In this paper, we used data about arable land yield potential, arable land rent proportion and arable land rent price as indicators for soil quality and soil governance through tenancy agreements to test the abovementioned theoretical assumptions. We used the yield potential estimates for arable land in Germany of the German Federal Institute for Geosciences and Natural Resources [24
]. They are based on the Müncheberger Soil Quality Rating (MSQR), which is a visual procedure for yield potential estimation, taking soil structure and soil degradation threats into account [4
]. It integrates eight basic soil indicators with 13 hazard indicators into a rating of soil quality on an ordinal scale of 0 to 102, with higher values indicating higher yield potential. The eight soil indicators are: Substrate, A-horizon depth, topsoil structure, subsoil structure, rooting depth, profile available water, wetness and ponding, slope, and relief. The 13 hazard indicators are: Contamination, salinization, sodification, acidification, low total nutrient status, shallow soil depth above hard rock, drought, flooding and extreme waterlogging, steep slope, rock and surface, high percentage of coarse texture fragments, unsuitable soil thermal regime, and miscellaneous hazards (e.g., exposure to wind and water erosion) [4
]. The procedure is an up-to-date, internationally acknowledged and applied method to assess soil quality [3
]. Because of its focus on soil structure and soil degradation characteristics, most (albeit not all) of the indicators are sensitive to improper agricultural soil management, which makes it particularly useful for the scope of this study. Overall, the MSQR compiles parameters on a uniform basis that can be spatially processed. We combined the yield potential data with arable land rent price and rent proportion data. The source of both data sets is the German Statistical Agency of the Federal States, which provided German-wide data from the Agricultural Census in 2010 [20
]. In Germany, the rent prices and the rent proportion rate increased constantly during the last decades, with significant differences between East Germany (former GDR) and West Germany (former BRD) [19
]. Because of this situation, we analyzed the data for Germany in total as well as for East and West Germany separately. For all three indicators, we were able to access data on a uniform basis at the county level and therefore focus our empirical analysis at this level.
3.1. Data Acquisition: Arable Land Rent Proportion, Rent Price and Yield Potential
The most relevant source for data on arable land rent proportion and arable land rent price is the Agricultural Census for Germany from 2010. It comprises data from the main Land Use Survey, the Survey on Agricultural Production Methods, and the Agricultural Structure Survey. It is commonly used for German and EU administrative purposes at different political and administrative levels and includes data with spatial information on Federal State-level, NUTS2- and NUTS3-level (the abbreviation NUTS is explained in the next paragraph) per farm operating business (according to a threshold of 5 ha or more and based on the locality of the operating business). The Agricultural Census comprises (based on uniform table formats) data on land use, livestock, labor forces, acquisition of agricultural production methods and “further survey characteristics” that constitute the legal form, place of farm operating business, owner and tenancy information, land under tenure and rent prices for arable land [18
]. Comparisons with previous Agricultural Census data are difficult because of changes in data collection processes and definitions such as the size of farm operating business considered in the statistics [20
For the purpose of this analysis, the smallest but nation-wide uniformly assessed spatial unit for Germany is the statistical unit NUTS-3 (county level). The “Nomenclature des Unités territoriales statistiques—NUTS” represents statistical regions within the EU and facilitates the supranational statistical comparison of such regions. NUTS-3 regions represent the statistically (based on the population) smallest regions. In Germany, NUTS-3 regions represent counties and corporate cities and number 402 in total [39
]. To ensure a nation-wide uniform analysis of data based on the Agricultural Census from 2010, data were requested from the Statistical Agency of Germany. Statistical data from all federal states of Germany were requested with a response rate of 15 out of 16 (federal state statistical offices). After a data preparation and cleansing process, 389 out of 402 NUTS-3 regions were used for further analysis. In addition to four corporate city regions, the data for six NUTS-3 regions from one federal state were not available. For the purpose of our analysis, data on arable land rent proportion at the NUTS-3-level could not be derived directly from the acquisition, but data on rent prices could be derived directly for NUTS-3 regions.
Aside from the statistical data sources used to assess arable land rent proportion and arable land rent price, spatial data on the arable land yield potential was assessed with the aim to visualize the linkage and spatial distribution of those three variables. Here, two basic types of spatial datasets form the basis for the visualization, shapefiles and a raster dataset. The shapefiles mostly represent administrative borders. The raster dataset comprises data on arable land yield potential that is available for arable land in Germany at a scale of 1:1,000,000 (BÜK 1000) [24
The arable land rent price as well as the arable land rent proportion were calculated and visualized for NUTS-3 regions, which reflects the highest spatial resolution that can be acquired at the national scale for these data. Rent proportion and rent price data were not normally distributed and thus were classified for the spatial visualization based on Jenks (natural breaks), which is a common and suggested method given an uneven distribution (Figure 1
a,b). The method orientates itself on natural data gaps and classifies the data in such a way that variations within classes are as low as possible while differences between classes are as large as possible [42
]. For arable land yield potential, the mean value has been visualized for NUTS-3 regions. Heterogeneities within those regions can therefore not be visualized. Figure 1
c represents the frequency distribution and the breaks within the arable land yield potential data based on the classification of the BGR [24
] (Table 2
), reflecting heterogeneities among the raster data set of almost 2 million points for Germany.
3.2. Data Calculation and Visulization: Arable Land Rent Proportion, Rent Price and Yield Potential
The arable land rent proportion was determined by the share of rented arable land out of the total arable land (in %) at the county level (=NUTS-3 regions) from the Agricultural Census of 2010. Arable land rent prices were directly taken from the Agricultural Census for NUTS-3 regions. The calculated values for arable land rent proportion and rent prices as well as the average arable land yield potential for NUTS-3 regions were correlated with each other. The three variables were tested for normality in their distribution using the “Kolmogorov-Smirnov-Test”, and the Spearman rank correlation rs
was chosen as none of the variables are normally distributed (Figure 1
a–c). In this method, the correlations are based on the rank of the values that are ranked beforehand [43
]. The scatter plots provide a visual representation of the characteristics of the cases and support the identification of patterns and the description of correlations. They are not intended as an analysis in inferential statistical terms.
For the visualization of the data in a spatial map, the raster dataset was transformed into a vector dataset. The vector dataset allows for the attachment of the NUTS-3 information on arable land rent proportion, rent prices for arable land, and the average (mean) value for the arable land yield potential. The classification of the arable land yield potential (Table 2
) guides the map of arable land yield potential of German soils by the BGR [24
] (Figure 5).
During data processing, data for some NUTS-3 regions were removed and therefore do not appear in the map nor were they considered for the statistical analysis. The data set used therefore only includes data comprising information on arable land rent proportion and price. Where the arable land yield potential is not visualized (Figure 5), no arable land is determined by the MSQR. In these cases, the dominant land use type might be, for example, grassland.