Joint Research Centre, Institute for Environment and Sustainability, Ispra (VA), Italy

^*Corresponding author: Dr. Fethi Şaban ÖZBEK, Joint Research Centre, Institute for Environment and Sustainability, Ispra (VA), Italy, Ph: +39 333 1917803, Postal address: Via Fermi 2749, TP 266/023I-21027 ISPRA (VA), Italy, E-mail: fethiozbek@yahoo.com

Article Information: Submission: 15/04/2015; Accepted: 09/05/2015; Published: 15/05/2015

Copyright: © 2015 Fethi Şaban ÖZBEK, et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

This paper presents the analysis of relationship between nitrogen surplus for agriculture (NS) and some socio-economic properties among the regions of Turkey for the period of 2007-2011. The correlations of NS with gross domestic product (GDP) per capita (0.66), permanent meadows and grassland share in utilized agricultural area (UAA) (-0.55), population density (0.49), out-migration rate (-0.42), the export value of agriculture and forestry (0.35), illiterate share (-0.31), number of villages (-0.31), arable land share in UAA (0.29), the value of livestock per capita (-0.26) and permanent area share in UAA (0.23) were observed as statistically significant. A model was established by using C5.0 algorithm to define NS levels of the regions according to socio-economic properties of the regions. According to the model results; permanent meadows and grassland in UAA, GDP per capita, population density, and illiterate share were determined as the important variables to define NS levels of the regions. It can therefore be concluded that the permanent meadows and grassland share in UAA and GDP per capita are the main socio-economic properties having noticeable effect on the environmental quality and human welfare related to NS.

Agriculture; Gross Domestic Product; Nitrogen Budget; Permanent Grassland; Turkey

Ozbek and Leip [1] summarized the importance of nitrogen (N) to living things as follows; nitrogen (N) is an important source of nutrition for plants; while nitrogen deficiency negatively affects plant growth, nitrogen surplus for agriculture (NS) can cause important problems that affect environmental quality and human welfare [2-4]. These problems can be listed as negative effects on biodiversity, eutrophication and nitrates accumulation in waters, acidification, nitrous oxide emission, corrosion of ozone layer and risks to human health due to exposure to nitrous oxide, ozone and particles [3,5]. The agricultural sector is defined as the main source of nitrogen contamination in underground, surface and air levels [6-10].

There are many studies focused on the relationship between environmental degradation and socio-economic properties, especially the relationship between environmental degradation and income level [11-14]. An important number of these studies on this relationship were about the environmental Kuznets curve (EKC). In literature, there are limited number of studies on the relationship between NS and socio-economic properties [15,16]. In these studies, only income level was analyzed as socio-economic indicator. This paper will ensure a great contribution to the literature in terms of examining the relationship between NS and socio-economic properties among the regions with different climates and socio-economic properties.

The NS values range vastly among the regions in Turkey [1]. The socio-economic differences among the regions in Turkey, which is a geographically big country, are also high. The purpose of this study is to analyze the relationship between NS values and some socio-economic properties (the number of villages, population density, crop patterns (the shares of arable land, permanent area, permanent meadows and grassland in utilized agricultural area (UAA)), organic area share in UAA, value of livestock per capita, value of crop products per capita, the export value of agriculture and forestry, gross domestic product (GDP) per capita, out-migration rate, illiterate share)among Turkey Nomenclature of Territorial Units for Statistics (NUTS2) regions. The correlation analysis between NS and socio-economic properties of NUTS2 regions was carried out, and a model was established by using C5.0 algorithm to define NS levels of the regions according to their regional socio-economic properties.

The nitrogen budget methodology used in this study is based on the methodology recommended in Eurostat/OECD common guideline [17]. In this methodology, NS was estimated by using equation 1.

The reference area (Aref) is UAA (arable land, permanent crop land, and permanent grassland). The inputs (Ninput) and the outputs (Noutput) used in NS estimation, and the methodology and the data sources used in the estimations of these inputs and outputs are presented in Table 1. In order to minimize the impact of regional differences in Turkey, where different climates are observed, NUTS2 division was used in the estimations [1,18].

A Pearson correlation analysis between NS and socio-economic properties of NUTS2 regions was carried out by using SAS package software. Turkey’s NUTS2 NS data from 2007-2011 was used in the analysis. Socio-economic properties used in the analysis are presented in Table 2.

A model was established by using C5.0 algorithm to define NS levels of the regions according to their regional socio-economic properties. The analysis was carried out using SPSS Clementine 12.0 package software. The C5.0 algorithm is a new generation of Machine Learning Algorithms based on decision trees, which is an important model to realize the classification [19]. C5.0 model works by splitting the sample based on the field that provides the maximum information gain [20]. Information gain (IG(Y|X)) was calculated by using entropy (H(Y)) (Equations 2 and 3) [21].

In these equations, pm is equal to P(Y=Vm), and it is supposed that Y can have one of m values (V1, V2… Vm). H(YǀX) is the average of H(YǀX=v)the entropies of Y among only those records in which X has value v.

In C5.0 algorithm, the dependent variable is the target variable that we are trying to understand and/or classify, and the input variables are used for this aim. The input variables were determined as socio-economic properties of the regions at NUTS2 level, and NS was determined as target variable in this study. Target variable used in C5.0 algorithm should be categorical. So that, the target variable NS in numeric type was transformed to the symbolic type by grouping the regions according to NS values. The regions were classified according to their NS values in four groups in a way that the interval of each group was equal. These groups were denominated as low NS (L: NS< 21 kg N ha^-1 yr^-1), lower medium NS (LM: 21 kg N ha^-1 yr^-1< NS < 56 kg N ha^-1 yr^-1), upper medium NS (HM: 56 kg N ha^-1 yr^-1< NS < 91 kg N ha^-1 yr^-1), and high NS (H: NS > 91 kg N ha^-1 yr^-1).

Regional GDP per capita and permanent meadows and grassland data set of 2007-2011 period are absent for Turkey. So that, the regional GDP shares for the year of 2000 were used in the analysis for estimating regional GDP per capita values of 2007-2011 period. The regional shares of permanent meadows and grassland in 2001 General Agricultural Census were used for estimating regional permanent meadows and grassland data set of 2007-2011 period. 2008-2011 data set was used for out-migration rate and illiterate share in the analysis as the data of the year 2007 for these properties are absent.

The correlations of NS with GDP per capita (0.66), permanent meadows and grassland share in UAA (-0.55), population density (0.49), out-migration rate (-0.42), the export value of agriculture and forestry (0.35), illiterate share (-0.31), number of villages (-0.31), arable land share in UAA (0.29), the value of livestock per capita (-0.26) and permanent area share in UAA (0.23) were observed as statistically significant (Table 3). The correlation coefficients of other socio-economic properties were relatively low, and their relationships with NS were not statistically significant.

It is shown from the evaluation graph (Figure 1) indicating the accuracy of the model formed by using C5.0 algorithm that the best line and the model line are very close to each other. This shows that the model has high accuracy. The analyse accuracy ratio of the model is so high as 97.28%. According to the model results,permanent meadows and grassland share in UAA, GDP per capita, population density, and illiterate sharewere found as important variables (Figure 2). The scatter pilot diagrams of the important variables and NS are presented in Figure 3.

The first split of the decision tree was based on permanent meadows and grassland share in UAA. The following split was based on GDP per capita. In case permanent meadows and grassland share in UAA was lower than 27%, the following split was based on GDP per capita. In this case, if GDP per capita was higher than a definite value (13724$) NS was observed high, otherwise it was relatively low. It was also observed that the regions with low permanent meadows and grassland share in UAA, low GDP per capita, and lowpopulation density had relatively low NS (Figure 4).

All of the regions with permanent meadows and grassland share in UAA were higher than 59% presented in low NS group. The regions with permanent meadows and grassland share in UAA was between 27% and 59% were divided in the following split according to the illletrate share. It was observed that the regions with lower illterate share had relatively lower NS (Figure 4).

In the regions with high GDP per capita, population density and arable land share of Turkey, where intensification level in agriculture was also high, hence mineral fertilizer usage was high; it was observed that the NS values were also high. The reverse relationship of NS with permanent meadows and grassland can be explained as follows; in the regions of Turkey, where permanent meadows and grassland were high, the extensification level in agriculture in these regions was also high. Therefore, mineral fertilizer usage was low; it was observed that NS values of these regions were relatively low.

It was shown that the relationship between environmental degradation and income level was the form of inverted U. The relationship between N surplus and GDP per capitain this study was similar to that curve (Figure 3). This indicates us that the relationship between N surplus and GDP per capita ensures EKC curve similar to some environmental issues (eg. SO₂ emission [22]; suspended particulate matter, total deforestation etc. [11]. This result is also parallel to the study made by Shen et al. [15] and Zhang et al. [16] showed the pollution from NS and economic growth relationship followed an inverse-U shape.

All the regions with permanent meadows and grassland shares in UAA were higher than 27% in the group of low NS or low medium NS. This indicates us that the regions / countries with high permanent meadows and grassland shares have relatively lower NS.

12 of EU countries with permanent meadows and grassland shares in UAA was higher than 27% had low or low medium NS [23] similar to the model results in this study (Figure 4). But, 5 of EU countries had upper medium or high NS. It is a fact that the regionalisation of NS estimations gives more sound results compared to the nonregionalisation estimations. Therefore, the most important reason for the bias is the non-regionalisation of NS estimations in Eurostat methodology. The model shows us that the more GDP per capita the more NS for the regions. The GDP per capita has increased in TR in recent years [24]. It was speculated that in the future if current trends in GDP growth continue the environmental risks related to NS can increase.

Although the effect of population density and illiterate share on the NS were relatively low compared to the effect of permanent meadows and grassland share in UAA and GDP per capita, some indirect relationships were observed between these properties and NS. According to the model results, the NS values of the regions with higher illiterate shares and lower population density were higher than the NS of the regions with lower illiterate share and higher population density. It was also observed from the model that the more population density the more NS for the regions (Figure 4). The direct relationship of NS with population density and the inverse relationship of NS with illiterate share can be explained as follows; in the regions where population densities were low and illiterate shares were high, the extensification level in agriculture was also high and consequently low mineral fertilizer usage, hence NS values of these regions were relatively low [18]. It was therefore speculated that the regions with low population densities and high illiterate shares have a low risk for environmental degradation related to NS in comparison to other regions.

Although NS is not only one or main indicator for measuring environmental degradation, it is well known that nitrogen surplus can cause important problems that affect environmental quality and human welfare [25]. It can therefore be concluded that the permanent meadows and grassland shares in UAA and GDP per capita are the main socio-economic properties having the noticeable effect on the environmental quality and human welfare related to NS.