data availability and objectives have modified the calculation of some flows.

TABLE 8
Average nutrient balances of some sub-Saharan African countries

Country

N

P

K


1982-84

2000

1982-84

2000

1982-84

2000


(kg/ha/year)

Benin

-14

-16

-1

-2

-9

-11

Botswana

0

-2

1

0

0

-2

Cameroon

-20

-21

-2

-2

-12

-13

Ethiopia

-41

-47

-6

-7

-26

-32

Ghana

-30

-35

-3

-4

-17

-20

Kenya

-42

-46

-3

-1

-29

-36

Malawi

-68

-67

-10

-10

-44

-48

Mali

-8

-11

-1

-2

-7

-10

Nigeria

-34

-37

-4

-4

-24

-31

Rwanda

-54

-60

-9

-11

-47

-61

Senegal

-12

-16

-2

-2

-10

-14

United Republic of Tanzania

-27

-32

-4

-5

-18

-21

Zimbabwe

-31

-27

-2

2

-22

-26

Nutrient-balance studies in Africa, IFDC approach

The methodological approach used by the International Fertilizer Development Center (IFDC) to estimate nutrient balances, depletion rates and requirements combines information on agricultural production, soil characteristics and biophysical constraints with methods and procedures designed for making such estimates (Henao and Baanante, 1999). The information and data related to agricultural production include land use, population-supporting capacity of land, crop production, and use of mineral and organic fertilizers. The approach uses attribute and geographic database systems in conjunction with empirical and mechanistic models to produce information for analyses and monitoring.

The approach builds upon previous work on nutrient balances (Stoorvogel and Smaling, 1990; Smaling and Fresco, 1993; Smaling, Stoorvogel and Windmeijer, 1993). This building on previous work involves the linking of methods and procedures for estimating nutrient balances with attribute databases and geographic information systems (GIS). It integrates data and information in a common geo-referenced base, and illustrates in the form of maps and graphs estimates of nutrient balances and rates of nutrient depletion from soils of agricultural lands at country and regional levels. Figure 3 presents a flowchart of the approach used to integrate the various components into a geo-referenced system for estimating nutrient depletion and nutrient requirements.

Attribute data include crop areas and levels of production, as well as nutrient uptake for ten crop groups that include 90 major food and industrial crops. The crops included in the database account for about 95 percent of the total cultivated area in Africa. Uptake rates for N, P and K for each crop are estimated using data from field studies. The database includes time series data on crop production and areas for the period 1961-1995 (FAO, 1994) and on mineral fertilizer consumption by country and region for the period 1985-1995. Information on organic fertilizer use and practices is also a component of the database. Combined with information on crop and soil management systems, soil constraints, soil characteristics and climate by region and country, these data have been assembled into a database management system.

FIGURE 3
Geo-referenced system for estimating nutrient depletion and requirements

Methodology

A simple specification of the balance of nutrients (N, P and K) in soils of agro-ecosystems at a country or regional scale is given by the following equation:

(1)

where: Rntn is the quantity of inorganic and organic nutrients remaining in the soil at time tn; APt is the soil inorganic and organic nutrients present at time t; ARDt is the inorganic and organic nutrients added or returned to the soil during the time interval Dt. The RMDt estimate is the plant nutrients removed with the harvested product and residue management during the time interval Dt, and LDt is the inorganic and organic nutrients lost during the time interval Dt. The value of t represents the beginning time period, tn represents the ending time period, and Dt is the time interval between t and tn.

Equation 1 states that if the amounts of nutrients removed from the soil (outflows) are greater than the additions (inflows) either by fertilization or management practices, then the reservoir or stock of nutrients within the soil pool will decline. Exact determination of different soil nutrient pools is difficult because of the complex dynamic and stochastic nature of nutrient transformation processes in the soil system.

The production of crop outputs and residues is used to calculate total crop nutrient uptake from soils. Nutrient depletion and requirements are assessed by calculating and using estimates of nutrient gains attributable to the application of mineral and organic fertilizers and to biophysical processes of deposition, sedimentation and fixation. Information on weather, soil constraints, soil characteristics and AEZs is used to estimate soil nutrient losses resulting from erosion, leaching and volatilization (gaseous losses). Estimates of nutrient gains and losses are developed from assumed soil nutrient transfer functions and from estimation of empirical statistical models.

Empirical nutrient loss models and transfer functions are estimated and used to calculate removal and assess nutrient losses through various mechanisms and processes. Further research and improvements in data should enhance the reliability of these models as predictors of nutrient transfers and losses through various processes. The specification and estimation of these models are described below.

Harvested product (Nu)

The harvest of crop outputs and removal (export) of crop residues are major mechanisms of nutrient removal. Average values of N, P2O5 and K2O uptake were obtained from the literature and experimental data. The nutrient uptake (Nu) in harvested product j and country i was calculated by multiplying total crop production (Cpij) by the crop nutrient uptake index (NIj), expressed in kilograms per tonne:

Nuij = Cpij (NIj)

(2)

Values of crop nutrient uptake indices (NIj) were derived from the literature and from experimental results. These indices were estimated for crop yields of traditional and improved crop varieties under average management conditions.

Crop residues (Nr)

Indices of content of N, P2O5 and K2O in crop residues were obtained from references and field studies. The nutrient removed from the soil by crop residues was calculated by multiplying the nutrient content in the residue (NI) by the crop production data (Cp) for countries and regions, the harvest index (HI) and the approximated percentage of residue left on the soil after crop harvesting (Ref). Thus, the amount of nutrient uptake in the residue removed from soil for a given crop (j) in a country/region (i) is determined by:

Nrij = Cpij (1-HIj) NIj Refj

(3)

where Nrij represents the nutrient uptake in crop residues, in tonnes or kilograms per hectare, depending on the crop production values. Estimates of the amount of residue left on the soil after harvesting and grazing were obtained from references and country reports. The harvest index (HI) measures the proportion of the economically produced part of the biomass that is actually harvested.

Leaching of nutrients (Nl)

Most of the literature on nutrient leaching is confined to information on point observations for N and K, which are variable and difficult to extrapolate. The literature reveals that N leaching can be predicted reliably in an African environment on the basis of information on rainfall, soil moisture content and nutrient content of the soils. Regression models have been estimated to predict nutrient leaching at country and regional levels. The general specification of this model includes as variables: the fertility of the soils expressed as soil fertility class (Fc), the average rainfall (R) for the region/site, and the nutrients applied (Cn). The model was specified as follows:

(4)

where: 100 < R < 3 300; Nli is the amount of leaching of N or K at site i, expressed as percentage of the quantity applied; the parameter estimates a, b1, b2, b3, and b4 measure the effects of site management, soil fertility class (Fc), rainfall (R), and nutrient applied in the form of mineral and organic sources (Cn), respectively. The soil fertility class Fc is included to account for the fertility and management of the soil as determined by soil classification and the availability of soil nutrients. This is assessed broadly as: 1 = low; 2 = moderate; and 3 = high. The parameter is the error associated with the estimation of the model.

Nitrogen gaseous losses (Ng)

Experimental data were used to predict denitrified soil N in Kenyan soils. Losses of N through ammonia volatilization can also occur in tropical areas with high use of fertilizer and organic sources of N. Such losses are influenced mainly by: soil texture, pH, and climate factors. Nutrient losses through both mechanisms are included in calculating N balances. A model has been specified to predict these N losses. This model includes as variables: rainfall, soil fertility class to account for soil factors, and the quantity of nutrients applied as a proxy of N availability. The estimating model used has the same form as that in Equation 4. N losses in the model are measured as percentage of the total N uptake. Parameter estimates a, b1, b2, b3, and b4, have a similar interpretation and meaning as in Equation 4 but, for this purpose, with respect to the measure of N loss (Ng).

Soil erosion (Ne)

There is abundant information in the literature on the amount of soil eroded by water in different areas and soil types of Africa. Many different factors interact to determine the amount of soil loss occurring at a particular time and place. The Universal Soil Loss Equation (USLE) describes the impact of the most important factors (Wischmeier and Smith, 1978). Estimates of soil erosion have been obtained by using the USLE and available data. This model estimates soil erosion in tonnes per acre per year as a function of: a rainfall erosivity index, a soil erodibility factor, topographic factors of slope gradient and length, and a land cover and crop management factor. The cropping and management factor is a composite of: the effects of crops and crop sequence, tillage practices, and the interaction between these factors and the timing of rainfall through the year.

Wind and water can transport soil. Erosion by wind is noticeable in the dry areas of Africa (North and sub-Saharan). Empirical equations have been derived to estimate soil erosion caused by wind. These equations require data on wind velocity, precipitation and moisture indices (Lal, 1985; FAO, 1976). General functional relationships between factors that affect wind erosion have been included in the wind erosion equation. This equation specifies soil loss in tonnes per acre year as a function of: a soil erodibility index, a soil-ridge roughness factor, a climate factor, the field length along the prevailing wind erosion direction, and an index of vegetative cover.

Where reliable information is available, estimations of soil erosion by water can be derived using the soil loss erosion models. However, in this study, very few data were available to use the wind equation or to estimate soil erosion by wind. Enrichment values (nutrient adsorbed on soil particles) were used from empirical models and table of references to convert soil erosion losses to nutrient losses. Estimates of nutrient losses due to erosion were obtained for country and regional levels by using the following regression function model to adjust and predict the amount of nutrient eroded (Ne):

(5)

where: Nei is the percentage of nutrient loss through soil erosion in the selected crop/region; and a, d1 and d2 are parameters measuring the effects of factors that are not included in the models but characterize the Sudano-Sahelian, humid, and subhumid regions, respectively. These factors characterize and are specific to each of the countries/regions. The parameters b1, b2 and b3 measure the effects of the soil fertility class (Fc) and the mineral and organic nutrients applied each cropping season (Cn) on the amount of nutrient eroded. The variable Îi is a random error.

Assessment of nutrient inputs and inflows

In order to assess the use of mineral fertilizers (Mf), information on nutrient applications per country in tonnes of N, P2O5 and K2O was obtained from the FAO database (FAO, 1996). Weighting factors and GIS routines were used to calculate fertilizer use at higher levels of aggregation (region, soil class, land use class, AEZ, etc.).

The data required to calculate organic nutrient inputs (Of) (mainly in the form of animal manure) include: the livestock population; the amount of manure reaching arable land; and the nutrient content of the manure at the time of application. However, additional information is required to estimate recycling of household waste and industrial refuse. Some of these data are often not readily available at country and regional levels.

Information from the literature on type of manure and organic products, rates of application by farmers, and livestock production practices in selected regions and countries was used to estimate the amounts of nutrient inputs provided by the use of organic fertilizers.

Country-level estimates of the amount of nutrient returned to the soil in the form of solid manure were calculated on the basis of: the amount of residue left on the field that is grazed, the nutrient content of the residue, and the fraction of nutrients from the residue that remains inside the animal. The value of this fraction used in the estimations presented in this paper was 10 percent.

The amounts of nutrients that return to the soil by deposition (Nd) are difficult to estimate. Deposition is associated mainly with the levels of nutrients used (and produced) and with the amount of rainfall. Wet and dry depositions were evaluated for selected sites using transfer functions. A model was estimated by using forms of empirical functions from other studies (Stoorvogel and Smaling, 1990; Smaling and Fresco, 1993). In those studies, nutrient deposition is specified as a function of the square root of average annual rainfall. Therefore, the following model was estimated and evaluated in this study:

(6)

where: Ndi is nutrient deposition as a percentage of total nutrients; a, d1, d2 and d3 are parameters of discrete variables included to account for variability due to regional factors; b1 is the parameter measuring the effect of soil fertility on nutrient deposition; b2 is the parameter measuring the effect of rainfall on nutrient deposition; and Îi is the error term.

The mechanism concerning inputs of nutrients due to soil sedimentation (Ns) is particularly important in irrigated areas and on naturally flooded soils. Quantification is a difficult task because of the lack of sufficient information on the nutrient content of sediments. Because of this limitation, values in kilograms per hectare per year of the amounts of nutrients in irrigation water were used for selected regions and crop systems.

Regarding N inputs due to N fixation, information in the literature about the nature of N uptake by crops was used to identify three basic distinctive scenarios determined by the nature of N uptake by crops:

Assessment of nutrient depletion and requirements

The quantity or rate of nutrient depletion is estimated as the difference between the amount of nutrients exported annually from cultivated fields and the amount added or imported annually in the form of fertilizers, manure, fixation, and the physical processes of deposition and sedimentation. The balance of nutrient inflows and outflows (Nbi) per year or nutrient depletion in kilograms per hectare per year for a country (i) and crop (j) is assessed and estimated as follows:

Nbi = Sij (Mfij, Ofij, Nfij) + Si (Ndi, Nsi) - (Sij (Nuij, Nrij) + Si (Nli, Ngi, Nei))

(7)

The calculation of nutrient requirement is indicated by:

Nuri = Sij (Cpij) (NIj) + Sij Nrij + Si (Nli, Ngi, Nei)

(8)

The nutrient requirement (Nuri) is calculated as the amount of nutrient uptake required to achieve a specific target yield without depleting the soil nutrient. The calculated nutrient uptake requirements are minimum requirements. A crop could take up more than Nuri and this would result in increased production or yield or improved quality of the product. As necessary, the model is adjusted by the available soil nutrient content. Furthermore, in order to estimate the amount of a fertilizer product required, the nutrient requirement is adjusted to account properly for the fraction of fertilizer nutrient that is actually taken up by the crop (fertilizer use efficiency).

Average rates of nutrient depletion and nutrient requirements were estimated initially at a macroscale for each country in Africa. Because of significant variability within countries, estimates were calculated for selected areas within countries using more elaborated transfer functions, empirical response models, and geostatistical routines.

Results

Nutrient depletion rates were calculated for all African countries (Table 9). The balances were negative for all countries except Mauritius, Réunion and the Libyan Arab Jamahiriya. The nutrient balance ranged from -14 kg NPK/ha/year for South Africa to -136 kg NPK/ha/year for Rwanda. The N and K losses were associated primarily with leaching, soil erosion and low recycling of crop residues. Losses of P were associated mainly with soil erosion.

TABLE 9
Average level of NPK balances, 1993-95


High (> 60)

Medium (30-60)

Moderate/low (< 30)

(kg NPK/ha/year)

Burkina Faso

Mali

Benin

Algeria

Burundi

Mozambique

Cape Verde

Angola

Cameroon

Nigeria

Central African Republic

Botswana

Côte d’Ivoire

Rwanda

Chad

Egypt

Dem. Rep. Congo

Senegal

Congo

Morocco

Ethiopia

Somalia

Equatorial Guinea

South Africa

Gambia

Swaziland

Gabon

Tunisia

Ghana

Uganda

Lesotho

Zambia

Guinea

United Republic of Tanzania

Mauritania


Guinea-Bissau


Niger


Kenya


Sierra Leone


Liberia


Sudan


Madagascar


Togo


Malawi


Zimbabwe


Discussion

The methodological approach of the study by Henao and Baanante (1999) was based on Stoorvogel and Smaling (1990). The same nutrient flows were used and the calculation of the flows was similar although a different notation was used. The innovative aspect of this study is the link with a database and GIS system, which makes the nutrient-balance calculations much easier and faster. Calculations can be made for each year as the data is based only on FAOSTAT and GIS maps, while Stoorvogel and Smaling (1990) used a unique data set with soil/water classes and LUSs. However, the calculation is still on a country basis and differences within the country are not shown. The GIS and database system offer the possibility to link with decision-support systems and crop growth models, but this has not been done yet.

National soil surface nitrogen balances, OECD

The issue of agricultural nutrient use has been a priority issue for the Organisation for Economic Co-operation and Development (OECD) in developing a set of agri-environmental indicators as part of the analysis of the interactions between agriculture and the environment and the impact of changes in agricultural policy on the environment.

The major environmental issues associated with N surpluses from agriculture include pollution of surface water, groundwater and air. However, a deficiency of soil N can also impair the resource sustainability of agriculture through soil degradation and soil mining, resulting in declining fertility in areas under crop or forage production. In cooperation with the statistical office of the European Communities (EUROSTAT), the OECD is in the process of improving and updating the N balances presented here (OECD, 2001a). The work is also being extended to cover P balances.

FIGURE 4
The nitrogen cycle

Note: Grey arrows represent N inputs and black arrows represent N outputs. The different forms of N are in bold text and the processes of N transformation are in italics.
Source: OECD, 2001a.

Methodology

The OECD soil surface N balance calculates the difference between the total quantity of N inputs entering the soil and the quantity of N outputs leaving the soil annually, based on the N cycle (Figure 4). Therefore, N loss directly from livestock (e.g. ammonia volatilization from stored manure) is not included in the balance, although livestock manure production is a major source of N input; this affects the balance. The excess or surplus N may remain in the soil, leach into groundwater and volatilize into the air.

The estimate of the annual total quantity of N inputs for the soil surface N balance includes the addition of:

FIGURE 5
The main elements in the OECD soil surface N balance

Source: OECD, 2001a.

The estimate of the annual total quantity of N outputs, or N uptake, for the soil surface N includes the addition of:

The OECD soil surface balance calculation is not a gross calculation of all N losses from agriculture (Figure 5). This is because the focus is on N losses to soil and water as volatilization of ammonia from stored manure and livestock housing is excluded from the calculation.

The basic data in the database are preliminary, and data definitions may vary across countries following the definitions in the original surveys. For example, although crop production data generally refer to the normal state of the specific crop unless otherwise stated (e.g. dry weight for cereals, fresh weight for vegetables), forage production may refer to weights with different moisture contents across countries.

The coefficients used for the calculation are preliminary and their derivation may vary across countries. In any case, the definition of coefficients should meet the definition of the corresponding basic data.

The database consists of four parts (Figure 6):

FIGURE 6
Summary of database structure

Basic data on fertilizer/
headage/crops

Coefficients

Quantity of nitrogen

Nitrogen balance

Fertilizers: inorganic and
organic products (excluding
livestock manure)

Fertilizers kg nitrogen/
tonne of fertilizer

Fertilizers

Calculation of the total
nitrogen balance and the
balance/ha

Livestock (number of live
animals)

Livestock manure kg
nitrogen/head/year

Livestock manure


Manure withdrawals from
agriculture, manure stocks
and imports

Manure withdrawals stocks
and imports kg nitrogen/
tonne of manure

Manure withdrawals, stocks
and imports


Harvested crop production

Harvested crops kg nitrogen
uptake/tonne of crop

Nitrogen uptake by
harvested crops


Forage production

Forage kg nitrogen uptake/
tonne of forage

Nitrogen uptake by forage


Seeds and planting
materials

Seeds and planting
materials kg nitrogen/
tonne of material

Nitrogen contained in seeds
and planting materials


Area of legume crops

Biological nitrogen fixation
kg of nitrogen/ha of
legume crop area and
total agricultural area
respectively

Biological nitrogen fixation


Agricultural land use area

Atmospheric deposition kg
nitrogen/ha of agricultural
land

Nitrogen fixed from
atmospheric deposition


The classification system of crops and livestock draws on the original data sets, i.e. national sources, EUROSTAT (for European Union member countries) and FAO.

Disaggregated data are provided where possible, especially for crop and livestock series, in order to facilitate a more accurate estimate of the N balance (e.g. piglets and sows), plus the relevant subtotal (e.g. total pigs). However, where disaggregated data do not exist, then aggregated data are provided (e.g. total pig numbers), together with the corresponding coefficients to convert these data into N composition and quantities.

Countries use different classification systems to record the numbers of live animals, especially for cattle, pigs and poultry.

Fertilizers

This category covers data on apparent inorganic fertilizer consumption and on other organic fertilizers applied to agricultural land, excluding livestock manure, which is treated separately.

Inorganic fertilizer consumption includes:

Organic fertilizers include:

Livestock numbers

This category covers the total livestock inventory of live animals required in the calculation of the N content of livestock manure production. The numbers of live animals include those recorded for a given census day in the year, and do not include the total numbers of animals slaughtered in a given year. The total numbers of livestock slaughtered in a year are reflected in the coefficients used to convert livestock numbers into N content in manure. The livestock categories covered include:

Manure withdrawals, stocks and imports

This category covers data on: livestock manure withdrawn and not used on agricultural land (including manure exports); the increase or decrease of manure stocks intended for use on agricultural land; and manure imported into a country for use on agricultural land. This information provides the basis for calculating the ‘net’ input of livestock manure on agricultural land in given year as follows:

Net input = livestock production - withdrawals + change in stocks + imports

Manure withdrawals represents the amount of manure withdrawn from agriculture and not applied to agricultural land. The volatilization of ammonia and mineralization of N after applying manure to the soil are regarded as a part of nutrient losses (or nutrient surplus) and are not included in this category. On the other hand, destruction of manure and volatilization of ammonia from stored manure, livestock housing and manure-spreading operations are excluded from the balance. The manure categories are:

Harvested crops and forage

This category covers data on: harvested crop production from arable field crops (e.g. cereals); permanent crops (e.g. citrus fruits); forage production, including both harvested fodder crops (e.g. fodder beets); and pasture production from temporary grassland and permanent pasture. The definitions and categories of crops and forage follow closely those used by FAO. While many countries have disaggregated fruit and vegetable production data, these are included only where coefficients exist to convert the particular fruit or vegetable into its nutrient content and composition.

Harvested crops, regardless of their final destination, include those for human consumption, livestock feed, industrial use and seeds:

Crop residues

Where possible, the calculation of the soil surface N balance includes the ‘actual’ utilization or consumption of vegetation from pasture, and excludes that vegetation not utilized by livestock and remaining on pasture. Few countries regularly collect data related to pasture consumption by livestock. However, statistics are more commonly available on pasture area and pasture production, which includes both pasture vegetation consumed by livestock and that remaining in the field. For those countries with data on pasture area alone, pasture production was estimated using an assumed pasture yield figure.

For most countries, pasture consumption was estimated using the number of grazing livestock and average consumption levels per animal, or using pasture production and the consumption-production ratio.

The inclusion of crop residues in the soil surface N balance requires further research. In particular, examination is required with respect to the use of N conversion coefficients, i.e. uptake coefficients cover the N content not only in harvested cereal grains but also in other parts of the plant, which may or may not be removed from the field. Data are not provided in this entry at this stage of OECD work on the balances.

Seeds and planting materials

This category includes data on the major categories of seeds and planting materials covering:

Biological nitrogen fixation

This category covers the planted area of legume crops that contribute to BNF, mainly pulses, soybeans, clover and alfalfa. It is the planted area and not the harvested area of legumes that is relevant as BNF occurs regardless of whether the crop is harvested or not. For example, leguminous crops are often not harvested but ploughed into the field to provide soil N.

This category also covers the land area data, i.e. arable and permanent cropland and permanent pasture, for calculating BNF by free-living micro-organisms in the soil.

Land use

This category covers agricultural land, which is subdivided into arable and permanent cropland and permanent pasture.

Coefficients to convert basic data to N content and composition vary over time and among countries. Where the availability of national N conversion coefficients is limited, the following approach is provisionally used to obtain a consistent set of coefficients:

Fertilizers

This category provides the N composition coefficients to convert quantities of inorganic and organic fertilizers. From its definition (expressed in N contents, not in weight of fertilizer), nitrogenous inorganic fertilizer has a fixed N conversion coefficient of 1 000 kg/tonne. Livestock manure is not included in this category.

Livestock manure production

This category provides the coefficients to convert livestock numbers into N composition in annual manure production. However, regarding these coefficients:

Manure withdrawals, stocks and imports

This category provides N composition coefficients for manure withdrawals (including manure exports), changes in stocks and imports.

Harvested crops and forage

This category provides the N uptake coefficients for converting the production of harvested crops and forage into quantities of N uptake from the field. However, regarding these coefficients:

Seeds and planting materials

This category provides coefficients for converting the quantities of seeds and planting materials into their N composition. Coefficients in this group are not the same as those for crops, which do not concern N composition but uptake (including uptake for by-products, such as stems and leaves).

Biological nitrogen fixation

This category provides coefficients for calculating the BNF from the planted area of leguminous crops and BNF by soil micro-organisms on all agricultural land.

Atmospheric deposition

This category provides the coefficients for calculating atmospheric deposition of N on all agricultural land.

Denitrification

The denitrification process on agricultural land is important for Japan and the Korean Peninsula, where rice production is dominant in the agricultural systems. This process is the release of mineralized nitrogen as gaseous nitrogen (N2), which is deemed to be harmless to the environment as it is a major component of the atmosphere.

Quantity of nitrogen

This category provides the total N content of the inputs and outputs in the soil surface balance in terms of tonnes of N. The N content data in these tables are derived basically from the multiplication of the basic data (fertilizers/headage/crops) by the N coefficients.

The calculation of the soil surface N balance is:

Results

The methodology developed by the OECD (OECD, 2001b) has been converted to a software and database program. The database includes data from all OECD countries for the period 1985-1998. The user can select the data required and calculate the nutrient balances.

Discussion

This case concerns mainly surpluses, which makes it different from other cases that are more typical in Africa. The data needs for the nutrient-balance model are high, which makes a well-functioning statistical office necessary. This may not be a problem for developed countries, but data availability is usually much lower for developing countries.

On the output side, OUT3 (leaching), OUT4 (gaseous losses) and OUT5 (erosion) are not included, which makes the figures in the balance strongly positive. For gaseous losses, denitrification is taken into account, but N2O and NH3 losses from animals, volatilization and burning are not included. On the other hand, sewage sludge and seed and planting material are included.

Soil nutrient audits for China

Sheldrick, Syers and Lingard (2002) have developed a model of the various input and output components of the nutrient cycles of N, P and K that allows national-level nutrient audits and balances to be carried out quickly and with sufficient accuracy to give meaningful results. Sheldrick, Syers and Lingard (2003a) have used this model to calculate nutrient output and input relationships, nutrient balances and nutrient depletion rates between 1961 and 1997.

Methodology

Conceptually, the model is a mass balance in which nutrients exported in crops and livestock are compared with nutrients imported into the soil. It considers the following outputs: arable crops, arable crop residues, animal products and livestock excreta. Inputs are: mineral fertilizers, crop residues, manure, animal feeds, non-livestock waste, BNF and atmospheric deposition. Obtained from FAOSTAT databases, the input information included annual crop production for both the arable and livestock sectors, fertilizers, land use and population.

The model defines nutrient efficiency as the percentage of nutrient input that is recovered as nutrient output in the crop. A nutrient balance is achieved when nutrient output no longer increases with increasing nutrient input. The study estimated nutrient efficiency using input and output data from the model. Based on nutrient audits for 197 countries for 1996, the nutrient efficiency for China for N, P and K was set at 50, 40 and 80 percent, respectively.

Of the crop residues, it was estimated that 40 percent was returned, 25 percent was consumed as animal fodder and 35 percent went to other uses or was lost from the soil nutrient cycle. N fixation was estimated at 65 percent of the total N uptake for pulses and groundnut and 50 percent for soybean. For green manure, 0.6 percent of the total N input was estimated. N fixation by Azolla in paddy rice was neglected. Atmospheric deposition was considered only for N and was set at 20 kg/ha/year. Non-livestock waste was estimated as function of population: 1 000 kg N, 250 kg P and 250 kg K per 1 000 people.

The nutrient audit model contains a detailed submodel to estimate the quantities of livestock excreta produced and recovered as manure (Sheldrick, Syers and Lingard, 2003b). The study considered different livestock categories (cattle, pigs, sheep, goats, horses and poultry). Total numbers of live animals were multiplied by the respective coefficients for the quantity of nutrient contained in excreta per animal per year. The numbers and average weights of animal slaughtered in each country were also reflected in the coefficients used to convert livestock numbers into the quantity of nutrients in livestock excreta.

Nutrient losses such as leaching, gaseous losses and erosion are not estimated directly in the model, but calculated as the difference between nutrient inputs plus nutrients depleted from the soil, and nutrient outputs in the crop. After nutrient depletion rates have been determined from the model, the total nutrient loss can be calculated.

Results

The N balance for China was calculated between 1961 and 1997 (Figure 7). First there was an increasing depletion of N, but owing to the use of large quantities of N fertilizers, this depletion subsequently decreased and came more or less into equilibrium. For P and especially K the balances grew increasingly negative. K depletion increased from 28 kg/ha in 1961 to 62 kg/ha in 1997. Table 10 shows the nutrient input and output flows for China. From the table, it appears that K depletion is highest at 41 percent of total K inputs.

Discussion

The model can readily be used for any country and year because it uses only readily available national databases. However, the model contains several major simplifications that make the results less reliable. The coefficients for crop residue removal are the same for all crops, while, for example, crop residues of cereals are generally used more intensively than those of perennial crops. The main limitation of the model is that it depends on the assumption that nutrient efficiency is a direct function of nutrient input. Nutrient concentrations cannot change as nutrient inputs increase and the model does not allow for the effects of interactions between N, P and K. This makes the total nutrient balance less reliable, because the other losses (erosion, leaching and gaseous losses) are based indirectly on this nutrient efficiency.

FIGURE 7
Nitrogen balances for China, 1961-1997

TABLE 10
Nutrient input and output flows in arable farming in China, 1997

Input flows

N

P

K

million
tonnes

%

million
tonnes

%

million
tonnes

%

Fertilizer

23.3

64.7

4.1

56.8

2.8

14.1

Crop residues

2.8

7.8

0.4

5.5

5.2

26.1

Manure

5.2

14.4

1.8

25.1

3.4

17.1

N fixation

1.0

2.8





N deposition

1.4

3.8





Sewage

1.2

3.4

0.3

4.3

0.3

1.5

From soil

1.1

3.1

0.6

8.3

8.2

41.2

Output flows







Arable crops

12.0

33.4

2.3

31.2

4.6

23.0

Crop residues

7.0

19.4

1.0

13.8

12.9

65.0

Losses

17.0

47.1

4.0

55.0

2.4

12.0

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

The study was carried out as an FAO-commissioned research activity by Wageningen University, the Netherlands, in collaboration with national partners in three African countries (FAO, 2003). The purpose was to revisit and synthesize studies on soil nutrient stocks, flows and balances at macrolevel and microlevel, and to calculate them at mesolevel for a few SSA countries. The project’s ultimate objective was to provide a methodology for national and subnational planners and other mesolevel stakeholders to better articulate and target scale-specific soil-fertility-enhancing measures. The mesolevel part of this study is discussed later in this chapter.

The study examined three countries: Ghana, Kenya and Mali. These countries covered major AEZs and landscapes in SSA with different farming systems. The methodology has been developed in such a way that it can be applied to all SSA countries, because the input data (continental GIS maps and FAOSTAT data) are available for each country. The calculation was performed for N, P and K based on averaged data for the period 1997-99.

Methodology

The methodology is based on Stoorvogel and Smaling (1990), with five inflows and five outflows, but has been updated and made spatially explicit. The data set on LUSs was a unique data set; only FAOSTAT data is now available, which is on a country and crop basis. This made it necessary to create a new approach based on a land use map. As land use is the main driver of the nutrient flows and balance, it was chosen to form the basis for the methodology. A procedure was developed to create a land use map based on suitabilities and showing the most likely crop distribution. This grid map with a cell size of 1 km was combined with other spatial data needed for the nutrient-balance calculation.

The methodology for land use mapping is based on three key steps:

  1. Identify land units with similar topography, climate and soil conditions.
  2. Match properties of the land units with crop requirements.
  3. Disaggregate harvested areas from FAOSTAT over the land units.

For the creation of the land use maps the following input data were used:

IN1: Mineral fertilizer

The input of mineral fertilizer was calculated per crop. A fraction of the total fertilizer consumption per nutrient was given to each crop (total is one). The factors were based on data of the fertilizer use per crop (IFA/IFDC/FAO, 2000). These data were not available for every country. For Ghana and Mali, this study used data from surrounding countries within the same AEZ. The FAOSTAT database yielded the figures for total fertilizer consumption per country.

IN2: Organic inputs

Livestock density maps were available for the major livestock classes, i.e. cattle, small ruminants and poultry. FAO and the environmental research group of Oxford Limited developed the cattle and small ruminant maps (FAO, 2000). The poultry density map was based on the rural population of SSA (FAO, 2001). The number of poultry was presumed to have the same spatial distribution as the rural population. The livestock densities were multiplied by the excretion per animal per year and the nutrient content of the manure. This generated the total amount of nutrients produced per livestock class.

The calculation procedure per grid cell was:

IN2 = livestock density × factor manure × factor management (during grazing) + livestock density aggregated × factor manure × factor management (application from bomas etc.)

where:

This calculation was performed for each livestock group (cattle, small ruminants and poultry) and the values summed.

IN3: Atmospheric deposition

Nutrient input by deposition consists of two parts: wet deposition related to rainfall; and dry deposition related to Harmattan dust. Factors for nutrient contents were calculated based on literature. A map with Harmattan dust deposition values was created by interpolation, based on several literature sources and wind-stream patterns. The amount of dust was derived from this map, whereas the amount of precipitation was derived from the IIASA rainfall map (Leemans and Cramer, 1991).

IN4: N fixation

Input by BNF consists of different parts, i.e.: symbiotic N fixation by leguminous crops, non-symbiotic N fixation and N-fixing trees. From the literature (Giller, 2001; Danso, 1992; Giller and Wilson, 1991; Hartemink, 2001), the percentages of total N uptake through symbiotic N fixation were:

For wetland rice, cyanobacteria fix N, and this study used a value of 15 kg N/ha per year. This value is somewhat lower than most experiments reveal, but the effect of the cyanobacteria is overestimated and it does not occur in all fields. This N fixation occurs only in wetland rice, but in Africa more than 50 percent of the rice area is under upland rice. However, FAOSTAT does not differentiate between wetland and upland rice. Therefore, the amount of N fixation by cyanobacteria was multiplied by a factor for wetland rice. Ghana has 15 percent wetland rice, Kenya 25 percent and Mali 95 percent (Nyanteng, 1986). Not much literature is available for non-symbiotic N fixation and N-fixing trees. This input was estimated on the basis of the amount of rainfall using the following equation (N fixed is expressed in kilograms of N per hectare, and rainfall in millimetres per year):

IN5: Sedimentation

This flow consists of two parts: input of nutrients by irrigation water; and input of sediment as a result of erosion. FAO and the University of Kassel, Germany, have developed a worldwide map of irrigation areas (Döll and Siebert, 2000). The nutrient input was calculated by combining this map with the estimated amount of irrigation water (Stoorvogel and Smaling, 1990), set at 300 mm/ha/year, and the nutrient content of irrigation water (N: 3.3 mg/litre, P: 0.43 mg/litre and K: 1.4 mg/litre). The input by sedimentation was calculated by the “LandscApe ProcesS modelling at mUltidimensions and Scales” (LAPSUS) model, which also provided a feedback between IN5 and OUT5. The model output was the net sedimentation in metres. It was possible to convert this value into nutrient input by combining it with bulk density and nutrient content.

OUT1: Crop products

This study calculated the output of nutrients by crop products by multiplying yield by the nutrient content of the crops. FAO statistics (FAOSTAT) provided data on harvested area, production and, hence, yield for each country.

OUT2: Crop residues

This study calculated the output of nutrients by nutrients in crop residues by multiplying yield with nutrient content of the crop residues and a removal factor. The latter is crop and country specific, and it is based on scarce literature and expert knowledge. The removal factor reflects the type of management. Removal factors for central Kenya, with a high population density and many animals, are higher than those for southern Ghana, where livestock are relatively unimportant. A special form of residue removal is ‘burning’. It is difficult to determine the extent of burning at this macrolevel. Therefore, burning was considered solely for cotton, because farmers normally burn these residues in order to prevent pests and diseases. All N is lost by volatilization and an estimated 50 percent of all K is lost directly through leaching.

OUT3: Leaching

Leaching can be an important outflow for N and K. De Willigen (2000) developed a regression model to estimate the amount of leached N. This model is based on an extensive literature search and is valid for a wide range of soils and climates. A new regression model for K leaching was developed, based on the same data set:

N leaching = (0.0463 + 0.0037 × (P / (C × L))) × (F + D × NOM - U)
K leaching = -6.87 + 0.0117 × P + 0.173 × F - 0.265 × CEC

where:

P =

annual precipitation (mm);

C =

clay (percent);

L =

layer thickness (m) = rooting depth, derived from FAO (FAO, 1998);

F =

mineral and organic fertilizer nitrogen (kg N/ha);

D =

decomposition rate (= 1.6 percent per year);

NOM =

amount of N in soil organic matter (kg N/ha);

U =

uptake by crop (kg N/ha);

CEC =

cation exchange capacity (cmol/kg).

The N-leaching regression model is based on 43 different measurements, where 67 percent of the variance is accounted for (De Willigen, 2000). The equation was edited slightly for perennial crops by multiplying the amount of N in soil organic matter by 0.5. This prevented overestimation of N leaching, because perennials can take up N throughout the year. The K-leaching regression model is based on 33 representative experiments and has an R2 value of 0.45.

OUT4: Gaseous losses

This study developed a regression model to estimate gaseous nitrogen losses. The equation losses through denitrification, consisted of two parts: one regression model for the N2O and NOx and a direct loss factor for volatilization of NH3. The equations were based on literature data for tropical environments. These were derived from a larger data set compiled for a recent study to estimate global gaseous emissions of NH3, NO and N2O from agricultural land (IFA/FAO, 2001). The N2O regression model was based on a data set of 80 experiments and had an R regression model was based on 36 different measurements and had value of 0.45. The NOx an R2 emissions, 73 measurements were available. Of all fertilizer N value of 0.91. For NH3 applied, 11.3 percent is lost, with a standard deviation of 6.2 percent.

OUT4 = (0.025 + 0.000855 × P + 0.01725 × F + 0.117 × O) + 0.113 × F

where:

P = annual precipitation (mm);
F = mineral and organic fertilizer nitrogen (kg N/ha);
O = organic carbon content (percent).

OUT5: Erosion

This study used the LAPSUS model to estimate erosion (Schoorl, Sonneveld and Veldkamp, 2000; Schoorl, Veldkamp and Bouma, 2002). This model simulates the amount of erosion and sedimentation at landscape scale. This method has several advantages: quantitative data is generated, erosion is considered, and sedimentation is taken into account. Main input parameters for the grid-based LAPSUS model are topographical potentials (slope gradients) from a DEM and the evaluation of the rainfall surplus that will generate the overland flow.

The study used the following input data:

The outcome of the model is a net erosion-sedimentation map with units in metres, convertible to tonnes per hectare. It is possible to calculate the loss or gain of nutrients by multiplying soil erosion by the soil nutrient contents and an enrichment factor. Based on literature, the study used the following enrichment factors: 2.3 for N, 2.8 for P, and 3.2 for K (FAO, 1984; FAO, 1986; Khisa et al., 2002). It is possible to derive the nutrient content of the soil from the soil map. As a result of erosion, the rooting depth zone is extended, which means that new nutrients come within reach of plant roots. This study assumed that 25 percent of P and K, which is lost because of erosion, was gained at the rooting zone through this process.

Fallow

The amount of fallow land was calculated by subtracting the total sum of harvested areas from the total arable land. IN1 and OUT1 are not relevant for fallow. IN2 and OUT2 are related, as they comprise grazing animals and the same defecating animals. It is not known whether IN2 should be larger or smaller than OUT2. Not all manure is left on the field (only about 57 percent), but, on the other hand, a lot of animal feedstuff is obtained from sources other than crop residues, and from roadside grazing. Hence, for fallow land, the amount of nutrient input by manure (IN2) was presumed to be equal to the amount lost by grazing (OUT2). All other nutrient flows can be treated equally as for other crops.

Calculation of soil nutrient stocks

The World Inventory of Soil Emission potentials (WISE) database, developed by the International Soil Reference and Information Centre (ISRIC) was the source of all soil data for the macrolevel (Batjes, 2002). 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.

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.

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 available amount of 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, 83 percent of all analyses were performed according to the P-Olsen method. The Bray method was used for 6 percent and the Truong method for 3 percent of all analyses. 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 ppm is high, 200-1 000 ppm is medium and < 200 ppm is low (Landon, 1991). The following regression equation was developed to relate available P (Pa) to total P:

Ptotal = 13 × Pa1.5

Results

TABLE 11
Total nutrient balances


N

P

K


(kg/ha)

Ghana

-27

-4

-21

Kenya

-38

0

-23

Mali

-12

-3

-15

Table 11 presents the total nutrient balance for the three countries. Figure 8 highlights the differences in the nutrient flows between the countries, and shows that erosion is the main cause of the negative nutrient balance for Kenya. The results from the calculation can be linked to the original land use map, which makes it possible to present the results spatially (Figure 9). Linking the database system with a GIS makes it easy to analyse results per crop, nutrient flow and region.

FIGURE 8
Overall nitrogen flows for Ghana, Kenya and Mali

Discussion

FIGURE 9
Nitrogen balance per 1-km grid cell for Ghana

The macrolevel calculation procedure used in this study has undergone a number of important methodological improvements compared with the original continental study by Stoorvogel and Smaling.

First, the methodology was spatially explicit. This made it possible to take the spatial variation of soils and climate into account. It also provided the possibility to indicate areas with high and low nutrient depletion within the country. The procedures for calculating the nutrient flows also underwent significant improvement (Table 12). Finally, the soil nutrient stocks were quantified for each soil unit instead of the three discrete, soil fertility classes based on FAO soil classification orders.

TABLE 12
Improvements in calculation procedure compared with the 1990 study

Flow

Methodological improvements

IN1

Fertilizer use data per crop (IFA/IFDC/FAO, 1999) available

IN2

Livestock density maps and differentiation between cattle, small ruminants and poultry included

IN3

Harmattan deposition map and more literature values available

IN4

N fixation percentages based on much more literature

IN5

Feedback between erosion-sedimentation from LAPSUS model introduced

OUT1

Comparable with the 1990 study

OUT2

Comparable with the 1990 study

OUT3

New leaching models, based on much more data, especially for N (De Willigen, 2000)

OUT4

New regression model, based on much more data from IFA/FAO (2001)

OUT5

Erosion simulated with a dynamic landscape model LAPSUS (Schoorl, Veldkamp & Bouma, 2002)


Previous Page Top of Page Next Page