Chapter 3

General discussion

This chapter discusses important aspects in nutrient-balance studies that were not covered in Chapter 2. These issues are: uncertainties in nutrient balances; sampling for nutrient-balance studies; available nutrients versus nutrient flows; the use of spatial data; upscaling; and the impact of negative nutrient balances.

Uncertainties

The required accuracy and precision of a nutrient balance depend on the objectives and the originators of the study. The achievable accuracy and precision depend to a large extent on the complexity of the ecosystem and on the understanding of nutrient cycling and nutrient transformation processes. Biases and errors can introduce uncertainties. Bias is defined as systematic deviation and error as random variation. Five possible sources of bias exist: personal biases, sampling biases, measurement biases, data manipulation biases (including guesses), and fraud.

Sampling errors and measurement errors are sources of error. Sampling errors originate from spatial or temporal variations. Soils, crops and animal waste are notoriously variable in space and time and require well-designed sampling strategies. Measurement errors originate from variations introduced during the determinations of the sample volume and composition. The measurement error is usually much smaller than the sampling error. Table 31 presents an example of relative errors of nutrient flows for N and P budgets of farms in the Netherlands (Oenema and Heinen, 1999).

TABLE 31
Approximate values for the relative errors of N and P balances of farms, the Netherlands

Sampling

To improve nutrient balances with validation and better input data, more field measurements are necessary. The fact that soil properties in particular are highly variable, highlights the need for good sampling strategies. A new technique for rapid estimation of soil properties, developed at the International Center for Research in Agroforestry (ICRAF), might prove useful in this respect. A scheme was developed for using soil spectral libraries for the rapid non-destructive estimation of soil properties based on diffuse reflectance spectroscopy. A diverse library of more than 1 000 archived topsoils from eastern and southern Africa was used to test the approach.

A portable spectrometer (0.35-2.5 µm) with an artificial light source scans air-dried soils. Integrated indicators of soil quality that relate directly to plant productivity and soil enrichment/depletion processes (e.g. organic inputs and erosion) can be derived using visible-near-infrared reflectance spectroscopy.

The following soil properties can be determined: clay content, silt content, sand content, pH, organic carbon, exchangeable Ca, exchangeable Mg, exchangeable K, effective CEC, extractable P and N mineralization potential. Such indicators need to be readily measurable in order to permit monitoring of actual impacts of alternative farming practices on soil quality. This non-destructive technique allows large numbers of soil samples to be characterized rapidly (2 000 samples/week). Geo-referenced observations of the spectral quality index can also be interpolated spatially over large areas (> 1 000 km²) using satellite imagery (Shepherd and Walsh, 2002).

In addition to a sound sampling scheme, correct sampling methodologies and measurements are important. High resolution data are required in order to assess accurately changes induced by INM strategies, which are often changes of 20 percent or less. These changes are detectable given the correct statistical design. However, systematic errors introduced in soil sampling methods and laboratory analysis generate data that are either always greater than or less than the actual sample mean. In soil sampling, one of the most widespread cause of systematic errors is sampling a soil to a given depth increment and assessing changes in that increment. Even minor changes in bulk density, which commonly occur during a trial as a result of natural processes or INM interventions, change the mass of a soil being sampled in a given depth increment. If the soil is compacted during a trial, this will result in an overestimation of nutrient stocks in a given depth increment, whereas if the soil is de-compacted, an underestimation will occur. Errors of 10-15 percent are not uncommon. Similar systematic errors can be introduced in the laboratory analysis. When combined with soil sampling errors, these generate misleading data and erroneous conclusions. Soil mass sampling eliminates sampling errors caused by depth sampling, but it requires methods not commonly employed (Wendt, 2003).

Available nutrients

TABLE 32
Estimated fractions of available nutrients in each nutrient flow

Source: Janssen, 1999.

Available nutrients are conceived as the nutrients that are present in the soil solution at the beginning of the growing season or that will enter the soil solution during the season. In general, OUT1-OUT4 flows consist solely of available nutrients. OUT5 comprises flows of nutrients that are not immediately available, because they are preen in solid organic matter and inorganic particles (erosion) and flows of dissolved and, hence, available nutrients (runoff). The situation is more complex for Inflows. The availability of IN1 and IN2 nutrients depends on the composition of the fertilizers and manure; it is affected by weather conditions, length of growing season and soil life. IN3 consists of direct available nutrients from precipitation and not direct available nutrients from dry deposition. For IN4 and IN5, nutrients via symbiotic N fixation and irrigation are directly available, while nutrients via non-symbiotic N fixation and sedimentation are not. Table 32 shows estimates of available fractions of each nutrient flow.

Source: Rijpma and Fokhrul Islam, 2003.

Most nutrient-balance studies focus on the macronutrients N, P and K. However, plant growth depends on the most limiting nutrient, which might also be one of the micronutrients. Especially in countries with higher fertilizer use, the deficiency of nutrients changes from macronutrients to micronutrients because most mineral fertilizers consist of combinations of only N, P and K. This is illustrated in Figure 18 for Bangladesh, e.g. boron (B) deficiency in wheat, Mg deficiency in potato or maize and zinc deficiency in rice.

Spatial data

The understanding of spatial variation in crop response to environment and management is an essential component of agronomic research. The increasing availability of tools for spatial analysis, especially GISs, provides researchers with opportunities to improve analyses of spatial variation inherent to agronomic research. Benefits might include: improved selection of research sites or treatments; more quantitative assessments of the impact of climate and edaphic factors; and enhanced appreciation and improved presentation of how responses might vary over a target region. Typically, mesoscale variation would be of interest in field research conducted at one or more locations over a region where relevant map scales are of the order of 1:10 000 to 1:500 000. This might range from the county or district level to state or province level (White, Corbett and Dobermann, 2002).

Upscaling

Important for the upscaling of nutrient balances is first the determination of the system boundaries. Two methods for upscaling can be used: generalization and aggregation. With a generalization, a representative individual describes the characteristics of a group or population, e.g. fertilizer application data for one farmer is used to describe the fertilizer use of the whole village. Aggregation uses the information obtained for individuals to describe a population.

This involves grouping farms on the basis of one or more common properties. Frequency distributions can be used to describe the variability on a group scale. The temporal scale should also be taken into account. For this aspect, the size of the population is important because small systems (e.g. individual farms) change more rapidly and drastically than large systems (e.g. national livestock population).

The paradox is that upscaling and loss of information are connected very closely. Information for the lower scale facilitates the constructing of nutrient balances for the higher scale. However, when the nutrient balance of the higher scale is determined, information is lost. Therefore, when presenting nutrient balances, the recommendation is to provide the information from the lower scales used for constructing the nutrient balance of the scale considered. Any scaling exercise is embedded in the data. Quantitative data with adequate spatial and temporal resolution on the use and management of fertilizers, animal excreta and environmental data are often very sparse (Van der Hoek and Bouwman, 1999).

Impact

The impact of a negative nutrient balance cannot be seen independently from actual soil fertility, i.e. the nutrient stocks. A negative nutrient balance on a rich soil will not affect yield in the short term, while on a poor soil, crop yield may decline each year as a result of nutrient depletion. At some stage in marginal areas, a negative nutrient balance may no longer affect production as yields reach a bottom-line level where natural inputs such as atmospheric deposition make up for losses.

Figure 19 shows an example of assessing the impact of nutrient depletion. Maize grown on a poor soil (N stocks of 1 500 kg/ha) without mineral or organic fertilizer inputs has a yield of 2 000 kg/ha. The yield will start to decline when the amount of available nutrients (mineralization rate of 3 percent) becomes lower than the necessary N uptake. This will happen after five years on the poor soil, while nutrient depletion can continue without affecting yield for 55 years on the richer soil. In this scenario, the long-term equilibrium will be reached with very low yields (400 kg/ha) and N stocks (270 kg/ha). Hence, one could say that nutrient depletion often does not manifest itself clearly, but problems are likely to occur for the ‘future generations’ of the Brundtland definition (Brundtland, 1987).

Source: FAO, 2003.

Declines in yield and nutrient stocks can also be expressed in economic terms. Yield decline is a private (farmer) cost, whereas the decline in nutrient stocks is a social cost. Farmers will normally adapt their management when they experience yield decline. Where they do not have the means to increase fertilizer or manure use, they can adapt their management, e.g. make more efficient use of their fertilizer, use higher yielding varieties, make use of microvariability in their fields, and apply other INM techniques.