The paper "Classification of Paleosols" by Mack et al., (1993, hereafter called "the authors") [see related abstract on p. 9], read from a soil scientist's scope, evokes considerable criticism, some of which follows below.
The purpose of classifying soils. It is true that paleopedologists so far have not been able to overcome the problem of classification of paleosols. Attempts to use Soil Taxonomy (Soil Survey Staff, 1975, 1990, 1992) and other classification systems always leave considerable doubt because of missing data. Modern soil classification systems, such as Soil Taxonomy and FAO (1974, 1988, 1990) are meant to provide a link between unaggregated soil properties and a soil's potential uses. Because potential use depends heavily on climate, nutrients, and water availability, climatic factors (aridisols, boreal subgroups), fertility aspects (base saturation) and water holding properties (thickness of humic surface horizons) are emphasized when they are considered to be more or less extreme in the positive or negative sense.
Classifying paleosols serves a different purpose: the comparison of ancient soils with modern ones with the aim of obtaining information on the fossil environment. Therefore, the emphasis in the classification of paleosols will necessarily be different. In paleopedology, climate frequently is a factor that needs to be reconstructed from the soil properties and that can therefore not be part of the classification of paleosols. As the authors state, base saturation of many paleosols cannot be measured or even reconstructed (see below). Humose surface horizons may have lost much of their organic matter or have been removed by erosion. Clay minerals may have recrystallized into a different assemblage. Therefore, the idea to classify paleosols by properties that can still be measured after burial and that have not suffered from diagenesis, is a valid one.
In their paper, the authors state that their proposal is meant for all paleosols, irrespective of age. They therefore exclude from the classification all properties that may undergo major changes during diagenesis. Nevertheless, excluding properties from the classification because they may have changed in some soils implies that we, unnecessarily, throw in the waste basket a lot of useful information. This severely limits the applicability and the resolution of their system. Many paleosols, especially the Tertiary and Quaternary ones, have not intensely suffered from late diagenesis and still preserve most of their clay mineralogy, including "unstable" materials such as allophane. Not using such information to its full extent is comparable to abandoning Milankovitch's curve for Quaternary glaciations because the resolution of the record of the Permian glaciations is too low to allow its recognition. Also in the study of paleosols we suffer from the reverse telescope effect: the older the soils, the more reluctantly they reveal their secrets. It might therefore be more useful to distinguish between diagenetically altered paleosols and those without major diagenetic alteration before attempting any classification, so that the higher resolution of properties in the latter is recognized and used.
Adapting existing classification names. Although it may be difficult to apply existing classification systems to paleosols, an additional problem arises when existing soil classification names are used with changed definitions. To change definitions of a commonly used term, or, even worse, to use side by side two different definitions for the same feature or group of features, creates confusion where the authors want to create order. This is, of course, the case with the terms borrowed from Soil Taxonomy, but also with some of the proposed "new" names, such as Calcisols, Gypsisols, and Gleysols. The latter are by no means new terms in soil science. The term Gleysols was introduced in the FAO Legend for the World Soil Map (FAO, 1974), and the terms Calcisols and Gypsisols are part of the revised legend (FAO, 1988, 1990). Similarly to Soil Taxonomy, the FAO legend/classification is used worldwide in soil classification and mapping.
Using the Order names, but different definitions for paleosols and recent soils would create considerable confusion and the purpose of paleopedology is not served: The comparison between recent and fossil soils becomes hampered by a difference in meaning of the same names. We should remember that the classification of paleosols should serve a purpose, and that purpose is not the mere giving of names.
Paleosols are not restricted to buried sediments, but they occur extensively as relicts at and close to the present surface. Such relicts usually show overprints of two or more strongly different processes of soil formation (polycyclicity). It would be unwise to use different "grades" of criteria for the classification of recent and fossil soil formation in such soils. Soil scientists have long recognized the polycyclicity of many soils and have incorporated the combined characteristics in their classification systems, without weighing the criteria of the subsequent phases of soil formation. In the study of paleosols, however, the sequence in which processes acted upon a soil may be critical to the reconstruction of the paleo-environment.
Also the choice of criteria for the various Orders recognized by the authors leaves quite a few loose ends.
Calcisols. As mentioned above, Calcisols are a unit in the FAO legend. The FAO (1988, 1990) definition is very close to that proposed by the authors, and includes the recognition of Calcisols with clay illuviation. It appears that the FAO unit is fully suitable for the authors' purposes.
Histosols. The adoption of the Histosol order for paleosols appears a logical decision. The only disagreement between the author's proposal and Soil Taxonomy is that in the latter a detrital origin of the organic matter is allowed. Therefore, high carbonaceous oil shales need not be excluded. For paleosols, however, it may be desirable to distinguish between in situ and transported material at a lower level of the classification.
Spodosols. The authors state that it is necessary to demonstrate a down-profile increase into the spodic horizon of both organic matter and iron. This criterion has two major setbacks: 1) spodosols that have lost their eluvial horizon by erosion cannot be identified, and 2) hydromorphic sandy spodosols, in which the B horizon is characterized by accumulation of organic matter and aluminum (and where iron is removed) are excluded. Soil Taxonomy recognizes these hydromorphic profiles. Chemically, the proposed criterion is not sound, because iron accumulation can occur due to lateral water movement and not to podzolization, e.g., bog iron. By far the best criterion for recognition of spodic material is the presence of coatings or pellets of organic matter. The latest proposal for a new definition of "spodic materials," on which the Spodosol order is based (ICOMOD, 1991) is sufficiently general to include most of the fossil podzols.
Aluminum in the B horizons of podzols is not normally present as "amorphous or extremely finely crystalline aluminum hydroxide," but in organic complexes and as hydroxy-interlayers in clay minerals. Organic matter content in spodic horizons is not readily determined by combustion, unless the sediment consists of quartz sand. Dry combustion gives an overestimate of organic matter because dehydration water of layer-lattice clays is a significant part of the weight loss.
Iron in podzol-B horizons is usually goethite and not hematite. If hematite is found, this is almost certainly due to diagenetic alteration. It should be kept in mind, that podzols do not normally form in fine-grained sediments, unless the climate is boreal. Therefore podzol-like horizonation in fine-grained fluvial sediments of non-boreal climates is probably not due to podzolization but to gleying (see also Buurman, 1993; comment on the paleosols recognized by Pratt & Kelly, 1992).
Oxisols. Soil Taxonomy does not exclude soils with an oxic horizon that reaches to the soil surface from Oxisols. It is therefore not necessary to prove that the oxic horizon was a subsurface horizon. Actually, surface horizons of oxisols usually have oxic properties in the mineral fraction.
Amorphous iron and aluminum oxides and aluminum-silicon oxides (?allophane) are not common in modern oxic horizons and they are certainly not a criterion in the recognition of such soils. Iron usually has hematite or goethite (maghemite in soils from ultramafic rocks) mineralogy, while gibbsite is the only common aluminum mineral. These minerals cannot be termed amorphous (apart from the fact that, by definition, a mineral cannot be amorphous). It is a typical property of Oxisols that only a small part of the "free" iron can be extracted with acid oxalate. This acid oxalate fraction is usually considered to indicate the poorly crystalline iron minerals. Amorphous aluminum-silicon oxides, better known as allophane and imogolite, are virtually absent because they are unstable in the leaching environment that leads to the formation of Oxisols.
If we exclude the chemical and (clay) mineralogical criteria for Oxisols, it will hardly be possible to identify Oxisols, because they tend to have very little horizon differentiation. In situ Oxisols can usually be distinguished from transported oxic material (pedolith) by the presence of remnants of sedimentary structure and abrupt color transitions in the latter. Because Oxisols are soils of old landscapes and usually have a considerable horizontal extent, the quest for unweathered lateral equivalents will usually be fruitless. In most cases, comparison with underlying material will be more promising.
Vertisols. Although the authors want to exclude from their classification, criteria that cannot be measured, they forgot to do so in case of Vertisols. Vertisols are defined by a number of criteria, among which "open cracks at a depth of 50 cm that are at least 1 cm wide, at some time in most years" is one. Cracks are hardly ever preserved in paleosols, although their presence can be guessed or reconstructed. The wedge-shaped peds and slickensides that are typical of Vertisols are also formed by tectonic pressure. If the smectitic clays have disappeared during diagenesis, it is doubtful that such structural features should have survived. By implying that "pedoturbation" is equivalent to mixing due to swelling and shrinking of smectite clays, the authors forget that in most soils "pedoturbation" is due to animal activity.
Gypsisols. As Calcisols, this is a unit in the FAO (1988, 1990) legend. The definition of Gypsisols by FAO is much simpler than that proposed by the authors, but it is clear that diagenetically intruded gypsum should be excluded when paleosols are classified.
Argillisols. This order would include Alfisols and Ultisols of Soil Taxonomy, and Luvisols, Lixisols, Acrisols, Alisols, and Podzoluvisols of FAO, besides parts of some other units that have clay illuviation as a subordinate property, such as Chernozems, Kastanozems, Calcisols, Gypsisols, etc. Recognition in thin section, of (>1%) illuviated clay is by far the best criterion for the recognition of illuviated clay. The use of a textural difference with the topsoil or the subsoil is quite tricky, especially in sedimentary environments. Moreover, a difference in texture with the topsoil can be due to surface removal of fine material by erosion (compare the kandic horizon of Soil Taxonomy; SSS 1990, 1992).
Investigations on modern soils indicate that there is no clearcut relationship between base status and mineralogy of such soils. Even in the case of high amounts of weatherable ("unstable") minerals, base status can be lower, especially in high-rainfall climates. Consequently, we have no tool to guess at the base status.
Gleysols. Here the proposal of the authors is somewhat at variance with the pedological interpretation of gley. FAO defines gleyic properties as properties related to strong reduction of a soil (horizon), and Gleysols should have gleyic properties within 50 cm of the surface. Above the strongly reduced zone is frequently a mottled zone. This mottled zone actually points to the presence of soil formation: a reduced (gray) layer in a sedimentary sequence, without the overlying mottled zone or humic topsoil is not indicative of soil formation; many under-water sediments are gray without showing any evidence of alteration by plants. FAO recognizes mottled horizons to have gleyic properties as long as gray colors dominate, which means that Gleysols do not necessarily have a completely reduced horizon. It would be impossible to identify Gleysols in alluvial sedimentary systems if only the reduced horizon were used as a criterion: the superposition of many subsequent profiles frequently leads to the absence of gray-colored horizons. In FAO systems, subunits of other "orders" are gleyic if they have gleyic properties within 1 meter depth.
Protosols. The authors use the Protosol for all those soils that do not have spodic, oxic, or argillic horizons, or accumulations of calcite or gypsum. Because the authors eliminated the mollic (and umbric) horizons, this order would contain not only the Entisols and Inceptisols, but also the greater part of the Mollisols, and all Andisols. Such lumping of widely different soils induces a significant loss of information at order level.
Eliminating soil orders. Dependence of soil classification on data of soil climate is undesirable, both for present-day and for fossil soils. I therefore heartily agree with the author's elimination of Aridisols. In the FAO classification the climatic factor was excluded as well, and soils of arid climates are classified by measurable features only. Eliminating the soil orders Mollisols and Andisols, however, is a different thing. Mollisols are common fossil soils, e.g., in the U.S. and European Pleistocene loess sequences, such important markers would end up in the author's dustbin of "Protosols." Fossil Andisols are common in all areas with volcanic deposits. Actually, Andisols may be among the most common fossil soils; they occur in almost all terrestric volcanic sequences, all over the world. Although volcanic successions occur throughout the geological record, paleosols, in Quaternary volcanics most catch the eye and are most intensely studied. These soils have not suffered from strong diagenesis and still preserve their allophanic and glass materials. Andisols are a reality, both of the present and of the past. They may lose their characteristics in highly-altered sequences, but certainly not in Quaternary, and probably not in Tertiary sequences.
Combination of adjectives. The combination of adjectives in one classification name, such as "argillic, gleyed, calcic Spodosol" suggests that such soils actually exist. Although this is probably a bad example, I think that, in general, this kind of nomenclature is misleading if it is not placed in the context of polycyclicity. The evidence of illuviation of clay (argillic), groundwater influence (gleyic), calcium carbonate redistribution may occur together in one horizon but may supersede each other in a polygenetic profile. For a correct interpretation of such a profile, the sequence of processes is crucial and may have to be expressed in the classification.
An alternative solution. The criticisms uttered above might suggest that the authors will come up with a better solution. I do want to make a suggestion, but with the marginal note that it remains to be tested. In order to preserve the resolution in soils that have not been severely changed by diagenesis, it may be useful to distinguish between diagenetically altered and unaltered paleosols before attempting a classification. For diagenetically altered soils, the applicability of classification names remains to be investigated. For relatively unaltered paleosols, the FAO system, which I personally do not advocate for detailed mapping of modern soils, may be a better option than Soil Taxonomy. None of the FAO units or subunits is distinguished on the basis of climatic criteria, but they are closely linked to climatic zones. The FAO system is almost fully based on measurable criteria (some of which, however, may have been lost in paleosols). The FAO system allows the combination of adjectives that indicate specific properties, such as advocated by the authors.
I would like to caution against abandoning the classification criteria that are used for recent soils without properly testing them for paleosols. Classification systems of recent soils are based on far more information than will ever be available for paleosols and this, in reverse, will allow a better interpretation of paleosols. The classification of diagenetically altered paleosols will always remain stuck at the higher level of generalization than that of soils that still retain many of their original properties. Using a completely different classification for such soils would sever the link with modern soils and soil environments, which is one of the main purposes of the study of paleosols. Paleopedologists will have to live with the fact that paleosols cannot be classified in as much detail as modern soils, at least if the link with paleoenvironment is to be preserved.