Soil quality has been defined as “the capacity of a soil to function within ecosystem and land-use boundaries to sustain biological productivity, maintain environmental quality and promote plant and animal health”, as quoted by Bünemann et al. (2018) in a comprehensive review of more than 250 scientific papers covering soil quality. The authors conclude that, in contrast to the quality of water, air and nature, there still is no universally accepted method for measuring soil quality. This is a serious problem, and it limits application in practice and in environmental rules and regulations.

Soil health has been defined in the US as “the continued capacity of the soil to function as a vital living ecosystem that sustains plants, animals and humans”. Indicators for soil health have been defined in the USA, with 19 by Cornell University (Moebius-Clune et al., 2017), 31 by the National Soil Health Institute (http://soilhealthinstitute.org, last access: 17 September 2020; Norris et al., 2020) and 11 by the US Department of Agriculture (USDA, 2019). How these indicators are combined into a single soil health parameter for a given soil is presented by the Cornell protocol. Only three texture classes of soils are distinguished, namely coarse, medium and fine. For each texture class, measurements for each indicator are assembled for soils at different locations in that particular texture class, and a frequency curve of values is constructed. Obviously, such curves become more diagnostic as more data become available. When placed on the frequency curve, any new observation of the indicator will obtain a number between 0 and 100. This procedure is repeated for every indicator, and in the end, all numbers will be averaged to produce one characteristic number for soil health for that particular soil, which is quite attractive for communication purposes. The frequency curve also allows the distinction of a threshold frequency value above which the particular indicator exceeds a critical environmental threshold value, which is sometimes defined by environmental laws and regulations. In their reporting, red, orange, yellow and green labels are used to indicate whether or not this occurs. A red label indicates that a given threshold is exceeded, and that action is needed, possibly to be based on favourable management experiences obtained elsewhere in soils of the same texture class or by new research. This is attractive because it can directly result in management advice. In an example presented by Moebius-Clune et al. (2017, p. 73), values for 12 indicators are presented, three of which with have red labels for “surface hardness”, “aggregate stability” and “active carbon content”, suggesting a need for corrective measures. But what does this imply for soil health? Does this mean a soil is unhealthy only if one or more indicators are red? And how does one interpret an average value for all 12 – quite different – indicators with different colours?

Also, a question can be raised about the large number of indicators for soil health in the three US systems. Why not primarily consider demands by roots as they link plants with the soil? A number of conditions do not allow root growth, for example, the presence of excessive amounts of chemical pollutants, salty soils (solonchack), alkaline soils (solonetz) and very acid soils with low pH values. Soils with such properties are clearly unhealthy. Otherwise, roots require (i) temperatures that allow growth, (ii) soil structure that allows easy accessibility of the entire soil volume, allowing roots to reach their genetically determined depth, (iii) adequate water, air and nutrient availability during the growing season, (iv) adequate infiltration rates of water at the soil surface, and (v) adequate organic matter content and the associated biological activity that is essential for many soil functions, including nutrient uptake by plants. These five parameters can be measured at a given time and place, and the reports by Moebius Clune (2017) and USDA (2019) contain detailed descriptions of measurement methods.

Parameters to be measured at a given point in time should have a semi-permanent character to be diagnostic. Temperature and nutrient status are quite variable, with the latter being high at the moment of fertilisation and increasingly lower as the crop adsorbs nutrients. Of course, this is different in areas where inherent nutrient contents are important to allow particular types of vegetation to develop. However, nutrient deficiencies in agricultural soils can be rapidly corrected by fertilisation, and the nutrient status, though essential for root growth, is therefore less suitable as a parameter in agricultural soils. Soil structure, excluding a limited period after soil tillage, is more permanent and governs infiltration rates and soil water and air regimes as a function of weather conditions and groundwater dynamics. Soil structure is therefore suitable as a parameter. Aggregate stability is a measure for soil resistance to deformation, but the method has been criticised as being unrepresentative (e.g. Baveye, 2020). The use of penetrometers may be more effective for measuring mechanical resistance affecting root penetration. Biological activity is subject to an even longer time span than compaction; increasing the organic matter content of soils may take several years. The organic matter content is, therefore, a suitable parameter, and many measurement methods are available, including rapid methods applying proximal sensors (e.g. Priori et al., 2016; Duda et al., 2017). More detailed measurements of biodiversity have been defined by Moebius-Clune (2017) and for the Land Use/Cover Area frame statistical Survey (LUCAS) soil database (Orgiazzi et al., 2018), requiring laboratory measurements.

In conclusion, parameters for soil health for a given soil type at a given time and place, are (i) soil structure, expressed by descriptions in soil survey reports and supported by bulk density values and measured infiltration rates and, possibly, by penetrometer values; (ii) water and air regimes, as estimated by drainage class in soil survey reports, that can be expressed indirectly by the widely used, but static, parameter, namely “available water”, for defining the water content between two pressure heads, which, however, poorly represent natural dynamic soil water and air regimes (dynamic modelling presents more realistic data as will be discussed later; e.g. Bouma, 2018; Bonfante et al., 2019); and (iii) organic matter contents.

Nevertheless, the procedure based on the three parameters mentioned above produces three separate values. Back, therefore, to the definition of soil health that mentions the “functioning of soils”, whereby soil contributions to biomass production are a key function, among six other defined functions (EC, 2006). The degree to which biomass production is affected by the three separate parameters remains unclear. An integrated approach is therefore needed and can be obtained by simulating the soil–water–atmosphere–plant system.

The soil health concept offers one basic problem. A sandy soil and a clay soil can both be healthy, but they obviously have quite different water and nutrient regimes and use potentials. Such differences among soils can be expressed by the soil quality concept when considering the inherent properties of soils as expressed in soil classification by defining soil types (soil series at the most detailed level in the USA). In fact, Moebius-Clune et al. (2017) express and classify soil health for three texture classes and, in so doing, express the effects of inherent soil properties in a very general manner that does not reflect soil properties, as defined in soil classification, that are likely to strongly affect soil behaviour. Their procedures for defining soil health are different for each texture class as they define three different frequency curves.

In the analogy with human health, soil health for a given soil at a given time expresses the actual condition expressed by the parameters discussed above, just like a doctor assesses the health of a patient at a given time after conducting a set of tests. We propose that the soil health concept is determined in the same way for all soils, emphasising its specific identity at a given location and point in time. Next, soil quality expresses the fact that different health values can be found in the same soil type as a function of past management, leading to, for example, compaction, organic matter depletion, soil crusting followed by runoff, erosion, etc., as illustrated in the Italian case study presented below. However, the range of such soil health values is characteristically different for every soil type and can, therefore, function as a measure of soil quality for that particular soil type. Droogers and Bouma (1997) have distinguished genoforms, expressing a given soil classification, but also phenoforms of that particular genoform as a function of different forms of management, with strong effects on soil functioning (e.g. organic matter depletion, erosion, compaction, crust formation, etc.). Each phenoform can be characterised with a soil health value, as shown in the Italian case study below. Traditional soil survey interpretations are based on so-called “representative profiles” for each mapping unit on the soil map, based on permanent taxonomic soil criteria, correctly ignoring, in the context of soil classification, the effects of management which would lead to highly variable classifications. But different phenoforms of a given genoform can, however, function quite differently, and this cannot be ignored when considering soil health. Just considering a soil type, as such, in terms of a representative profile is inadequate for reflecting soil behaviour that determines soil health.

Application of simulation models of the soil–water–atmosphere–plant (SWAP) system can integrate the values of the parameters mentioned above as they function as input data for the model, producing a single integrated value for biomass production. Many operational models are available (e.g. Reynolds et al., 2018; SWAP by Kroes et al., 2017; SWAP-WOFOST by Hack-ten Broeke et al., 2019; ICASA by White et al., 2013; the Agricultural Production Systems sIMulator (APSIM) by Holzworth et al., 2018; and Ma et al., 2012, and others). These models use rooting depth, weather data and, when the required hydraulic conductivity and moisture retention data are not available, these values can be estimated with pedotransfer functions using texture (as defined by the soil type), organic matter and bulk density as input data, defining the soil health parameters identified above (Bouma, 1989; Van Looy et al., 2017). So, rather than having sets of separate parameters for soil health, an integrated expression is obtained by the model that directly addresses a key soil function, which is its contribution to the ecosystem service of biomass production. The term “contribution” needs to be emphasised as “biomass production” is not determined by soils alone but by many other factors and, certainly, by management. Applying modelling, an alternative procedure for defining soil health was proposed by Bonfante et al. (2019) in which biomass production forms the starting point. Following the agronomic yield gap programme (van Ittersum et al., 2013), yields are calculated by simulation models of the soil–water–atmosphere–plant system, i.e. Yp = potential production, determined for a representative crop considering radiation and temperature regimes in a given climate region, assuming that adequate water and nutrients are available and pests and diseases do not occur. This is a science-based value that applies everywhere on Earth and yields unique, quantitative and reproducible data. Yw is the water-limited yield, as is Yp, but expressing the effect of the actual soil water regime under local conditions, and Ya is the actual yield. The yield gap is Yw–Ya. These parameters of the yield gap programme can be applied to define soil health and soil quality parameters (to be discussed in the next section) but need to be modified to express the specific impact of the soil.

Simulation modelling offers the possibility of expressing soil functioning, as mentioned in the definition of soil health, as an interdisciplinary modelling effort with input by agronomists, hydrologists and climatologists, who each provide basic data for the models. This yields one number, based on an interdisciplinary analysis, which is preferable to a series of separate numbers for soil parameters only as in the US systems. The soil science discipline presents the parameters, mentioned above, to the interdisciplinary research team in the context of a well-defined soil type that defines moisture regimes and rooting patterns. In this way, the soil type functions as a “carrier of information” or a “class–pedotransfer function” (Bouma, 1989).

Moreover, and more importantly, modelling is the only option for exploring the possible future effects of climate change on soil health and soil quality, as will be demonstrated below. Procedures for defining single soil health and soil quality parameters will be presented in the materials and methods section of the paper.

The discussion of soil health and soil quality so far focused on the soil and the way it functions, mentioning goals such as “biological productivity and environmental quality” (soil quality) and “vital soils that sustain plants, animals and humans” (soil health). As mentioned in the introduction, since 2015 a total of 193 countries have made a UN-initiated commitment to reach the 17 Sustainable Development Goals (SDGs). The European Union launched its Green Deal in 2019. The soil quality and soil health concepts are not meaningful goals by themselves and can obtain societal significance when linked to the SDGs and the EU Green Deal. But there is no direct link, if only because soil management plays a key role in achieving the SDGs and the goals of the EU Green Deal. The challenge for soil science is to explore ways in which healthy soils can contribute to improving a number of key ecosystem services that, in turn, contribute to the SDGs (e.g. Bouma, 2014; Keesstra, 2016). This is important because SDGs and the goals of the EU Green Deal are not only determined by ecosystem services but also by, for example, socioeconomic and political factors that are beyond the control of sciences studying crop growth. Attention to the SDGs and the EU Green Deal implies attention to not only biomass production (SDG 2 – zero hunger) but also to other ecosystem services that relate directly to environmental quality, such as the quality of ground and surface water (SDG 6 – clean water and sanitation), carbon sequestration and reduction of greenhouse gas emissions for climate mitigation (SDG 13 – climate action) and biodiversity preservation (SDG 15 – life on land). That is why the following definitions of soil health and soil quality are proposed:

Soil health is the actual capacity of a particular soil to function, contributing to ecosystem services.

The four ecosystem services mentioned above have a different character. Biomass production (SDG 2) is governed by climatic conditions and soil water regimes, as characterised by modelling that yields quantitative and reproducible results for Yp and Yw. Management plays a key role in determining Ya, and the other ecosystem services, and is characteristically different for different soil types. Clean water (SDG 6) can, for example, be obtained by precision fertilisation, minimising nutrient leaching to the groundwater, while combatting erosion can minimise surface water pollution. But there are, in contrast to Yp or Yw values for biomass production, no theoretical reference values for this ecosystem service – only threshold values of water quality by environmental laws and regulations. This also applies to carbon sequestration and the reduction of greenhouse gas emissions (SDG 13) and to life on land (SDG 15) for which, as yet, no environmental laws have been introduced. Different soils in different climate zones will offer different challenges and opportunities to be met by appropriate management.

The soil–water–atmosphere–plant (SWAP) model (Kroes et al., 2017) was applied to solve the soil water balance during maize cultivation under estimated climate change and soil percentage of soil organic matter (SOM) scenarios of Ap horizons. SWAP is an integrated, physically based simulation model of water transport in the saturated–unsaturated zone in relation to crop growth. It assumes unidimensional vertical flow processes and calculates the soil water flow through the Richards equation. Soil water retention θ(h) and hydraulic conductivity k(θ) relationships, as proposed by Van Genuchten (1980), were applied. The unit gradient was set as the condition at the bottom boundary. The upper boundary conditions of SWAP in agricultural crops are generally described by the potential evapotranspiration (ETp), irrigation and daily precipitation. Potential evapotranspiration was then partitioned into potential evaporation and potential transpiration according to the leaf area index (LAI) evolution, following the approach of Ritchie (1972). The water uptake and actual transpiration were modelled according to Feddes et al. (1978), where the actual transpiration declines from its potential value through the parameter α, varying between 0 and 1 according to the soil water potential.

Application of the soil–water–atmosphere–plant simulation model and the yield gap parameters results in the following four characteristics: i.

A measure for actual soil health of a given soil type in a given climate zone at a given time by the following soil health (SH) index:1SH=(Yw-phenoform/Yw-ref)×100,where Yw-phenoform expresses Yw for a given phenoform and Yw-ref represents the undisturbed soil phenoform. This index expresses the effect of the soil on the measured yield Ya, a value that is affected by many other factors than the soil.

Six prominent Italian soil series were analysed to illustrate the proposed method for defining soil health and soil quality. Because of space constraints, the results of three soils will be discussed in this paper. The modelling process and the background of the IPCC scenarios have been presented elsewhere (Bonfante et al., 2019, 2020; Bonfante and Bouma, 2015) and will be summarised below.

The maize was simulated from May (emergence) to the end of August (harvest), with a peak of leaf area index (LAI) of 5.8 m2 m-2. Finally, the above ground biomass (AGB) for determining the yield values (Yw) was estimated using the normalised water productivity (WP) concept (33 g m-2 for maize; Steduto et al., 2012).

The simulation runs were performed for six selected soils using a future climate scenario of a site in southern Italy (Destra Sele plain) where half of the analysed soils occur. The future climate scenarios were obtained by using the high-resolution regional climate model (RCM) consortium for small-scale modelling coupled with climate limited area modelling (COSMO–CLM; Rockel et al., 2008), with a configuration employing a spatial resolution of 0.0715∘ (about 8 km), which was optimised over the Italian area. The validations performed showed that model data agree closely with different regional high-resolution observational data sets, in terms of both average temperature and precipitation (Bucchignani et al., 2015) and in terms of extreme events (Zollo et al., 2015).

The severe Representative Concentration Pathway (RCP) 8.5 scenario was applied, based on the IPCC modelling approach, to generate greenhouse gas concentrations (Meinshausen et al., 2011).

The results were performed on reference climate (RC; 1971–2005) and RCP 8.5, with the latter divided into three different time periods (2010–2040, 2040–2070 and 2070–2100). Daily reference evapotranspiration (ET0) was evaluated according to the Hargreaves and Samani (1985) equation.

Under the RCP 8.5 scenario, the temperature in Destra Sele is expected to increase by approximately 2 ∘C, respectively, every 30 years to 2100, starting from the RC. The differences in temperature between RC and the period of 2070–2100 showed an average increase in the minimum and maximum temperatures of about 6.2 ∘C (for both minimum and maximum temperatures over the year). The projected increase in temperatures produces an increase in the expected ET0. In particular, during the maize growing season, an average increase in ET0 of about 18 % is expected until 2100 (Bonfante et al., 2020).

Simulations were run considering an undisturbed soil (the reference) and three phenoforms, with two expressing degradation phenomena (erosion and compacted plough pan) and one considering an increase in the percentage of OM in the first soil horizon (Ap) as a possible result of combatting a low percentage of OM due to soil degradation.

In particular, we considered the following: i.

The compacted plough layer was applied at 30 cm depth (10 cm thickness) with the following physical characteristics: θ s = 0.30 cm3 cm-3, n=1.12, α=0.004 and k0=2 cm d-1, following the notation of Van Genuchten (1980). Roots were restricted to the upper 30 cm of the soil.

Water-limited yields (Yw) for four climate periods and three phenoforms for each soil are shown in Fig. 1a for P4, Fig. 1b for P5 and Fig. 1c for P6. Yw values drop for all soils and their phenoforms in the period from the RC to the 2070–2100 climate scenario, particularly for climate scenarios beyond 2040, but, due to relatively high standard deviations, not all differences are significant. However, each soil shows significant drops in Yw for the erosion and plough pan phenoforms, again particularly beyond 2040, when comparing values with Yw undisturbed. Soils P4 and P5 show rather identical behaviour, but soil P6 has significantly higher values for Yw for the erosion and plough pan phenoforms beyond 2040. An increase in percentage OM has a minimal effect, as explained by Bonfante et al. (2020), when considering hydraulic conductivity and moisture retention data.

The soil health (SH) index applies to soil health parameter measurements for a given soil at a given time, defining actual conditions, with reference to the particular production potential of the soil type that is present, as expressed by Yw calculated with optimal soil parameters as discussed above. Yw-phenoform conveys conditions expressed by the three soil parameters observed at the site. When Yw-phenoform is equal to Yw, the soil health value will be 100, but this is highly improbable. Lower values indicate room for improvement but offer no information as to factors that lead to these low values (see the next section). Calculated SH indexes for three Italian soil series in four climate periods are reported in Table 2. In this study, four soil conditions were simulated that are common in the field and four climate periods were considered, namely a non-degraded soil characterised by optimal soil parameters (producing Yw-ref) and two Yw-phenoform values, i.e. erosion of topsoil, formation of a plough pan and an increase to 4 % OM. As actual conditions are discussed here, the current climate of 2010–2040 should be considered. Erosion reduces SH to approximately 88, while the plough pan has much stronger effect, with significantly different values of 55 (P4), 66 (P5) and 75 (P6). Increasing the percentage of OM does not deviate from the value of 100, which corresponds with data reported in Fig. 1a–c.

To determine the health index at a given time and place in a given soil, the three soil parameters discussed above are measured, and the model is used to calculate a (Yw-phenoform) value that is then compared with the Yw-ref value calculated with optimal soil parameter values for that particular soil. Management practices that have resulted in the Yw-phenoform being considered should be documented.

The SH index, mentioned in the previous section, characterises soil health at a given time and location, as measured in a particular soil type. A gap may become obvious between Yw-phenoform and Yw-ref, but it is not clear what can be done to close the gap. Soil health values for a given soil series can also be obtained at different locations in the same climate zone in which different forms of management have resulted in different phenoforms representing a characteristic range of values that can be seen as a measure for inherent soil quality (SQp). Figure 2 shows a range of values obtained for a given soil type assuming, in this case, the occurrence of only three phenoforms. This only illustrates a principle, and many observations in the field can and should extend the number of points for Yw-phenoform. This range offers a point of reference for each observation, as discussed in the previous section, and allows conclusions regarding advisable management procedures associated with the different phenoforms that, together, determine the observed ranges in Fig. 2.

Figure 2 shows a decreasing sensitivity to soil degradation moving from soil P4 to soil P6. Soil health ratios change from 56 (P4) and 66 (P5) to 78 (P6). The effects of climate change on the index are, again, strongest for soil P4. Figure 2 shows that not only are the ranges of the health index significantly different for the three soils but also their resilience to climate change. A particular soil health measurement in a given soil, as described in the previous section, can now be placed into the bars shown in Fig. 2, indicating possible room for improvement. As every measurement is combined with an assessment of soil use and management that has resulted in the particular phenoform being observed, the system allows the generation of useful management information for the land user.

So far, particular soil types have been considered. The analysis can be extended to all soils in a given region and climate zone, and this comparison of different soils can be valuable for regional land use planning. This requires the definition of Yp for the area that is used for the simulations. For the Italian soils being considered, Yp = 18 tons ha-1, and this value is maintained for all climate scenarios considered, implicitly assuming that other factors affecting biomass production will not change. Table 3 shows significant differences among the soils, providing a valuable quantitative assessment. Differences are maintained when different climate periods are considered. Soil P4 scores the lowest values again, with soil P5 scoring intermediate values and soil P6 with the highest values, but even this soil has a low score of 50 for the last climate period when a plough pan is present.

Questions about potential food production in future, considering the effects of climate change, require a mechanism for comparing different soils in the world in their capacity to produce biomass. Assuming a maximum production to be achieved in the world (Ymax), considering theoretical photosynthesis under particular climate conditions, values of Yp and Yw can be expressed as a function of Ymax. Use of Yw will produce the most realistic values in view of the limited water availability in many areas of the world. Areas with relatively high values have a higher potential than areas with low values, and this analysis can be helpful input from soil science, contributing to global food production scenarios. Based on current evaluations, a Ymax of 20 tons ha-1 is used here as a reference, and this results in SQw values that can also be expressed for various phenoforms, showing the effects of different forms of degradation Table 4. As in Table 3, differences between the three soils are significant. How these values are to be judged will depend on comparable values assembled for other areas of the world.