Main authors: | Luis Garrote, David Santillán, Ana Iglesias |
iSQAPERiS Editor: | Jane Brandt |
Source document: | Garrote L., Santillán D., Iglesias A. (2018) Report on key management practices affecting soil quality and their applicability in various farming systems. iSQAPER Project Deliverable 7.2 140 pp |
1. Soil quality indicators and ecosystem services considered
One of the key aspects of the iSQAPER model is the identification and quantification of the functional relations established between soil threats, agricultural management practices, soil quality indicators, soil functions and soil ecosystem services. These relationships may be very complex, as recognized by Bünemann et al. (2018) in their review of soil quality, conducted in the framework of iSQAPER project. Every soil threat affects several soil functions, which are in turn linked to several ecosystem services. This complexity needs to be simplified for establishing the effect of agricultural management practices in different farming systems at the continental scale.
The approach followed in the upscaling model has been to include functional relations that are essential to establish the linkage between policy choices and soil ecosystem services and that can be supported by the science developed within the iSQAPER project. Scientific analyses are based on the input provided by the case study sites and on the elaboration of data collected in the LTEs over a long period of time. These studies focused mainly on the relationship between agricultural management practices and soil quality indicators. These functional relations are based on science and are readily available within the project, and therefore they have been selected to build the upscaling model.
The nature of the functional relations is outlined in Figure 4. For each farming system, the long-term evolution of soil quality indicators in determined by local conditions and the management practices adopted for farming. As shown in the LTEs, sustained application of beneficial practices has a measurable impact on soil quality indicators that may be quantified, at least to a first approximation.
Figure 4
2. Linking scientific results at the local level to the evaluation at the continental scale
The selected indicators that represent the effect of agricultural management practices in soil ecosystem services were detailed in »Effect of farming on soil quality. To quantify their effect, our approach is to upscale the scientific results at the local level to the evaluation at the continental scale.
Bai et al. (2018) (»Management practices and soil quality) evaluated paired response ratios for different soil quality indicators and management practices. In total, 354 paired observations were analysed. The table shown in Figure 5, taken from Bai et al. (2018), summarizes the results, showing the number of pairs compiled and relevant parameters of the distribution (mean, median, standard deviation, skewness, quartiles, maximum and minimum. The indicators with more data are Yield and SOM, while Earthworms shows the least data availability.
Table 1 and Figure 5 of Bai et al. (2018), showing descriptive statistics for impact of selected management practices on specific soil quality indicators (response ratios, dimensionless).
Table 1. Descriptive statistics for impact of sectoral management practices on soil quality (Bai et al., 2018)
The results were also graphically represented as “flower petals”, as shown in Figure 5. The median impact for each management practice is represented in polar coordinates. Values greater than one indicate positive effects, with a colour code to identify the intensity of the impact: orange, median ≤1; light green, 1 < median<1.5; and dark green, median>1.5.
Figure 5
Results obtained in the LTE sites are the core of the functional relations proposed in the upscaling model. They provide a solid description of the long-term influence of agricultural management practices on soil quality indicators, based on a large number of experimental measurements. They also analysed other factors, like the expected dispersion of results for various local conditions, that are relevant for implementing the upscaling model.
However, the upscaling model requires further information to account for different farming systems or agro-climatic zones. This additional information was derived from iSQAPER case study sites through personal interviews during the Second General Assembly held in Tartu in June 2018 and through a questionnaire that was distributed to iSQAPER partners. The objective of the interaction with case studies was twofold: validation of the general framework of the upscaling model and identification of singularities for each farming system under the local conditions of each case study site. The initial input provided during the interviews was incorporated to the upscaling functional relations, which were later validated through the questionnaire.
3. Deriving functional relations for ecosystem services
Functions that relate agricultural management practices and soil quality indicators are defined from the results compiled for the LTE sites. We start from the reference values published in Bai et al., 2018 and adapt them to different farming systems accounting for the variability of local conditions.
The results of the LTE sites for the soil quality indices selected for upscaling are summarized in Figure 6. It shows the “flower graphs” for Yield, Water Holding Capacity and Soil Organic Matter. Each graph includes the effect of the five agricultural measures under analysis. In the case of Yield, three measures produce positive effects (organic matter addition, crop rotation and irrigation) and two measures produce negative impact (no tillage and organic farming). For the two other indices, all measures produce positive effects. The most relevant effect is no tillage, which produces a response ratio of 1.46. These values are taken as reference conditions, accounting for the variability of effects.
Figure 6
Table 2, derived from Table 1 of Bai et al. (2018) (»Management practices and soil quality), shows the median values of the response ratios, together with the standard deviation of the results obtained for the Relevant combinations of soil quality indicators and agricultural management practices. Most of the results show large uncertainty, represented by high values of the standard deviation.
Table 2. Relevant results (response ratios) of Long Term Experiment sites
Yield | Earthworms | Soil Organic Matter | ||||
Median | SD | Median | SD | Median | SD | |
Organic matter | 1.37 | 1.19 | 1.69 | 1.67 | 1.29 | 0.33 |
No tillage | 0.98 | 0.12 | 1.53 | 0.62 | 1.20 | 0.69 |
Crop rotation | 1.17 | 0.40 | 1.73 | 1.55 | 1.25 | 0.61 |
Organic farming | 0.89 | 0.30 | 1.93 | 0.37 | 1.12 | 0.56 |
4. Integrating knowledge from scientists’ stakeholders
Functional relations are formulated in qualitative terms. The objective is to account for the positive or negative effects of management practices on soil quality indicators linked to soil ecosystem services and thus assess the projected impact of alternative policies in future scenarios. The proposed qualitative domain was defined in Deliverable 7.1 and was inspired on the Likert scale. Likert scaling is a bipolar scaling method, measuring either positive or negative response to a statement. Experts were asked to fill a questionnaire about the impact of management practices on soil quality for the farming systems available at their case study site. Based on their responses and on the analyses carried out in WP3, the effect of the management practice for every farming system was classified in the following categories:
- Positive (++): This category means that the management practice applied to the local farming system will certainly improve the soil quality indicator with respect to the reference value obtained in the LTE, with effects larger than 10%. We adopt 12.5%.
- Beneficial (+): This category means that the management practice applied to the local farming system has potential to improve the soil quality indicator with respect to the reference value obtained in the LTE, but the effects may depend on additional factors. The improvement will be between 5% and 10%. We adopt 7.5%.
- Neutral (=): This category represents a neutral impact of the management practice applied to the local farming system on the soil quality indicator under analysis with respect to the reference value obtained in the LTE. It corresponds to a positive or negative effect of less than 5%. We adopt that the practice has no effect.
- Unfavourable (-): This category means that the management practice applied to the local farming system may degrade the soil quality indicator with respect to the reference value obtained in the LTE, but the effects may depend on additional factors. The degradation will be between 5% and 10%. We adopt -7.5%.
- Negative (--): This category means that the management practice applied to the local farming system will certainly degrade the soil quality indicator with respect to the reference value obtained in the LTE, with effects larger than 10%. We adopt -12.5%.
The resulting values of the application of management practices to farming systems are presented in the following tables.
Table 3. Effect of agricultural management practices on crop yield
Organic matter | No tillage | Crop rotation | Organic farming | |
Cereals | = | + | ++ | + |
Rice | = | n.a. | n.a. | + |
Maize | = | = | + | = |
Soybean | = | = | + | = |
Vegetables | + | = | = | + |
Pasture | + | + | n.a. | + |
Permanent crops | + | + | n.a. | + |
Mean | 1.67 | 0.99 | 1.31 | 0.96 |
Median | 1.37 | 0.98 | 1.17 | 0.89 |
St. Dev | 1.19 | 0.12 | 0.40 | 0.30 |
Mean, Median and St. Dev values taken from »Management practices and soil quality
Table 4. Effect of agricultural management practices on soil organic matter
Organic matter | No tillage | Crop rotation | Organic farming | |
Cereals | = | ++ | ++ | + |
Rice | = | n.a. | n.a. | + |
Maize | = | + | ++ | = |
Soybean | = | + | + | = |
Vegetables | + | + | = | + |
Pasture | ++ | ++ | n.a. | + |
Permanent crops | = | = | n.a. | + |
Mean | 1.39 | 1.46 | 1.41 | 1.31 |
Median | 1.29 | 1.20 | 1.25 | 1.12 |
St. Dev | 0.33 | 0.69 | 0.61 | 0.56 |
Mean, Median and St. Dev values taken from »Management practices and soil quality
Table 5. Effect of agricultural management practices on earthworms
Organic matter | No tillage | Crop rotation | Organic farming | |
Cereals | = | + | ++ | + |
Rice | = | n.a. | n.a. | + |
Maize | = | = | + | = |
Soybean | = | = | + | = |
Vegetables | + | = | = | + |
Pasture | + | + | n.a. | + |
Permanent crops | + | + | n.a. | + |
Mean | 2.45 | 1.53 | 1.73 | 1.75 |
Median | 1.69 | 1.53 | 1.73 | 1.93 |
St. Dev | 1.67 | 0.62 | 1.55 | 0.37 |
Mean, Median and St. Dev values taken from »Management practices and soil quality
5. Spatial analysis
The objective of the upscaling model is to produce maps of improvement of soil environmental footprint under different policy scenarios. Therefore, the model needs to account for spatially-explicit representation of soil processes. The foundation of the spatial representation is the data catalogue introduced and discussed in »Effect of farming on soil quality. The data catalogue is a compilation of variables under a unified structure and spatial resolution: a gridded data structure of 0.05 spatial degrees’ resolution. Sources of information are heterogeneous, including a diversity of variables with different resolutions. The data catalogue finally selected is summarized in Figure 7. Variables are clustered in tables according to the Local Conditions, the Farming Systems, the Management Practices and the Soil Quality Indicators.
Figure 7
6. Drivers of change
The science developed in the iSQAPER project provides the basic building blocks for upscaling, but the intensity of the actions is determined by external factors that should be evaluated separately. These drivers of change will determine the extent of adoption of beneficial management practices and the corresponding improvement of soil quality indicators. Drivers of change may be natural, such as climate change, or man-made, such as socioeconomic development or policy priorities.
The upscaling model will identify a reduced set of scenarios where these drivers of change will be identified and characterized. The identification will include a diversity of factors, ranging from climatic factors or population dynamics to policy formulation. As a result of the analysis, a policy portfolio will be defined, consisting on a spatial representation of the degree of adoption of the different management practices considered (see »Policies and environmental footprint).
7. Spatio-temporal analysis
The unit of computation is the model cell, which corresponds to a spatial resolution of 0.5 minutes (approximately 9 km at the Equator). Information about the grid cell includes the climate zone, the soil type, the cropping patterns within the cell (there may be several), the soil status described by the available soil quality indicators and the current degree of implementation of each category of agricultural management practice in the region. The scenario determines the additional degree of implementation of each agricultural management practice to be achieved in the time frame of the analysis. Upscaling functional relations are applied to appropriate grid cells where each agricultural management practice is considered to be implemented. This leads to a modification of the soil quality indicators, which is the initial output of the upscaling model.
The basic approach of the spatial analysis is illustrated on Figure 8. The domain is divided in grid cells with the same resolution as the data catalogue: 5 minutes. The model accounts for the current local values of the soil quality indicators and the current degree of implementation of management practices, if available. The external forcing is described through a scenario of policy drivers, that determine an additional implementation of certain management practices. Model inference estimates the changes in soil quality indicators as a result of the policy drivers.
Figure 8
A differential response is expected as a result of local conditions: farming system, current values of soil quality indicators and current degree of implementation influence the extent to which the soil react changes in management practices. These local effects are analysed in detail in the following section.
8. Quantification of soil ecosystem services in each point
In order to estimate the effect of management practices on soil quality indicators, it is essential to account for values of each point in the coarse-scale geographical analysis. The basic rationale of the upscaling model is that the influence of soil management practices will be larger on areas with relatively lower values of soil quality indicators. Assuming that the rest of conditions are equal, the fact that a local point shows a low value of the soil quality indicators may be explained by poorer soil management practices.
Local conditions were established based on the variable considered most relevant for each soil quality indicator. Yield was linked to climate zone, soil biomass was linked to biome and soil organic carbon was linked to soil type.
The local variable selected for Yield is climate zone, taken from the Köppen-Geiger climate classification system. The basic variable for zonation is the World Map of Köppen-Geiger Climate Classification distributed by the University of Vienna (Rubel and Kottek, 2010). Local yield for a certain farming system is compared to the distribution of yields for the same farming system obtained from all cells in the same climatic zone. Figures 9 and 10 show the climate zones for Europe and China and the yield distribution for the “Rice” farming system in the Arid (B) climatic zone in China.
Figure 9
Figure 10
The local variable selected for Soil Organic Carbon is soil type. The basic variable for zonation is the Digital Soil Map of the World distributed by FAO (Version 3.6, completed January 2003). Local Soil Organic Carbon for a certain model cell is compared to the distribution of SOC obtained from all cells of the same soil type. Figure 14 shows the soil types for Europe and China and the SOC distribution for the Podzols (P) soil type in Europe.
In order to account for local conditions, soil quality indices are re-scaled to standardized variables that compare local values to conditions for the same local group. The “Standardized Soil Quality Index” is defined applying the following equation:
Where SSQI is the standardized soil quality index for a certain local group (for instance, cereal yield in Arid (B) climate); μ is the average value of the soil quality index in all cells in the same local group and σ is the standard deviation of the soil quality index values of all cells in the same local group.
The following categories may be defined according to the values of the standardized soil quality index:
- Very small: The standardized soil quality index is less than -1.5
- Small: The standardized soil quality index is larger than -1.5, but less than -0.5
- Average: The standardized soil quality index is larger than -0.5, but less than 0.5
- Large: The standardized soil quality index is larger than 0.5, but less than 1.5
- Very large: The standardized soil quality index is larger than 1.5
Figure 11 illustrates this classification for the local group of rice yield in arid (B) climate in China.
Figure 11
The response of soil quality indicators to the susained application of the management practice is based on the conclusions of the analysis of the LTE sites. The main value is the response ratio, RR, defined as:
Where SQI0 is the value of the soil quality indicator in the reference condition and SQIMP is the value of the soil quality indicator after the application of the management practice. The results of the long term experiments show that there is a significat uncertainty in the response ratios observed in different locations. The distributions of the response ratios were characterized in Table 5 through their median values and their standard deviation. These two values are taken as input for the local influence models. Local conditions are accounted through the standardized soil quality indicator.
The local influence model determines the response ratio for the individual cell as a function of the standardized soil quality indicator. The effect of the measure is considered to be larger or smaller values of the standardized soil quality indicator, according to the following function definition.
- If the standard soil quality indicator is smaller than -1, the response ratio is considered to be equal to the Median value plus the Standard Deviation of the distribution.
- If the standard soil quality indicator is greater than 1, the response ratio is considered to be equal to the Median value minus the Standard Deviation of the distribution.
- If the standard soil quality indicator is greater than -1 and smaller than 1, the response ratio is computed with the following equation:
where RR is the response ratio, SSQI is the standardized soil quality indicator and SD is the standard deviation of the distribution.
The response function is illustrated in Figure 12.
Figure 12
Note: For full references to papers quoted in this article see