N ratios in European agricultural

Science of the Total Environment 627 (2018) 975–984

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

GEMAS: CNS concentrations and C/N ratios in European agricultural soil Jörg Matschullat a,⁎, Clemens Reimann b, Manfred Birke c, Debora dos Santos Carvalho d, the GEMAS Project Team: S. Albanese, M. Anderson, R. Baritz, M.J. Batista, A. Bel-Ian, D. Cicchella, A. Demetriades, B. De Vivo, W. De Vos, E. Dinelli, M. Ďuriš, A. Dusza-Dobek, O.A. Eggen, M. Eklund, V. Ernsten, K. Fabian, P. Filzmoser, D.M.A. Flight, S. Forrester, U. Fügedi, A. Gilucis, M. Gosar, V. Gregorauskiene, W. De Groot, A. Gulan, J. Halamić, E. Haslinger, P. Hayoz, J. Hoogewerff, H. Hrvatovic, S. Husnjak, F. Jähne-Klingberg, L. Janik, G. Jordan, M. Kaminari, J. Kirby, V. Klos, P. Kwećko, L. Kuti, A. Ladenberger, A. Lima, J. Locutura, P. Lucivjansky, A. Mann, D. Mackovych, M. McLaughlin, B.I. Malyuk, R. Maquil, R.G. Meuli, G. Mol, P. Négrel, P. O'Connor, K. Oorts, R.T. Ottesen, A. Pasnieczna, V. Petersell, S. Pfleiderer, M. Poňavič, C. Prazeres, S. Radusinović, U. Rauch, M. Sadeghi, I. Salpeteur , R. Scanlon, A. Schedl, A. Scheib, I. Schoeters, P. Šefčik, E. Sellersjö, I. Slaninka, J.M. Soriano-Disla, A. Šorša, R. Svrkota, T. Stafilov, T. Tarvainen, V. Tendavilov, P. Valera, V. Verougstraete, D. Vidojević, A. Zissimos, Z. Zomeni a

Interdisciplinary Environmental Research Centre, TU Bergakademie Freiberg, Brennhausgasse 14, 09599 Freiberg, Germany Geological Survey of Norway (NGU), P.O. Box 6315, Sluppen, 7491 Trondheim, Norway c Federal Institute for Geosciences and Natural Resources (BGR), Stilleweg 2, D-30655 Hannover, Germany d PEA – Departamento de Engenharia de Energia e Automação Elétricas, Av. Prof. Luciano Gualberto, Travessa 3, no 158, São Paulo, SP CEP 05508-010, Brazil b

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• The distribution of total C, N and S, and TOC in European agricultural soil is presented. • First complete dataset with extensive quality control at the European scale • Continental-scale processes control regional TC, TN and TS and their distributions become visible. • Soil status (TC, TN, TS) is dominated by natural drivers (geology, climate) at the continental scale.

a r t i c l e

i n f o

Article history: Received 8 December 2017 Received in revised form 21 January 2018 Accepted 21 January 2018 Available online xxxx Editor: F Frederic Coulon

a b s t r a c t A reliable overview of measured concentrations of TC, TN and TS, TOC/TN ratios, and their regional distribution patterns in agricultural soil at the continental scale and based on measured data has been missing – despite much previous work on local and the European scales. Detection and mapping of natural (ambient) background element concentrations and variability in Europe was the focus of this work. While total C and S data had been presented in the GEMAS atlas already, this work delivers more precise (lower limit of determination) and fully quantitative data, and for the first time high-quality TN data. Samples were collected from the uppermost 20 cm of ploughed soil (Ap horizon) at 2108 sites with an even sampling density of one site per 2500 km2 for

⁎ Corresponding author. E-mail address: [email protected] (J. Matschullat).

https://doi.org/10.1016/j.scitotenv.2018.01.214 0048-9697/© 2018 Elsevier B.V. All rights reserved.

976 Keywords: Carbon Nitrogen Sulphur Total organic carbon (TOC) Spatial distribution GEMAS project Arable soil

J. Matschullat et al. / Science of the Total Environment 627 (2018) 975–984

one individual land-use class (agricultural) across Europe (33 countries). Laboratory-independent quality control from sampling to analysis guaranteed very good data reliability and accuracy. Total carbon concentrations ranged from 0.37 to 46.3 wt% (median: 2.20 wt%) and TOC from 0.40 to 46.0 wt% (median: 1.80 wt %). Total nitrogen ranged from 0.018 to 2.64 wt% (median: 0.169 wt%) and TS from 0.008 to 9.74 wt% (median: 0.034 wt%), all with large variations in most countries. The TOC/TN ratios ranged from 1.8 to 252 (median: 10.1), with the largest variation in Spain and the smallest in some eastern European countries. Distinct and repetitive patterns emerge at the European scale, reflecting mostly geogenic and longer-term climatic influence responsible for the spatial distribution of TC, TN and TS. Different processes become visible at the continental scale when examining TC, TN and TS concentrations in agricultural soil Europe-wide. This facilitates large-scale land-use management and allows specific areas (subregional to local) to be identified that may require more detailed research. © 2018 Elsevier B.V. All rights reserved.

Fig. 1. a. Sampling sites for the GEMAS Ap materials. Each dot represents an individual site, where a composite sample was taken. b. Total carbon (Ctot) distribution across the European continent (kriging based). Concentrations (wt%) in Ap soil material (0–20 cm depth). c. Organic carbon (Corg) distribution across the European continent (kriging-based). Concentrations (wt%) in Ap soil material (0–20 cm depth). d. Total sulphur (Stot) distribution across the European continent (kriging based). Concentrations in Ap soil material (0–20 cm depth).


1. Introduction Life on Earth depends on the three major elements carbon (C), nitrogen (N) and sulphur (S). Their concentrations in soil and their ratios (especially the TOC/TN ratio) determine soil health and productivity. Nitrogen and sulphur together with calcium, magnesium, potassium and phosphorus belong to the most important plant macronutrients in soil. They are a major limiting factor for productive agriculture and healthy forests, if deficient. Until now the GEMAS dataset provided an overview of total organic carbon (TOC) and of total carbon and sulphur concentrations (TC and TS used here throughout to distinguish from results based on extractions). However, total nitrogen (TN) concentrations were missing and in consequence the important TOC/TN ratio (similar to the less precise yet frequently used ‘C/N ratio’) and its regional distribution patterns at the continental scale (land-use class ploughed agricultural land across most of Europe). More than 4100 soil samples were collected in 33 European countries, covering an area of 5.6 million km2 (Reimann et al., 2014a, 2014b; Fig. 1(a)). Soil samples were taken from two different land-use types, regularly ploughed agricultural fields (0–20 cm; 2108 Ap samples) and land under permanent grass cover (0–10 cm; 2024 Gr samples), according to the European REACH Regulation (Registration, Evaluation, Authorisation and Restriction of Chemicals; EC No 1907/ 2006, n.d.) guideline documents (https://echa.europa.eu/documents/ 10162/13632/information_requirements_r16_en.pdf - Table R16-11) and the ECHA guideline for the assessment of environmental exposure (ECHA, 2008). This paper focuses on new analytical results obtained for TN, TC and TS on the Ap samples. The main objective of the GEMAS project was to detect and map natural background element variation at the European scale. The project has resulted in a geochemical atlas and a series of peer-reviewed publications with detailed interpretations of the regional distribution of single elements or groups of elements and of other properties (e.g. Saaltink et al., 2014). Analytical results for TC, TOC and S (total and aqua regia are available) have been presented in the GEMAS atlas (Baritz et al., 2014; Reimann et al., 2014a), whereas N, one of the most important elements for agriculture, has been missing until now. Both total C and S were reanalysed here to overcome a too high detection limit for S (25% of all samples returned an analytical result below the detection limit for the dataset used in the GEMAS atlas), or relatively poor precision at very low concentrations for C (original GEMAS dataset). The new data can also be viewed as an additional and independent quality control. Another more model-based approach to mapping soil organic carbon with a different sampling strategy is available from the LUCAS survey (Jones et al., 2005; Tóth et al., 2013, 2016; Brogniez et al., 2015; Yigini and Panagos, 2016, see http://eusoils.jrc.ec.europa.eu/esdb_ archive/eusoils_docs/other/EUR26102EN.pdf). Unfortunately, those data cannot be compared directly with ours because five different land-use classes were used for the LUCAS survey and differentiation into a single land-use class covering all of Europe evenly (e.g. agricultural soil) is not provided. A comparison of their approach with the one discussed here has the potential to improve model-based mapping. Related background information from maps of geographical, geological, climatological and land-cover information is easily available (e.g., Reimann et al., 2014a), and not repeated here. Additional maps were used, such as the soil map of Europe from Stanners & Bourdeau (1995, p. 149) to verify soil types based on FAO data and the nitrogen supply from manure and from fertilizer maps for comparison with N distribution found in our study. The main aim of this paper is to present the new results for total nitrogen (TN) concentrations and TOC/TN ratios for European agricultural soil to derive typical background concentrations for different parts of Europe, and to illustrate the regional distribution of TN and TOC/TN at the European-scale. In addition, the new datasets for TC and TS are presented and used; all data are available online; see Supplementary

977

Information. The samples collected for the GEMAS project are an excellent base for studying soil chemical variation related to features such as lithology, climate, topography, soil texture and anthropogenic impact at the continental scale. 1.1. Importance and role of total nitrogen Little is known about N concentrations in rocks (Holloway and Dahlgren, 2002; Stevenson, 1962), yet the total average concentration of N in the Earth's upper continental crust (UCC) has been calculated to be 83 mg/kg (0.008 wt%; Wedepohl, 1995). In comparison with the UCC, the European Ap soils are considerably enriched in N (Ap/UCC ratio = 21; this study), reflecting soil fertility and soil formation processes. Soil from N-rich substrates such as coal beds or carbonaceous shales (England, Belgium, Ruhr area in Germany, central western Europe) or subject to hydrothermal activity (Italy) may show elevated N concentrations (Holloway and Dahlgren, 2002). It is, therefore, important to determine the true European-scale distribution of N in agricultural soil. Nitrogen is applied with fertilizers in various forms: (i) ammonia solutions and ammonium nitrate (NH4NO3) as ‘straight’ fertilizer, (ii) binary fertilizers with a combination of N with potassium and or phosphorous, the latter as mono-ammonium phosphate (MAP: NH4H2PO4) or di-ammonium phosphate (DAP: [NH4]2HPO4), and (iii) three-component NPK fertilizer, providing various percentage concentrations of the three components NPK (e.g. 10-10-10 or 16-4-8). Excessive use of fertilizers may lead to both eutrophication (nutrient overload) in soil and surface waters and to soil and water acidification through hydrolysis and formation of nitric acid from excessive nitrate. France (2.5), Germany (2.0), United Kingdom (1.3), and Spain (1.2) are among the top ten countries worldwide in their consumption (megatons per year) of N-based fertilizers (FAO, 2009). When calculating the annual per-hectare consumption, Luxemburg and Ireland consume almost 400 kg ha−1 year−1, whereas countries such as Austria, Slovakia, Greece, the Baltic countries, Hungary, Sweden, and Romania use b100 kg ha−1 year−1. On average, the 28 EU member states consume about 150 kg of fertilizer ha−1 of arable land per year (World Bank, 2012). The expected correlation between fertilizer usage and TN distribution and concentrations is clearly visible. 1.2. Importance of the TOC/TN (C/N) ratio in agriculture The TOC/TN ratio is the proportion of total organic carbon (TOC) to total nitrogen (TN) in soil; it is important for managing soil and crop nutrient recycling. The TOC/TN ratio provides information on the status of decomposition and potential nutrient availability in organic matter. Large ratios (above 25) indicate slow rates of decomposition of organic material, whereas small ratios (below 25) represent increasingly faster rates of decomposition. The soil TOC/TN ratios are used in agriculture as an indicator of soil fertility and to determine at what point N-fixing plants such as legumes should be grown in a rotation. An increasing amount of TN is available with a smaller TOC/TN ratio. This concept has limitations in soil with very small absolute TOC or TN concentrations (very coarse soil types with large pore spaces, typical for Scandinavian soils, and many types of soil under subtropical and tropical conditions, similar to parts of the Mediterranean region). The ratio can be applied quite well under a broad range of more temperate and humid conditions. Soil microbial communities show average ratios of 8, green cuttings and clippings between 7 and 15, decaying organic matter around 20 and raw humus between 25 and 40. Straw may have TOC/TN ratios above 70. Under temperate conditions, TOC/TN ratios in grassland soil are often around 11, whereas for arable soil TOC/ TN ratios are about 14 and forest soil has ratios of around 21 and above (Blume et al., 2008). To explore the TOC/TN ratio in European soil and to assess regional differences is as an important task for largescale land-use planning and sustainable agricultural development.

978


2. Materials and methods 2.1. Sampling and sample preparation Sampling took place from summer 2008 to early spring 2009. In total, 2108 samples (plus field duplicates) of agricultural soil (Ap horizon, 0–20 cm) were taken at an average sampling density of one site per 2500 km2 (50 km × 50 km grid) across Europe (Fig. 1(a)). A field handbook describes the sampling procedure in detail (EGS, 2008). All samples were sent to a central laboratory for preparation (air-drying, sieving to b2 mm with nylon screens, randomization, insertion of analytical duplicates, project standards and aliquots). Sample preparation is explained in detail in Mackových and Lučivjanský (2014). 2.2. Analytical methods Total C, N and S were analysed quantitatively (after sample grinding to b63 μm) by elemental analysis (EA) at TU Bergakademie Freiberg (Elementar Vario EL Cube, Hanau, Germany). The technique is based on ‘purge and trap’ separation (C, N, [S]) with additional infrared (IR) detection [S], following high temperature incineration (induction furnace) and constant temperatures above 1150 °C for the sample, with tungsten(IV)oxide (WO3) as catalyst. Helium was used as the carrier gas. Concentration ranges for the calibration were 0.0458–21.33 mg C, 0.0006–3.108 mg N and 0.00085–1.665 mg S; equivalent to 0.35–43.59 wt% C, 0.08–1.18 wt% N, and 0.04–4.90 wt% S. The daily adjustment factor to adjust the calibration curve was determined with sulphanilic acid; this factor was used later in concentration calculations. Each 20 mg of Ap-soil sample (b63 μm) was mixed with 60 mg of WO3 and weighed on a piece of tinfoil, which was then folded into a closed sample container. These vessels were incinerated in a pure oxygen atmosphere. Here, C, N and S were oxidized to CO2, SO3, SO2, NO2, NO and molecular nitrogen. Nitrogen dioxide and NO were reduced to molecular N on a copper surface and SO3 was reduced to SO2. Molecular N and CO2 were detected by thermal conductivity (TCD); SO2 was quantified by an IR detector plus TCD (depending on concentration). All 2108 samples could be fully quantified because the instrument allows for low detection limits: C, 0.04 wt%; N and S, 0.003 wt%. Total organic carbon (TOC) was determined earlier according to ISO standard 10694 at a commercial laboratory (Fugro, now Kiwa, Hamburg, Germany; Reimann et al., 2014a). To remove any inorganic C, 1 g of sample was treated with hydrochloric acid (4 mol l−1) and left standing for 4 h at room temperature. Thereafter, the samples were dried at 70 °C in an oven for 16 h. Then 100–200 mg per sample was placed into the furnace and TOC determined by IR spectroscopy. 2.3. Quality control For quality control purposes, a field duplicate was taken at every 20th site, an analytical duplicate was prepared from each field duplicate and inserted near the original sample with a different sample number, a project standard was inserted at a rate of 1 in 20 samples and all samples were randomized prior to submission to the laboratory. In addition, standards (control reference material) from an Australian continentalscale mapping program (de Caritat and Cooper, 2015), the North American Soil Geochemical Landscapes project (Smith et al., 2013) and the BraSol-2010 pedogeochemical background of north-eastern Brazil (Schucknecht et al., 2012) were inserted regularly. The two project standards of the GEMAS project, Ap and Gr and the BraSol-2010 in-house standard underwent international ring testing to be able to estimate the accuracy of the results. The Freiberg laboratory included the certified materials MRG, GBW 07601 and 07602, KH, TS and IVA 33802150 as unknown materials although not all of these materials give certified values for all three components (only recommended values).

Analytical precision for TC, TN and TS (based on the duplicate results) is in the range of ±10% for all three properties. Precision for the project standard Ap, analysed 124 times, was in the range of 2–3% for TC and TN and 8% for TS. The smallest sample values were all above the detection limits. An analysis of variance (unbalanced design) based on field and analytical duplicates (for details see Reimann et al., 2014a) returned 98% (TC), 96% (TN) and 95% (TS) on the degrees of ‘natural variation’. The results suggest that the data are well suited to produce regional distribution maps. 2.4. Data analysis and mapping Exploratory data analysis techniques (EDA; Tukey, 1977) were used throughout this paper. Geochemical data are compositional (closed) data (Aitchison, 1986; Filzmoser et al., 2009). Consequently, correlation analysis and other more advanced data analysis techniques should be used with care (Filzmoser et al., 2009, 2010). It is important to note that statistical parameters such as the mean or standard deviation of compositional data do not plot in Euclidian space, but rather on the Aitchison simplex. Statistics presented here are calculated from percentiles and the standard deviation is replaced by the power, i.e. the order of magnitude of the variation in the data ranges (Filzmoser et al., 2014; Reimann et al., 2012). Geochemistry aims to quantify the chemical composition of the Earth and its parts, and to identify the properties and processes that control the distribution of individual elements (Goldschmidt, 1937, 1954). Geochemical maps based on the original (measured) data are needed, and one could state simply that the principal idea is to detect the unexpected at any scale. In contrast, in soil science there has been a tendency to produce maps of soil classes that express several variables, usually with a given range, or interpolated maps such as those by inverse weighted distance or in recent decades by kriging. This has probably resulted from the need for maps that show the general patterns of variation in the soil and individual properties to aid land management, and in particular agriculture. The methods smooth the variation in maps of both soil classes or interpolated soil variables. Nevertheless, as with geochemists, soil scientists always pay attention to the original data at the sampling points, and relate this to the maps. A comparison between model-based maps with those derived from original data can then help to assess the reliability of the former and to improve the quality of any models that might underlie such maps. Point source black and white maps provide the most honest view of the data because every measurement and data gap remains visible (Reimann, 2005; Reimann et al., 2008, 2014a). Interpolated maps, however, often represent the overall variation more clearly and are used increasingly in modern geochemistry and soil science. A set of black and white point source maps based on EDA symbols (see Reimann et al., 2008 for a discussion of the different mapping techniques) were produced. The colour-plot maps are shown here, the EDA-based maps are provided in the Supporting Information, jointly with the original data. Summary statistics from the new analytical results are given in Table 1. Concentrations of chemical elements in bedrock, soil, sediments, plants and water are reported in relative units like wt%, mg/kg or μg/L. These units indicate and imply that earth, soil and environmental scientists are dealing with compositional data (CoDa) in practically all their investigations. Aitchison (1986) was the first to discuss the important consequences of CoDa for statistical data analysis. In short, CoDa values have a finite range (e.g. 0–100%, or 0–106 mg/kg) and the sum of all variables is less or equal to the maximum value. Mathematically this implies that CoDa vectors do not belong to a linear infinite Euclidean space, but are constrained to the Aitchison simplex. Because most standard statistical techniques are based on Euclidean spaces, and Euclidean distances between data vectors, their application to compositional data are outright false. Notably, some important statistical quantities still have a well-defined meaning for CoDa. These include individual concentration mean values and all concentration quantiles. However, for


979

Table 1 Statistical parameters for TC, TOC, TN, and TS (all wt%) and TOC/TN-ratios (C/N) in GEMAS Ap soil samples from all across Europe. Variable/unit

n

DL

Min.

Q10

Q25

Q50

Q75

Q90

IQR

Max.

Power

TC/wt% TOC/wt% TN/wt% TOC/TN TS/wt%

2108 2095 2108 2095 2108

0.04 0.1 0.003 n/ab 0.003

0.212 0.40a 0.018 1.8 0.008

0.374 0.90 0.086 8.0 0.008

1.44 1.20 0.121 8.9 0.026

2.20 1.80 0.169 10.1 0.034

3.51 2.60 0.239 11.8 0.048

3.52 3.90 0.333 15.0 0.048

2.07 2.70 0.212 6.1 0.022

46.3 46.0 2.64 252 9.74

2.1 2.1 2.2 2.1 3.1

a

The larger minimum value for Corg than for Ctot is probably related to analytical differences between the Freiberg labora tory and the Fugro laboratory (see Methods section, 2.2). n/a, not applicable. TC, total carbon; TOC, total organic carbon; TN, total nitrogen; TS, total sulphur. DL, detection limit; Min., minimum value; Max., maximum value; Q10 to Q90, quantiles; IQR, interquartile range; Power, orders of magnitude variation between Min. and Max. b

example, the standard deviation of these concentrations has lost its mathematical background (Euclidean distance or normal distribution), because the concept of standard deviation is implicitly based on the false assumption that concentrations can take arbitrary real values. Thus Table 1 shows exclusively percentiles and uses the powers (orders of magnitude variation) as a measure of spread. 3. Results and discussion The large spread of all data values of more than two orders of magnitude between the smallest and largest values reflects the diversity of soil types and climate regimes in Europe. Two (TC, TN) and three (TS) orders of magnitude between the smallest and largest element concentrations is not unusual for natural distributions of chemical elements (e.g. Reimann et al., 2003; Salminen et al., 2004, 2005; De Vos and Tarvainen, 2006). The distribution of sample locations across Europe is shown in Fig. 1 (a) and maps of the properties determined previously, TOC, TC and TS, are presented in Fig. 1(b–d), respectively. The TC and TS maps in Fig. 1 (c,d) are based on the new data reported here, in contrast to the maps provided in the GEMAS atlas (Reimann et al., 2014a). Fig. 2 presents a comparison between the old and the new data. A few samples were identified that needed to be reanalysed; see sample numbers in Fig. 2. In general reproducibility for TC was excellent. The results for TS emphasize the problem of the too high detection limit (25% of all samples bDL) that was used for the old dataset (horizontal symbols at 0.005 wt %). 3.1. Total nitrogen (TN) The statistical distribution of TN data is almost symmetric on a logarithmic scale (Fig. 3). Several samples with either unusually small or large TN concentrations are visible in both the one-dimensional scatterplot and boxplot. The spatial distribution for TN in the Ap horizon shows large regional differences (Fig. 4) that are similar to those depicted in Fig. 1(b,c) for TC and TOC, however, with a more narrow range of values as compared with TC (10th to 90th percentile;

Table 1). The spatial variation for TN is only partly visible when compared with the N maps for fertilizers of Stanners & Bourdeau (1995: Figs. 22.1 and 22.2). For example, Fig. 4 shows larger values for TN, e.g. in the Netherlands, and Lower Saxony, Germany, at the boundary between Switzerland and Austria, in Galicia, Spain, southwestern England, Northern Ireland and the area around Uppsala, Sweden. The differences between the maps of fertilizer N and Fig. 4 confirm the hypothesis of important influences related to the spatial distribution of N that require additional local and small scale regional studies. For example, ‘hotspots’ of TN are evident in the western Netherlands and in the region around Osnabrück in Germany that are probably related to intensive livestock management, similar to those of TOC. The relatively small TN concentrations in much of Spain's intensive agricultural land are possibly the result of a more rapid rate of biomass uptake and export (harvest) of N because of the higher average temperatures in the Mediterranean. Total N decreases with increasing average temperatures and increases with precipitation, albeit the relation is weak (Fig. 6b, c, respectively), whereas there is no discernible effect of altitude or pH (Fig. 6a, d, respectively). The median values of soil-type groups are similar to each other, and the more fine-grained soils show most of the maxima and sandy soils the minima (not shown, yet see supplementary box plot on clay). Fig. 7 shows boxplots of TN concentrations of the different countries covered by the GEMAS survey for comparison. The countries are sorted according to decreasing TN median values and the plots show among others a strong climatic effect. Most of the northern countries (Ireland, United Kingdom, Norway, Sweden) have large TN concentrations, whereas the southern countries (Spain, Portugal, Greece) have predominantly small TN values. Interestingly, the median value for Poland is towards the small values of the plot. Here climate is not the cause, but rather the coarse grained sandy soil formed on top of the sediments of the last glaciation. The majority of the countries show considerable variation in TN concentrations from the spread of values beyond the boxes. Unlike the lithological source of carbon in soil, nitrogen enrichment or depletion reflects microbial life, soil–plant interaction and fertilizer regime. Atmospheric deposition may also play a role, albeit secondary (Dämmgen et al., 2013) because atmospheric N input is dominated by

Fig. 2. Scatter plots of the original (GEMAS) and new measurements for: (a) TC and (b) TS plotted on logarithmic scales.

980


the strongly water-soluble species. Human interference with the N cycle is strong, mainly through biomass removal and related fertilizer applications in agricultural soil management. Pastureland with high animal density per hectare might also be over fertilized through animal excrements as suggested for the sites in the Netherlands and in Lower Saxony. Both management practices are known to result in excess additions of N locally. However, large TN concentrations in organic soil such as in large parts of northern Europe, including Ireland and the United Kingdom are natural and reflect organic-rich soils, mostly bogs and mires. They correspond with the distribution of organic carbon (TOC).

3.2. Organic carbon to nitrogen ratio (TOC/TN = Corg/Ntot = C/N-ratio)

Fig. 3. Distribution of data for total nitrogen (TN) in a combined histogram, density trace, one-dimensional scatterplot and boxplot diagram. The x-axis shows the concentrations (wt%) on a logarithmic scale in Ap samples, analysed by El Cube (see Methods).

Organic-rich soil (Fig. 1c) tends to show considerably larger TN concentrations (wetter and more Nordic environments, plus mountainous areas) while quartz-rich soil (coarse) tends to have the smallest concentrations, reflecting its higher permeability, good drainage and small organic material content (Figs. 4, 5; Négrel et al., 2018: 1280). This is not corroborated by average precipitation data, where, when comparing boxplots of different “average precipitation” classes, median values of all classes are almost identical and the largest spread emerges for the lowest annual precipitation levels. Conversely, TOC/TN clearly increases

Fig. 4. Total nitrogen (Ntot) distribution across the European continent (kriging based). Concentrations in Ap soil material (0–20 cm depth).


981

Fig. 5. Organic carbon-nitrogen ratios (C/N) distribution across the European continent (kriging based).

with larger annual average temperatures (Supplement Information boxplots). The TOC/TN ratios, similar to TN concentrations, show that finer-grained soils have a larger spread of values and more large values than coarser soils because of their higher permeability. The largest TOC/ TN ratios correlate with the maximum soil clay content (Supplement Information boxplots). The TOC/TN ratios show some strong correlations with natural conditions, albeit not uniform across the continent. Fig. 5 shows two regions with large values, much of Spain and Scandinavia. An area with larger values occurs to the north of the southern extension of the Pleistocene glaciations (Négrel et al., 2018: 1280), separating the younger postglacial soil in the north from the potentially much older soil to the south of the glacial extent. Although this might not necessarily reflect the effect of parent material alone, this region, particularly to the south of the Baltic Sea, is characterized by comparatively young soils in a mature and fertile state with a good buffering capacity. In Scandinavia, cooler average temperatures and limited precipitation lead to limited decomposition over time. However, the larger TN concentrations of Ireland, southwestern UK and the (interrupted) stretch from northwestern France to Bulgaria are the reverse for the TOC/TN ratio, with smaller average ratios in this “belt” and larger values to its north and southwest. The summary statistics show a large spread of values in the TOC/TN ratio from 1.8 to 252, with a median value of 10.1 (Table 1). This median value resembles the TOC/TN ratio of soils under temperate forests and of humus material, which suggests temperate conditions with potentially

natural vegetation. Most sites have relatively small TOC/TN ratios of between 8 and 15, which indicate moderate rates of biological decomposition and related healthy soil life. Even on the logarithmic scale used here, the data are clearly negatively skewed with a significant number of samples showing larger TOC/TN ratios (Fig. 5; Table 1) suggesting insufficient and slow rates of decomposition. Areas with large TOC/TN ratios can be explained by below average TOC concentrations, for example central and eastern Spain. Consolidated calcareous rocks underlie this area. A third area of large TOC/TN ratios occurs in central Italy west of the Abruzzi Mountains, again an area with calcareous rocks. Improvement in soil management appears to be the best option to improve TOC/TN ratios where necessary by increasing the organic matter content of the soil. The largest soil calcium values relate to the largest TOC/TN ratios, whereas this was only weakly evident for phosphorus (Reimann et al., 2014a: 335). Both signals indicate fertilization effects apart from lithology (Fabian et al., 2014). This is corroborated further by the patterns of variation in calcium and TOC/TN ratios and pH (Supplement Information boxplots). In the United Kingdom large TOC/TN values occur in some areas with old siliciclastic sedimentary rocks. While climate variables such as temperature and precipitation, and the climate zones have some effect on the decomposition of organic matter, their patterns of variation do not necessarily correspond with those of TOC/TN ratios throughout the EU. Sarmatic and Baltic mixed forest environments (around the southern Baltic Sea) have above-

982


Fig. 6. Scatter plots of total nitrogen against altitude (m above sea level), mean annual temperature (°C), mean annual precipitation (mm) and pH (CaCl2).

average TOC/TN ratios that extend into the East European forest steppe. The higher mountain ranges, Pyrenees, Alps and Carpathians, clearly stand out with below average TOC/TN ratios. Similar relations continue into lower altitude forest biomes, such as the northwest Iberian montane forest (Spain), and the Rhodope montane mixed forests (Bulgaria and Greece), but are not evident in the Apennine deciduous montane zone forests (Italy) where lithological forces seem to dominate. In general, agricultural practices over the past several thousand years seem to have left less of a legacy of visible change in TOC/TN ratios than the natural boundary conditions of ecozones and biomes. In addition, Fig. 5 shows several areas with unusually large TOC/TN ratios, some of which do not follow the general pattern described above: for example, near to Geilo (NOR), Norrköping (SWE), the Hebrides (UNK),

northwestern Ireland, Newcastle area (UNK), Meppen (GER), Reims (FRA), Terni and Terracina (ITA) and various locations within the eastern half of Spain. These areas often appear to represent organic soils such as former bog and moor environments now under cultivation. Seventeen samples have TOC/TN ratios above 30 (32 samples have TOC/TN ratios ≥25). Of these 17 ratios, 10 are in Spain. It appears unlikely that lithological parent material explains these anomalies. It might be that the method of TOC quantification (nSpain = 9, see Methods section) on strongly calcareous soil affects this pattern. High pH values, large cation exchange capacities (CEC) with a small chemical index of alteration (CIA) might also occur at other places, yet have different TOC/TN ratios. The most prominent example is central western Ukraine, whereas the large TOC/TN

Fig. 7. Box plots of TN concentrations in the European countries covered by the GEMAS survey, determined by El Cube (see Methods). Countries have been sorted according to decreasing median concentration per country (MON, Montenegro; IRL, Ireland; UNK, United Kingdom; NOR, Norway; SWE, Sweden; AUS, Austria; SLO, Slovenia; BOS, Bosnia; SIL, Switzerland; SKA, Slovakia; EST, Estonia; SRB, Serbia; BEL, Belgium; CZR, Czech Republic; FIN, Finland; CRO, Croatia; UKR, Ukraine; NEL, The Netherlands; HUN, Hungary; FRA, France; DEN, Denmark; GER, Germany; LAV, Latvia; ITA, Italy; BUL, Bulgaria; CYP, Cyprus; HEL, Greece; FOM, Former Yugoslavian Republic of Macedonia; LIT, Lithuania; POL, Poland, PTG, Portugal; LUX, Luxembourg; SPA, Spain). Numbers at the top of plot indicate the number of samples taken in the country.


ratio to the south of Reims (FRA) relates strongly with the large ratios in Spain (similar values). Several very small TOC/TN ratios occur in Europe, namely near Kilkenny (IRL), Northampton (UNK), Laval (FRA), around Lake Balaton (HUN), around Andorra in the Pyrenees, Córdoba (SPA), around Rome and on north-eastern Sicily (ITA). These locations have above-average TN concentrations (0.18 wt%) in the Ap horizon, probably because of local or sub-regional agricultural practices (Fig. 5). 3.3. Additional remarks on TS The Freiberg data for total sulphur are of considerably better quality than the original TS data reported in the GEMAS atlas (Reimann et al., 2014a), see Fig. 2. The overall interpretation of the map (Fig. 1(d)) does not change, however, compared to that of aqua regia available S in the GEMAS atlas (Reimann et al., 2014a). The interpretation provided in the atlas, e.g. very low S-values in coarse-grained soils (Supplement Information boxplots) remains valid. Fig. 1(d) shows the distribution of TS in European agricultural soils at the continental scale based on the new data. This can for example be used to study whether many decades of acidic precipitation and long-term application of S fertilizers are visible. A distinct correlation of increasing TS values with phosphorus and with calcium (except for the largest Ca concentrations; see Supplement Information) emerges. This suggests the effects of fertilizer application. The spatial distribution of European SO2 sources (Stanners and Bourdeau, 1995: Figs. 4.1a and 4.2) only partly overlaps with some of the larger soil TS values. The hypothesis that long-term acidic precipitation across many parts of Europe is reflected in the soil has to be largely rejected, however. There is neither a correlation with soil pH classes nor with the TS concentrations in agricultural soil in parts of European that are remote from direct sources of S (Fig. 1(d)). Both phenomena can be explained by the soluble nature of the sulphate anion that drains rapidly from the soil to groundwater. 3.4. Comparison of model-based versus raw data maps Several maps of estimates of soil carbon have been published (e.g. Jones et al., 2004, 2005; Lettens et al., 2004; Brogniez et al., 2015). Jones et al.'s (2004, 2005) first map was based on pedotransfer function rules applied to the European soil database. It shows the distribution of soil organic carbon concentrations in the 0–30-cm layer. A more recent assessment of national soil carbon data showed that directly measured data are needed rather than estimates to produce a reliable map at the European scale (Panagos et al., 2013). Baritz et al. (2014) used the GEMAS data for the Ap and the Gr samples to produce two highresolution TOC maps. These estimates and those by Jones et al. (2004, 2005) indicate that the modelled results substantially overestimate TOC in northeastern Europe and Sweden and underestimate it in Norway, for example compared with the raw data. According to Baritz et al. (2014), strong local variation makes predictions quite difficult in parts of Europe. Interpolated maps appear smooth compared with measured values at the sampling sites. However, the former are well suited to studying the large-scale variation in background TOC values and regional scale differences. The original geochemical GEMAS TOC maps show clearly that large sub-regional deviations from estimated values must be expected in some places. Following up on such differences between model-based maps versus representatively distributed databased maps will lead to a better understanding of processes that cause the variation of TOC in agricultural soil across Europe. The apparently high-resolution maps based on modelled data can thus easily be misleading. The more recent maps based on the LUCAS survey results provided by Brogniez et al. (2015), are quite comparable (use of raw data) although several of the countries covered by GEMAS are missing here, and different land-use types were blended. The apparently high resolution of these maps is, however, not based on original measured data but on often rather poor correlations between TOC and other parameters in

983

the model – especially the high-resolution datasets based on topography and climate. Correlation coefficients that the models are built around lie most often in the b0.3 r2 range. That may be a “highly significant correlation”, according to statistical tests. Surprisingly it is often ignored that these statistical tests do not test for the strength of correlation but rather test whether the correlation is significantly different from 0. That is the case for very low correlation coefficients indeed, given the large datasets with thousands of samples at hand. A statistically highly significant correlation can thus be highly insignificant in practice, where correlation coefficients should in fact be N0.7 for good and reliable models. Models built around rather weak correlations will always have a strong smoothing effect on the distribution of the modelled variable in a map. One of the key messages of the GEMAS map – that a very high local variability of TOC concentrations has to be expected especially in northern Europe is practically lost in the modelled LUCAS map. That is the very reason why geochemists will always want to look at distribution maps of real, measured variables first before possibly continuing with modelled maps depending on purpose. 4. Conclusions The data from 2108 Ap horizon samples suffice to provide a detailed characterization of European agricultural soil on a subregional level. Similar to previous experience from other inorganic pedogeochemical parameters, TC, TN and TS concentrations vary by two to three orders of magnitude across the continent (Fig. 7). The interquartile range (IQR) of TN distribution is much more narrow as compared to TC (Table 1). Relative TN enrichment or depletion reflects microbial activity, soil–plant interaction and fertilization practise (including mass animal husbandry). Human interference with the N cycle is strong; mainly through biomass removal and related fertilizer application in agricultural soil management. The TOC/TN ratio median value of 10.1 resembles soils under temperate forests, comprised of humus material, indicating healthy soil conditions. Most data represent ratios between 8 and 15, indicating good biological decomposition rates. Despite this relative homogeneity, distinct elevated ratios (TOC/TN N 15) occur in various parts of Europe. Generally speaking, agricultural practices over the past several thousand years seem to have left less of a legacy of visible change in TOC/TN ratios than the natural boundary conditions of ecozones and biomes. Lithological conditions dominate the S distribution, too, despite many decades of acidic deposition. Since atmospheric deposition consists mostly of dissolved sulphate, this input will percolate soils and influence seepage and groundwater chemistry much more than the soils after recent years of relatively low atmospheric S input. All evidence considered it becomes obvious that despite more than a century of acidic deposition and many centuries of fertilization, the concentration distribution of carbon, nitrogen and sulphur is most strongly influenced by natural (geogenic) processes and much less by anthropogenic ones – at the investigated scale. This is very good news for agricultural soils in Europe. At the same time, partly serious disturbances may dominate soil chemistry on a local scale, “invisible” to the lowresolution mapping that underlies this project. Acknowledgements The authors are very grateful to Elvira Rüdiger, lab technician in the Institute of Mineralogy of TU Bergakademie Freiberg, for her dedicated contribution to ensure the highest analytical standards, and to B·Sc. Melanie Vierling, who dedicated her thesis to complete the data set. A thank you goes to Dr. Alexander Plessow for proofreading and a special appreciation to Anne Marie de Grosbois for language editing. Two anonymous referees and Prof. Dr. Margaret A. Oliver (University of Reading) have contributed significantly to

984


increase clarity; Thank you. The authors have no conflict of interest to declare. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.scitotenv.2018.01.214. References Aitchison, J., 1986. The Statistical Analysis of Compositional Data. Chapman & Hall, London. Baritz, R., Ernstsen, V., Zirlewagen, D., 2014. Carbon concentrations in European agricultural and grazing land soil. In: Reimann, C., Birke, M., Demetriades, A., Filzmoser, P., O'Connor, P. (Eds.), Chemistry of Europe's Agricultural Soils–Part B: General Background Information and Further Analysis of the GEMAS Data Set. Geologisches Jahrbuch, Schweizerbarth, Hannover, pp. 117–129. Blume, H.-P., Brümmer, G.W., Schwertmann, U., Horn, R., Kögel-Knabner, I., Stahr, K., et al., 2008. Scheffer/Schachtschabel Lehrbuch der Bodenkunde. Fifteenth edition. Spektrum Akademischer Verlag, Heidelberg. Brogniez, D., de Ballabio, C., Stevens, A., Jones, R.J.A., Montanarella, L., van Wesemael, B., 2015. A map of the topsoil organic carbon content of Europe generated by a generalized additive model. Eur. J. Soil Sci. 66, 121–134. Dämmgen, U., Matschullat, J., Zimmermann, F., Strogies, M., Grünhage, L., Scheler, B., et al., 2013. Emission reduction effects on bulk deposition in Germany – results from longterm measurements. Part 3. Sulphur and nitrogen compounds. Gefahrstoffe – Reinhaltung der Luft 73, 330–339. de Caritat, P., Cooper, M., 2015. A continental-scale geochemical atlas for resource exploration and environmental management: the National Geochemical Survey of Australia. Geochem. Explor. Environ. Anal. 16, 3–13. De Vos, W., Tarvainen, T. (Eds.), 2006. Geochemical Atlas of Europe. Part 2. Interpretation of Geochemical Maps, Additional Tables, Figures, Maps and Related Publications. Geological Survey of Finland, Espoo (690 p). EC No 1907/2006: Regulation (EC) No 1907/2006 of the European Parliament and of the Council of 18 December 2006 concerning the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH), establishing a European Chemicals Agency, amending Directive 1999/45/EC and repealing Council Regulation (EEC) No 793/93 and Commission Regulation (EC) No 1488/94 as well as Council Directive 76/769/ EEC and Commission Directives 91/155/EEC, 93/67/EEC, 93/105/EC and 2000/21/EC. Off. J. Eur. Union, Luxemburg, L396: 1-849, http://eur-lex.europa.eu/LexUriServ/ LexUriServ.do?uri=OJ:L:2006:396:0001:0849:EN:PDF (accessed on 5 September 2017). ECHA, 2008. Guidance on information requirements and chemical safety assessment. Chapter R.16: Environmental Exposure Estimation. European Chemicals Agency, Helsinki 136 pp. https://echa.europa.eu/documents/10162/13632/information_requirements_r16_en.pdf, Accessed date: 9 May 2017. EGS, 2008. EuroGeoSurveys Geochemistry Working Group. EuroGeoSurveys geochemical mapping of agricultural and grazing land in Europe (GEMAS) – field manual. Norges Geologiske Undersøkelse Report, 2008.038 46 pp. http://www.ngu.no/upload/ Publikasjoner/Rapporter/2008/2008_038.pdf. Fabian, C., Reimann, C., Fabian, K., Birke, M., Baritz, R., Haslinger, E., Project Team, G.E.M.A.S., 2014. GEMAS: spatial distribution of the pH of European agricultural and grazing land soil. Appl. Geochem. 48, 207–216. FAO, 2009. Livestock's role in climate change and air pollution. Chapter 3 of Livestock's Long Shadow: Environmental Issues And Options. Food and Agriculture Organization of the United Nations ftp://ftp.fao.org/docrep/fao/010/a0701e/a0701e00.pdf, Accessed date: 9 May 2017. Filzmoser, P., Hron, K., Reimann, C., 2009. Univariate statistical analysis of environmental (compositional) data – problems and possibilities. Sci. Total Environ. 407, 6100–6108. Filzmoser, P., Hron, K., Reimann, C., 2010. The bivariate statistical analysis of environmental (compositional) data. Sci. Total Environ. 409, 4230–4238. Filzmoser, P., Reimann, C., Birke, M., 2014. Univariate data analysis and mapping. Chapter 8. In: Reimann, C., Birke, M., Demetriades, A., Filzmoser, P., O'Connor, P. (Eds.), Chemistry of Europe's Agricultural Soils – Part A: Methodology and Interpretation of the GEMAS Data Set. Geologisches Jahrbuch, Schweizerbarth, Hannover, pp. 67–81 (B 102). Goldschmidt, V.M., 1937. The principles of distribution of chemical elements in minerals and rocks. J. Chem. Soc. 655–673 London. Goldschmidt, V.M., 1954. Geochemistry. Oxford at the Clarendon press (730pp). Holloway, J.M., Dahlgren, R.A., 2002. Nitrogen in rock: occurrences and biogeochemical implications. Glob. Biogeochem. Cycles 16 (4) (65-61–65-17). Jones, R.J.A., Hiederer, R., Rusco, E., Loveland, P.J., Montanarella, L., 2004. The map of organic carbon in topsoils in Europe, version 1.2 September 2003: explanation of special publication Ispra 2004 no.72 (S.P.I.04.72). European Soil Bureau Research

Report No.17, EUR 21209 EN. Office for Official Publications of the European Communities, Luxembourg (26 pp). Jones, R.J.A., Hiederer, R., Rusco, E., Montanarella, L., 2005. Estimating organic carbon in the soils of Europe for policy support. Eur. J. Soil Sci. 56, 655–671. Lettens, S., Van Orshoven, J., Van Wesemael, B., Muys, B., 2004. Soil organic and inorganic carbon content of landscape units in Belgium for 1950–1970. Soil Use Manag. 20, 40–47. Mackových, D., Lučivjanský, P., 2014. Preparation of GEMAS project samples and standards. Chapter 4. In: Reimann, C., Birke, M., Demetriades, A., Filzmoser, P., O'Connor, P. (Eds.), Chemistry of Europe's Agricultural Soils – Part A: Methodology and Interpretation of the GEMAS Data Set. Geologisches Jahrbuch, Schweizerbarth, Hannover, pp. 37–40 (B 102). Négrel, P., de Vivo, B., Reimann, C., Ladenberger, A., Cicchella, D., Albanese, S., Birke, M., de Vos, W., Dinelli, E., Lima, A., O'Connor, P.J., Salpeteur, I., Tarvainen, T., GEMAS Project Team, 2018. U-Th signatures of agricultural soil at the European scale (GEMAS): distribution, weathering patterns and processes controlling their concentrations. Sci. Total Environ. 622–623, 1277–1293. Panagos, P., Hiederer, R., Van Liedekerke, M., Bampa, F., 2013. Estimating soil organic carbon in Europe based on data collected through a European network. Ecol. Indic. 24, 439–450. Reimann, C., 2005. Geochemical mapping – technique or art? Geochem. Explor. Environ. Anal. 5, 359–370. Reimann, C., Siewers, U., Tarvainen, T., Bityukova, L., Eriksson, J., Gilucis, A., Gregoraiskiene, V., Lukashev, V.K., Matinian, N.N., Pasieczna, A., 2003. Agricultiural soils in Northern Europe: A geochemical atlas. Geologisches Jahrbuch, Sonderhefte, D, SD 5. Schweizerbart, Stuttgart, p. 279. Reimann, C., Filzmoser, P., Garrett, R.G., Dutter, R., 2008. Statistical Data Analysis Explained. Applied Environmental Statistics with R. Wiley, Chichester (343 pp). Reimann, C., Filzmoser, P., Fabian, K., Hron, K., Birke, M., Demetriades, A., et al., 2012. The concept of compositional data analysis in practice – total major element concentrations in agricultural and grazing land soils of Europe. Sci. Total Environ. 426, 196–210. Reimann, C., Birke, M., Demetriades, A., Filzmoser, P., O'Connor, P. (Eds.), 2014a. Chemistry of Europe's Agricultural Soils – Part A: Methodology and Interpretation of the GEMAS Data Set. Geologisches Jahrbuch, Schweizerbarth, Hannover B 102. (528 pp). Reimann, C., Birke, M., Demetriades, A., Filzmoser, P., O'Connor, P. (Eds.), 2014b. Chemistry of Europe's Agricultural Soils – Part B: General Background Information and Further Analysis of the GEMAS Data Set. Geologisches Jahrbuch, Schweizerbarth, Hannover B 103. (352 pp). Saaltink, R., Griffioen, J., Mol, G., Birke, M., the GEMAS Project Team, 2014. Geogenic and agricultural controls on the geochemical composition of European agricultural soils. J. Soils Sediments 14 (1), 121–137. Salminen, R., Chekushin, V., Tenhola, M., Bogatyrev, I., Glavatskikh, S.P., Fedotova, E., et al., 2004. Geochemical Atlas of the Eastern Barents Region. Elsevier, Amsterdam (548 pp). Salminen, R., Batista, M.J., Bidovec, M., Demetriades, A., De Vivo, B., De Vos, W., et al., 2005. FOREGS Geochemical Atlas of Europe, Part 1 – Background Information, Methodology, and Maps. Geological Survey of Finland, Espoo 690 pp. http://weppi.gtk.fi/ publ/foregsatlas/, Accessed date: 9 May 2017. Schucknecht, A., Matschullat, J., de Caritat, P., Melo, G., Mello, J., Plessow, A., et al., 2012. Pedogeochemistry in NE-Brazil, compared to Australia and Europe. Sci. Total Environ. 438, 342–356. Smith, D.B., Cannon, W.F., Woodruff, L.G., Solano, F., Kilburn, J.E., Fey, D.L., 2013. Geochemical and Mineralogical Data for Soils of the Conterminous United States: U.S. Geological Survey Data Series 801. 19 p. http://pubs.usgs.gov/ds/801/, Accessed date: 9 May 2017. Europe's environment. In: Stanners, D., Bourdeau, P. (Eds.), The Dobříš Assessment. Earthscan Publications, London (676 p.). Stevenson, F.J., 1962. Chemical state of the nitrogen in rocks. Geochim. Cosmochim. Acta 26, 797–809. Tóth, G., Jones, A., Montanarella, L., 2013. The LUCAS topsoil database and derived information on the regional variability of cropland topsoil properties in the European Union. Environ. Monit. Assess. 185, 7409–7425. Tóth, G., Hermann, T., Da Silva, M.R., Montanarella, L., 2016. Heavy metals in agricultural soil of the European Union with implications for food safety. Environ. Int. 88, 299–309. Tukey, J.W., 1977. Exploratory Data Analysis. Addison Wesley, Reading (506 pp). Wedepohl, K.H., 1995. The composition of the continental crust. Geochim. Cosmochim. Acta 59 (7), 1217–1232. World Bank, 2012. Indicators of Agriculture & Rural Development: Fertilizer Consumption (Kilograms per Hectare of Arable Land). http://data.worldbank.org/indicator/AG. CON.FERT.ZS/countries?order=wbapi_data_value_2007%20wbapi_data_ value&sort=desc&display=default, Accessed date: 9 May 2017. Yigini, Y., Panagos, P., 2016. Assessment of soil organic carbon stocks under future climate and land cover changes in Europe. Sci. Total Environ. 557–558, 838–850.