New Approach to Experimental Soil Health Definition Using Thermogravimetric Fingerprinting

Abstract

Degradation and sealing are still frequent in soil management today despite intensive research. An unsatisfactory assessment of soil key components and soil health still limits sustainable land use. For the future evaluation of soil health, soils under productive use have been compared with natural and semi-natural soils using thermogravimetric fingerprinting of air-dried soil samples. This approach has led to a more precise quantification of known relationships and the discovery of several new ones between soil components that have evolved over thousands of years of soil formation without human intervention, each changing in a specific way due to land use. The use-related deviations from the natural soil condition allow a distinction between natural soils, disturbed soils, and soil-like carbon-containing mineral mixtures (e.g., compost, horticultural substrates). Carbon added to soils with fresh organic residues or from anthropogenic (soot, slag) or geological (coal) sources can be distinguished from soil organic matter (humus) accumulated during soil genesis, regardless of extreme chemical heterogeneity. The degree of carbon sequestration in soils is easy to quantify. Using near-natural soils as a reference, considering bound water seems to be a suitable starting point for the experimental definition of soil health. An elucidation of the causal relationships between the soil components used should accompany it.

1. Introduction: Challenges for Soil Health as a Base for Sustainable Land Use

Soils are an undeniable foundation of human existence. Historians have linked the survival of civilizations to the regional management of soils [1]. Today, the spread of land use technologies is accompanied by challenges related to the global climate, economy, and social equity. This development has also changed the way soils are viewed [2]. They are now considered non-renewable resources with multiple functions [3].

The changing understanding of soils is reflected in the definitions of soil, soil health, and soil management activities. According to Doran et al. (1996) [4], for example, soil health mainly serves to “maintain plant and animal productivity, maintain or improve water and air quality, and promote plant and animal health”. Lehmann et al. (2020) [5] see productive soil functions as only one part of soil health, while soil health should document many functions that have developed as a common good over millennia.

These different views can lead to various misunderstandings [6]. The modern view of soils is at odds with current policies and trends in land management. Short-term land use for urban development, infrastructure, industrial settlement, and mining still seems more economically important, politically attractive, and democratically enforceable than maintaining soil health. Anthropogenically modified and artificial soils, such as “anthrosols”, “technosols”, or urban soils (see the World Reference Base of Soil Resources [7]), appear to be comparable to unmodified soils in terms of productivity [8]. Today’s soil research focuses on short-term, local issues, such as productive land use, nature conservation, or urban development, reflecting a simplistic understanding of soil functions and health. It remains open to the extent to which the prevention of erosion, salinization, compaction, or contamination [9] is sufficient to adequately consider soil functions in the biosphere for sustainable land use. Current standards for land reclamation after coal mining, for dealing with soil contamination, erosion, or sealing, assume that soils and all their functions will be restored within months, years, or a few decades, not over millennia of ecosystem development. This may underestimate the global and long-term importance of the diversity and complexity of natural soil functions that have evolved throughout life on Earth, as reflected in the diversity of interconnected activities of organisms, the characteristics of the current biosphere, the heterogeneity in soil components, or the superposition of biotic and abiotic processes at different temporal (minutes to millennia) and spatial (microscopic to global) scales [10].

However, even the productivity of natural ecosystems untouched by human activity remains largely unknown. It is rarely studied except in the humid tropics. Figure 1 shows an example of extreme biomass production in temperate forests on watersheds with no human influence and no access to groundwater under temperate continental climate conditions on old glacial deposits with apparently poor soils (retisols). Despite the similarities between such rare, undisturbed-by-human activity sites and the humid tropics (e.g., stratified vegetation structure, distribution of nitrophilous species and lianas, presence of evergreen plants despite extended snow cover, absence of permanent organic litter on the soil surface, etc.), no studies have been conducted on the high assimilation activity as a basis for long-term land use.

Similar considerations apply to undisputed soil-specific key components, such as soil organic carbon (SOC) or soil organic matter (SOM). SOM influences many physical, chemical, and biological soil properties [11], and its dynamics are significantly influenced by land use and management practices. Therefore, it has been proposed as a component of all indicator sets for soil and soil health [12,13,14]. However, many studies focus on enriching soils with charcoal, biochar, plastics, or other organic residues to increase soil fertility, sequester carbon, or combat climate change rather than maintaining carbon stocks in geological deposits. Moreover, despite the improved capabilities of soil analysis through technological innovations in recent decades and different regional, temporal, political, economic, or academic foci, no single approach to soil organic matter (SOM) assessment has yet been adopted in land use practice [15]. Difficulties in distinguishing the SOM of natural soils (humus) from organic admixtures of geological or anthropogenic origin result in the lack of reliable thresholds for SOM content [16] and hinder the use of soil functions [17] and the benefits of higher yields from organic fertilization [18,19]. Despite its recognized importance, using SOM as a basis for soil health is challenging. At the same time, the “persistence of soil organic carbon through functional complexity” [5,20] poses challenges that are difficult to address with current methodological developments in the foreseeable future, while climate change increases the need for action.

Any understanding of soil as the basis for sustainable land use [21] raises the question of how to develop methods to assess soil health [22,23]. There should be several meaningful indicators [5], while a single indicator seems inappropriate due to the multiple interactions between soil and its environment [22].

We believe that soil health as a goal requires an experimentally definable baseline. According to Montgomery (2012) [1], such a baseline should ensure the long-term survival of human civilizations. Since soil formation is based on evolutionary processes in ecosystems over thousands of years, and since humanity has evolved over the same period, knowledge of soils that have not been altered by human activity could guarantee survival. However, while using unmodified reference objects is taken for granted in other fields of knowledge, using natural soils without human influence is not common in soil science. The assessment of human interventions in landscapes is reflected, for example, in efforts to define the near-naturalness, naturalness, or hemeroby of vegetation [24,25]. Such approaches are not yet known for soils. In medicine, variations in body temperature are used to detect undesirable changes in complex control systems (disease, disability). Even if simple indicators of deviations from the original or normal state do not usually provide information about the causes, they help develop therapies or answer other survival challenges.

Similarly, comparing natural and managed soils should provide easily applicable indicators of deviations from the original state, even if knowledge of causal relationships is limited. This study, therefore, documents an attempt to experimentally define soil health using a combination of simple analytical techniques with a thermogravimetric fingerprinting approach on samples from sites with different levels of human impact. The focus is on the relationships and quantitative proportions between soil components that evolve during long-term natural soil development from a changing network of regulatory processes (ecosystems) as a specific feature of soil formation.

2. Experimental Base

2.1. Sample Collection

Obtaining soil samples with low, or better, without human intervention, as reference objects for experimental soil health definition is challenging. Protected areas to preserve the natural diversity of soil functions were nowhere to be found. Modifying the soil and vegetation for food and raw materials production has been the main objective of land use over the millennia of civilization. Areas with the highest natural fertility in temperate climate conditions were generally used first. The resulting decline in fertility during land use was only partially compensated by human migration and later by fertilization. In industrialized countries, a long history of human settlement, acid rain, contaminations, and other influences make large-scale changes more likely than undisturbed soil formation. Therefore, natural soils undisturbed by humans can only be observed or studied as reference objects in exceptional cases (e.g., van Breemen 2002 [26]).

Our sampling efforts focused on regions of apparently undisturbed natural vegetation in protected or remote areas. These were supplemented by samples from native forests, long-term monitoring sites, and long-term agricultural field experiments for comparison (natural and disturbed) (see Figure 1; details in Siewert 2001 [27] and 2004 [28] and Siewert and Kučerík 2015 [29]). Soils from all climatic zones were accepted to increase the chances of finding near-natural soils and to be able to distinguish between use-related differences and natural regional soil formation peculiarities (Figure 2).

2.2. The Methodical Approach

Despite the complexity of biological regulation and adaptation, their consequences are often easy to see. In organisms, for example, they can be seen in the definable proportions of limbs or organs, as well as in the regulation of metabolism, body temperature, and chemical composition, the rules governing the structure of genetic information, and similar indicators. Examples in soils are the ratio of clay to organic matter [30] or the proportion of organic carbon in soil organic matter. Both can be readily determined using a combination of modern analytical tools. However, the limitations of traditional organic matter determination by loss on ignition [31,32,33,34] raise the question of how these relationships can be more accurately determined for assessing soil health. Modern thermobalances offer the possibility to consider the influence of bound-to-clay surface water on the determination of organic matter and thus compensate for a disturbing factor [30,31].

These considerations led to using thermogravimetry as a reliable methodological basis, as a modern extension for determining SOM by loss on ignition, and as a supplement to the standard determination of SOC, nitrogen, and clay contents. In contrast to the traditional applications of thermogravimetry in soil science, mass losses in the temperature range below 200 °C were included to take into account the importance of the clay-dependent water content in air-dried soils for the amount of soil organic matter. The evaluation of the dynamics of thermally induced mass loss was less concerned with peak analysis to identify components (easily degradable organic matter, water, mineral decomposition, etc.) than with use-related changes in the relationships between soil components of natural soils, analogous to fingerprinting approaches from, for example, the early days of genetic biology. Accordingly, all our evaluations on mass losses refer to the furnace temperature, while in standard applications, the sample temperature always serves as a reference.

2.3. Sample Preparation and Used Methods

Soil samples were collected from 1993 to now as composite samples from different depths, primarily up to 30 cm, according to soil horizons or the Ap-horizon in agricultural soils with a minimum of 30 individual boreholes. The samples were gently air-dried at room temperature immediately after collection and then sieved to 2 mm. To better compare the samples from different climatic conditions and to take into account the relationship between soil chemical and physical properties as a specific feature of soil formation, the water content of the samples was homogenized. For this purpose, all air-dried samples were stored in thin layers in an atmosphere of 76% relative humidity over saturated NaCl solution for at least two weeks before thermal analysis. This conditioning is intended to reduce the risk of irreversible changes in biological and other properties due to prolonged dehydration.

Thermal analyses were conducted using a Mettler Toledo GmbH thermobalance (Gießen, Germany, TGA 851). Unlike previous applications of thermogravimetry (TG) in soil science (e.g., Ondruch et al., 2019 [35]), a temperature range of 25 °C to 950 °C was used to account for bound water loss, mineral alteration, and thermal transformation of soil organic matter [28,36] as part of the fingerprinting of whole soil samples. All samples were analyzed using large ceramic (Al₂O₃) crucibles with a volume of 0.8 mL and approximately 1 g of the soil samples. Mass loss was recorded every 4 s or per 0.3 K increase in temperature at a heating rate of 5 K/min. During heating, the samples were flushed with an airflow of at least 200 mL/min at 76% relative humidity and 20 °C. Details of the sampling sites, the samples used, and the sample preparation can be found in Siewert (2001) [27], Kučerík et al. (2016) [37], Tokarski et al. (2019) [38], and Tokarski et al. (2020) [39].

Standard methods were used to analyze soil properties, including organic and mineral carbon content, total nitrogen (elemental analyzer), mineralized nitrogen (DIONEX), pH, CEC, and texture (clay content, pipette method). Additionally, the biodegradability of organic matter was measured through soil respiration in laboratory incubation experiments using Nordgren Innovations RESPICOND instruments (versions III, VIII, and X, Bygdeå, Sweden) [27,40,41].

For data evaluation, the recorded mass losses of the thermal analysis were averaged for every 10 °C increase in temperature. During incubation, the hourly values were initially evaluated, and after 1–3 weeks, the daily average values of CO₂ release were evaluated. After various statistical tests for outliers, analytical artifacts, normal distribution (with subsequent normalization if necessary), etc., R was used to search for differences between near-natural and used soils using all data. Correlation and regression analyses using Microsoft Excel (version 2019) and “R statistics” (version 4.1.3) were used to look for relationships between all the data from the methods used. For example, all mass losses from thermal analysis (about 95 datasets for temperatures in 10 °C increments from 30 to 950 °C) were compared with all soil respiration data (about 50 per dataset for experiments over several weeks) for all soil properties studied. Different reduced sample compositions were used to compensate for uncertainties in assessing soil nativeness and to evaluate the influence of land use on the detectability of interrelations. All statistical analyses were repeated with different reduced sample compositions to compensate for uncertainties in assessing soil nativeness and evaluate the influence of land use on the detectability of interrelations.

3. Thermal Decay Dynamics and Detectability of Relationships

Soil mass loss changes dynamically with warming (Figure 3). However, in contrast to various spectroscopic methods, the evaluation of peaks does not usually provide directly usable information about soil properties or components. Only the summation of mass losses over larger temperature ranges indicates dominant features [36,42,43].

Thermal mass losses (TMLs) around 100 °C are often associated with the loss of bound water. With increasing temperature, they are superimposed by mass losses due to the thermal transformation of organic matter [29]. Because mass losses of organic matter dominate only above 200 °C, thermo-analytical data on soil organic matter were only evaluated above this temperature in the past (e.g., Beyer et al. 1998 [44], Palasser et al. 2013 [45]). If air-dried samples are conditioned at 76% relative humidity, the mass losses around 100 °C can be included in the data evaluation for indicator development.

Mass losses above 200 °C usually show individual fluctuations with several shoulders or smooth peaks, mainly caused by the decomposition of organic matter. However, mineral soil components may also be involved. At temperatures above 550 °C, the decomposition of mineral carbon (carbonates) can increase TMLs. Thermally stable organic matter derived from transformation at lower temperatures (e.g., soot, charcoal) or carbon of geological origin (e.g., coal) may superimpose these mass losses [46].

When comparing the average TML dynamics, cultivated soils often exhibit lower mass loss than adjacent natural soils, particularly in the 200–550 °C temperature range. However, these comparisons were difficult because of the preferred use of fertile soils and the lack of sites comparable to natural vegetation. As a result, no generalizable differences between natural and cultivated soils in terms of TML dynamics or peak values could be identified using standard or modern statistical evaluation methods.

A different perspective emerged when evaluating the TML dynamics independently of the peaks in predefined steps of a 10 °C temperature increase, like a fingerprinting approach. Several very close relationships between TMLs and soil properties such as organic carbon, nitrogen, clay, and carbonate content have been demonstrated (Siewert 2001 [27], 2004 [28], examples in Figure 3). Additionally, several TMLs of distant temperature ranges show correlations with each other [27,29] (examples below). Several correlations between soil properties and TMLs and between the TMLs of remote temperature ranges confirm known relationships (e.g., between clay and organic carbon content or between clay and bound water content). However, conditioning samples by storage at 76% relative humidity provided significantly higher accuracy and demonstrated validity independent of the origin of the sample material. Also, using the sample temperature instead of the oven temperature as a reference for evaluating thermal mass loss allowed for the detection of closer relationships.

These relationships were more evident in natural soils with little or no human intervention. In agricultural soils, they were only detectable to a limited extent, mainly in plots of long-term field experiments. In contrast, soil-like carbonaceous mineral mixtures such as composts, horticultural soils, anthrosols, technosols, or urban soils did not show such relationships or only in selected, rare cases [29,47]. This is also applicable to soil samples with added lithogenic (e.g., coal) or anthropogenic carbon (e.g., soot, slag, charcoal), as stated by Tokarski et al. (2018) [48]. No relationships were found in the soil-like carbonaceous samples from arid or cold desert conditions, which are characterized by the dominance of geological processes such as landslide, freeze–thaw cycles, and actively moving sandy dunes over ecosystem succession with continuous vegetation development and soil formation [49,50,51].

Reducing the datasets to samples that were near-natural gave the closest relationships. In contrast, no relationships could be found for the soils far from nature and used for agricultural or urban purposes. Unfortunately, the number of samples that were near-natural was minimal. Therefore, despite higher coefficients of determination, it is impossible to make quantitative statements about the importance of naturalness for most relationships. Including long-term observation plots and forest soils from reserves, soils from highly fertilized areas from long-term agricultural experiments, etc., increased the significance of the relationships. However, the quantitative assessment of the closeness of the relationships fails due to the lack of a reliable quantitative assessment of the closeness to the nature of these sites.

Recent laboratory studies indicate that the relationships between soil components become weaker depending on the type and amount of fresh organic matter added. Long-term microbiological degradation of admixed organic compounds under laboratory conditions strengthens the correlations in proportion to respiration activity [29,48]. Fresh, biodegradable organic matter may, therefore, confound the relationships found. Other possible reasons for deviations from the relationships have been investigated in other studies (examples in the following sections).

4. Example Results for the Development of Soil Health Indicators

4.1. Soil Organic Matter and Clay-Dependent Thermal Mass Losses

An example of the confirmed known relationships between components in natural soils is the importance of clay for stable organic matter fractions. It was first quantified by Körschens et al. (1998) [30] in long-term agricultural field experiments on unfertilized and black fallow plots. This relationship is reflected in our experiments in the close relationship between the clay content and mass losses of thermally stable organic matter (mass losses between 510 °C and 520 °C, TML₅₂₀) [27,28], regardless of the origin of the soil samples.

The closeness of the relationship allows for a theoretical prediction of the clay content in soils to an accuracy of ±5% clay from thermal mass loss. However, another aspect seems to be more critical. Carbon in charcoal, ash, soot, slag, asphalt, biochar, or coal is typically thermally stable, depending on its origin. Adding such organic matter to soil samples increases the TML at high temperatures and, therefore, the TML₅₂₀. Consequently, the measured TML₅₂₀ is higher than expected from the clay content of natural soils. As more material is added, the differences between the actual clay content and that calculated by TML₅₂₀ will increase. Laboratory incubation experiments confirmed the low microbial degradability of thermally stable organic materials of geological (e.g., coal) or anthropogenic (e.g., charcoal) origin. Consequently, the incubation of soil samples under laboratory conditions did not reduce the deviations from the relationship between clay content and TML₅₂₀ [38,48].

The relationship between the clay content and the amount of bound water in soils has also been demonstrated [3]. Using soil samples conditioned at a defined humidity (by storage at 76% relative humidity), the clay content of soils correlates much more closely with TML₁₃₀ (TML between 120 and 130 °C) [27,28,52]. This allows the clay content to be determined by analogy with the relationship between TML₅₂₀ and the clay content (see above). However, fresh organic residues added to soils may also contain bound water unrelated to the soil’s clay content. This results in TML₁₃₀ values higher than expected from the clay content. Since added fresh residues are biodegradable, the associated deviations decrease proportionately to the measured respiratory activity or biological degradation rate [38,39,48]. Fresh, unconverted, biodegradable organic matter, regardless of its heterogeneous chemical composition, can, therefore, be indirectly quantified based on the deviations between the traditionally determined clay content and that derived from TML₁₃₀.

As TML₁₃₀ and TML₅₂₀ depend on the soil’s clay content, both mass losses show a stochastic relationship (Figure 4, modified from Siewert and Kučerík 2015 [29]).

Adding organic matter causes deviations between the calculated and actual clay content for both mass losses. The direction of these deviations depends on the quality of the materials added, while the amount of organic matter determines the magnitude of the deviation. However, variations can be masked and misinterpreted by adding thermally stable and fresh organic matter together (red-bordered arrows in Figure 4). For this reason, thermal analysis cannot replace standard methods in determining the clay content. Their use can only extend the existing standard analytical capabilities to detect non-regulated soil constituents.

4.2. Correlating Thermal Mass Losses with Carbon and Nitrogen Contents

Another example of a known relationship between soil constituents is the C/N ratio in the SOM. This ratio is typically 12–14 for arable land [3]. The C and N contents of soils determined by elemental analysis (EA, dry combustion) correlate with several TMLs in soils. For instance, the organic carbon content is correlated with TML₃₃₀ (320–330 °C), TML₃₄₀ (330–340 °C), TML₃₅₀ (340–350 °C), TML_250–440, and TML_105–550, with decreasing accuracy and precision of the estimate [27,28,53]. Deviations from these relationships can be observed in used soils, particularly in samples with added organic matter or overlapping geological processes, such as erosion, landslides, and water saturation.

Similar to the clay content, the temperature of the affected relationships provides information on the interfering organic compounds. Similar possibilities arise when comparing the nitrogen content determined by EA with TML_300–380, TML_260–440, TML_300–500, TML_320–330, or TML_400–410. Combining all the information from all the relationships provides a wide range of information on deviations from undisturbed soils. The importance of individual land use practices (e.g., fertilization, tillage), climatic and geological influences, and other factors on these deviations have not yet been investigated, nor has their importance for sustainable land use.

4.3. Mass Loss on Ignition and Level of Organic Carbon Sequestration

Long-term agricultural field trials have shown that carbon sequestration has its limits. Once a site-specific equilibrium level is reached, the carbon content does not increase further, even with a continuous high supply of organic matter. The minimum carbon content or lowest level of carbon sequestration caused by decades of black fallow land cannot be reduced further [11,18]. However, quantifying the level of C sequestration in an unknown soil or determining the level of organic matter supply has been challenging until now.

Despite the undeniable disadvantages of the loss on ignition (MLI) method for determining the organic matter in soils, the MLI can be well predicted from the clay and carbon content of the soil. This requires conditioning the samples at a specific, but not mandatory, humidity of 76% [54]. This makes the relationship applicable to soils with a clay content and clay mineral compositions. With a constant heating rate of 5 K/min, the correlation between MLI and the carbon and clay contents becomes closer when the mass losses of the traditional temperature range (105–550 °C) are replaced by mass losses from 110 to 550 °C (TML_110–550). In addition, replacing the carbon and clay contents with TML strengthens the relationship shown in Figure 5.

Deviations between the measured and calculated (from clay and carbon content) MLI were found when plots with different fertilization levels in long-term agricultural experiments were analyzed instead of natural soils. The magnitude of the deviations was related to the long-term level of soil organic matter sequestration, independent of the site conditions of the long-term field experiment [29].

A reduction in soil organic matter content, as measured by biological CO₂ release in laboratory incubation experiments, resulted in anticipated alterations in the deviations of the measured MLI from the calculated values [37]. Conversely, adding fresh organic matter caused opposite changes [52].

5. Discussion

5.1. Importance of Sample Preparation

The better quantification of known relationships between soil components using thermogravimetry was only possible by homogenizing air drying through storage at 76% humidity. Although other moisture values seem applicable [53], we see this as confirmation of the need for a uniform, better-defined, careful preparation of soil samples while avoiding very low air humidity or temperatures above 30 °C.

Bound water may be necessary to define soil health due to plausible relationships with soil organic matter. The ability of organic matter to bind water depends on the type and position of functional groups, which can affect pH and other properties [55,56]. Large, biologically stable organic molecules typical of soil organic matter tend to oxidize on the surface to form acidic functional groups. The associated increase in water binding reduces microbial sensing [57] and promotes biofilm formation as an expression of inhibited degradation of organic matter. Accordingly, uniformly wet conditions should favor soil carbon accumulation, as observed in several studies. In contrast, a single episode of water deprivation could cause adjacent carboxyl groups to release CO₂ during abiotic decay. This changes the polarity of the organic substance, its water binding, pH, etc. The function of biofilms or mucous membranes can be disturbed, and soil surfaces become hydrophobic. Peat mineralization after drainage of acidic, nutrient-rich, oxygen-rich raised bogs, the disappearance of organic layers after logging in spruce plantations, the controversial phenomenon of the “priming” or “Birch” effect [58,59,60], increasing microbiological mineralization of organic matter in soils with lower water-holding capacity, and many other phenomena are in line with these considerations in soil science, but they have not yet been investigated systematically.

From this point of view, a possible functional relevance of bound water cannot be excluded either. The inclusion of collecting factors for carbon sequestration [11,61] seems appropriate. If soil health is to be based on an understanding of soil functions developed with the evolution of life during the history of Earth, more hints can be found.

5.2. Origins of the Relationships Found Between Bound Water and Soil Components

If soil health is to be sustainable, long-term processes in the Earth’s history and the evolution of life must be considered. According to standard theories, primary carbon accumulation by the first organism-like life forms during the history of the Earth was most likely triggered or accompanied by modified water binding on solid surfaces (see mica theory, e.g., in Hansma 2010 [62]). Since Cairns-Smith (1982) [63], similar processes have been discussed as the origin of genetic information in evolving organismal species. Proteins in living organisms today have a structure stabilized by bound water, indicating the importance of bound water for life functions. The known loss of protein function upon denaturation (e.g., heating) highlights the importance of bound water. This supports proposals to use soil proteins as a component of soil health indicators [64]. From the same point of view, the formation of biofilms on mineral surfaces can be understood as an expression of altered water binding with associated changes in the biodegradability of organic matter [65,66]. Similar processes may be reflected in ecosystem succession on larger scales of space and time, associated with carbon accumulation depending on geological parent material and modified water storage with soil genesis over millennia. Interactions between organisms in the biosphere during Earth’s history are thought to have led to increasing CO₂ sequestration in mineral (carbonates) or organic carbon reservoirs (coal, oil, gas), together with changes in atmospheric water content, oxygen concentration, and climate [67,68]. From this perspective, the known soil functions may reflect the same global rules of co-evolutionary changes in the abiotic environment by life, as the properties of solid surfaces are modified by organic matter. However, the logical fit of these considerations to the open questions in Section 5.4 does not indicate a correct understanding of the relationships. Methodological artifacts, stochastic chance, and error cannot be excluded. Experimental proof of the causality of the relationships is therefore required before use.

5.3. Consideration of Naturalness in Soils

The inclusion of semi-natural soils was also a prerequisite for the detection of new relationships between soil components. If soil-specific relationships between thermal mass losses and organic carbon content disappear due to the addition of fresh, biologically degradable substances, organic fertilizers, or carbon from geological (e.g., coal) and anthropogenic sources (soot, slag, biochar) and are re-established by the biological degradation of the added substance, then regulatory processes of the carbon budget in soils could be responsible for these relationships. The degree of deviation from these relationships due to human intervention could then be interpreted as an indicator of more or less effective regulatory mechanisms and describe soil conditions in terms of soil health. This seems to be the case for the deviations in the relationships between clay content and mass loss (see Section 4.1) and many other relationships. Experimentally quantifiable thresholds for soil definition are also conceivable.

The use of deviations from relationships between soil components seems to work independently of climatic conditions (such as temperature and precipitation) and soil type, unlike the site-specific approach of Nunes et al. (2021) [12]. It requires only the addition of thermogravimetric analysis to the usual determination of soil properties, not complex measurements of many physical, chemical, and biological soil properties for an indicator, which becomes more accurate with region-specific scoring [13,14].

5.4. Challenges of Thermogravimetric Soil Analysis

The more precise quantification of the known and the discovery of several new relationships between soil components using thermogravimetric analysis raise many questions. For example, it remains open to why TML₁₃₀ correlates with soil clay content when thermally induced water release is superimposed by organic matter decay processes [29]. The exact correlation between clay and thermal mass losses seems to hold for soils with different clay mineral compositions (see the origin of samples in Figure 1). Carbon-free clay minerals do not show this correlation due to lower and different water binding [69]. The limited ability to quantify SOM content via MLI in clayey soils cannot explain the correct assignment of organic carbon contents in soil from different regions via tiny subsets of SOM reflected in TML at temperature increments of 10 °C. In contrast, similar capabilities do not exist in carbonaceous mineral mixtures, including organic fertilizers, composts, and other substrates. The dependence of MLI on carbon and clay contents in soils [37] does not explain why the degree of carbon sequestration is reflected in differences between calculated and measured MLI. In addition, thermogravimetry is a method that has known drawbacks. Traditional peak interpretation with assignment to organic matter fractions is not meaningful [36]. Conclusions on the biodegradability of SOM based on its thermal stability are unlikely [42,43].

However, these and similar limitations seem less relevant to determining relationships between soil components as practiced here when comparing soils of different origins. This does not exclude the possibility that other methods, such as heat flow measurements, spectral analysis, or other approaches, may be more appropriate. Their use can enhance or replace the options discussed in the previous sections.

5.5. Expected Benefits and Practical Application Perspectives

Despite the complexity of the above considerations, they are consistent with the traditional understanding of clay–humus complexes as a unique feature of soil formation. Organic matter modifications in the surface properties of the mineral soil components, especially clay, over thousands of years of soil genesis can explain most of the relationships found, assuming a functional significance of bound water analogous to the biology of organisms.

This makes a clear distinction between SOM quality and soil health a challenge and offers several possibilities for extension and application. Initial studies indicate possible predictions for nitrate release from biological decomposition processes of organic matter. They document close relationships between thermogravimetric data and soil water-holding capacity, which the functionality of organic matter can explain. The publication of these data is expected after validation and understanding in the context of extreme natural fertility (see Figure 1). We hope there will be economic interest in experimentally determined soil health, with statements about expected yield security, mineral fertilizer, and energy savings, a reduction in the environmental impact of current land use, and other benefits. Increased knowledge of soils with little or no human impact could be as crucial as maintaining higher standards of sample preparation.

Our approach differs from other indicators of soil health in using near-natural soils as a reference, considering water binding as a possible functional basis of soil development control processes, and recording these control mechanisms via proportions between soil components. The thermogravimetric fingerprint has proven effective for this purpose. However, detected changes can only be used for land use management if the causes are understood. A summarizing indicator would have to weigh the deviation from individual correlations. A comparison with other approaches to soil health is difficult, as thermogravimetry, for example, cannot provide indications of the complex interactions of biological, chemical, and physical processes. A combination with other methods for assessing soil health will therefore remain the focus of future studies, also to meet the changing requirements for practical and sustainable soil health management.

6. Conclusions

A modification of the surface properties of mineral and mainly clayey soil components by organic substances over thousands of years of soil formation has not yet been proven. It could explain most of the relationships found with a functional significance of bound water, as in biology. The validation of these relationships offers a starting point for the first experimental definition of internationally comparable, location-independent soil health indicators. Deviations in proportions between soil components provide a starting point for this. This requires the inclusion of near-natural soils and possibly also the consideration of homogenized sample preparation concerning bound water. It remains unclear to what extent thermal analyses can contribute to the experimental definition of soil health or should be replaced by other methods. Also, the necessity of using the furnace temperature instead of the sample temperature as a reference for thermal mass loss relationships has not yet been analyzed. Further experiments are needed to clarify the causality of the relationships found. They must help quantify the degree of deviation from natural status caused by soil use practices, which should be prioritized in assessing soil health.