Previous Page Table of Contents Next Page


Microlevel

NUTMON - nutrient monitoring for tropical farming systems

NUTMON (NUTrient MONitoring) is an integrated, multidisciplinary methodology that targets different actors in the process of managing natural resources in general and soil nutrients in particular. With the NUTMON methodology, farmers and researchers jointly analyse the environmental and financial sustainability of tropical farming systems. The NUTMON-toolbox (manual plus accompanying software) has been developed to integrate the assessment of nutrient stocks and flows with economic farm analyses. It has been tested and applied in diverse AEZs in close cooperation with partners from Kenya, Ethiopia, Uganda, Burkina Faso, China and Viet Nam (Vlaming et al., 2001). More information and the toolbox are available at: www.nutmon.org.

Participatory research techniques, such as resource flow mapping, matrix ranking and trend analysis, are used to obtain the farmers’ perspective. A quantitative analysis generates import indicators such as nutrient flows, nutrient balances, cash flows, gross margins and farm income. Both the qualitative and quantitative analyses are then used to improve or design new technologies that tackle soil fertility management problems and can help improve the financial performance of the farm.

The problem of soil fertility management has biophysical, economic and socio-cultural aspects. From a biophysical standpoint, soil fertility depletion relates to low and untimely or inefficient application of manure and fertilizer, farm management practices that lead to high losses through leaching and erosion, and to the lack of integration of livestock. From an economic standpoint, soil fertility decline relates to short-term economic considerations of farm households, insecure climate and market environment, poor property rights, limited infrastructure and risk management. Socio-cultural aspects are also important because they influence the decision-making of farmers. Farmers’ perceptions, knowledge, creativity and competence are essential elements for the adoption of new technologies. Gender issues also play an important role. Female-headed households often have less access to fertilizers because of cash constraints or because extension systems and marketing organizations ignore them. In order to tackle the different problems of soil fertility decline effectively, the integration of disciplines (soil science, agronomy, animal husbandry, economy and sociology) is a prerequisite, as is the integration of formal science and farmers’ knowledge.

Defined as the judicious manipulation of nutrient stocks and flows in a way that leads to satisfactory and sustained production from environmental, financial and socio-cultural standpoints, integrated nutrient management (INM) is seen as the way ahead. This represents a major shift from traditional fertilizer trials aimed at increased production towards comprehensive solutions in the field of integration of organic and inorganic fertilizers, integration of livestock, soil water conservation, agricultural policies and marketing.

Methodology

The NUTMON approach distinguishes two phases: the diagnostic phase and the development phase (Figure 12). Multidisciplinarity and integration of knowledge systems are important in both phases.

Diagnostic phase

The diagnostic phase is carried out at farm level as farm management decisions are taken at this level. The goal of the diagnostic phase is a participatory analysis of the current situation regarding soil nutrient depletion and economic performance. It entails the application of the various tools in the NUTMON-toolbox, preceded by participatory techniques, such as a participatory rural appraisal (PRA) and participatory resource flow mapping. The NUTMON-toolbox plays a central role in this phase as it quantifies the nutrient flows between soils, crops and livestock. Flows are expressed in kilograms of N, P and K (nutrient flows), but also in monetary values (financial flows). The quantified nutrient flows explain which activities within a farm are nutrient consuming and which are accumulating nutrients, and how and when nutrients flow from one activity to another. The quantified financial flows give insight into the profitability of activities (crops, livestock, fishponds and compost pits) and labour demands.

Soil sampling and analysis provides essential information concerning the current nutrient status of the soils. A variety of existing participatory tools can be used to collect and analyse the perceptions of other stakeholders concerning the current soil fertility problems. The quantitative results of the NUTMON-toolbox, combined with the often qualitative information from the other stakeholders, provide a solid basis for an appropriate, thorough and participatory diagnosis. Products of this phase are: quantified nutrient flows and stocks; financial performance indicators; flow diagrams; ranking of problems and possible solutions; and historical descriptions of farm management. During the process, the perceptions and strategies of various stakeholders (farmers, researchers, extensionists) and biophysical and economic boundary conditions surface, resulting in a common understanding of the problem.

FIGURE 12
Overview of the NUTMON approach and the role of the NUTMON-toolbox

Development phase

The development phase that follows can be executed at two different scales. At farm level, a process of participatory technology development is launched with the aim of identifying and developing technologies to address the problems identified in the diagnostic phase. Based on the diagnosis, farmers prioritize technologies, which are tested on-farm. For example, negative nutrient balances caused by large outflows of erosion and leaching may call for soil and water conservation technologies. A situation where low application levels of external inputs has caused negative nutrient balances may call for changes in the marketing infrastructure to make external inputs more attractive.

The NUTMON-toolbox plays an important role in monitoring and evaluating the impact of applied technologies by providing scientific and quantitative information. Similar to the diagnostic phase, other tools and methods are applied to arrive at an impact evaluation by farmers (De Jager, Nandwa and Okoth, 1998; Vlaming, Gitari and Van Wijk, 1997). At regional level, a participatory policy development process can be launched. The results of a farm diagnosis of the major farming systems in a region are scaled up to regional level and presented to policy-makers. Policy interventions are defined in discussions between farmers, scientists and policy-makers.

In both phases, knowledge and experiences are tapped from both science-based and local knowledge systems in order to arrive at the most appropriate solutions. The process of integration of these knowledge systems results in research capacity building for farmers (learning how to conduct applied research) and researchers (increasing knowledge of farm management practices).

NUTMON-toolbox

The NUTMON-toolbox consists of four modules and two databases that together facilitate nutrient monitoring at the level of individual farmers’ fields and farms as a whole.

The four modules consist of:

The two databases are:

Conceptual framework

Because the complexity of farms does not usually allow for quantification of all nutrient flows and stocks, a conceptual framework has been developed. The framework simplifies reality to the extent that major nutrient flows and pools are included and minor flows and pools are neglected. The framework consists of four major components:

The boundaries of the farm coincide with the physical borders. The lower boundary is the depth to which leached nutrients are assumed to be lost from the system, as defined in the leaching transfer function.

FIGURE 13
The farm concept as used in NUTMON

The farm concept (Figure 13) differentiates the farm in farm section units (FSUs), primary production units (PPUs), redistribution units (RUs) and secondary production units (SPUs). An FSU is a continuous field within the farm, which is assumed to have homogeneous soil properties, slope, flooding regime and land tenure. The FSUs are defined because soil and land characteristics determine some of the nutrient flows (e.g. leaching and erosion). A PPU is a crop activity consisting of one or various crops grown deliberately in one field within the farm. It can include annual and perennial crops, pasture or fallow. An RU is a location within the farm where nutrients are collected or accumulated and from where nutrients are redistributed, e.g. stables, corrals, fish ponds, compost pits and latrines. An SPU is defined as a group of animals of the same breed/species managed by the farmer.

Quantifying nutrient flows

Figure 14 presents the Inflows and outflows that are accounted for at the farm level in NUTMON. These flows are quantified using four methods: (i) asking the farmer; (ii) using transfer functions; (iii) livestock mode; and (iv) assumptions. All nutrient flows are determined in kilograms of nutrient per hectare per year.

IN1 (mineral fertilizer) is determined by asking the farmer and combining the applied quantities with the nutrient contents from the background database.

IN2 (organic inputs) is determined by asking the farmer and combining the applied quantities with the nutrient contents from the background database.

IN3 (atmospheric deposition) is determined using three transfer functions:

N: IN3 = 0.14 × P
P: IN3 = 0.023 × P
K: IN3 = 0.092 × P

where P = annual precipitation (mm/year).

IN4 (BNF) consists of two parts: symbiotic and non-symbiotic N fixation. Non-symbiotic N fixation is determined using the mean annual precipitation P:

IN4 = 2 + (P - 1 350) × 0.005

The symbiotic N fixation is assumed to be a crop-specific percentage of the total N uptake of leguminous species (annual or perennial). The total N uptake is defined as the sum of the amounts of N in the crop product and the crop residues.

IN5 (sedimentation) is the amount of irrigation multiplied by the nutrient content of the irrigation water.

IN6 (subsoil exploitation) is normally ignored because of the difficulties in determining this flow and its small contribution to the total nutrient balance.

OUT1 (farm products) is obtained from the questionnaires and is multiplied by the nutrient content of the crops from the background database.

OUT2 (other organic outputs) is also obtained by asking the farmer and the quantities are multiplied by the nutrient content from the background database.

OUT3 (leaching) is determined by transfer functions. For N leaching, one can choose between the ‘De Willigen 2000 model’ and the ‘Smaling 1993 model’. The De Willigen 2000 model is based on an extensive literature review (De Willigen, 2000).

OUT3 = 21.37 + (P/C × L) × (0.0037 × Nf + 0.0000601 × Oc - 0.00362 × Nu)

where:

P =

annual precipitation (mm/year);

C =

clay content (percent);

L =

rooting depth (m);

Nf =

mineral fertilizer N;

Oc =

organic carbon content of the soil (percent);

Nu =

N uptake by the crop (kg/ha/year).

The Smaling 1993 model is a simple transfer function based on soil and fertilizer N (Smaling, 1993):

OUT3 = (Ns + Nf) × (0.021 × P - 3.9)

C < 35 percent

OUT3 = (Ns + Nf) × (0.014 × P + 0.71)

35 percent < C < 55 percent

OUT3 = (Ns + Nf) × (0.0071 × P + 5.4)

C > 55 percent

where:

Ns =

amount of mineralized N in the upper 20 cm of the soil;

Nf =

amount of N applied with mineral and organic fertilizers;

P =

annual precipitation (mm/year);

C =

clay content of the topsoil (percent).

For K leaching, only the Smaling 1993 model can be used:

OUT3 = (Ke + Kf) × (0.00029 × P + 0.41)

C < 35 percent

OUT3 = (Ke + Kf) × (0.00029 × P + 0.26)

35 percent < C < 55 percent

OUT3 = (Ke + Kf) × (0.00029 × P + 0.11)

C > 55 percent

where:

Ke =

exchangeable K (cmol/kg);

Kf =

amount of K applied with mineral and organic fertilizers;

P =

annual precipitation (mm/year);

C =

clay content of the topsoil (percent).

OUT4 (gaseous losses) consists of two parts: gaseous N losses from the soil, and gaseous N losses related with storage of organic inputs. Gaseous N losses from the soil are calculated as a function of the clay percentage and the precipitation:

OUT4 = (Ns + Nf) × (-9.4 + 0.13 × C + 0.01 × P)

where:

Ns =

mineralized N in the rootable zone (kg/ha);

Nf =

N applied with mineral and organic fertilizer (kg/ha);

C =

clay content (percent);

P =

mean annual precipitation (mm/year).

Gaseous N losses related to the storage of organic inputs (manure and compost) are calculated with a user-defined percentage based on roofs, etc.

OUT5 (erosion) is calculated using the USLE. A hypothetical soil loss per FSU is calculated based on slope, slope length, rainfall, soil characteristics and the presence of soil conservation measures. For each PPU, the hypothetical soil loss (in kilograms per hectare per year) is multiplied by a crop cover factor, the nutrient content of the soil and an enrichment factor.

OUT6 (human excreta) are calculated with a user-defined amount per consumer unit. The human excreta can be distributed into a PPU or RU or are completely lost in the case of a deep latrine.

A livestock model has been developed to estimate: (i) the amount and type of feed consumed by livestock; (ii) the amount and composition of the manure excreted by livestock; and (iii) the distribution of the excreted manure over the various units within the farm. The model can be used for all animal types, but is more elaborated for cattle. The model makes no distinction between nutrients excreted in urine and manure.

FIGURE 14
Nutrient and economic flows that influence the nutrient balance and farm income

Results

The NUTMON methodology has been applied in several studies and projects and many copies of the NUTMON-toolbox have been distributed to institutes in the tropics. Descriptions of projects and results are available on the NUTMON Web site (www.nutmon.org). Results from nutrient monitoring in three districts in Kenya are described in Van den Bosch et al. (1998) and De Jager, Nandwa and Okoth (1998). The sustainability of low-external-input farm management systems was assessed using the NUTMON approach for a case study in Kenya in De Jager et al. (2001). The VARINUTS project (SC-DLO et al., 2000) used the NUTMON approach to determine variations in soil fertility management in five AEZs in Embu District, Kenya.

Discussion

Although based on Stoorvogel and Smaling (1990), the NUTMON approach has been developed out completely at the farm level. This means that it can also serve as a tool for monitoring nutrient flows on farms. The methodology has been converted to a software program with databases and questionnaires. Therefore, the methodology is in widespread use in many farm-level projects. However, high data needs make the model less suitable for a rapid inventory.

Participatory nutrient management in southern Mali

Distinct farming systems associated with different ethnic subregions characterize the agropastoral community of southern Mali. Varying reliance on livestock, inorganic fertilizers and bush fallowing account for significantly different nutrient balances. While nutrient balances are indispensable, they are methodologically complex tools. The study reviewed here outlines some of the potential difficulties arising from assumptions made about soil processes and both spatial and temporal system boundaries (Ramisch, 1999).

The study area was the village of Lanfiéla in southern Mali, an area with sandy loam soils and a relatively high annual rainfall (1 100 mm). Intensive agriculture based on cotton and draught power and large cattle herds coexist within its boundaries. The study divided the area into three groups: village, hamlet and Fulani. Village refers only to the cluster of interconnected compounds at the centre of the cultivated plain; Fulani refers to the semi-sedentary Fulani residents; and hamlet covers all non-Fulani households whose compounds are surrounded by their own cultivated fields.

Methodology

The nutrient balance was based on Stoorvogel and Smaling (1990), but combined with a new participatory approach. This methodological approach is called participatory learning and action research (PLAR) (Defoer, 2000; Defoer, 2002). It consists of four phases (Figure 15); the cycle is repeated on a crop seasonal basis and forms the heart of the long-term engagement between farmers and researchers. The PLAR approach can be compared with the farmer field school (FFS) approach. However, the FFS approach does not deal explicitly with diversity and it does not build on a long-term engagement of farming communities. The PLAR approach is based on four principles that are applicable in each of the phases:

Two levels of investigation are distinguished: the village community or group of farmers, and the farm household. The diagnostic phase consists of eight steps:

FIGURE 15
Participatory Learning and Action Research process

Source: Defoer, 2002.

One of the main elements of the diagnostic phase is the making of a resource flow map by the farmers themselves. This map depicts farm fields and other farm elements, such as kraals and compost pits (Figure 16). The resource flows between fields and other farm elements are drawn as are resources leaving or entering the farm, e.g. crop products and mineral fertilizer. The resource flow maps provide the starting point for constant monitoring and evaluation of the fields over the season.

Steps in the planning phase are:

In the implementation phase the farmers are assisted with:

FIGURE 16
Example of a resource flow map

The last phase of the PLAR process, the evaluation phase has three steps:

The nutrient-balance model

The model determines net surpluses or deficits of nutrients by measuring and summing all of the imports and exports of resources from a given plot (Table 17). Exports that are management-influenced are all those concerning the fate of crop residues. These include whether to: (i) stock them for livestock feed or litter; (ii) compost them directly with other organic waste; or (iii) burn them in the fields (immediately after harvest or later in the season). Residues left in the fields unburnt are often grazed in situ by livestock, and then allowed to decompose under the influence of termites and other processes. Where relevant, the study distinguished between grazing by the household’s own animals and that by animals from other farms. Grazing by one’s own animals is, like stocking or composting, a transfer that potentially keeps nutrients within the same field-herd-household, while grazing by other animals exports nutrients completely outside of that system.

On the import side of the balance sheet, management-related transfers involve all the ‘intentional’ movements of organic matter to the fields from livestock pens or compost pits, as well as the application of inorganic fertilizers. The manure from livestock herds corralled on fields in the dry season is also a management-related input. ‘Management’ also influences the movement of livestock within and across the fields, determining the nutrients introduced ‘in passing’ by grazing animals allowed to use the field as corridors across the landscape even after the residues have been consumed. The environmental transfers are determined largely by regional climate, especially the inputs via atmospheric deposition (dust and rainfall), asymbiotic fixation, and the weathering of parent material. The exports are also driven by factors external to management, but which interact with the management transfers. For example, while largely a function of slope, soil type and rainfall, erosion is also influenced by crop cover and human management.

TABLE 17
Variables considered in the nutrient-balance calculations


Exports

Imports

Management

OUT1 Harvested crop

IN1 Inorganic fertilizer


OUT2 Crop residues

Complex (NPK + SB)


Stocked

Urea


Composted

IN2 Transported to field


Grazed in situ

Compost


Burnt

Household waste


Left in fields

Pen manure



Manure deposited by corralled animals



Manure deposited by grazing animals

Environmental

OUT3 Leaching

IN3 Atmospheric deposition


OUT4 Denitrification & volatilization

IN4 Biological fixation


OUT5 Erosion

IN5 Parent material (sedimentation)

Mineral leaching and gaseous nitrogen losses (through volatilization and denitrification) are also a function of the quantities of nutrients applied. Owing to logistic constraints in the field, these transfers were estimated using the criteria listed in Table 18.

TABLE 18
Mean nutrient values retained for environmental transfers

Transfer

N

P2O5

K2O

IN3 - Atmospheric deposition

5 kg/ha

1.2 kg/ha

3.5 kg/ha

IN4 - Biological fixation (symbiotic)

50% of uptake



IN4 - Biological fixation (asymbiotic)

2 kg/ha



IN5 - Weathering


1 kg/ha

5 kg/ha

OUT3 - Leaching





Cotton

7 kg/ha

1 kg/ha

16 kg/ha


Legumes

15 kg/ha

1 kg/ha

16 kg/ha


Maize

6 kg/ha

1 kg/ha

16 kg/ha

Millets/sorghum

1.5 kg/ha

1 kg/ha

16 kg/ha

OUT4 - Volatilization

(Soils too acid)



OUT4 - Denitrification

10 kg/ha + 30%
applied - 10%
uptake



OUT5 - Erosion

0.76 kg/tonne
sediment

0.26 kg/tonne
sediment

0.46 kg/tonne
sediment

TABLE 19
Sample-wide nutrient balances


Entire sample

Village

Hamlets

Fulani


(256 ha)

(n = 191)

(n = 59)

(n = 13)


(kg/ha)

N

-8.2

-11.9

-4.7

23.3

P2O5

19.5

26.5

35.1

39.4

K2O

8.9

3.3

20.8

74.5

Results

Table 19 summarizes the nutrient balances for cultivated plots. The balances for N and K were significantly higher in the hamlets than in the village. The highest balances were among the Fulani. This system fared well because the cultivated plots were smaller and large cattle herds were able to supply them with abundant manure. Households in the hamlets used larger doses of mineral fertilizer than villagers did, and living directly adjacent to their fields allowed them to nurture their crops better and obtain higher yields. Therefore, the hamlet residents could devote a greater proportion of their cotton income to fertilizers.

Discussion

The nutrient balance was again based on Stoorvogel and Smaling (1990) and it included all five Inflows and all five outflows. The innovative aspect of the study was the participatory approach, where the focus was on the perceptions of farmer groups and not on the INs and OUTs per se. Furthermore, the farmers determined their own nutrient stocks and flows diagram.

Nutrient balances for niche management

Soil fertility management in southern Ethiopia

The broad objective of the study was to examine soil nutrient balances at small spatial scales. Earlier survey reports had indicated declining crop yields, and the farmers attributed this to declining soil fertility. The study set out to explore whether there was any evidence of negative balances of major plant nutrients (N and P) in the area, and whether the balance was related to the AEZ and the socio-economic status of the farmer (Elias, Morse and Belshaw, 1998).

Four case-study farms were selected in each of two AEZs (highland and lowland), representing four socio-economic groups of farmers in terms of their resources: rich, medium, poor and very poor. Differentiation of households into socio-economic groups was carried out by the farmers utilizing a wealth-ranking exercise based on local criteria centred primarily on draught oxen ownership and livestock herd size. Draught oxen ownership is the major local indicator of wealth and is a central criterion in any attempt to classify households. The differentiation of households into socio-economic groups was:

Methodology

In order to assess declining soil fertility, this study used the nutrient balance rather than the technically more difficult approach of comparing changes in soil nutrient stocks (Pieri, 1983). With the nutrient-balance approach, the quantities of nutrients entering and leaving a field are estimated, and the balance (input - output) calculated. Balances were calculated for all fields within the farms, and the farm balances were calculated by aggregating input and output data for all fields. This study examined only N and P as they are the two nutrients identified as particularly deficient in Kindo Koisha soils. The N and P balances were calculated from a combination of four input and five output processes. The input flows were:

Sedimentation, identified as IN5 in the original model, is not relevant as there are no irrigation schemes or flood plains in Kindo Koisha.

The output flows were:

The eight farmers included in the case studies used diagrams to identify the perceived key nutrient input and output flows on their farms. These flows were measured over one production year in order to produce a nutrient balance sheet for each field. Quantification of N and P in the input and output flows was achieved through a combination of different methods: field measurement, use of empirical quantitative relations (i.e. transfer functions), and assumptions based on secondary data from a variety of sources. Tables 20 and 21 summarize the type of data required and the method of quantification for each of the input and output functions.

For both N and P, primary data were obtained as applicable on type and quantity applied of fertilizer, manure, households refuse and leaf litter. The number of baskets of manure transported and the site of application were monitored on a daily basis, and the fresh weight of manure per basket was measured. Composite samples of fresh manure from the livestock pen were collected and analysed for moisture content and composition of N and P at the International Livestock Research Institute (ILRI). The manure samples were oven dried at 105 °C before analysis, and moisture contents of 60 percent (highland) and 50 percent (lowland) were used to convert manure input into nutrient input.

Local data on atmospheric deposition (IN3) were not available in the area, hence atmospheric deposition of N and P were calculated as the square root of the average annual rainfall using the regression equation derived by Stoorvogel and Smaling (1990). The regression coefficients were 0.14 for N and 0.023 for P. Haricot bean provided the only N input from biological fixation (IN4). Grain and residue yields of haricot bean were measured in the field, and the nutrient composition of these products was determined through chemical analysis at the ILRI. Like Smaling, Stoorvogel and Windmeijer (1993) working with this crop on Nitosols in Kenya, it was assumed that 50 percent of the bean N requirement was derived from biological fixation.

TABLE 20
Type of data required and quantification method for the four input processes employed in calculating N and P balances

Input
process

Code and
nutrients

Data required

Method of
quantification

Mineral fertilizer

IN1 (N & P)

Type of fertilizer applied

Field measurement



Amount of fertilizer applied

Field measurement

Manure

IN2a (N & P)

Amount of manure applied

Field measurement



Nutrient content of manure

Laboratory analysis

Leaf litter

IN2b (N & P)

Amount of leaf litter collected

Farmer estimation



Types of trees used for leaf
collection

Field observation



Nutrient content of litter

Laboratory analysis

Deposition

IN3 (N & P)

Average annual rainfall

Rainfall records



N and P deposition in rainfall

Transfer functions

BNF

IN4 (N only)

Type of legume grown

Field observation



Grain and residue yield of legume

Field measurement



Nutrient content of grain and residue

Laboratory analysis



Percentage of uptake attributed to
symbiotic fixation

Secondary data

TABLE 21
Type of data required and quantification method for the five output processes employed in calculating N and P balances

Output process

Code and
nutrients

Data required

Method of quantification

Harvested product

OUT1

Crop yield

Field measurement


(N & P)

Nutrient content of product

Combination of laboratory
analysis and estimations

Crop residue

OUT2

Residue yield

Field measurement


(N & P)

Destination of residue

Field observation



Nutrient content of residue

Combination of laboratory
analysis and estimations

Leaching &
denitrification

OUT3 & OUT4

Average annual rainfall

Rainfall records


(N only)

N in applied fertilizer

Field measurement



N in applied manure

Field measurement records



Leaching & denitrification of
soil N & applied N

Estimate (transfer function)

Erosion

OUT5

Average annual rainfall

Rainfall records


(N & P)

Erodibility (K)

Secondary data



Slope length (L)

Estimate based on field
measurement



Slope gradient (S)

Secondary data



Land cover (C)

Secondary data



Management factor (P)

Secondary data



Nutrient content of sediments

Secondary data

Removal of nutrients in crop products (OUT1) and residues (OUT2) were quantified through a combination of primary data and estimates based on secondary data. Subsamples of harvested products and residues of maize, enset, teff and haricot bean were analysed for N and P content at the ILRI laboratory. This was necessary because reported values for N and P composition of maize vary considerably in the literature, and secondary data were not available for N and P content of enset, teff and haricot bean. The N and P composition of sweet potato, taro and sorghum were estimated using FAO data. Removal of nutrients in crop residue was quantified by taking into account the fraction of residues removed from the field for feed and fuel. In the highlands, about 80 percent of crop residue was completely removed from the field, but in the lowland the proportion was 30-50 percent. The high and low ranges of nutrient composition of maize were determined using the mean values of the lowest and highest quartiles of 15 data points from several countries in SSA.

No quantitative information was available on leaching and denitrification within the study area or in comparable AEZs nearby. Therefore, N loss through leaching (OUT3) and denitrification (OUT4) were estimated using the transfer function of Smaling, Stoorvogel and Windmeijer (1993). These authors derived multiple regression equations for OUT3 and OUT4 using the generally accepted determinants of rainfall, soil texture (clay content), soil N and application of fertilizer (IN1) and organic matter (IN2). The multiple regression equations are of the form:

OUT3 = 2.3 + (0.0021 + 0.0007 x F) x R + 0.3 x (IN1 + IN2) - 0.1 x UN

where:

F = soil fertility class, highland soils assumed moderate (2) and lowland soils low (1);
R = rainfall (annual average, in millimetres);
UN = total N uptake (in kilograms per hectare);

OUT4 = X + 2.5 x F + 0.3 x (IN1 + IN2) - 0.1 x UN

where X = ‘relative wetness’, an LWC specific fixed value estimated at 5 kg N/ha/year for uncertain rainfall areas of Africa such as Kindo Koisha.

Volatilization of ammonia and burning can also cause gaseous nitrogen losses. However, volatilization is generally recognized as negligible in crop fields of highly-weathered acidic soils of east Africa. Therefore, it was not included in this study. In the study area, crop residue is used intensively as feed and not burnt in the field. Therefore, nutrient losses through burning were assumed to be negligible.

Soil erosion only occurs in the highlands of Kindo Koisha as the lowlands are mostly fl at. Soil loss from erosion was estimated using the simplified and adapted version of the USLE (Hurni, 1985). The equation predicts soil loss as a function of rainfall erosivity, soil erodibility, slope length, slope gradient, land cover and land management (as explained above).

Nutrient loss in the eroded sediment was calculated using the total N and P composition of eroded sediments determined by Belay (1992) for the study area: 0.22 percent total N and 0.07 percent total P.

Values for N and P balances estimated using the procedures described above are regarded as the most probable because they have been calculated using the most likely assumptions for Kindo Koisha. However, some of the parameters and processes, such as atmospheric deposition (IN3), leaching (OUT3), denitrification (OUT4) and erosion (OUT5), were estimated from secondary data, within which there is some variability. In order to incorporate some of this uncertainty, ‘optimistic’ and ‘pessimistic’ values (based on assumptions considered to be extreme for the area) were also calculated. The result was an uncertainty range likely to encompass the ‘real’ value. The procedure for quantifying optimistic and pessimistic values follows the one used by Van der Pol (1992). The optimistic balance is calculated by combining high estimates of nutrient Inflows and low estimates of nutrient exports. Conversely, the pessimistic values combine high values for export with low values for input.

Because of the lack of data for other crops, the high and low estimates of N and P export in harvested products and residues of maize were used to estimate ranges for OUT1 and OUT2. The optimistic and pessimistic values of atmospheric deposition and leaching were derived using high and low ranges of rainfall in the regression equation. The high and low values for the rainfall erosivity factor (R) adapted for the area was used to calculate the optimistic and pessimistic ranges of erosion (OUT5). The optimistic value was calculated by using the lower range of the rainfall, which corresponds with an R factor of 441, and the pessimistic value was calculated using the high rainfall range, which gave an R factor of 890.

TABLE 22
Farm nutrient balances for different household groups


Households

N

P



(kg/ha)

Highland

Rich

-47

11.7


Medium

-51

4.8


Poor

-19

3.6


Very poor

-6

1.1

Lowland

Rich

-49

30.5


Medium

-41

17.3


Poor

-55

3.8


Very poor

-20

-1.6

Results

The N balances were negative for all household groups, while the P balance was positive for most farms (Table 22). Poorer farmers had lower N depletion rates, which may seem contradictory. However, they can compensate for lower mineral fertilizer inputs by intensive soil enriching and nutrient conserving practices, including: rational use of available manure; systematic management and recycling of crop residues; collection of leaf litter; and improved soil conservation. The differences between the farm components were very large. The enset, taro and darkoa (homestead) fields received many inputs and therefore had a positive or neutral balance, but the shoka (out fields) had very negative balances owing to low inputs.

Discussion

The methodology for the nutrient-balance calculation was based on Stoorvogel and Smaling (1990), and all five Inflows and all five outflows were calculated. The additional value of this study is the calculation per household group and per farm component (enset garden, taro root, darkoa and shoka fields). This shows the diversity and complexity of the farming system in the Kindo Koisha area of Ethiopia and the impact of the different management of each social class.

Banana-based land use system in the northwest of the United Republic of Tanzania

Farmers in Bukoba District, in the northwest of the United Republic of Tanzania, are facing a continuous decline in crop productivity. Nutrient flows in land use systems are not well documented. The study, supported with data collection, presents the nutrient balances for various AEZs. The objectives of the study were: (i) quantify nutrient flows of the home garden; (ii) assess the sustainability of the banana-based land use system in different AEZs; and (iii) identify possibilities and set strategies for increasing nutrient use efficiency (Baijukya and Steenhuijsen de Piters, 1998).

The Bukoban agro-ecosystem is characterized by a combination of: a banana-based home garden (kibunja); small fields with annual crops (kikamba), usually of no permanent character; and grasslands (rweya). The study considered nutrient balances of the banana-based land use systems in the kibanja because this LUS produces the vast majority of agricultural produce of the farm households.

Methodology

The authors of the study adopted the nutrient-balance calculation model of earlier literature-based studies (e.g. Janssen, 1993) and verified or modified their data. Nutrient flows were quantified according to the nutrient-balance model proposed by Stoorvogel and Smaling (1990).

Data on nutrient inputs (IN1 and IN2) and outputs (OUT1 and OUT2) through harvested crops (bought, consumed and sold), and the use of grass and ash produced in the households, were collected in three villages in three AEZs. These villages represented variations found in the district. Data on farm size, bean and coffee production were available from earlier work in these areas.

Fifteen farmers per village were selected for data collection; six were monitored closely and the others were visited regularly as ‘check farmers’. Cattle and non-cattle owners were considered as distinct categories of farmers, which were included in the study. The closely monitored farmers were provided with scales (to weigh the bananas and root crops they harvested and grasses they used) and ledgers (to keep records of crops and grasses they used). The ‘check farmers’ were interviewed on grass use and banana production, and samples of banana bunches and grass bundles were weighed for verification. The closely monitored farmers were visited at two-week intervals from August 1993 until October 1994.

Different samples of banana (pulp, peal and stalk), root and tuber crops, and grasses (categorized into mulch, carpet and brewing) were collected from the respective villages. The collected samples were dried to determine their dry matter, and a part of the dried samples were analysed on total N, P, K, Ca, Mg and sulphur (S). Data on nutrient input through application of farmyard manure were available from other studies conducted in the area. Local data on wet deposition were not available. Data on average nutrient contents of four rainwater samples were collected at the research station in order to predict wet deposition. Nutrient input was linked to the concentration of the individual element in rainwater and the mean annual rainfall received in different zones. Inputs through dry deposition were assumed to be negligible given the humid environment.

Common bean (Phaseolus vulgaris) is the only leguminous species grown in the kibanja. N contribution by beans through biological N2-fixation (IN4) was estimated as 50 percent of total plant uptake in aboveground biomass. The contribution to the N balance through asymbiotic N fixation was estimated using annual rainfall data of each zone. Input of nutrients through sedimentation (IN5) was not considered important in the perennial home gardens.

Data on nutrient losses of home gardens through leaching (OUT3) were available for the Bukoban high rainfall zone (Van der Eijk, 1995). Nutrient losses were calculated on the basis of nutrient concentrations in the percolating water. The rainfall, rain days, rain months and potential evapotranspiration (PET) data were used to calculate the percolation water in different AEZs.

The PET for Bukoba was reported to average 3.5 mm/day, and the rain days 260, 220 and 180 for the respective zones. For the Bukoban high rainfall zone, the percolating water was found to be 990 mm/year and the leaching index was assumed to be 1. For the Karagwe-Ankolean low rainfall zone (annual rainfall of 900 mm/year), the percolating water was found to be 270 mm/year and the leaching index 0.27. Applying the same procedure, the leaching index for the Bukoban medium rainfall zone was estimated at 0.64. Nutrient losses via leaching for the Bukoban high rainfall zone were extrapolated to the other zones using the calculated leaching indices. S losses were estimated on the basis of Ca:Mg:S ratios of Van der Eijk (1995) and Umoti, Atage and Isnemila (1983).

Denitrification was considered to be the most important process through which gases are lost (OUT4). Gaseous losses through volatilization were not considered important as alkaline soils are rare found in Bukoba. The percentage of mineralized soil N was calculated first by determining the fraction of soil organic matter that decomposes annually (k) and the humification coefficient of fresh organic matter (h). It has been reported (Janssen, 1984; Janssen, 1993) that k is dependent on temperature and h on the nature of fresh organic matter. The mean annual temperature in Bukoba is 21 °C and the k value was assumed to be 5 percent. The dominant fresh organic matter forms applied in the home garden are banana residues, grass and farmyard manure. Their h values were assumed to be 0.2, 0.3 and 0.5, respectively. The relationship between soil organic matter (SOM), effective organic matter (OM), fresh organic matter (FOM), k and h was reported to be:

OM = k × SOM = h × FOM

Using the above information, and by assuming a soil bulk density of 1.25 g/cm3 and a carbon to nitrogen ratio of 11, the mineralization for home garden soils in different zones was calculated.

The denitrified soil N (DN soil; percentage of mineralized N) was calculated using a transfer function:

DN = -9.4 + 0.13 × clay content + 0.01 × annual rainfall

For soil with N mineralization of 330 kg/ha/year, 25 percent clay and 1 900 mm of rain, the DN is 13 percent. Thus, the N loss is 42 kg/ha/year.

Nutrient output through erosion (OUT5) was not considered important because of the absence of traces of erosion in farmers’ fields.

The major nutrient fluxes in home gardens in the different AEZs were translated into a general input-output model (Table 23). The nutrients considered in the nutrient-balance calculations were N, P, K, Ca, Mg and S. Except for P, which is available in abundance, these nutrients were reported limiting in most Bukoban soils. The determinants used to calculate the balances were mostly scale-neutral. Therefore, they can be used to calculate the balances at plot, farm and village levels.

Results

Table 24 shows that nutrient balances were negative for home gardens without cattle and positive for those with cattle. These results suggest that intensification of cattle is a solution for declining soil fertility. However, the results are misleading to a certain extent. The home garden is only one component of the farming system, and grasslands have to produce the vast amount of cattle feed required, which causes exhaustion of the soil.

TABLE 23
Nutrient flows at farm level

Flows



Nutrients

Input

IN1


Mineral fertilizers


IN2


Organic inputs



IN2a

Grass (mulch, carpet and brew)



IN2b

Concentrates for dairy cattle



IN2c

Fodder grasses fed to dairy cattle



IN2d

Manure from indigenous cattle grazing outside the farm


IN3


Atmospheric deposition in rain


IN4


BNF by beans and free-living bacteria


IN5


Sedimentationa


IN6


Subsoil exploitation by coffee and other perennial treesb

Output

OUT1


Harvested crops, banana, coffee, beans, roots and tubers


OUT2


Crop residues and manure leaving the farma


OUT3


Leaching below the rootzone


OUT4


Gaseous losses


OUT5


Runoff and erosiona


OUT6


Human faeces in pit latrinesb

a Not relevant in kibanja system.
b Not considered in the present study.

TABLE 24
Nutrient balances of banana farms

Zone

Farm nutrient management
level *

N

P

K

(kg/ha/year)

Bukoban high rainfall

1

-76.2

-4.9

-50.0


2

-73.9

4.2

-41.2


3

-7.5

10.8

-6.4


4

7.0

12.3

15.5


5

80.5

42.8

198.7

Bukoban medium rainfall

1

-49.0

-1.7

-39.8


2

-45.0

-1.0

-22.8


3

-6.7

8.0

-4.8


4

1.7

8.8

4.3


5

30.8

23.5

90.9

K-A low rainfall

1

-27.9

-2.7

-30.1


2

-25.1

-2.0

-20.6


3

-8.7

1.6

-15.1


4

-3.9

2.4

-8.8


5

11.0

8.9

32.1

* Banana farm management level: 1 = farm with no cattle and without brewing; 2 = farm without cattle but brewing; 3 = farm with indigenous cattle but use no bedding; 4 = farm with indigenous cattle and use bedding; 5 = farm with improved (zero-grazing) cattle.

Discussion

This study is an example of niche management, in this case the banana-based farming system in the United Republic of Tanzania. The methodology was based on Stoorvogel and Smaling (1990) and not adapted significantly. Again, a differentiation between farmers was made, in this case between the intensity of cattle keeping in combination with the banana-based system.

Land use types in eastern and central Uganda

The aim of the study was to estimate the nutrient balances at the crop, LUT and farm levels, and to estimate the impact of adopting alternative practices on nutrient balances and productivity. Nutrient balances were estimated for small-scale farming systems at four subhumid, medium-altitude locations in eastern and central Uganda. Nutrient flows were estimated using data from several sources, including farmer interviews, observations of the farming systems, soil analyses, and the output of simulation models (Wortmann and Kaizzi, 1998).

A survey was carried out in four districts of central and eastern Uganda during the second season of 1995 in order to gather the data needed to estimate nutrient balances at the field and farm levels. The characteristics of the locations differ but there are similarities: two major cropping seasons with mean annual precipitation of 1 050-1 300 mm, similar mean temperature, and similar crops (although their relative importance varies). Land use was divided into seven categories: banana-based systems, annual cropping systems, fallow, pasture, tree lots, napier grass plantings, and home gardens.

Methodology

Nine or ten farmers were interviewed at each location; where feasible, the validity of their responses was verified through observations. Their farms were mapped in order to show the size and use of various parcels of land. Detailed observations made on three parcels per farm included: slope, slope length, soil physical and chemical properties (texture, organic carbon, pH with 1:1 water, Olsen P, CEC, and total amount of N, P and K) at 0-20 cm and texture at 20-40 cm depth. Farmers were interviewed in detail concerning the use of the parcels of land and nutrient flows to and from the parcels. Observations were made and questions asked about other aspects of nutrient flows on a whole-farm basis including management and utilization of household wastes and farmyard manure, and sale and purchase of different commodities.

Farmers were asked to give mean yield estimates for the crops growing on parcels studied in detail. Their estimates covered an unrealistically wide range and were considered to be generally unreliable. Therefore, mean yield estimates were made based on government statistics and researchers’ experiences with the crops in these areas. Nutrient contents for some commodities were determined through the analysis of materials collected in central Uganda. For other commodities, values reported elsewhere were used.

Soil erosion losses were estimated using the USLE. Nutrient enrichment of the runoff was assumed to be 1.5. No attempt was made to estimate sedimentation although it may be significant in the fallow and pasture LUTs. Leaching, volatilization and denitrification losses were estimated using the CERES Maize model (Ritchie et al., 1989) with three soil profile descriptions, four seasons of typical rainfall, and sowing on 1 March and 15 August. The CERES Maize model is able to capture nuances in daily weather data and estimate their effects on N flows as a function of characteristics of a one-dimensional, multilayered soil profile and crop management conditions. Rainfall during the five-month periods of simulation ranged from 374 to 591 mm. The estimates of N losses were used in N balance calculations for all crops. The Quantitative Evaluation of the Fertility of Tropical Soils (QUEFTS) model (Janssen et al., 1990) was used in the interpretation of soil test data. The QUEFTS model estimates nutrient availability during a season and soil productivity in terms of maize yield using data for organic carbon, soil pH in water, Olsen P, exchangeable K, and the totals for N, P and K.

It was assumed that ash and dry household waste produced per family were 20 and 100 kg/ year. For grazing livestock, 50 percent of the faeces and urine were estimated to be deposited in the grazing areas, where N loss to volatilization was 10 percent. Nutrient losses from the farmyard manure were estimated to be: 50 percent of N, and 20 percent of P and K for livestock kept in open pens; and 20 percent of N, P, and K for those confined in a covered structure. The burning of manure resulted in the loss of 80 percent of the N and 20 percent of the P and K; 25 percent of the remainder was lost due to erosion and leaching.

Annual human consumption rates of N, P and K were assumed to be 4.00, 0.36 and 6.00 kg per capita. Although some of these nutrients are recycled through plant growth, no attempt was made to estimate the amount. Nutrients consumed by people were considered lost to the system. Burning of bean and soybean crop residues is common in Palissa District after threshing the crop at home. The ash is commonly used in cooking. The burning of crop residues was estimated to result in the loss of 80 percent of the N and 20 percent of the P and K.

Results

TABLE 25
Nutrient balance for major crops

Crops

N

P

K


(kg/ha/year)

Banana

-13.2

1.2

-35.7

Maize

-104.2

-13.6

-82.4

Bean

-40.4

-8.8

-42.7

Sweet potato

-71.3

-13.2

-78.9

Soybean

-121.5

-16.4

-68.3

Fallow

33.2

-1.5

-13.7

Pasture

19.2

-3.3

-30.7

Home garden

3.0

-1.8

-18.9

The small-scale farms in eastern and central Uganda were biologically, agronomically and economically diverse. However, the nutrient balances were negative in all locations, showing that the current systems are not sustainable even with the low productivity. The nutrient balances in the banana-based LUT were near neutral (Table 25), because of transfer of organic material from other LUTs. The annual crops had high nutrient losses because of removal with harvested products and erosion.

Discussion

The nutrient balance was based on Stoorvogel and Smaling (1990), but some flows were calculated differently. Leaching and gaseous losses were estimated with the CERES Maize model, which uses local soil data. This probably results in a better estimation than by using transfer functions as it is based on the local circumstances. This study made a greater differentiation in the organic inputs; mulching or application of crop residues, farmyard manure, ash and household waste were treated separately.

A sisal plantation in the United Republic of Tanzania

Hartemink (2001) describes several case studies on soil fertility decline in the tropics. The study emphasizes the importance of hard data on soil property changes in relation to soil fertility decline. It compares the results of the nutrient balances with actually measured soil changes. The case study on sisal plantations is based on experimental work in the Tanga Region of the United Republic of Tanzania. Sisal is an introduced fibre crop and is grown mainly on large plantations (Hartemink and Van Kekem, 1994).

Methodology

Two different approaches were used to monitor soil chemical properties. In one approach, soil dynamics were monitored over time at the same site. This approach is called chronosequential sampling or Type I data. Type I data show changes in a soil chemical property under a particular type of land use over time. In the other approach, soils under adjacent different land use systems were sampled at the same time and compared. This approach is called biosequential sampling or Type II data. The underlying assumption is that the soils of the cultivated and uncultivated land are the same soil series, but that differences in soil properties can be attributed to the differences in land use.

Soil properties of permanently cropped fields were compared with historical data from the 1950s or 1960s from the same field (Type I data). Topsoil samples were taken in sisal fields and in similar soils immediately outside the plantation that had never been cropped (Type II data).

A nutrient balance was calculated for a sisal field that had been cropped permanently since 1957. Yield and soil data (Rhodic Haplustox) were available from 1966 to 1990. The balance included the following nutrient inputs: wet deposition, non-symbiotic N fixation, and nutrients added with the planting material. Mineral fertilizer or organic inputs were not applied on the sisal and were therefore not taken into account. The wet deposition (part of IN3) and non-symbiotic N fixation (part of IN4) were calculated according to Stoorvogel and Smaling (1990). The input with planting material is necessary for sisal as at the beginning of a cycle thousands of small sisal plants (about 2 kg each) are brought to the field. The only nutrient output that could be fairly well quantified was removal with the harvested products. Crop residues were not removed from the field. Erosion was negligible, because sisal is a perennial crop with a grass cover between the rows.

Results

The resulting nutrient balances were negative for all nutrients (Table 26), especially K and Ca. The negative balance was confirmed by the decline in nutrient in the topsoil (0-20 cm) for all nutrients. For most nutrients, the nutrient balance was more negative than the actual soil changes. Only for N were the soil changes much more negative. This might be explained by the omission of important outflows, i.e. leaching and gaseous losses. Inclusion of these flows should make the nutrient balance for N more negative, while these flows are not so important for the other nutrients.

TABLE 26
Nutrient balance and soil nutrient content of a sisal field, 1966-1990


N

P

K

Ca

Mg

Input with rainfall (kg/ha)

115

19

75

213

105

Input with BNF (kg/ha)

19

0

0

0

0

Input with planting material (kg/ha)

32

10

35

87

13

Output with yield (kg/ha)

491

100

1 067

1 400

605

Difference (kg/ha)

-326

-71

-957

-1 100

-487

Nutrient balance (kg/ha/year)

-13

-2.8

-38

-44

-19

Content in 1966 (kg/ha)

5 764

52

369

996

355

Content in 1990 (kg/ha)

3 144

8

82

271

97

Difference (kg/ha)

-2 620

-44

-287

-725

-258

Soil changes (kg/ha/year)

-104

-1.8

-11

-29

-10

Source: Hartemink, 2001.

Discussion

This study is one of the few that deal with plantation crops. Soil fertility improving measures may have more impact on plantation crops because of better investment opportunities. Another interesting aspect of this study is the comparison of measured soil changes with nutrient balances. The study concludes that hard data is necessary for validating nutrient balances and improving understanding of soil processes. It also emphasizes the importance of long-term field experiments. The nutrient balance was based on Stoorvogel and Smaling (1990), but several flows were not included because of data availability issues (leaching and gaseous losses) and irrelevance (mineral fertilizer, organic inputs, erosion and sedimentation).

Soil fertility management in southern Mali

This study by Kanté (2001) is described in “Scaling soil nutrient studies” (FAO, 2003) together with the VARINUTS project (SC-DLO et al., 2000) as representative microlevel studies. Microlevel studies provide a picture of the variation within a mesolevel unit. Relevant management factors can be included, and monitoring can check whether changes in nutrient management have a bearing on nutrient balances and farm income. In this particular study, farmers were classified in three ‘soil fertility management’ classes, instead of the ‘average’ farmer. The study focused on two villages in the cotton zone of southern Mali (M’Peresso and Noyaradougou), with strong and moderate pressure on land respectively.

Methodology

The study followed a participatory approach according to the PLAR methodology (explained earlier). Farm households classified one another into three nutrient management groups (1 = good management, 3 = poor management). The partitioning largely reflected the number of household members, possession of animals and manure, and carts. The classification was evaluated annually with farmers being promoted or relegated to another class.

The nutrient balance was based on Stoorvogel and Smaling (1990), but used mainly ‘partial balances’ for comparisons between farmers and villages. The partial balance included IN1 (mineral fertilizer), IN2 (organic inputs), OUT1 (harvested products) and OUT2 (crop residues). These flows are the ones that are most management related and they are also called ‘easy to measure’ nutrient flows. These ‘easy to measure’ flows can be quantified from farm survey data and they can also be expressed in monetary or labour units. The ‘difficult to measure’ flows (IN3, IN4, IN5, OUT3, OUT4 and OUT5) are not normally measured but estimated with transfer functions. IN2 was subdivided into animal manure and compost, and OUT2 was subdivided into crop residue removal by animals, removal by households, and burning.

Results

At first glance, both villages have comparable farming systems. Cotton is the basic cash crop and cereals such as maize, sorghum and millet the major food crops. Livestock is very important. However, a closer look shows that the pressure on land is considerably higher in M’Peresso (higher population density, higher ratio of cultivated land to total land) and, as a consequence, the management of crop residues is more intensive in M’Peresso (Table 27). Similarly, one might say Noyaradougou has a labour shortage, which does not allow the villagers to recycle all crop residues. Manure application is higher in M’Peresso, while farmers in Noyaradougou use more mineral fertilizers to compensate (Table 28). Therefore, the partial nutrient balance is more positive in Noyaradougou.

TABLE 27
Observed differences between two villages, Mali


M’Peresso

Noyaradougou

Fallow/cultivated land ratio

0.6

1.4

Total N in soil (g/kg)

0.20

0.31

Total P in soil (mg/kg)

126

171

Availability organic manure (tonne)

26

11

Mineral fertilizer use on cotton (kg/ha)

102

155

Crop residues as animal feed (%)

35

15

Crop residues as compost (%)

16

43

Crop residue burning (%)

3

16

Partial N balance for cotton (kg/ha)

58

22

Partial N balance for maize (kg/ha)

-30

2

TABLE 28
Partial nutrient balances for two villages, Mali


M’Peresso

Noyaradougou


N

P

K

N

P

K


(kg/ha)

IN1

15.3

4.0

4.3

41.9

8.3

10.4

IN2

16.8

3.3

22.7

10.8

2.0

14.6

OUT1

18.7

2.2

4.7

25.2

3.3

6.3

OUT2

14.1

1.2

36.7

16.7

1.1

21.1

Partial balance

-0.7

4.0

-14.4

10.7

6.0

-2.4

Discussion

The study is a good example of INM in the cotton zone of Mali. The participatory approach and the division into farmer classes make the results more useful, because the ‘average’ farmer does not exist. The partial balance is useful for comparing different nutrient management strategies. However, in order to know more about the sustainability of the system, one should use a complete nutrient balance. This is because the partial nutrient balance does not express indirect related nutrient losses, e.g. leaching and gaseous losses.

Integrated smallholder agriculture-aquaculture in Asia

A study by Dalsgaard and Prein (1999) applied a nutrient modelling approach to show how the combination of crops, trees, livestock and fish, that is integrated agriculture-aquaculture (IAA), helps in optimizing nutrient flows in Asian rice-based agro-ecosystems. Smallholder IAA is defined as diversification of agriculture in the sense that aquaculture (fish farming) is developed as a subsystem on a farm with existing crops, trees or livestock subsystems, or a combination thereof. A comparative on-farm study of integrated and non-integrated rice farming investigated N flows of four Philippine smallholder agro-ecosystems (Dalsgaard and Oficial, 1997).

FIGURE 17
ECOPATH flow diagram of a theoretical IAA farm system, kg N/ha/year

Note: B = average standing biomass, P = production, Q = consumption.
Source: Dalsgaard and Prein, 1999.

Methodology

TABLE 29
Published values of N flows into and out of fertilized rice agro-ecosystems

Inflows


Dry and wet atmospheric deposition

1.5 kg/ha/year

Run-on with irrigation water

10 kg/ha/crop

BNF:


Associative fixation in the rice rhizosphere

4 kg/ha/crop

Heterotrophic fixation associated with rice straw

2-4 kg/tonne straw

Heterotrophic fixation in flooded planted soil associated with organic debris

10-30 kg/ha/crop

Photodependent fixation by cyanobacteria

27 kg/ha/crop

Outflows


Ammonia volatilization and denitrification

50-75% of fixed N

Erosion and runoff

Unknown

Leaching

Unknown

Source: after Dalsgaard and Prein, 1999.

The ECOPATH approach and software (Lightfoot et al., 1993) were used for the modelling and analysis of the agro-ecosystems. This mass-balance framework provides a good basis for exploring the characteristics of nutrient flows and budgets in rice agro-ecosystems. ECOPATH diagrams individual farm components as boxes and indicates their biomass, production and consumption parameter values and linkages to other components, including detritus that denotes the soil resource base (Figure 17). The values for the different Inflows and outflows were obtained from farm surveys and literature (Table 29).

TABLE 30
Agro-ecological performance indicators for four Philippine smallholder farm systems


Fertilizer input and rice system


High; monoculture

High; diversified

Low; diversified & integrated


Farm (A)

Farm (B)

Farm (C)

Farm (D)

Surplus N (kg/ha/year)a

190

152

58

62

N balance (kg/ha/year)

-2

72

1

-9

N efficiencyb

0.19

0.17

0.40

0.38

N yield (kg N/ha/year)

43

45

39

33

Gross margin (US$/ha/year)

250

750

625

600

a = lost from the farm system primarily in gaseous form and to a lesser extent through erosion/runoff.

b = ratio of system N harvest over all N inputs.

Source: Dalsgaard and Oficial, 1997.

Results

The on-farm investigation showed that economically attractive, productive and balanced systems can be generated and maintained through integrated natural resources management (Farms C and D - Table 30). It also showed that high application rates of mineral fertilizer are not necessarily associated with a positive nutrient balance (Farm A), but rather with high flows through the rice-based agro-ecosystem and high losses to the environment. High input diversification systems (Farm B) as yet have the highest gross margin.

Discussion

This case shows that nutrient balances can also be determined for other farming systems, such as the integrated agriculture-aquaculture system. N fixation is very important in these rice-based farming systems. However, outflows from leaching/deep percolation and erosion/runoff are almost unknown in such systems.


Previous Page Top of Page Next Page