Interpreting Soil Test Results: What Do All the Numbers Mean?

1

Soil sampling issues: aspects of design

and implementation of soil investigations

1.1 The fundamental problem of sampling to test soils

A fundamental reality facing soil testing is that the amount of soil in 1 ha of soil to a depth of 10 cm is 1400 tonnes. A 50 ha paddock can have 70 000 tonnes of soil to 10 cm. For a soil test, as little as 0.5 g or 1.0 g of soil may be tested in the laboratory (Price 2006). Therefore it is essential that every effort is made and precaution is taken to ensure that the small amount of soil that is tested is representative of the 70 000 tonnes of soil in the paddock or assessment area. If adequate procedures are undertaken, this can usually be done (Price 2006).

To characterise soils at a site, a suitable sampling design is required. Sampling design depends on the:

•  landscape or location from which samples are being taken;

•  purpose for which samples are being taken;

•  resources available to take and test the samples.

The basic questions to be considered are:

•  When is the best time to sample?

•  Where should the soil be sampled?

•  How many samples are required?

•  What spatial pattern should be used to take the samples?

•  What depths should be sampled?

For general sampling purposes, the basis for making these decisions is discussed in Petersen and Calvin (1986), Beattie and Gunn (1988), Rayment and Higginson (1992), McBratney (1993), Brown (1999), Price (2006), NEPC (2013a), McKenzie et al. (2008).

Another difficulty in sampling is the problem of temporal variability, where results for samples or measurements taken at one time may be different from results for samples or measurements taken at another time. Some obvious examples are:

•  measuring infiltration in a tilled paddock – the infiltration is much higher before rainfall compacts the soil and crusts the surface;

•  measuring salinity on a site after a large rainfall event – the soil solution may be diluted, compared with the solution measured after a dry period;

•  sampling soils immediately after addition of fertiliser or soil ameliorants such as lime or gypsum;

•  sampling soils in paddocks in which sheep and cattle graze, resulting in change in nutrients over time.

These difficulties need to be considered when sampling soils or making measurements on soils and in interpreting the results of any tests carried out. Sampling through time is required, or, alternatively, the conditions when measurements are made or samples are taken should be standardised (or at least recorded).

When interpreting soil test results it is necessary to consider the origin and nature of the samples of soil being taken and the purpose for which interpretations are being made. For example, often a soil sample that is sent for analysis can include a relatively large volume of soil of 1 kg size or larger. Such a sample includes a large range of soil materials on the scale of a plant root. A plant root is often of the order of 1 mm in size or less. There may be microenvironments within the soil that have quite different chemical and physical properties from the bulked soil property measured during analyses. Hence plant root behaviour or plant growth may not always directly reflect the soil properties measured by a large sample. A typical example of this is the comparison of the dispersion behaviour of small aggregates (often 5–10 mm) to the bulk soil properties of exchangeable sodium percentage. At the microscale of soil structural units there may be considerable variation in exchangeable sodium percentage and dispersion behaviour. This microvariation can be masked by the measurement of bulked soil properties. This emphasises the need for adequate sampling of soils and the risks of relying on a single sample to make recommendations.

1.2 Purposes of soil sampling

Soils at a site are sampled and tested for a wide variety of reasons including:

•  diagnosis of soil constraints for agricultural production, often in a specific paddock for a specific crop;

•  diagnosis of plant nutrition problems and formulation of fertiliser and soil amelioration programs;

•  monitoring changes or trends in soil chemical properties including pH, aluminium levels, nutrients, salinity and soil organic carbon concentrations;

•  soil testing for engineering and soil stabilisation purposes before the construction of buildings and infrastructure;

•  testing of soils for the occurrence of contamination and identification of the type and concentration of contamination;

•  estimation of soil carbon stocks for the purpose of determination of potential carbon credits;

•  characterisation of soils for mapping and the identification of general soil properties for natural resource management and assessment of land and soil capability and suitability.

Each of these circumstances can result in a different set of requirements for a sampling pattern, intensity of sampling, and accuracy and degree of confidence expected in the results from the soil testing program. The sampling program can include the following fundamental patterns (NSW EPA 1995a; Laslett and McBratney 1995; Brown 1999; McKenzie et al. 2008; NEPC 2013a):

•  judgemental – selection of sample sites is based on prior knowledge of the site and professional judgement. There is a risk that sampling using this method can be subject to bias, even if unintentional, reducing the confidence in any statistical analysis using these results. However, it can be a very efficient method for sampling and minimising costs of analysis. Much depends on the experience and knowledge of the person taking the samples. This can be most useful in preliminary and exploratory investigations;

•  systematic – selection of sample sites is based on regularly spaced sites on a grid or transect. Various patterns of transects may be used including:

  zig-zag patterns or herringbone patterns which can ensure adequate spatial coverage of the target area and minimise the likelihood of ‘hot spots’. The intensity of sampling depends on the purpose of the sampling. For example, to obtain an estimate of the need for liming on a paddock, the sampling intensity needs to be sufficient to obtain an overall estimate of the soil pH. In investigating potential soil contamination, the sampling intensity must ensure that any ‘hotspots’ are detected;

  series of transects within a paddock or assessment area;

  transects are based on strata defined as;

•  stratified – the assessment area is separated into non-overlapping sub-areas or strata from which a planned number of samples are taken. Within the strata, the sampling locations are often chosen at random to give a stratified random sampling pattern. The strata are sometimes chosen so that the area within the strata may have some degree of uniformity. Alternatively, the strata may be derived by simply dividing the assessment area into equal area strata (Brus et al. 1999).

If it is desirable to undertake statistical analysis on the samples, it is necessary that there is a randomised element to the selection of the sampling sites (Laslett and McBratney 1995; McKenzie et al. 2008).

Once collected, samples for nutrient analyses may be bulked to give a composite sample. Generally, bulking should be done only when the samples come from a relatively uniform area, or what is thought to be a relatively uniform area. Petersen and Calvin (1986), Tiller (1992), McBratney (1993) and Laslett and McBratney (1995) discuss the limitations of bulking samples and the recommended procedures to follow. Brown (1999) provides the following guidelines for bulking:

•  the population of soil samples to be bulked are from a uniform population, i.e. all the subsamples are taken from a uniform part of the assessment area. The subsamples do not come from, say, a creek flat and also a hilltop, or from parts of a paddock with different soil types;

•  uniform amounts from each subsample contribute to the composite sample analysed;

•  standard sampling depth and equipment are used for each subsample contributing to the composite sample;

•  no interactions occur that would affect the results;

•  the only objective is to obtain a single, unbiased mean value of the soil property being analysed.

Bulking can be a helpful and cost-saving method of obtaining useful information on soil properties across a large area, but needs to be used carefully to ensure reliable results.

1.3 Specialised sampling methodologies for soils

When soils are to be tested for specific purposes or specific problems, often it is necessary to use specialised sampling methodologies. For example, specialised sampling methodologies are required for the following:

•  for best practice sampling procedures for acid sulfate soil refer to Dear et al. (2014);

•  sampling soils for contaminated sites need to follow guidelines set out in standard procedures (see NSW EPA 1995a; Ministry for the Environment 2010; NEPC 2013a; Murphy and Hazelton 2014);

•  in sampling soils to estimate soil carbon stocks to determine carbon credits there are set guidelines (Department of Environment 2014a,b; Chappell et al. 2013; Murphy et al. 2013);

•  protocols for the identification of biophysical strategic agricultural land for planning purposes in NSW require specific guidelines to be followed, as set out in the published document (NSW Government 2013);

•  sampling soils for rapidly metabolised chemical species such as nitrate (NO3–) (see Peverill et al. 1999).

1.4 Statistical requirements for sampling

Sampling soils for many purposes requires that statistically based inferences be made from the data produced from the sampling and analysis (Brus et al. 1999; McKenzie et al. 2008; NEPC 2013a; Department of Environment 2014a,b; Chappell et al. 2013). In these circumstances, it is essential that the sampling program meets the basic guidelines of:

•  effective spatial coverage of the area to be investigated (Brus et al. 1999);

•  the selection of sampling sites is randomised in some way to ensure that standard statistical procedures can be applied to the data and results. The sampling patterns may be completely randomised or a stratified randomisation pattern be used, but the selection of sites must be randomly selected in some way. The operator cannot subjectively select the site in the field. The object is to avoid bias in the selection of sampling sites, even if unintentional;

•  if the sampling program has effective spatial coverage and is randomly based, valid values of the mean, variance, standard deviation, standard errors and confidence intervals for the results can be estimated (Henderson et al. 2008).

1.5 Sampling for soil mapping

1.5.1 The number of samples required to produce a soil map

The number of samples or ground observations needed to produce a soil map or to undertake an investigation will vary with the local characteristics of the site. Factors that influence the number of samples required include:

•  geology;

•  landform;

•  land use history;

•  purpose for which the investigation is carried out.

There are general guidelines on the number of samples required to produce a reliable map (Reid 1988; Schoknecht et al. 2008; Chapman and Atkinson 2007). It should be remembered that these are general guidelines and their applicability will be influenced by the factors outlined in this chapter. The number of samples required is usually expressed as the number of samples or observations per cm² of the map.

Table 1.1 is a general guide to the minimum number of samples required for 1 ha or 1 km² of land at different map scales. These can be converted to a sampling intensity of observations or samples per ha or per km² (see Tables 1.1 and 1.2). When developing a soil map, different types of soil observations are made with different levels of information and rigour (DLWC 2000; Schoknecht et al. 2008) (see Table 1.3). For soil mapping, some observations can be quite brief and will collect minimum data, but other observations are rigorous, collecting the maximum amount of soil information including obtaining samples for laboratory analysis. The proportion of different kinds of observations will vary with the scale of mapping and the purpose for developing the soil map.

The relationship between soil survey effort and map scale has been derived by Dent and Young (1981), and Gunn et al. (1988) estimates the effort required in days in the field to develop a soil map at different scales (Table 1.4). The estimates by Schoknecht et al. (2008), a more recent reference, are substantially less. These more recent estimates possibly take account of modern methods of soil survey using remote sensing and geographic information systems. The actual effort will vary depending on such factors as existing information, the complexity and predictability of the soil patterns and difficulties of access. These estimates also do not consider the purpose for which a soil map is being developed. They are intended to indicate the effort required for a detailed soil map at the appropriate scale and thus represent a maximum value. Therefore, these are only broad guidelines.

Table 1.1. Number of observations recommended for a published soil map given in number per km² (1000 m × 1000 m) unless otherwise specified

Source: Reid (1988) and see Schoknecht et al. (2008).

Table 1.2. Soil map types based on scale and sampling intensity

Source: Adapted from Rossiter (2000) and Gallant et al. (2008).

Table 1.3. Recommended intensities of investigation based on map scale

Source: DLWC (2000).

Table 1.4. Soil survey effort as minimum number of days in the field for different scales

1.5.2 Sampling density and frequency for mapping acid sulfate soil

The number of soil sampling locations required will depend on the nature, depth and size of the disturbance proposed. Determination of the sampling strategy will depend on the future land use and development stages (DSE 2010; Tulau 2007; Dear et al. 2014).

The number of sampling points needs to be sufficient to generate a detailed map of soil net acidity and stratigraphy at a better than 1:10 000 scale, which is considered adequate to accurately and efficiently plan engineering works and manage the disturbance of acid sulfate soil (DSE 2010; Tulau 2007; Dear et al. 2014).

1.5.3 Use of technology in soil mapping: digital soil mapping

Soil mapping can be achieved using computer technology and remotely sensed data such as digital elevation models, satellite imagery and radiometric data. This data is often used to predict the spatial distribution of individual soil properties (McBratney et al. 2003; Minasny and McBratney 2016) and has been used to develop a global soil map of individual soil properties (Arrouays et al. 2014; Grundy et al. 2015). Digital soil mapping can develop outcomes more rapidly than conventional mapping methods but the quality and reliability of the outputs for a specific area are dependent on the availability of data.

1.6 Assessment of soil health, soil quality, soil condition or soil capability

In some circumstances it may be desirable to assess the overall soil health, soil quality, soil condition or soil capability. Several schemes provide a list of soil properties to be used in the assessment of the overall quality or condition of soils. These can also relate to soil security (McBratney et al. 2014) or land and soil capability (OEH 2012) but several schemes also use specific sets of soil properties to assess soil quality. Numerous systems of assessing soil health/soil quality/soil condition are available (Karlen et al. 2003; MacEwan 2007). These include those of Crawford et al. (2015) for tilled soils and more generally Palm et al. (2007) to assess the capacity of soil to provide ecosystem services and Sanchez et al. (2003) for the fertility of tropical soils. The NSW Office of Environment and Heritage (OEH 2012) also has a list of land and soil attributes that are used to assess the land and soil capability at a site.

1.7 Investigations for landscape design and development: some guidelines

Landscape design and development is undertaken where there is substantial site disturbance or land requires rehabilitation for stabilisation. The soils formed in landscape design are anthropic or man-made soils:

Landscape soil is an anthropic soil profile that is modified from a natural ‘in situ’ soil or manufactured and installed using artificial components for the purpose of sustaining vegetation chosen for landscape design or land rehabilitation (Leake and Haege 2014).

Clear specifications are needed for the most common types of soil investigations, soil reconstruction processes and constructed soil specifications used in landscape design projects. Soil management guidelines also need to be established for the implementation of the design and its long-term success. These specifications and guidelines are summarised in Table 1.5.

Table 1.5. Common soil types and construction management used in landscape design projects