- Email: [email protected]
Taxonomic distance between South African diagnostic horizons and the World Reference Base diagnostics
Catena 113 (2014) 276–280
Contents lists available at ScienceDirect
Catena journal homepage: www.elsevier.com/locate/catena
Taxonomic distance between South African diagnostic horizons and the World Reference Base diagnostics C.W. van Huyssteen a,⁎, E. Michéli b, M. Fuchs b, I. Waltner b a b
Department of Soil, Crop and Climate Sciences, University of the Free State, P.O. Box 339, Bloemfontein 9300, South Africa Department of Soil Science and Agricultural Chemistry, Szent István University, 2103 Gödöllő, Páter Károly u. 1, Hungary
a r t i c l e
i n f o
Article history: Received 2 March 2012 Received in revised form 26 July 2013 Accepted 12 August 2013 Keywords: Classification Diagnostics Euclidean distance Morpho-genetic Property-based
a b s t r a c t A myriad of soil classifications exist internationally. These usually cater for unique national variations and conditions. These different classification systems, however, hinder international communication. This paper attempted to relate the South African Soil Taxonomic (SAT) soil classification system with the World Reference Base for Soil Resources (WRB) through taxonomic distance classification. A probability matrix of the presence of selected identifiers of the diagnostic elements (properties, horizons, materials) of the South African classification system and the WRB was constructed to determine the taxonomic relationships between them. Euclidean distance calculation on these data enabled numeric expression of the taxonomic similarities and dissimilarities between the South African and WRB diagnostics. Results proved encouraging and some recommendations can be made. For example, a N 20% OC family for the organic O, as well as stagnic and gleyic families for the G horizon is proposed. It is further proposed that the WRB consider recognition of red apedal B, yellow-brown apedal B, and lithocutanic B horizons. Since the compared units are the basic building blocks of the two systems, the results presented here can be useful in the relation of soil classification in the South African Soil Taxonomy to the WRB. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Humans classify the objects in their environment to create order, to reduce variability, to increase understanding, and to inventorise (Krasilnikov et al., 2009). A myriad of soil classification systems were developed, probably due to the relative young nature of the science, the unique local variation of soils, and lack of means of easy information exchange. A further consideration might be that scientists in each region consider different soil properties with varied interest. This situation creates obvious challenges for comparing soils and international communication. The United States Department of Agriculture Soil Taxonomy (Soil Survey Staff, 2010) and the World Reference Base for Soil Resources (WRB; IUSS Working Group WRB, 2006) are used for international communication. The WRB has, however, been adopted as preferred soil correlation system by the International Union of Soil Sciences and the European Union (Jones et al., 2005). The WRB is based on diagnostic horizons, properties, and materials, each with strict differentiating quantitative criteria and definitions. The WRB has two tiers: 32 reference soil groups (RSGs), determined by a key and qualifiers (that are accommodated as prefixes or suffixes to the RSG). Although it is stated that the differentiating criteria “should ⁎ Corresponding author. Tel.: +27 51 401 9247; fax: +27 51 401 2212. E-mail addresses: [email protected] (C.W. van Huyssteen), [email protected] (E. Michéli), [email protected] (M. Fuchs), [email protected] (I. Waltner). 0341-8162/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.catena.2013.08.010
be measureable and observable in the field” the reality is quite different and in many cases almost complete chemical and physical analysis of the soil profile (horizons) is required to perform a classification. This is especially true where the local soil properties are not known. Soil Classification — A Taxonomic System for South Africa (SAT) is used exclusively in South Africa (Soil Classification Working Group, 1991). The system follows a morpho-genetic approach, similar to that proposed by Kubiëna (1953). Very few chemical or physical analyses are therefore required to classify the soil. This feature makes the classification system uniquely suited to the low-technology (resource poor) environment experienced in Africa. The South African Taxonomy defines five topsoil and 25 subsoil diagnostic horizons, combinations of which give rise to 73 soil forms. The soil forms are subdivided into soil families, based on 19 sets of distinguishing properties. The final classification should also include the soil depth and topsoil texture. The challenge is, however, to relate the South African Taxonomy with the WRB. Efforts in this regard are severely hampered by the differences in approaches (principles) between the two systems (morphogenetic vs. property-based). Taxonomic distance calculations, first promoted by Adanson (1763), are based on measures of similarity often applied in the phenetic view of numerical taxonomy, where the relative similarities or dissimilarities are measured based on different attributes without a priory weighting (Dunn and Everitt, 1982; Jardine and Sibson, 1971; Sneath, 1962; Sokal and Sneath, 1963). Taxonomic distance calculation in soil science was first proposed by Hole and Hironaka (1960). Numerical soil classification has been applied
C.W. van Huyssteen et al. / Catena 113 (2014) 276–280
277
Table 1 Physical, chemical, and morphological properties of soils selected from the WRB diagnostic criteria that were used as dominant identifiers and coded against SAT and WRB diagnostics. Physical properties
Chemical properties
Morphological properties
Consolidated material b20% Cracks N10 cm Horizontal crack spacing Vertical cracks N10 cm apart b50% (v/v) Rock structure b80% Gravel Irreversible hardening N50% Indurated Indurated by CaCO3 b10% Dipersible clay b10% Weatherable minerals ≥4.5 MPa penetration resistance ≥4 MPa penetration resistance ≥50% Slaking Clay increase Doubling of clay content over 7.5 cm ≥8% Clay ≥30% Clay Very fine sand, loamy very fine sand or finer texture Loamy sand or finer texture
Organic material saturated b30 consecutive days Organic material saturated ≥30 consecutive days Organic carbon accumulation b0.5% Organic carbon N0.6% Organic carbon b5.9 pH b50% BS N50% BS Effervescence with HCl Strong effervescence with HCl N0.50% AlO+FeO N15% Exchangeable sodium percentage N2% Calcium carbonate equivalent Calcium carbonate equivalent ≥15% Cation exchange capacity b16 cmolc kg−1 clay Si accumulation
≥5% Red/black nodules N15% Nodules N40% Nodules Indurated nodules or plates N5% Oximorphic colours N90% Reductimorphic colours or ≥15% Red mottles or Dark/red colour Darker than overlying Grey dry colour Grey moist colour Munsell value and chroma ≤3 Pale pedfaces and bright interiors Evidence of clay illuviation (clay bridging/films) ≥5% (v/v) Secondary carbonates N25% Stratification ≥10% (v/v) Si nodules/fragments Under albic Wedge-shaped peds Slickensides Massive, blocky, columnar or prismatic structure ≥10 cm Peds
in multiple studies since, including some with distance metrics. Most of these studies, however, focused on smaller areas and/or datasets and therefore had only limited applicability or scope (Bidwell and Hole, 1964a,b; McBratney et al., 2000; Sarkar et al., 1966). National and international application of the method has not been published for almost another decade (McBratney et al., 2009). Carré and Jacobson (2009) incorporated distance metrics into their OSACA model application and used it to allocate soils to existing classifications or to derive centroids and create new classifications based on clustering. Minasny et al. (2009) were first to apply taxonomic distance metrics at an international level for the WRB Reference Soil Groups. Their concept-based approach focused on the dominant identifiers, or diagnostic criteria, of the system instead of deriving centroids. With the centroid-based approach the taxonomic distance metrics are generally based on actual numerical data, where a centroid is calculated for the group in question and for each selected attribute (“dominant identifier”). However, distance calculations can also be based on the presence or absence of certain features, derived from scientifically sound concepts (Minasny et al., 2009). These concept- and centroidbased approaches were enhanced for the semi-quantitative analysis and correlation of different classification systems (Láng et al., 2010, 2013). Probably the simplest and most common method to calculate taxonomic relationships is the Euclidean distance (Dunn and Everitt, 1982; Webster, 1977). The taxonomic distance (dij) is based on Pythagoras' theorem, so for points xi and xj with two variables it can be expressed as: dij ¼
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi r 2 2 2 xi1 −x j1 þ xi2 −x j2
ð1Þ
Extending the same principles to multiple dimensions (representing multiple variables) gives: vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u p 2 X u 2 xik −xjk dij ¼ t
ð2Þ
k¼1
where p is the number of dimensions observed. In application to soil classification, the taxon being considered forms the centre and the distance of the other taxa from the centre are calculated based on the selected soil properties (diagnostic criteria in this instance). Interpretation can therefore not be made between the different
taxa, but only between the two (i.e. the central and one additional) taxa under consideration. For example, two taxa may have the same distance from the centre, but may be equally far from each other, because the direction of the departure is not specified. The purpose of this paper was to explore the relationship between diagnostic horizons in the South Africa Soil Taxonomy and the diagnostic horizons, materials, and properties in the WRB. 2. Methods This taxonomic distance calculation focussed only on relating 33 possibly related WRB diagnostic horizons, properties, and materials with the SAT diagnostic horizons. The WRB qualifiers and SAT family criteria were therefore excluded. In addition the fulvic and melanic horizons, andic and vitric properties, and tephric material were excluded, because recent pyroclastic deposits are not known in South Africa. All the anthropogenic diagnostics in the WRB and SAT were excluded. Firstly because the hortic, irriagric, plaggic, terric or anthraquic horizons are not known to exist in South Africa and secondly because the diagnostics for the man-made soil deposit are not scientifically well developed. The physical, chemical, and morphological properties selected as dominant identifiers are presented in Table 1. The taxonomic distance calculation was done on the concept based approach (Láng et al., 2010; Minasny et al., 2009), by constructing a coded matrix table. The matrix table expressed the probability that a property (selected as dominant identifier) must be present (code: 1), cannot be present (code: 0), or is likely to be present (code: 0.5) in the compared diagnostic elements. The codes are the probability values (0, 0.5, 1) of the presence of the selected dominant identifier (physical, chemical, and morphological criteria) of the selected diagnostics. The assignment of codes was based on the soil properties and diagnostic criteria in the compared systems (IUSS Working Group WRB, 2006; Le Roux et al., 1999; Soil Classification Working Group, 1991), personal experience, and expert judgement (the coded matrix and calculated taxonomic distances are available as online supplementary material.) Taxonomic distances were calculated, based on the matrix table, as the Euclidean distance between the different taxa by using the R software package (Baier and Neuwirth, 2007). Mathematically a taxonomic distance value of 0.0 indicates an exact similarity. Assessment of the data has shown that values less than 1.0 indicate large similarities, while values approaching 2.0 and above indicate large differences.
278
C.W. van Huyssteen et al. / Catena 113 (2014) 276–280
3. Results and discussion For the purpose of discussion only the WRB diagnostic horizons, properties, and materials closest, based on the calculated taxonomic distance, to the SAT diagnostic horizons are considered here. The three closest WRB diagnostics and the most important similarities and dissimilarities between them could be elucidated, based on the taxonomic distance values (Table 2). The results therefore not only serve to highlight the WRB diagnostics closest to the SAT diagnostic horizons, but also indicate the relation of the SAT diagnostic to other WRB diagnostics, enlightening some of the underlying processes and principles. The SAT organic O, Vertic A, and Melanic A topsoil horizon diagnostics are in general agreement with their WRB counterparts. Some diagnostics can, however, be included at the SAT family level to improve these relationships. Consideration should, for example, be given to the inclusion of 10–20% OC and N 20% OC families for the Champagne form. The latter family would then better relate with OC requirement of the WRB organic material. This would enable better relation with the WRB histic horizon (because the SAT organic O is saturated with water) and also with the definition of peat (Joosten and Clark, 2002; Rydin and Jeglum, 2006), whilst also keeping the current N10% OC diagnostic as a family criterion. The humic A horizon related best with the WRB spodic horizon. Although this relation is not strong and nonsensical, further research in this regard might be valuable, even though the pedogenic processes between these two are different. The melanic A must have dark colours in the moist and dry state, while only in the moist state for the mollic horizon. The orthic A horizon related best, but not strongly, with the albic horizon and albeluvic tonguing. These are subsurface horizons and the relation is therefore nonsensical. The lack of a clear relationship of the orthic A with WRB diagnostics may, at least partly be because WRB currently does not define an ochric
horizon. This should not be problematic because the orthic A is in reality not a diagnostic in the SAT. The E horizon related well with the albic, while the G horizon related with the WRB reducing conditions as well as stagnic and gelyic colour pattern. The main difference between the E and albic horizons is that the albic is “grey” in the moist state, while the E may be “yellow” in the moist state and “grey” in the dry state. This is something the WRB might consider as a qualifier, because the yellow E horizon is interpreted as being drier than the grey E horizon in the same environment (Van Huyssteen and Ellis, 1997). The poor relationships between the G horizon and the WRB reducing conditions and stagnic and gleyic colour pattern were caused by the different approaches between the two systems. The G horizon is defined inter alia in terms of water saturation, colour, accumulation of colloidal matter, while the stagnic and gleyic colour patterns only refer to colour configuration and reducing conditions to a low redox potential (rH b 20). SAT should consider the inclusion of a “gleyic” and “stagnic” family criteria for the soil forms with G horizons to improve the water saturation inferences that can be made to facilitate relation with the WRB. The red and yellow-brown apedal B horizons related best with ferralic properties, colluvic material, and lithological discontinuity. The relationship with ferralic properties is not strong, but could make sense given that the red and yellow-brown apedal B horizons consist mainly of kaolinite clays. The relationships with colluvic material and lithological discontinuity are not logical. The poor relationships are probably because the WRB does not place as much emphasis on drainage status as the SAT, but rather focuses on clay illuviation and base status. In the same environment the red apedal B horizon is drier than the yellow-brown apedal B horizon (Van Huyssteen and Ellis, 1997; Van Huyssteen et al., 2004) and should therefore be considered by the WRB. Differentiation between these is especially advantageous in the
Table 2 SAT diagnostics, and their 3 closest correlations to WRB diagnostics based on the calculated taxonomic distance matrix (distance values indicated between brackets). SAT diagnostics
Closest WRB diagnostic(s) based on distance matrix
2nd closest WRB diagnostic(s) based on distance matrix
3nd closest WRB diagnostic(s) based on distance matrix
Organic O Humic A Vertic A Melanic A Orthic A E
Histic horizon (0.71) Spodic horizon (1.58) Vertic horizon (0.71) Mollic horizon (0.87) Albic horizon (1.66) Albic horizon (1.00)
Organic material (1.12) Umbric horizon (1.80) Vertic properties (1.12) Umbric horizon (1.32) Albeluvic tonguing (1.73) Albeluvic tonguing (1.12)
G Red apedal B Yellow brown apedal B Red structured B Soft plinthic B Hard plinthic B Prismacutanic B
Reducing conditions (1.73) Ferralic properties (1.41) Ferralic properties (1.22) Nitic horizon (0.87) Reducing conditions; Plinthic horizon; Stagnic colour pattern (1.22) Petroplinthic horizon (1.00) Argic horizon; Natric horizon (1.94)
Pedocutanic B Podzol B Placic pan Neocutanic B
Natric horizon (1.94) Spodic horizon (0.50) Spodic horizon (2.06) Reducing conditions (1.22)
Gleyic colour pattern (1.80) Colluvic material (1.50) Lithological discontinuity (1.41) Ferralic properties (2.12) Ferric horizon; Pisoplinthic horizon (1.32) Pisoplinthic horizon (1.22) Abrupt textural change; Vertic properties (2.50) Argic horizon (2.06) Sombric horizon (2.06) Sombric horizon (2.78) Colluvic material (1.41)
Folic horizon (1.73) Folic horizon; Histic horizon (1.87) Mollic horizon (2.12) Vertic properties (2.00) Colluvic material (1.80) Colluvic material; Reducing conditions; Ferralic horizon (1.87) Stagnic colour pattern (1.87) Lithological discontinuity; Fluvic material (1.73) Colluvic material (1.50) Lithological discontinuity (2.24) Albeluvic tonguing (1.66)
Neocarbonate B Soft carbonate
Calcaric material; Secondary carbonates (0.71) Secondary carbonates (0.87)
Hardpan carbonate Dorbank Regic sand Stratified alluvium Lithocutanic B
Petrocalcic horizon (0.87) Petroduric horizon (0.71) Lithological discontinuity (1.94) Fluvic material (0.87) Lithological discontinuity (1.87)
Saprolite Hard rock Unspecified material with signs of wetness
Lithological discontinuity; Ferralic properties (1.66) Continuous rock (0.00) Reducing conditions (1.22)
Calcic horizon (1.00) Calcaric material; Calcic horizon (1.12) Petroduric horizon (1.94) Petrocalcic horizon (2.00) Ferralic properties (2.06) Colluvic material (1.00) Ferralic properties; Albeluvic tonguing (2.00) Colluvic material (1.87) Lithological discontinuity (3.12) Stagnic colour pattern (1.41)
Plinthic horizon (2.60) Vertic horizon (2.55) Abrupt textural change; Vertic properties (2.50) Ferralic properties (2.24) Ferralic properties (3.00) Ferralic properties; Lithological discontinuity; Cambic horizon (1.50) Cambic horizon (2.06) Duric horizon (2.12) Calcic horizon (2.65) Duric horizon (2.35) Colluvic material; Albic horizon (2.12) Ferralic properties (1.80) Albic horizon; Reducing conditions (2.06) Fluvic material (2.06) Fragic horizon; Petroplinthic horizon (3.28) Gleyic colour pattern (1.50)
C.W. van Huyssteen et al. / Catena 113 (2014) 276–280
more arid environments, such as South Africa. The luvic B horizon family criterion of SAT can, however, be changed to better align the luvic red and yellow-brown apedal B horizons with the WRB argic horizon, because the luvic B criterion is very similar (5% vs. 3% clay increase and 1.3 times vs. 1.2 times more clay) to the clay increase required for the WRB argic horizon. Consideration should therefore be given to changing the luvic B diagnostics to align to that of the agric horizon. The red structured B related very well to the nitic horizon, but does not have a perfect relation with it, because the latter has additional chemical (texture and Fe) criteria, that may or may not necessarily be true for the former. Soft plinthic B horizons related with WRB reducing conditions, stagnic colour pattern, plinthic horizon, and gleyic colour pattern. Reducing conditions are not always expected in the soft plinthic B horizon, but definitely below the horizon. The duration of reducing conditions in the soft plinthic B horizon is expected to be approximately eight months per year (Van Huyssteen et al., 2005), but this is not specified for the WRB reducing conditions or stagnic colour pattern. It is quite possible that reducing conditions are present for only a few (b 14) days and therefore satisfy the criteria, but the soil fails to develop redox morphology (Vepraskas and Faulkner, 2001). The soft plinthic B horizon does not harden irreversibly upon drying and should therefore not be equated to the WRB plinthic horizon. The hard plinthic B can for all practical purposes be equated with the petroplinthic horizon. The major differences are that the hard plinthic B does not necessarily have a penetration resistance ≥4.5 MPa, Feo:Fed b0.10 or ≥10 cm thick. The prismacutanic B and pedocutanic B horizons both related weakly with the natric and argic horizons. Conceptually the prismacutanic B was expected to relate better with the natric horizon, although it does not necessarily have the required N 15% exchangeable sodium percentage or thickness of the natric horizon. Similarly, the pedocutanic B was expected to relate better with the argic horizon, but is not defined in terms of clay increase or thickness, resulting in a wider taxonomic distance. As expected, the podzol B related very well with the spodic horizon. The placic pan also related best, but poorly, with the spodic horizon. The chemical criteria differ between the podzol B and placic pan on the one hand and the spodic horizon on the other, resulting in wider relations than expected. Further the SAT placic pan is defined in the WRB as a placic qualifier and not as diagnostic horizon, property or material. The origin of the chemical criteria for the podzol B is unclear and attention should therefore be directed at aligning these with that of the WRB. Conceptually the neocutanic B horizon should relate with the cambic horizon. This was not the case, probably due to the stricter texture, structure, and colour criteria of the cambic horizon. The neocutanic B horizon related best, but not strongly, with reducing conditions and colluvic material, probably due to the lack of diagnostics for the neocutanic B horizon, implying the weak or moderate pedogenesis thereof. The neocutanic B horizon is diagnostic for inter alia the Tukulu and Oakleaf soil forms, both that should conceptually relate to the Regosols. The Regosols are, however, defined by exclusion and do not have any diagnostic horizons, properties or materials. Similarly there is no soil form in the SAT defined by exclusion. The neocarbonate B horizon had the best relation with the calcaric material, secondary carbonates, and the calcic horizon. Conceptually the neocarbonate B horizon was expected to relate better to calcaric material (any material that effervescences strongly with HCl), but because effervescence is also expected in the secondary carbonates and calcic horizon, the relationship did not quite make sense. The relationship therefore only holds true for calcaric material where lime does not dominate the soil morphology. The soft carbonate horizon related quite well with secondary carbonates, but also with calcaric material, and the calcic horizon. The weaker relation with the latter was attributed to the stricter calcium carbonate equivalent (N 15%) and thickness (15 cm) for the calcic horizon than for calcaric material (N2%).
279
Conceptually a better relation between hardpan carbonate and the petrocalcic horizon was expected. The diagnostics, requiring the occurrence of secondary lime, b20% vertical cracks, and N 10 cm thickness probably resulted in the lower than expected relation. It is, however, expected that most hardpan carbonate horizons will fulfil these criteria. Dorbank and petroduric had a similar wider than expected relation, probably due to the stricter diagnostics (N50% induration, vertical cracks b20%) in the WRB. Regic sand had no satisfactory relation with WRB diagnostics and will most probably classify as Arenosols in the latter. Arenosols are not defined by diagnostic horizons, properties or material, which led to the lack in relationships. It is proposed that this issue be addressed in the WRB. Stratified alluvium related best with fluvic and colluvic material. The age (recent or not) and volume in the soil occupied are, however, not specified for the stratified alluvium and may lead to some stratified alluvium not classifying as fluvic or colluvic material. Colluvic material is an unsatisfactory option, because it should have a human cause. In the latter case soils with stratified alluvium will probably key out as Fluvisols, Arenosols or Regosols. The lithocutanic B related best, but weakly, with the lithological discontinuity and more weakly with ferralic properties and albeluvic tonguing. Theoretically the lithocutanic B can be considered as a lithological discontinuity in some cases. The relationship with ferralic properties and albeluvic tonguing is, however, nonsensical. Conceptually there is no equal for the lithocutanic B in the WRB and it would, at least partly, classify as continuous rock. It is proposed that the WRB consider recognition of weathering rock in its diagnostics, because it greatly impacts on the land use interpretations that can be made. Saprolite related best to lithological discontinuity and ferralic properties, but also to colluvic material. Neither of these relations makes sense. The lack of relationships is probably due to the absence of weathering rock in the diagnostic horizons, properties or materials of the WRB, similar to the absence of the lithocutanic B equivalent. Hard rock in SAT related perfectly with continuous rock in WRB, although the latter has stricter criteria (b20% vertical cracks, N10 cm apart). These criteria will almost certainly always be true for hard rock. The SAT unconsolidated materials with and without signs of wetness are fairly similar to unspecified materials with and without signs of wetness. These were therefore considered in combination. Unspecified material with signs of wetness related best with reducing conditions, stagnic, and gleyic colour pattern, although the diagnostic criteria for the former are not as strict as in the WRB. Conceptually these horizons are quite similar though. 4. Conclusions Taxonomic distance metrics offered a simple, but quantitative procedure to evaluate the similarities between the diagnostics of South African Soil Taxonomy and the World Reference Base for Soil Resources. Only few SAT diagnostics, however, related directly with WRB diagnostics. This was due to the different approaches followed by SAT and WRB. Some diagnostics are, however, fairly similar, others are similar only in concept, while some have no corresponding diagnostic in the WRB. A few proposals are outlined below that could be applied in the SAT to facilitate relation with the WRB. These proposals can also be applied ad hoc when relations between SAT and WRB need to be done. In SAT 10–20% and N20% organic carbon families should be considered for the Champagne soil form. For the soil forms with G horizons family criteria should be considered to differentiate between G horizons with a stagnic and gleyic colour pattern. SAT steers away from thickness criteria for diagnostic horizons because “it is difficult to set limits that will be consistently relevant for all soils and all purposes” (Soil Classification Working Group, 1991). It might be fortuitous to revisit this argument and attempt aligning the SAT thickness criteria with those in the WRB.
280
C.W. van Huyssteen et al. / Catena 113 (2014) 276–280
In the WRB consideration should be given to the recognition of diagnostics to encapsulate the concept of a lithocutanic B horizon. The absence of diagnostic horizons, properties, and materials for some reference soil groups (e.g. Regosols) should also be addressed. Future research to relate the SAT to the WRB should consider utilising existing South African soil analytical databases to classify these soils into the WRB. Consideration should also be given to applying the methodology adopted here to determine the taxonomic distance of diagnostics in the WRB, as well as in the SAT. Acknowledgements Funding by the South African National Research Foundation (UID72383) and the Hungarian Economic Development Centre (TET 10-1-2011-0059) is gratefully acknowledged.
Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.catena.2013.08.010. References Adanson, M., 1763. Familles des plantes. Vincent, Paris. Baier, T., Neuwirth, E., 2007. Excel :: COM :: R. Comput. Stat. 22, 91–108. Bidwell, O.W., Hole, F.D., 1964a. Numerical taxonomy and soil classification. Soil Sci. 97, 58–62. Bidwell, O.W., Hole, F.D., 1964b. An experiment in the numerical classification of some Kansas soils. Soil Sci. Soc. Am. Proc. 26, 263–268. Carré, F., Jacobson, M., 2009. Numerical classification of soil profile data using distance metrics. Geoderma 148, 336–345. Dunn, G., Everitt, B.S., 1982. An Introduction to Mathematical Taxonomy. University Press, Cambridge. Hole, F.D., Hironaka, M., 1960. An experiment in ordination of some soil profiles. Soil Sci. Soc. Am. Proc. 24, 309–312. IUSS Working Group WRB, 2006. World reference base for soil resources. World Soil Resources Reports No. 103.FAO, Rome. Jardine, N., Sibson, R., 1971. Mathematical Taxonomy. John Wiley & Sons, New York. Jones, A., Montanarella, L., Jones, R., 2005. Soil Atlas of Europe. Joint Research Centre, Ispra. Joosten, H., Clark, D., 2002. Wise Use of Mires and Peatlands. International Peat Society, Devon.
Krasilnikov, P., Ibañez Marti, J.J., Arnold, R., Shoba, S., 2009. Soil terminology, correlation and classification. Earthscan, London. Kubiëna, W.L., 1953. Bestimmungsbuch und systematik der böden Europas. Enke-Verlag, Stuttgart. Láng, V., Fuchs, M., Waltner, I., Micheli, E., 2010. Taxonomic distance measurements applied for soil classification. Agrokém. Talajt. 59, 57–64. Láng, V., Fuchs, M., Watner, I., Michéli, E., 2013. Soil taxonomic distance, a tool for correlation: as exemplified by the Hungarian Brown Forest Soils and related WRB Reference Soil Groups. Geoderma 192, 269–276. Le Roux, P.A.L., Ellis, F., Merryweather, F.R., Schoeman, J.L., Snyman, K., Van Deventer, P.W., Verster, E., 1999. Guidelines for the Mapping and Interpretation of Soils in South Africa. University of the Free State, Bloemfontein. McBratney, A.B., Odeh, I.O.A., Bishop, T.F.A., Dunbar, M.S., Shatar, T.M., 2000. An overview of pedometric techniques for use in soil survey. Geoderma 97, 293–327. McBratney, A.B., Minasny, B., Viscarra Rossel, R., 2009. Numerical classification: a missed, but not a lost, opportunity. From the Dokuchaev School to numerical classification conference. 18 September 2009. Gödöllő, Hungary. Minasny, B., McBratney, A.B., Hartemink, A.E., 2009. Global pedodiversity, taxonomic distance, and the World Reference Base. Geoderma 155, 132–139. Rydin, H., Jeglum, J., 2006. The biology of peatlands. Biology of Habitats SeriesOxford University Press, Oxford. Sarkar, P.K., Bidwell, O.W., Marcus, L.F., 1966. Selection of characteristics for numerical classification of soils. Soil Sci. Soc. Am. Proc. 30, 269–272. Sneath, P.H.A., 1962. The construction of taxonomic groups. In: Ainsworth, G.C., Sneath, P.H.A. (Eds.), Microbial Classification. Cambridge University Press, Cambridge, pp. 289–332. Soil Classification Working Group, 1991. Soil Classification — A Taxonomic System for South Africa. Dept. of Agricultural Development, Pretoria (http://library.wur.nl/isric/ fulltext/isricu_i14123_001.pdf - Accessed 17 July 2013). Soil Survey Staff, 2010. Keys to soil taxonomy, 11th ed. USDA—Natural Resources Conservation Service, Washington, DC. Sokal, R.R., Sneath, P.H.A., 1963. Principles of numerical taxonomy. Freeman, London. Van Huyssteen, C.W., Ellis, F., 1997. The relationship between subsoil colour and degree of wetness in a suite of soils in the Grabouw district, Western Cape. I. Characterization of colour defined horizons. S. Afr. J. Plant Soil 14, 149–153. Van Huyssteen, C.W., Hensley, M., Le Roux, P.A.L., 2004. The relationship between soil morphology and soil water regime: preliminary results in the Weatherley catchment. S. Afr. J. Plant Soil 21, 240–244. Van Huyssteen, C.W., Hensley, M., Le Roux, P.A.L., Zere, T.B., Du Preez, C.C., 2005. The relationship between soil water regime and soil profile morphology in the Weatherley catchment, an afforestation area in the north-eastern Eastern Cape. WRC Report No K5/1317.WRC, Pretoria. Vepraskas, M.J., Faulkner, S.P., 2001. Redox Chemistry of hydric soils. In: Richardson, J.L., Vepraskas, M.J. (Eds.), Wetland Soils: Genesis, Hydrology, Landscapes, and Classification. Lewis Publishers, Washington DC, pp. 85–106. Webster, R., 1977. Quantitative and numerical methods in soil classification and survey. Monographs on Soil Survey.Clarendon Press, New York.
Contents lists available at ScienceDirect
Catena journal homepage: www.elsevier.com/locate/catena
Taxonomic distance between South African diagnostic horizons and the World Reference Base diagnostics C.W. van Huyssteen a,⁎, E. Michéli b, M. Fuchs b, I. Waltner b a b
Department of Soil, Crop and Climate Sciences, University of the Free State, P.O. Box 339, Bloemfontein 9300, South Africa Department of Soil Science and Agricultural Chemistry, Szent István University, 2103 Gödöllő, Páter Károly u. 1, Hungary
a r t i c l e
i n f o
Article history: Received 2 March 2012 Received in revised form 26 July 2013 Accepted 12 August 2013 Keywords: Classification Diagnostics Euclidean distance Morpho-genetic Property-based
a b s t r a c t A myriad of soil classifications exist internationally. These usually cater for unique national variations and conditions. These different classification systems, however, hinder international communication. This paper attempted to relate the South African Soil Taxonomic (SAT) soil classification system with the World Reference Base for Soil Resources (WRB) through taxonomic distance classification. A probability matrix of the presence of selected identifiers of the diagnostic elements (properties, horizons, materials) of the South African classification system and the WRB was constructed to determine the taxonomic relationships between them. Euclidean distance calculation on these data enabled numeric expression of the taxonomic similarities and dissimilarities between the South African and WRB diagnostics. Results proved encouraging and some recommendations can be made. For example, a N 20% OC family for the organic O, as well as stagnic and gleyic families for the G horizon is proposed. It is further proposed that the WRB consider recognition of red apedal B, yellow-brown apedal B, and lithocutanic B horizons. Since the compared units are the basic building blocks of the two systems, the results presented here can be useful in the relation of soil classification in the South African Soil Taxonomy to the WRB. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Humans classify the objects in their environment to create order, to reduce variability, to increase understanding, and to inventorise (Krasilnikov et al., 2009). A myriad of soil classification systems were developed, probably due to the relative young nature of the science, the unique local variation of soils, and lack of means of easy information exchange. A further consideration might be that scientists in each region consider different soil properties with varied interest. This situation creates obvious challenges for comparing soils and international communication. The United States Department of Agriculture Soil Taxonomy (Soil Survey Staff, 2010) and the World Reference Base for Soil Resources (WRB; IUSS Working Group WRB, 2006) are used for international communication. The WRB has, however, been adopted as preferred soil correlation system by the International Union of Soil Sciences and the European Union (Jones et al., 2005). The WRB is based on diagnostic horizons, properties, and materials, each with strict differentiating quantitative criteria and definitions. The WRB has two tiers: 32 reference soil groups (RSGs), determined by a key and qualifiers (that are accommodated as prefixes or suffixes to the RSG). Although it is stated that the differentiating criteria “should ⁎ Corresponding author. Tel.: +27 51 401 9247; fax: +27 51 401 2212. E-mail addresses: [email protected] (C.W. van Huyssteen), [email protected] (E. Michéli), [email protected] (M. Fuchs), [email protected] (I. Waltner). 0341-8162/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.catena.2013.08.010
be measureable and observable in the field” the reality is quite different and in many cases almost complete chemical and physical analysis of the soil profile (horizons) is required to perform a classification. This is especially true where the local soil properties are not known. Soil Classification — A Taxonomic System for South Africa (SAT) is used exclusively in South Africa (Soil Classification Working Group, 1991). The system follows a morpho-genetic approach, similar to that proposed by Kubiëna (1953). Very few chemical or physical analyses are therefore required to classify the soil. This feature makes the classification system uniquely suited to the low-technology (resource poor) environment experienced in Africa. The South African Taxonomy defines five topsoil and 25 subsoil diagnostic horizons, combinations of which give rise to 73 soil forms. The soil forms are subdivided into soil families, based on 19 sets of distinguishing properties. The final classification should also include the soil depth and topsoil texture. The challenge is, however, to relate the South African Taxonomy with the WRB. Efforts in this regard are severely hampered by the differences in approaches (principles) between the two systems (morphogenetic vs. property-based). Taxonomic distance calculations, first promoted by Adanson (1763), are based on measures of similarity often applied in the phenetic view of numerical taxonomy, where the relative similarities or dissimilarities are measured based on different attributes without a priory weighting (Dunn and Everitt, 1982; Jardine and Sibson, 1971; Sneath, 1962; Sokal and Sneath, 1963). Taxonomic distance calculation in soil science was first proposed by Hole and Hironaka (1960). Numerical soil classification has been applied
C.W. van Huyssteen et al. / Catena 113 (2014) 276–280
277
Table 1 Physical, chemical, and morphological properties of soils selected from the WRB diagnostic criteria that were used as dominant identifiers and coded against SAT and WRB diagnostics. Physical properties
Chemical properties
Morphological properties
Consolidated material b20% Cracks N10 cm Horizontal crack spacing Vertical cracks N10 cm apart b50% (v/v) Rock structure b80% Gravel Irreversible hardening N50% Indurated Indurated by CaCO3 b10% Dipersible clay b10% Weatherable minerals ≥4.5 MPa penetration resistance ≥4 MPa penetration resistance ≥50% Slaking Clay increase Doubling of clay content over 7.5 cm ≥8% Clay ≥30% Clay Very fine sand, loamy very fine sand or finer texture Loamy sand or finer texture
Organic material saturated b30 consecutive days Organic material saturated ≥30 consecutive days Organic carbon accumulation b0.5% Organic carbon N0.6% Organic carbon b5.9 pH b50% BS N50% BS Effervescence with HCl Strong effervescence with HCl N0.50% AlO+FeO N15% Exchangeable sodium percentage N2% Calcium carbonate equivalent Calcium carbonate equivalent ≥15% Cation exchange capacity b16 cmolc kg−1 clay Si accumulation
≥5% Red/black nodules N15% Nodules N40% Nodules Indurated nodules or plates N5% Oximorphic colours N90% Reductimorphic colours or ≥15% Red mottles or Dark/red colour Darker than overlying Grey dry colour Grey moist colour Munsell value and chroma ≤3 Pale pedfaces and bright interiors Evidence of clay illuviation (clay bridging/films) ≥5% (v/v) Secondary carbonates N25% Stratification ≥10% (v/v) Si nodules/fragments Under albic Wedge-shaped peds Slickensides Massive, blocky, columnar or prismatic structure ≥10 cm Peds
in multiple studies since, including some with distance metrics. Most of these studies, however, focused on smaller areas and/or datasets and therefore had only limited applicability or scope (Bidwell and Hole, 1964a,b; McBratney et al., 2000; Sarkar et al., 1966). National and international application of the method has not been published for almost another decade (McBratney et al., 2009). Carré and Jacobson (2009) incorporated distance metrics into their OSACA model application and used it to allocate soils to existing classifications or to derive centroids and create new classifications based on clustering. Minasny et al. (2009) were first to apply taxonomic distance metrics at an international level for the WRB Reference Soil Groups. Their concept-based approach focused on the dominant identifiers, or diagnostic criteria, of the system instead of deriving centroids. With the centroid-based approach the taxonomic distance metrics are generally based on actual numerical data, where a centroid is calculated for the group in question and for each selected attribute (“dominant identifier”). However, distance calculations can also be based on the presence or absence of certain features, derived from scientifically sound concepts (Minasny et al., 2009). These concept- and centroidbased approaches were enhanced for the semi-quantitative analysis and correlation of different classification systems (Láng et al., 2010, 2013). Probably the simplest and most common method to calculate taxonomic relationships is the Euclidean distance (Dunn and Everitt, 1982; Webster, 1977). The taxonomic distance (dij) is based on Pythagoras' theorem, so for points xi and xj with two variables it can be expressed as: dij ¼
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi r 2 2 2 xi1 −x j1 þ xi2 −x j2
ð1Þ
Extending the same principles to multiple dimensions (representing multiple variables) gives: vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u p 2 X u 2 xik −xjk dij ¼ t
ð2Þ
k¼1
where p is the number of dimensions observed. In application to soil classification, the taxon being considered forms the centre and the distance of the other taxa from the centre are calculated based on the selected soil properties (diagnostic criteria in this instance). Interpretation can therefore not be made between the different
taxa, but only between the two (i.e. the central and one additional) taxa under consideration. For example, two taxa may have the same distance from the centre, but may be equally far from each other, because the direction of the departure is not specified. The purpose of this paper was to explore the relationship between diagnostic horizons in the South Africa Soil Taxonomy and the diagnostic horizons, materials, and properties in the WRB. 2. Methods This taxonomic distance calculation focussed only on relating 33 possibly related WRB diagnostic horizons, properties, and materials with the SAT diagnostic horizons. The WRB qualifiers and SAT family criteria were therefore excluded. In addition the fulvic and melanic horizons, andic and vitric properties, and tephric material were excluded, because recent pyroclastic deposits are not known in South Africa. All the anthropogenic diagnostics in the WRB and SAT were excluded. Firstly because the hortic, irriagric, plaggic, terric or anthraquic horizons are not known to exist in South Africa and secondly because the diagnostics for the man-made soil deposit are not scientifically well developed. The physical, chemical, and morphological properties selected as dominant identifiers are presented in Table 1. The taxonomic distance calculation was done on the concept based approach (Láng et al., 2010; Minasny et al., 2009), by constructing a coded matrix table. The matrix table expressed the probability that a property (selected as dominant identifier) must be present (code: 1), cannot be present (code: 0), or is likely to be present (code: 0.5) in the compared diagnostic elements. The codes are the probability values (0, 0.5, 1) of the presence of the selected dominant identifier (physical, chemical, and morphological criteria) of the selected diagnostics. The assignment of codes was based on the soil properties and diagnostic criteria in the compared systems (IUSS Working Group WRB, 2006; Le Roux et al., 1999; Soil Classification Working Group, 1991), personal experience, and expert judgement (the coded matrix and calculated taxonomic distances are available as online supplementary material.) Taxonomic distances were calculated, based on the matrix table, as the Euclidean distance between the different taxa by using the R software package (Baier and Neuwirth, 2007). Mathematically a taxonomic distance value of 0.0 indicates an exact similarity. Assessment of the data has shown that values less than 1.0 indicate large similarities, while values approaching 2.0 and above indicate large differences.
278
C.W. van Huyssteen et al. / Catena 113 (2014) 276–280
3. Results and discussion For the purpose of discussion only the WRB diagnostic horizons, properties, and materials closest, based on the calculated taxonomic distance, to the SAT diagnostic horizons are considered here. The three closest WRB diagnostics and the most important similarities and dissimilarities between them could be elucidated, based on the taxonomic distance values (Table 2). The results therefore not only serve to highlight the WRB diagnostics closest to the SAT diagnostic horizons, but also indicate the relation of the SAT diagnostic to other WRB diagnostics, enlightening some of the underlying processes and principles. The SAT organic O, Vertic A, and Melanic A topsoil horizon diagnostics are in general agreement with their WRB counterparts. Some diagnostics can, however, be included at the SAT family level to improve these relationships. Consideration should, for example, be given to the inclusion of 10–20% OC and N 20% OC families for the Champagne form. The latter family would then better relate with OC requirement of the WRB organic material. This would enable better relation with the WRB histic horizon (because the SAT organic O is saturated with water) and also with the definition of peat (Joosten and Clark, 2002; Rydin and Jeglum, 2006), whilst also keeping the current N10% OC diagnostic as a family criterion. The humic A horizon related best with the WRB spodic horizon. Although this relation is not strong and nonsensical, further research in this regard might be valuable, even though the pedogenic processes between these two are different. The melanic A must have dark colours in the moist and dry state, while only in the moist state for the mollic horizon. The orthic A horizon related best, but not strongly, with the albic horizon and albeluvic tonguing. These are subsurface horizons and the relation is therefore nonsensical. The lack of a clear relationship of the orthic A with WRB diagnostics may, at least partly be because WRB currently does not define an ochric
horizon. This should not be problematic because the orthic A is in reality not a diagnostic in the SAT. The E horizon related well with the albic, while the G horizon related with the WRB reducing conditions as well as stagnic and gelyic colour pattern. The main difference between the E and albic horizons is that the albic is “grey” in the moist state, while the E may be “yellow” in the moist state and “grey” in the dry state. This is something the WRB might consider as a qualifier, because the yellow E horizon is interpreted as being drier than the grey E horizon in the same environment (Van Huyssteen and Ellis, 1997). The poor relationships between the G horizon and the WRB reducing conditions and stagnic and gleyic colour pattern were caused by the different approaches between the two systems. The G horizon is defined inter alia in terms of water saturation, colour, accumulation of colloidal matter, while the stagnic and gleyic colour patterns only refer to colour configuration and reducing conditions to a low redox potential (rH b 20). SAT should consider the inclusion of a “gleyic” and “stagnic” family criteria for the soil forms with G horizons to improve the water saturation inferences that can be made to facilitate relation with the WRB. The red and yellow-brown apedal B horizons related best with ferralic properties, colluvic material, and lithological discontinuity. The relationship with ferralic properties is not strong, but could make sense given that the red and yellow-brown apedal B horizons consist mainly of kaolinite clays. The relationships with colluvic material and lithological discontinuity are not logical. The poor relationships are probably because the WRB does not place as much emphasis on drainage status as the SAT, but rather focuses on clay illuviation and base status. In the same environment the red apedal B horizon is drier than the yellow-brown apedal B horizon (Van Huyssteen and Ellis, 1997; Van Huyssteen et al., 2004) and should therefore be considered by the WRB. Differentiation between these is especially advantageous in the
Table 2 SAT diagnostics, and their 3 closest correlations to WRB diagnostics based on the calculated taxonomic distance matrix (distance values indicated between brackets). SAT diagnostics
Closest WRB diagnostic(s) based on distance matrix
2nd closest WRB diagnostic(s) based on distance matrix
3nd closest WRB diagnostic(s) based on distance matrix
Organic O Humic A Vertic A Melanic A Orthic A E
Histic horizon (0.71) Spodic horizon (1.58) Vertic horizon (0.71) Mollic horizon (0.87) Albic horizon (1.66) Albic horizon (1.00)
Organic material (1.12) Umbric horizon (1.80) Vertic properties (1.12) Umbric horizon (1.32) Albeluvic tonguing (1.73) Albeluvic tonguing (1.12)
G Red apedal B Yellow brown apedal B Red structured B Soft plinthic B Hard plinthic B Prismacutanic B
Reducing conditions (1.73) Ferralic properties (1.41) Ferralic properties (1.22) Nitic horizon (0.87) Reducing conditions; Plinthic horizon; Stagnic colour pattern (1.22) Petroplinthic horizon (1.00) Argic horizon; Natric horizon (1.94)
Pedocutanic B Podzol B Placic pan Neocutanic B
Natric horizon (1.94) Spodic horizon (0.50) Spodic horizon (2.06) Reducing conditions (1.22)
Gleyic colour pattern (1.80) Colluvic material (1.50) Lithological discontinuity (1.41) Ferralic properties (2.12) Ferric horizon; Pisoplinthic horizon (1.32) Pisoplinthic horizon (1.22) Abrupt textural change; Vertic properties (2.50) Argic horizon (2.06) Sombric horizon (2.06) Sombric horizon (2.78) Colluvic material (1.41)
Folic horizon (1.73) Folic horizon; Histic horizon (1.87) Mollic horizon (2.12) Vertic properties (2.00) Colluvic material (1.80) Colluvic material; Reducing conditions; Ferralic horizon (1.87) Stagnic colour pattern (1.87) Lithological discontinuity; Fluvic material (1.73) Colluvic material (1.50) Lithological discontinuity (2.24) Albeluvic tonguing (1.66)
Neocarbonate B Soft carbonate
Calcaric material; Secondary carbonates (0.71) Secondary carbonates (0.87)
Hardpan carbonate Dorbank Regic sand Stratified alluvium Lithocutanic B
Petrocalcic horizon (0.87) Petroduric horizon (0.71) Lithological discontinuity (1.94) Fluvic material (0.87) Lithological discontinuity (1.87)
Saprolite Hard rock Unspecified material with signs of wetness
Lithological discontinuity; Ferralic properties (1.66) Continuous rock (0.00) Reducing conditions (1.22)
Calcic horizon (1.00) Calcaric material; Calcic horizon (1.12) Petroduric horizon (1.94) Petrocalcic horizon (2.00) Ferralic properties (2.06) Colluvic material (1.00) Ferralic properties; Albeluvic tonguing (2.00) Colluvic material (1.87) Lithological discontinuity (3.12) Stagnic colour pattern (1.41)
Plinthic horizon (2.60) Vertic horizon (2.55) Abrupt textural change; Vertic properties (2.50) Ferralic properties (2.24) Ferralic properties (3.00) Ferralic properties; Lithological discontinuity; Cambic horizon (1.50) Cambic horizon (2.06) Duric horizon (2.12) Calcic horizon (2.65) Duric horizon (2.35) Colluvic material; Albic horizon (2.12) Ferralic properties (1.80) Albic horizon; Reducing conditions (2.06) Fluvic material (2.06) Fragic horizon; Petroplinthic horizon (3.28) Gleyic colour pattern (1.50)
C.W. van Huyssteen et al. / Catena 113 (2014) 276–280
more arid environments, such as South Africa. The luvic B horizon family criterion of SAT can, however, be changed to better align the luvic red and yellow-brown apedal B horizons with the WRB argic horizon, because the luvic B criterion is very similar (5% vs. 3% clay increase and 1.3 times vs. 1.2 times more clay) to the clay increase required for the WRB argic horizon. Consideration should therefore be given to changing the luvic B diagnostics to align to that of the agric horizon. The red structured B related very well to the nitic horizon, but does not have a perfect relation with it, because the latter has additional chemical (texture and Fe) criteria, that may or may not necessarily be true for the former. Soft plinthic B horizons related with WRB reducing conditions, stagnic colour pattern, plinthic horizon, and gleyic colour pattern. Reducing conditions are not always expected in the soft plinthic B horizon, but definitely below the horizon. The duration of reducing conditions in the soft plinthic B horizon is expected to be approximately eight months per year (Van Huyssteen et al., 2005), but this is not specified for the WRB reducing conditions or stagnic colour pattern. It is quite possible that reducing conditions are present for only a few (b 14) days and therefore satisfy the criteria, but the soil fails to develop redox morphology (Vepraskas and Faulkner, 2001). The soft plinthic B horizon does not harden irreversibly upon drying and should therefore not be equated to the WRB plinthic horizon. The hard plinthic B can for all practical purposes be equated with the petroplinthic horizon. The major differences are that the hard plinthic B does not necessarily have a penetration resistance ≥4.5 MPa, Feo:Fed b0.10 or ≥10 cm thick. The prismacutanic B and pedocutanic B horizons both related weakly with the natric and argic horizons. Conceptually the prismacutanic B was expected to relate better with the natric horizon, although it does not necessarily have the required N 15% exchangeable sodium percentage or thickness of the natric horizon. Similarly, the pedocutanic B was expected to relate better with the argic horizon, but is not defined in terms of clay increase or thickness, resulting in a wider taxonomic distance. As expected, the podzol B related very well with the spodic horizon. The placic pan also related best, but poorly, with the spodic horizon. The chemical criteria differ between the podzol B and placic pan on the one hand and the spodic horizon on the other, resulting in wider relations than expected. Further the SAT placic pan is defined in the WRB as a placic qualifier and not as diagnostic horizon, property or material. The origin of the chemical criteria for the podzol B is unclear and attention should therefore be directed at aligning these with that of the WRB. Conceptually the neocutanic B horizon should relate with the cambic horizon. This was not the case, probably due to the stricter texture, structure, and colour criteria of the cambic horizon. The neocutanic B horizon related best, but not strongly, with reducing conditions and colluvic material, probably due to the lack of diagnostics for the neocutanic B horizon, implying the weak or moderate pedogenesis thereof. The neocutanic B horizon is diagnostic for inter alia the Tukulu and Oakleaf soil forms, both that should conceptually relate to the Regosols. The Regosols are, however, defined by exclusion and do not have any diagnostic horizons, properties or materials. Similarly there is no soil form in the SAT defined by exclusion. The neocarbonate B horizon had the best relation with the calcaric material, secondary carbonates, and the calcic horizon. Conceptually the neocarbonate B horizon was expected to relate better to calcaric material (any material that effervescences strongly with HCl), but because effervescence is also expected in the secondary carbonates and calcic horizon, the relationship did not quite make sense. The relationship therefore only holds true for calcaric material where lime does not dominate the soil morphology. The soft carbonate horizon related quite well with secondary carbonates, but also with calcaric material, and the calcic horizon. The weaker relation with the latter was attributed to the stricter calcium carbonate equivalent (N 15%) and thickness (15 cm) for the calcic horizon than for calcaric material (N2%).
279
Conceptually a better relation between hardpan carbonate and the petrocalcic horizon was expected. The diagnostics, requiring the occurrence of secondary lime, b20% vertical cracks, and N 10 cm thickness probably resulted in the lower than expected relation. It is, however, expected that most hardpan carbonate horizons will fulfil these criteria. Dorbank and petroduric had a similar wider than expected relation, probably due to the stricter diagnostics (N50% induration, vertical cracks b20%) in the WRB. Regic sand had no satisfactory relation with WRB diagnostics and will most probably classify as Arenosols in the latter. Arenosols are not defined by diagnostic horizons, properties or material, which led to the lack in relationships. It is proposed that this issue be addressed in the WRB. Stratified alluvium related best with fluvic and colluvic material. The age (recent or not) and volume in the soil occupied are, however, not specified for the stratified alluvium and may lead to some stratified alluvium not classifying as fluvic or colluvic material. Colluvic material is an unsatisfactory option, because it should have a human cause. In the latter case soils with stratified alluvium will probably key out as Fluvisols, Arenosols or Regosols. The lithocutanic B related best, but weakly, with the lithological discontinuity and more weakly with ferralic properties and albeluvic tonguing. Theoretically the lithocutanic B can be considered as a lithological discontinuity in some cases. The relationship with ferralic properties and albeluvic tonguing is, however, nonsensical. Conceptually there is no equal for the lithocutanic B in the WRB and it would, at least partly, classify as continuous rock. It is proposed that the WRB consider recognition of weathering rock in its diagnostics, because it greatly impacts on the land use interpretations that can be made. Saprolite related best to lithological discontinuity and ferralic properties, but also to colluvic material. Neither of these relations makes sense. The lack of relationships is probably due to the absence of weathering rock in the diagnostic horizons, properties or materials of the WRB, similar to the absence of the lithocutanic B equivalent. Hard rock in SAT related perfectly with continuous rock in WRB, although the latter has stricter criteria (b20% vertical cracks, N10 cm apart). These criteria will almost certainly always be true for hard rock. The SAT unconsolidated materials with and without signs of wetness are fairly similar to unspecified materials with and without signs of wetness. These were therefore considered in combination. Unspecified material with signs of wetness related best with reducing conditions, stagnic, and gleyic colour pattern, although the diagnostic criteria for the former are not as strict as in the WRB. Conceptually these horizons are quite similar though. 4. Conclusions Taxonomic distance metrics offered a simple, but quantitative procedure to evaluate the similarities between the diagnostics of South African Soil Taxonomy and the World Reference Base for Soil Resources. Only few SAT diagnostics, however, related directly with WRB diagnostics. This was due to the different approaches followed by SAT and WRB. Some diagnostics are, however, fairly similar, others are similar only in concept, while some have no corresponding diagnostic in the WRB. A few proposals are outlined below that could be applied in the SAT to facilitate relation with the WRB. These proposals can also be applied ad hoc when relations between SAT and WRB need to be done. In SAT 10–20% and N20% organic carbon families should be considered for the Champagne soil form. For the soil forms with G horizons family criteria should be considered to differentiate between G horizons with a stagnic and gleyic colour pattern. SAT steers away from thickness criteria for diagnostic horizons because “it is difficult to set limits that will be consistently relevant for all soils and all purposes” (Soil Classification Working Group, 1991). It might be fortuitous to revisit this argument and attempt aligning the SAT thickness criteria with those in the WRB.
280
C.W. van Huyssteen et al. / Catena 113 (2014) 276–280
In the WRB consideration should be given to the recognition of diagnostics to encapsulate the concept of a lithocutanic B horizon. The absence of diagnostic horizons, properties, and materials for some reference soil groups (e.g. Regosols) should also be addressed. Future research to relate the SAT to the WRB should consider utilising existing South African soil analytical databases to classify these soils into the WRB. Consideration should also be given to applying the methodology adopted here to determine the taxonomic distance of diagnostics in the WRB, as well as in the SAT. Acknowledgements Funding by the South African National Research Foundation (UID72383) and the Hungarian Economic Development Centre (TET 10-1-2011-0059) is gratefully acknowledged.
Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.catena.2013.08.010. References Adanson, M., 1763. Familles des plantes. Vincent, Paris. Baier, T., Neuwirth, E., 2007. Excel :: COM :: R. Comput. Stat. 22, 91–108. Bidwell, O.W., Hole, F.D., 1964a. Numerical taxonomy and soil classification. Soil Sci. 97, 58–62. Bidwell, O.W., Hole, F.D., 1964b. An experiment in the numerical classification of some Kansas soils. Soil Sci. Soc. Am. Proc. 26, 263–268. Carré, F., Jacobson, M., 2009. Numerical classification of soil profile data using distance metrics. Geoderma 148, 336–345. Dunn, G., Everitt, B.S., 1982. An Introduction to Mathematical Taxonomy. University Press, Cambridge. Hole, F.D., Hironaka, M., 1960. An experiment in ordination of some soil profiles. Soil Sci. Soc. Am. Proc. 24, 309–312. IUSS Working Group WRB, 2006. World reference base for soil resources. World Soil Resources Reports No. 103.FAO, Rome. Jardine, N., Sibson, R., 1971. Mathematical Taxonomy. John Wiley & Sons, New York. Jones, A., Montanarella, L., Jones, R., 2005. Soil Atlas of Europe. Joint Research Centre, Ispra. Joosten, H., Clark, D., 2002. Wise Use of Mires and Peatlands. International Peat Society, Devon.
Krasilnikov, P., Ibañez Marti, J.J., Arnold, R., Shoba, S., 2009. Soil terminology, correlation and classification. Earthscan, London. Kubiëna, W.L., 1953. Bestimmungsbuch und systematik der böden Europas. Enke-Verlag, Stuttgart. Láng, V., Fuchs, M., Waltner, I., Micheli, E., 2010. Taxonomic distance measurements applied for soil classification. Agrokém. Talajt. 59, 57–64. Láng, V., Fuchs, M., Watner, I., Michéli, E., 2013. Soil taxonomic distance, a tool for correlation: as exemplified by the Hungarian Brown Forest Soils and related WRB Reference Soil Groups. Geoderma 192, 269–276. Le Roux, P.A.L., Ellis, F., Merryweather, F.R., Schoeman, J.L., Snyman, K., Van Deventer, P.W., Verster, E., 1999. Guidelines for the Mapping and Interpretation of Soils in South Africa. University of the Free State, Bloemfontein. McBratney, A.B., Odeh, I.O.A., Bishop, T.F.A., Dunbar, M.S., Shatar, T.M., 2000. An overview of pedometric techniques for use in soil survey. Geoderma 97, 293–327. McBratney, A.B., Minasny, B., Viscarra Rossel, R., 2009. Numerical classification: a missed, but not a lost, opportunity. From the Dokuchaev School to numerical classification conference. 18 September 2009. Gödöllő, Hungary. Minasny, B., McBratney, A.B., Hartemink, A.E., 2009. Global pedodiversity, taxonomic distance, and the World Reference Base. Geoderma 155, 132–139. Rydin, H., Jeglum, J., 2006. The biology of peatlands. Biology of Habitats SeriesOxford University Press, Oxford. Sarkar, P.K., Bidwell, O.W., Marcus, L.F., 1966. Selection of characteristics for numerical classification of soils. Soil Sci. Soc. Am. Proc. 30, 269–272. Sneath, P.H.A., 1962. The construction of taxonomic groups. In: Ainsworth, G.C., Sneath, P.H.A. (Eds.), Microbial Classification. Cambridge University Press, Cambridge, pp. 289–332. Soil Classification Working Group, 1991. Soil Classification — A Taxonomic System for South Africa. Dept. of Agricultural Development, Pretoria (http://library.wur.nl/isric/ fulltext/isricu_i14123_001.pdf - Accessed 17 July 2013). Soil Survey Staff, 2010. Keys to soil taxonomy, 11th ed. USDA—Natural Resources Conservation Service, Washington, DC. Sokal, R.R., Sneath, P.H.A., 1963. Principles of numerical taxonomy. Freeman, London. Van Huyssteen, C.W., Ellis, F., 1997. The relationship between subsoil colour and degree of wetness in a suite of soils in the Grabouw district, Western Cape. I. Characterization of colour defined horizons. S. Afr. J. Plant Soil 14, 149–153. Van Huyssteen, C.W., Hensley, M., Le Roux, P.A.L., 2004. The relationship between soil morphology and soil water regime: preliminary results in the Weatherley catchment. S. Afr. J. Plant Soil 21, 240–244. Van Huyssteen, C.W., Hensley, M., Le Roux, P.A.L., Zere, T.B., Du Preez, C.C., 2005. The relationship between soil water regime and soil profile morphology in the Weatherley catchment, an afforestation area in the north-eastern Eastern Cape. WRC Report No K5/1317.WRC, Pretoria. Vepraskas, M.J., Faulkner, S.P., 2001. Redox Chemistry of hydric soils. In: Richardson, J.L., Vepraskas, M.J. (Eds.), Wetland Soils: Genesis, Hydrology, Landscapes, and Classification. Lewis Publishers, Washington DC, pp. 85–106. Webster, R., 1977. Quantitative and numerical methods in soil classification and survey. Monographs on Soil Survey.Clarendon Press, New York.