Characterising soil quality clusters in relation to land use and soil order in New Zealand: An application of the phenoform concept
Introduction
Soil survey, soil classification and land evaluation programmes have largely focused on inherent soil properties that change very slowly over time and are largely insensitive to land use. Conventional soil maps are made with reference to these relatively stable soil properties. This focus has traditionally been the domain of pedology. Soils are continually evolving and transforming within anthropogenic timescales (Richter et al., 2011), and the study of soil dynamic properties that are sensitive to land use has generally been the domain of agronomy, soil fertility, soil biology, soil quality, and soil ecology.
The state factor approach of Jenny (1941) firmly placed soil formation in an ecosystem context. A new emphasis in soil science is emerging with the recognition of soil as a crucial component of the earth's natural capital, that highlights not only the ecological integrity of soils but also the economic and social services that underpin the earth's ecosystems and economies (Robinson et al., 2013, Dominati et al., 2010). This growing area of research will force the integration of traditional sub-disciplines of soil science because the quantification and valuation of soil natural capital (Hewitt et al., 2012) and of soil services require the integration of data on both inherent and dynamic soil properties.
Droogers and Bouma (1997) have provided a useful conceptual framework to integrate inherent and dynamic soil properties that may bridge this gap. Borrowing from plant and animal ecology they coined the term ‘genoform’ for soil formed under native vegetation and ‘phenoform’ for the equivalent soil with similar inherent properties but with dynamic properties modified by the impacts of a specific land-use history. McBratney et al. (2014), discuss the genoform/phenoform concept in the context of soil capability and condition and suggest that the genoform represents a reference state that encompasses the inherent capability of the soil and condition under a specific long-term circumstance (e.g. natural vegetation). The phenoform reflects the condition due to specific management, but they also note that in a soil that passes a critical threshold, a phenoform may also become a new reference state.
Soil classifications and spatial soil databases generally exclude dynamic soil properties and are limited in their ability to support realistic spatial analyses of land use issues involving dynamic soil properties. Although there has been some movement toward linking knowledge of dynamic soil characteristics into soil survey, soil classification and land evaluation (see for instance Pennock and Veldkamp, 2006), progress has been slow. In contrast, soil quality is a subdiscipline that has focused almost exclusively on the variation of dynamic soil properties where indicators are deliberately chosen to represent key dynamic responses of soil natural capital to the impacts of human land use management. The soil quality literature provides many examples of relationships of individual soil quality indicators with land use and soil type (for example, Brejda et al., 2000, Sparling and Schipper, 2004, Cotching and Kidd, 2010). Sparling and Schipper (2002) examined an initial New Zealand soil quality data set using principal components.
Here, we examine in more detail the multivariate clustering of an expanded New Zealand soil quality data set that contains only dynamic soil properties. Cluster centroid classification has been proposed for soil classification systems of inherent soil properties (see McBratney and De Gruijter (1992) and Minasny et al. (2010)) and would also allow development of classifications that incorporate both dynamic and inherent soil properties — a feature particularly useful in the characterisation of soil natural capital. The motivation for our research was to utilise the genoform–phenoform concept to indicate where land-use change may significantly affect soil quality indicators for specific soils or groups of soils.
This paper considers three hypotheses:
- 1.
Managed sites (sites that are managed for a particular purpose and/or have had significant anthropogenic alteration) are statistically distinct from sites under native vegetation
- 2.
Managed sites are clustered into statistically distinct classes of related soil quality states
- 3.
Clusters have functional meaning as assessed by relationships to land use and soil type and their impress on soil quality states.
Section snippets
Soil quality database description
The New Zealand soil quality data set for regional-scale monitoring currently holds data for in excess of 700 sites, over 12 geographical regions within the country. The core set of soil quality indicators (pH, total C, total N, anaerobically mineralizable N, Olsen P, bulk density, and macroporosity) and their symbols, are listed in Table 1. There are very few soils containing carbonates in New Zealand (and of those that do, carbonates do not occur in the A horizon), so that total C is
Data exploration
Fig. 1 shows pair plots between the seven transformed soil indicator variables, where the transformation was applied to reduce the skew of the distribution. The lower diagonal plots show one indicator against a second, overlaid with a smooth (lowess) curve to show the trend. The upper diagonal of Fig. 1 indicates the correlation estimate between pairs of indicator variables; thus total carbon and total nitrogen are strongly correlated (correlation 0.79), while pH and total nitrogen are poorly
Conclusions
The evidence supports the definition of phenoforms and indicates that a taxonomy of phenoforms may be possible. Hypothesis one of our research proposed that managed sites were distinct from native sites. This was supported by discriminant analysis and permutational multivariate analysis of variance. Hypothesis two proposed that managed sites were clustered into statistically significant distinct classes. This was supported by principal components fuzzy-c means clustering, with recognition of
Acknowledgements
We gratefully acknowledge core funding to Crown Research Institutes, both directly to Landcare Research and through the Sustainable Land Use Research Initiative as a subcontract from Plant & Food Research.
References (37)
- et al.
Comparing different methods for estimating the soil moisture supply capacity of a soil series subjected to different types of management
Geoderma
(1999) - et al.
Soil quality evaluation and the interaction with land use and soil order in Tasmania, Australia
Agric. Ecosyst. Environ.
(2010) - et al.
A framework for classifying and quantifying the natural capital and ecosystem services of soils
Ecol. Econ.
(2010) - et al.
Effects of ecological soil management on workability and trafficability of a loamy soil in the Netherlands
Geoderma
(1996) - et al.
Soil quality monitoring in New Zealand: development of an interpretative framework
Agric. Ecosyst. Environ.
(2004) - et al.
Soil and informatics science combine to develop S-map: a new generation soil information system for New Zealand
Geoderma
(2012) - et al.
The dimensions of soil security
Geoderma
(2014) - et al.
Global pedodiversity, taxonomic distance, and the World Reference Base
Geoderma
(2010) - et al.
Advance in landscape-scale soil research
Geoderma
(2006) - et al.
Soil order and grazing management effects on changes in soil C and N in New Zealand pastures
Agric. Ecosyst. Environ.
(2014)